Maple7prguide PDF

Maple 7
Programming Guide
M. B. Monagan K. O. Geddes K. M. Heal

G. Labahn S. M. Vorkoetter J. McCarron
P. DeMarco

c 2001 by Waterloo Maple Inc.
ii •
Waterloo Maple Inc.

57 Erb Street West
Waterloo, ON N2L 6C2
Canada
Maple and Maple V are registered trademarks of Waterloo Maple Inc.

c 2001, 2000, 1998, 1996 by Waterloo Maple Inc.
All rights reserved. This work may not be translated or copied in whole
or in part without the written permission of the copyright holder, except
for brief excerpts in connection with reviews or scholarly analysis. Use in
connection with any form of information storage and retrieval, electronic
adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed is forbidden.
The use of general descriptive names, trade names, trademarks, etc.,

in this publication, even if the former are not especially identified, is not
to be taken as a sign that such names, as understood by the Trade Marks
and Merchandise Marks Act, may accordingly be used freely by anyone.
Contents
1 Introduction 1
1.1 Getting Started . . . . . . . . . . . . . . . . . . . . . . . . 2
Locals and Globals . . . . . . . . . . . . . . . . . . . . . . 7
Inputs, Parameters, Arguments . . . . . . . . . . . . . . . 8
1.2 Basic Programming Constructs . . . . . . . . . . . . . . . 11
The Assignment Statement . . . . . . . . . . . . . . . . . 11
The for Loop . . . . . . . . . . . . . . . . . . . . . . . . . 13
The Conditional Statement . . . . . . . . . . . . . . . . . 15
The while Loop . . . . . . . . . . . . . . . . . . . . . . . 19
Modularization . . . . . . . . . . . . . . . . . . . . . . . . 20
Recursive Procedures . . . . . . . . . . . . . . . . . . . . . 22
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.3 Basic Data Structures . . . . . . . . . . . . . . . . . . . . 25
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
A MEMBER Procedure . . . . . . . . . . . . . . . . . . . . . 28
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Binary Search . . . . . . . . . . . . . . . . . . . . . . . . . 29
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Plotting the Roots of a Polynomial . . . . . . . . . . . . . 31
1.4 Computing with Formulæ . . . . . . . . . . . . . . . . . . 33
The Height of a Polynomial . . . . . . . . . . . . . . . . . 34
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
The Chebyshev Polynomials, Tn (x) . . . . . . . . . . . . . 36
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
Integration by Parts . . . . . . . . . . . . . . . . . . . . . 37
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Computing with Symbolic Parameters . . . . . . . . . . . 39
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
1.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 42
iii
iv • Contents
2 Fundamentals 45
2.1 Evaluation Rules . . . . . . . . . . . . . . . . . . . . . . . 46
Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Local Variables . . . . . . . . . . . . . . . . . . . . . . . . 50
Global Variables . . . . . . . . . . . . . . . . . . . . . . . 51
Exceptions . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.2 Nested Procedures . . . . . . . . . . . . . . . . . . . . . . 54
Local Versus Global Variables . . . . . . . . . . . . . . . . 55
The Quick-Sort Algorithm . . . . . . . . . . . . . . . . . . 56
Creating a Uniform Random Number Generator . . . . . 59
2.3 Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Types that Modify Evaluation Rules . . . . . . . . . . . . 62
Structured Types . . . . . . . . . . . . . . . . . . . . . . . 66
Type Matching . . . . . . . . . . . . . . . . . . . . . . . . 68
2.4 Choosing a Data Structure: Connected Graphs . . . . . . 70
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
2.5 Remember Tables . . . . . . . . . . . . . . . . . . . . . . . 76
The remember Option . . . . . . . . . . . . . . . . . . . . 76
Adding Entries Explicitly . . . . . . . . . . . . . . . . . . 77
Removing Entries from a Remember Table . . . . . . . . . 78
2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3 Advanced Programming 81
3.1 Procedures Which Return Procedures . . . . . . . . . . . 82
Creating a Newton Iteration . . . . . . . . . . . . . . . . . 82
A Shift Operator . . . . . . . . . . . . . . . . . . . . . . . 85
3.2 When Local Variables Leave Home . . . . . . . . . . . . . 86
Creating the Cartesian Product of a Sequence of Sets . . 89
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.3 Interactive Input . . . . . . . . . . . . . . . . . . . . . . . 94
Reading Strings from the Terminal . . . . . . . . . . . . . 95
Reading Expressions from the Terminal . . . . . . . . . . 95
Converting Strings to Expressions . . . . . . . . . . . . . 97
3.4 Extending Maple . . . . . . . . . . . . . . . . . . . . . . . 98
Defining New Types . . . . . . . . . . . . . . . . . . . . . 99
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Neutral Operators . . . . . . . . . . . . . . . . . . . . . . 100
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Extending Certain Commands . . . . . . . . . . . . . . . 106
3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Contents • v
4 The Maple Language 111

4.1 Language Elements . . . . . . . . . . . . . . . . . . . . . . 113
The Character Set . . . . . . . . . . . . . . . . . . . . . . 113
Tokens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Token Separators . . . . . . . . . . . . . . . . . . . . . . . 118
4.2 Escape Characters . . . . . . . . . . . . . . . . . . . . . . 122
4.3 Statements . . . . . . . . . . . . . . . . . . . . . . . . . . 122
The Assignment Statement . . . . . . . . . . . . . . . . . 123
Unassignment: Clearing a Name . . . . . . . . . . . . . . 128
The Selection Statement . . . . . . . . . . . . . . . . . . . 130
The Repetition Statement . . . . . . . . . . . . . . . . . . 133
The read and save Statements . . . . . . . . . . . . . . . 138
4.4 Expressions . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Expression Trees: Internal Representation . . . . . . . . . 140
The Types and Operands of Integers, Strings, Indexed
Names, and Concatenations . . . . . . . . . . . . . 144
Fractions and Rational Numbers . . . . . . . . . . . . . . 147
Floating-Point (Decimal) Numbers . . . . . . . . . . . . . 148
Complex Numerical Constants . . . . . . . . . . . . . . . 151
Labels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Sets and Lists . . . . . . . . . . . . . . . . . . . . . . . . . 158
Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
The Arithmetic Operators . . . . . . . . . . . . . . . . . . 165
Non-Commutative Multiplication . . . . . . . . . . . . . . 168
The Composition Operators . . . . . . . . . . . . . . . . . 169
The Ditto Operators . . . . . . . . . . . . . . . . . . . . . 170
The Factorial Operator . . . . . . . . . . . . . . . . . . . 170
The mod Operator . . . . . . . . . . . . . . . . . . . . . . 171
The Neutral Operators . . . . . . . . . . . . . . . . . . . . 172
Relations and Logical Operators . . . . . . . . . . . . . . 173
Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
Ranges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
Unevaluated Expressions . . . . . . . . . . . . . . . . . . . 184
Constants . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
Structured Types . . . . . . . . . . . . . . . . . . . . . . . 187
4.5 Useful Looping Constructs . . . . . . . . . . . . . . . . . . 190
The map, select, remove, and selectremove Commands 191
The zip Command . . . . . . . . . . . . . . . . . . . . . . 194
The seq, add, and mul Commands . . . . . . . . . . . . . 195
vi • Contents
4.6 Substitution . . . . . . . . . . . . . . . . . . . . . . . . . . 197

4.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 200
5 Procedures 201
5.1 Procedure Definitions . . . . . . . . . . . . . . . . . . . . 201
Mapping Notation . . . . . . . . . . . . . . . . . . . . . . 202
Unnamed Procedures and Their Combinations . . . . . . 203
Procedure Simplification . . . . . . . . . . . . . . . . . . . 204
5.2 Parameter Passing . . . . . . . . . . . . . . . . . . . . . . 204
Declared Parameters . . . . . . . . . . . . . . . . . . . . . 206
The Sequence of Arguments . . . . . . . . . . . . . . . . . 207
5.3 Local and Global Variables . . . . . . . . . . . . . . . . . 208
Evaluation of Local Variables . . . . . . . . . . . . . . . . 210
5.4 Procedure Options and the Description Field . . . . . . . 212
Options . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
The Description Field . . . . . . . . . . . . . . . . . . . . 214
5.5 The Value Returned by a Procedure . . . . . . . . . . . . 215
Assigning Values to Parameters . . . . . . . . . . . . . . . 215
Explicit Returns . . . . . . . . . . . . . . . . . . . . . . . 218
Error Returns . . . . . . . . . . . . . . . . . . . . . . . . . 219
Trapping Exceptions . . . . . . . . . . . . . . . . . . . . . 221
Returning Unevaluated . . . . . . . . . . . . . . . . . . . . 225
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
5.6 The Procedure Object . . . . . . . . . . . . . . . . . . . . 227
Last Name Evaluation . . . . . . . . . . . . . . . . . . . . 227
The Type and Operands of a Procedure . . . . . . . . . . 228
Saving and Retrieving Procedures . . . . . . . . . . . . . 231
5.7 Explorations . . . . . . . . . . . . . . . . . . . . . . . . . 232
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
5.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 233
6 Programming with Modules 235

About This Chapter . . . . . . . . . . . . . . . . . . . . . 238
6.1 Syntax and Semantics . . . . . . . . . . . . . . . . . . . . 238
The Module Body . . . . . . . . . . . . . . . . . . . . . . 239
Module Parameters . . . . . . . . . . . . . . . . . . . . . . 240
Named Modules . . . . . . . . . . . . . . . . . . . . . . . 240
Declarations . . . . . . . . . . . . . . . . . . . . . . . . . . 242
Exported Local Variables . . . . . . . . . . . . . . . . . . 244
Module Options . . . . . . . . . . . . . . . . . . . . . . . . 249
Implicit Scoping Rules . . . . . . . . . . . . . . . . . . . . 249
Contents • vii
Lexical Scoping Rules . . . . . . . . . . . . . . . . . . . . 249

Modules and Types . . . . . . . . . . . . . . . . . . . . . . 251
Example: A Symbolic Differentiator . . . . . . . . . . . . 252
6.2 Records . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
6.3 Packages . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
What Is a Package? . . . . . . . . . . . . . . . . . . . . . 266
Example: The LinkedList Package . . . . . . . . . . . . . 268
Example: A Code Coverage Profiling Package . . . . . . . 275
Example: The Shapes Package . . . . . . . . . . . . . . . 282
6.4 The use Statement . . . . . . . . . . . . . . . . . . . . . . 291
Operator Rebinding . . . . . . . . . . . . . . . . . . . . . 294
6.5 Modeling Objects . . . . . . . . . . . . . . . . . . . . . . . 297
Example: Priority Queues . . . . . . . . . . . . . . . . . . 300
An Object-oriented Shapes Package . . . . . . . . . . . . 304
6.6 Interfaces and Implementations . . . . . . . . . . . . . . . 306
Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Example: Generic Graph Algorithms . . . . . . . . . . . . 312
Example: Quotient Fields . . . . . . . . . . . . . . . . . . 318
Example: A Generic Group Implementation . . . . . . . . 327
6.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 348
7 Debugging Maple Programs 349

7.1 A Tutorial Example . . . . . . . . . . . . . . . . . . . . . 349
Numbering the Procedure Statements I . . . . . . . . . . 350
Invoking the Debugger I . . . . . . . . . . . . . . . . . . . 351
Controlling Execution of a Procedure during Debugging I 353
Invoking the Debugger II . . . . . . . . . . . . . . . . . . 357
7.2 Maple Debugger Commands . . . . . . . . . . . . . . . . . 361
Numbering the Procedure Statements II . . . . . . . . . . 361
Invoking the Debugger III . . . . . . . . . . . . . . . . . . 362
Controlling Execution of a Procedure during Debugging II 371
Changing the State of a Procedure during Debugging . . . 371
Examining the State of a Procedure during Debugging . . 374
Using Top-Level Commands at the Debugger Prompt . . 379
Restrictions . . . . . . . . . . . . . . . . . . . . . . . . . . 379
7.3 Detecting Errors . . . . . . . . . . . . . . . . . . . . . . . 380
Tracing a Procedure . . . . . . . . . . . . . . . . . . . . . 380
Using Assertions . . . . . . . . . . . . . . . . . . . . . . . 385
Handling Exceptions . . . . . . . . . . . . . . . . . . . . . 388
Checking Syntax . . . . . . . . . . . . . . . . . . . . . . . 393
7.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 394
viii • Contents
8 Numerical Programming in Maple 395

8.1 The Basics of evalf . . . . . . . . . . . . . . . . . . . . . 396
8.2 Hardware Floating-Point Numbers . . . . . . . . . . . . . 399
Newton Iterations . . . . . . . . . . . . . . . . . . . . . . 402
Computing with Arrays of Numbers . . . . . . . . . . . . 405
8.3 Floating-Point Models in Maple . . . . . . . . . . . . . . . 407
Software Floats . . . . . . . . . . . . . . . . . . . . . . . . 408
Roundoff Error . . . . . . . . . . . . . . . . . . . . . . . . 408
8.4 Extending the evalf Command . . . . . . . . . . . . . . . 410
Defining Your Own Constants . . . . . . . . . . . . . . . . 410
Defining Your Own Functions . . . . . . . . . . . . . . . . 412
8.5 Using the Matlab Package . . . . . . . . . . . . . . . . . . 415
8.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 416
9 Programming with Maple Graphics 417

9.1 Basic Plot Functions . . . . . . . . . . . . . . . . . . . . . 417
9.2 Programming with Plotting Library Functions . . . . . . 421
Plotting a Loop . . . . . . . . . . . . . . . . . . . . . . . . 421
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423
A Ribbon Plot Procedure . . . . . . . . . . . . . . . . . . 423
9.3 Maple’s Plotting Data Structures . . . . . . . . . . . . . . 425
The PLOT Data Structure . . . . . . . . . . . . . . . . . . 427
A Sum Plot . . . . . . . . . . . . . . . . . . . . . . . . . . 429
The PLOT3D Data Structure . . . . . . . . . . . . . . . . . 432
9.4 Programming with Plot Data Structures . . . . . . . . . . 436
Writing Graphic Primitives . . . . . . . . . . . . . . . . . 436
Plotting Gears . . . . . . . . . . . . . . . . . . . . . . . . 438
Polygon Meshes . . . . . . . . . . . . . . . . . . . . . . . . 442
9.5 Programming with the plottools Package . . . . . . . . 444
A Pie Chart . . . . . . . . . . . . . . . . . . . . . . . . . . 445
A Dropshadow Procedure . . . . . . . . . . . . . . . . . . 446
Creating a Tiling . . . . . . . . . . . . . . . . . . . . . . . 448
A Smith Chart . . . . . . . . . . . . . . . . . . . . . . . . 450
Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451
Modifying Polygon Meshes . . . . . . . . . . . . . . . . . 451
9.6 Example: Vector Field Plots . . . . . . . . . . . . . . . . . 456
9.7 Generating Grids of Points . . . . . . . . . . . . . . . . . 466
9.8 Animation . . . . . . . . . . . . . . . . . . . . . . . . . . . 471
9.9 Programming with Color . . . . . . . . . . . . . . . . . . . 478
Generating Color Tables . . . . . . . . . . . . . . . . . . . 480
Adding Color Information to Plots . . . . . . . . . . . . . 482
Contents • ix
Creating A Chess Board Plot . . . . . . . . . . . . . . . . 485

9.10 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 487
10 Input and Output 489

10.1 A Tutorial Example . . . . . . . . . . . . . . . . . . . . . 489
10.2 File Types and Modes . . . . . . . . . . . . . . . . . . . . 493
Buffered Files versus Unbuffered Files . . . . . . . . . . . 493
Text Files versus Binary Files . . . . . . . . . . . . . . . . 494
Read Mode versus Write Mode . . . . . . . . . . . . . . . 495
The default and terminal Files . . . . . . . . . . . . . . 495
10.3 File Descriptors versus File Names . . . . . . . . . . . . . 495
10.4 File Manipulation Commands . . . . . . . . . . . . . . . . 496
Opening and Closing Files . . . . . . . . . . . . . . . . . . 496
Position Determination and Adjustment . . . . . . . . . . 498
Detecting the End of a File . . . . . . . . . . . . . . . . . 498
Determining File Status . . . . . . . . . . . . . . . . . . . 499
Removing Files . . . . . . . . . . . . . . . . . . . . . . . . 499
10.5 Input Commands . . . . . . . . . . . . . . . . . . . . . . . 500
Reading Text Lines from a File . . . . . . . . . . . . . . . 500
Reading Arbitrary Bytes from a File . . . . . . . . . . . . 501
Formatted Input . . . . . . . . . . . . . . . . . . . . . . . 502
Reading Maple Statements . . . . . . . . . . . . . . . . . 507
Reading Tabular Data . . . . . . . . . . . . . . . . . . . . 508
10.6 Output Commands . . . . . . . . . . . . . . . . . . . . . . 509
Configuring Output Parameters by using the interface
Command . . . . . . . . . . . . . . . . . . . . . . . 509
One-Dimensional Expression Output . . . . . . . . . . . . 509
Two-Dimensional Expression Output . . . . . . . . . . . . 510
Writing Maple Strings to a File . . . . . . . . . . . . . . . 513
Writing Arbitrary Bytes to a File . . . . . . . . . . . . . . 513
Formatted Output . . . . . . . . . . . . . . . . . . . . . . 514
Writing Tabular Data . . . . . . . . . . . . . . . . . . . . 518
Flushing a Buffered File . . . . . . . . . . . . . . . . . . . 519
Redirecting the default Output Stream . . . . . . . . . . 520
10.7 Conversion Commands . . . . . . . . . . . . . . . . . . . . 520
C or Fortran Generation . . . . . . . . . . . . . . . . . . . 520
LATEX Generation . . . . . . . . . . . . . . . . . . . . . . . 523
Conversion between Strings and Lists of Integers . . . . . 524
Parsing Maple Expressions and Statements . . . . . . . . 525
Formatted Conversion to and from Strings . . . . . . . . . 526
10.8 A Detailed Example . . . . . . . . . . . . . . . . . . . . . 527
x • Contents
10.9 Notes to C Programmers . . . . . . . . . . . . . . . . . . . 528

10.10Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 529
11 Using Compiled Code in Maple 531

11.1 Method 1: Calling External Functions . . . . . . . . . . . 532
External Definition . . . . . . . . . . . . . . . . . . . . . . 535
Type Specification . . . . . . . . . . . . . . . . . . . . . . 536
Scalar Data Formats . . . . . . . . . . . . . . . . . . . . . 536
Structured Data Formats . . . . . . . . . . . . . . . . . . 536
Specifying Argument Passing Conventions . . . . . . . . . 538
11.2 Method 2: Wrapper Generation . . . . . . . . . . . . . . . 539
Additional Types and Options . . . . . . . . . . . . . . . 539
Structured Data Formats . . . . . . . . . . . . . . . . . . 539
Enumerated Types . . . . . . . . . . . . . . . . . . . . . . 540
Procedure Call Formats . . . . . . . . . . . . . . . . . . . 540
Call by Reference . . . . . . . . . . . . . . . . . . . . . . . 540
Array Options . . . . . . . . . . . . . . . . . . . . . . . . 541
Non-Passed Arguments . . . . . . . . . . . . . . . . . . . 542
Argument Checking and Efficiency Considerations . . . . 543
Conversions . . . . . . . . . . . . . . . . . . . . . . . . . . 543
Compiler Options . . . . . . . . . . . . . . . . . . . . . . . 544
Evaluation Rules . . . . . . . . . . . . . . . . . . . . . . . 549
11.3 Method 3: Customizing Wrappers . . . . . . . . . . . . . . 550
External Function Entry Point . . . . . . . . . . . . . . . 551
External API . . . . . . . . . . . . . . . . . . . . . . . . . 556
11.4 System Integrity . . . . . . . . . . . . . . . . . . . . . . . 574
11.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . 575
A Internal Representation and Manipulation 579

A.1 Internal Organization . . . . . . . . . . . . . . . . . . . . 579
A.2 Internal Representations of Data Types . . . . . . . . . . 582
Logical AND . . . . . . . . . . . . . . . . . . . . . . . . . 582
Assignment Statement . . . . . . . . . . . . . . . . . . . . 583
Binary Object . . . . . . . . . . . . . . . . . . . . . . . . . 583
Break Statement . . . . . . . . . . . . . . . . . . . . . . . 583
Name Concatenation . . . . . . . . . . . . . . . . . . . . . 583
Complex Value . . . . . . . . . . . . . . . . . . . . . . . . 584
Communications Control Structure . . . . . . . . . . . . . 584
Type Specification or Test . . . . . . . . . . . . . . . . . . 584
Debug . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584
Equation or Test for Equality . . . . . . . . . . . . . . . . 585
Contents • xi
Error Statement . . . . . . . . . . . . . . . . . . . . . . . 585

Expression Sequence . . . . . . . . . . . . . . . . . . . . . 585
Floating-Point Number . . . . . . . . . . . . . . . . . . . . 586
For/While Loop Statement . . . . . . . . . . . . . . . . . 586
Foreign Data . . . . . . . . . . . . . . . . . . . . . . . . . 587
Function Call . . . . . . . . . . . . . . . . . . . . . . . . . 587
Garbage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 587
Hardware Float . . . . . . . . . . . . . . . . . . . . . . . . 588
If Statement . . . . . . . . . . . . . . . . . . . . . . . . . . 588
Not Equal or Test for Inequality . . . . . . . . . . . . . . 588
Negative Integer . . . . . . . . . . . . . . . . . . . . . . . 589
Positive Integer . . . . . . . . . . . . . . . . . . . . . . . . 589
Less Than or Equal . . . . . . . . . . . . . . . . . . . . . . 590
Less Than . . . . . . . . . . . . . . . . . . . . . . . . . . . 590
Lexically Scoped Variable within an Expression . . . . . . 590
List . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 591
Local Variable within an Expression . . . . . . . . . . . . 591
Member . . . . . . . . . . . . . . . . . . . . . . . . . . . . 591
Module Definition . . . . . . . . . . . . . . . . . . . . . . 592
Module Instance . . . . . . . . . . . . . . . . . . . . . . . 593
Identifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . 594
Next Statement . . . . . . . . . . . . . . . . . . . . . . . . 594
Logical NOT . . . . . . . . . . . . . . . . . . . . . . . . . 594
Logical OR . . . . . . . . . . . . . . . . . . . . . . . . . . 594
Procedure Parameter within an Expression . . . . . . . . 595
Power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 595
Procedure Definition . . . . . . . . . . . . . . . . . . . . . 596
Product, Quotient, Power . . . . . . . . . . . . . . . . . . 597
Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
Rational . . . . . . . . . . . . . . . . . . . . . . . . . . . . 598
Read Statement . . . . . . . . . . . . . . . . . . . . . . . . 598
Return Statement . . . . . . . . . . . . . . . . . . . . . . 598
Rectangular Table . . . . . . . . . . . . . . . . . . . . . . 598
Save Statement . . . . . . . . . . . . . . . . . . . . . . . . 600
Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 600
Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601
Statement Sequence . . . . . . . . . . . . . . . . . . . . . 601
Stop Maple . . . . . . . . . . . . . . . . . . . . . . . . . . 601
String . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601
Sum, Difference . . . . . . . . . . . . . . . . . . . . . . . . 602
Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 602
xii • Contents
Table Reference . . . . . . . . . . . . . . . . . . . . . . . . 602

Try Statement . . . . . . . . . . . . . . . . . . . . . . . . 603
Unevaluated Expression . . . . . . . . . . . . . . . . . . . 603
Use Statement . . . . . . . . . . . . . . . . . . . . . . . . 603
Polynomials with Integer Coefficients modulo n . . . . . . 604
A.3 The Use of Hashing in Maple . . . . . . . . . . . . . . . . 605
Hash Table . . . . . . . . . . . . . . . . . . . . . . . . . . 605
Hash Chain . . . . . . . . . . . . . . . . . . . . . . . . . . 605
The Simplification Table . . . . . . . . . . . . . . . . . . . 606
The Name Table . . . . . . . . . . . . . . . . . . . . . . . 607
Remember Tables . . . . . . . . . . . . . . . . . . . . . . . 607
Maple Language Arrays and Tables . . . . . . . . . . . . . 608
Maple Language Rectangular Tables . . . . . . . . . . . . 609
A.4 Portability . . . . . . . . . . . . . . . . . . . . . . . . . . . 609
Index 611
1 Introduction
As a Maple user, you may fall into any number of categories. You may
have used Maple only interactively. You may already have written many
of your own programs. Even more fundamentally, you may or may not
have programmed in another computer language before attempting your
first Maple program. Indeed, you may have used Maple for some time
without realizing that the same powerful language you regularly use to
enter commands is itself a complete programming language.
Writing a Maple program can be very simple. It may only involve
putting a proc() and an end proc around a sequence of commands that
you use every day. On the other hand, the limits for writing Maple pro-
cedures with various levels of complexity depend only on you. Ninety to
ninety-five percent of the thousands of commands in the Maple language
are themselves Maple programs. You are free to examine these programs
and modify them to suit your needs, or extend them so that Maple can
tackle new types of problems. You should be able to write useful Maple
programs in a few hours, rather than the few days or weeks that it often
takes with other languages. This efficiency is partly due to the fact that
Maple is interactive; this interaction makes it easier to test and correct
programs.
Coding in Maple does not require expert programming skills. Unlike
traditional programming languages, the Maple language contains many
powerful commands which allow you to perform complicated tasks with
a single command instead of pages of code. For example, the solve com-
mand computes the solution to a system of equations. Maple comes with a
large library of routines, including graphical display primitives, so putting
useful programs together from its powerful building blocks is easy.
The aim of this chapter is to provide basic knowledge for proficiently
writing Maple code. To learn quickly, read until you encounter some ex-
ample programs and then write your own variations. This chapter includes
many examples along with exercises for you to try. Some of them highlight
1
2 • Chapter 1: Introduction
important differences between Maple and traditional computer languages,

which lack symbolic computation capability. Thus, this chapter is also im-
portant for those who have written programs in other languages.
This chapter informally presents the most essential elements of the
Maple language. You can study the details, exceptions, and options in the
other chapters, as the need arises. The examples of basic programming
tasks for you to do come with pointers to other chapters and help pages
that give further details.
1.1 Getting Started

Maple runs on many different platforms. You can use it through a special-
ized worksheet interface, or directly through interactive commands typed
at a plain terminal. In either case, when you start a Maple session, you
will see a Maple prompt character.
>
The prompt character > indicates that Maple is waiting for input.
Throughout this book, the command-line (or one-dimensional) input
format is used. For information on how to toggle between Maple notation
and standard math notation, please refer to the first chapter of the
Getting Started Guide.
Your input can be as simple as a single expression. A command is
followed immediately by its result.
> 103993/33102;
103993
33102
Ordinarily, you complete the command with a semicolon, then press

enter. Maple echoes the result—in this case an exact rational number—
to the worksheet or to the terminal and the particular interface in use,
displaying the result as closely to standard mathematical notation as pos-
sible.1
You may enter commands entirely on one line (as in the previous
example) or stretch them across several lines.
1
section 10.6 discusses specific commands to control printing.
1.1 Getting Started • 3
> 103993
> / 33102
> ;
103993
33102
You can even put the terminating semicolon on a separate line. Noth-
ing evaluates until you complete the command. Maple may, however, parse
the command for errors at this stage.
Associate names with results by using the assignment statement, :=.
> a := 103993/33102;
103993
a :=
33102
Once assigned a value in this manner, you can use the name a as if
it were the value 103993/33102. For example, you can use Maple’s evalf
command to compute an approximation to 103993/33102 divided by 2.
> evalf(a/2);
1.570796326
A Maple program is essentially just a prearranged group of commands

that Maple always carries out together. The simplest way of creating such
a Maple program (or procedure) is to encapsulate the sequence of com-
mands that you would have used to carry out the computation interac-
tively. The following is a program corresponding to the above statement.
> half := proc(x)
> evalf(x/2);
> end proc;
half := proc(x) evalf(1/2 ∗ x) end proc
The program takes the input, called x within the procedure, and ap-
proximates the value of x divided by two. Since this is the last calculation
done within the procedure, the half procedure returns this approxima-
tion. Give the name half to the procedure using the := notation, just as
you would assign a name to any other object. Once you have defined a
new procedure, you can use it as a command.
> half(2/3);
.3333333333
> half(a);
1.570796326
> half(1) + half(2);
1.500000000
Merely enclosing the command evalf(x/2); between a proc(. . . )

and the words end proc turns it into a procedure.
Create another program corresponding to the following two state-
ments.
> a := 103993/33102;
> evalf(a/2);
The procedure needs no input.

> f := proc() local a;
> a := 103993/33102;
> evalf(a/2);
> end proc;
f := proc()
local a;
a := 103993/33102 ; evalf(1/2 ∗ a)
end proc
Maple’s interpretation of this procedure definition appears immedi-
ately after the command lines that created it. Examine it carefully and
note the following:
• The name of this program (procedure) is f.
• The procedure definition starts with proc(). The empty parenthesis

indicate that this procedure does not require any input data.
• Semicolons separate the individual commands that make up the pro-

cedure. Another semicolon after the words end proc signals the end
of the procedure definition.
• You see a display of the procedure definition (just as for any other
Maple command) only after you complete it with an end proc and a
semicolon. Even the individual commands that make up the procedure
do not display until you complete the entire procedure and enter the
last semicolon.
• The procedure definition that echoes as the value of the name f is

equivalent to but not identical to the procedure definition that you
entered.
• The local a; statement declares a as a local variable. This means

that the variable a within the procedure is not the same as the variable
a outside the procedure. Thus, it does not matter if you use that name
for something else. Section 1.1 discusses these further.
Execute the procedure f—that is, cause the statements forming the
procedure to execute in sequence—by typing its name followed by paren-
theses. Enclose any input to the procedure, in this case none, between the
parentheses.
> f();
1.570796326
The execution of a procedure is also referred to as an invocation or

a procedure call .
When you invoke a procedure, Maple executes the statements forming
the procedure body one at a time. The procedure returns the result of
the last computed statement as the value of the procedure call.
As with ordinary Maple expressions, you can enter procedure defini-
tions with a large degree of flexibility. Individual statements may appear
on different lines, or span several lines. You may also place more than one
statement on one line, though that can affect readability of your code.
You may even put extra semicolons between statements without causing
problems. In some instances, you may omit semicolons.2
2
For example, the semicolon in the definition of a procedure between the last com-
mand and the end proc is optional.
Sometimes you may not want Maple to display the result of con-
structing a complicated procedure definition. To suppress the display, use
a colon (:) instead of a semicolon (;) at the end of the definition.
> g := proc() local a;
> a := 103993/33102;
> evalf(a/2);
> end proc:
Sometimes you may find it necessary to examine the body of a pro-

cedure long after constructing it. For ordinary named objects in Maple,
such as e, defined below, you can obtain the actual value of the name
simply by referring to it by name.
> e := 3;
e := 3
> e;
If you try this with the procedure g, Maple displays only the name g
instead of its true value. Both procedures and tables potentially contain
many subobjects. This model of evaluation, referred to as last name
evaluation, hides the detail. To obtain the true value of the name g, use
the eval command, which forces full evaluation.
> g;
> eval(g);
proc()
local a;
a := 103993/33102 ; evalf(1/2 ∗ a)
end proc
To print the body of a Maple library procedure, set the interface
variable verboseproc to 2. See ?interface for details on interface
variables.
Locals and Globals

Variables that you use at the interactive level in Maple, that is, not within
a procedure body, are called global variables.
Variables that can be accessed only from the procedures in which
they are declared are called local variables . While Maple executes a pro-
cedure, a global variable by the same name remains unchanged, no matter
what value the local variables assume. This allows you to make tempo-
rary assignments inside a procedure without affecting anything else in
your session.
The scope of a variable refers to the collection of procedures and
statements which have access to the value of the variable. With simple
(that is, non-nested) procedures in Maple, only two possibilities exist.
Either the value of a name is available everywhere (that is, global ) or
only to the statements that form the particular procedure definition (that
is, local ). The more involved rules that apply for nested procedures are
outlined in Section 2.2.
To demonstrate the distinction between local and global names, first
assign a value to the global (that is, top-level) name b.
> b := 2;
b := 2
Next, define two nearly identical procedures: g, explicitly using b as

a local variable and h, explicitly using b as a global variable.
> g := proc()
> local b;
> b := 103993/33102;
> evalf(b/2);
> end proc:
and
> h := proc()
> global b;
> b := 103993/33102;
> evalf(b/2);
> end proc:
Defining the procedures has no effect on the global value of b. In fact, you
can even execute the procedure g (which uses local variables) without
affecting the value of b.
> g();
1.570796326
Therefore, the value of the global variable b is still 2. The procedure

g made an assignment to the local variable b which is different from the
global variable of the same name.
> b;
The effect of using the procedure h (which uses global variables) is

very different.
> h();
1.570796326
h changes the global variable b, so it is no longer 2. When you invoke

h, the global variable b changes as a side effect .
> b;
103993
33102
If you do not indicate whether a variable used inside a procedure is

local or global, Maple decides on its own and warns you of this. You can
always use the local or global statements to override Maple’s choice.
However, it is good programming style to declare all variables either local
or global.
Inputs, Parameters, Arguments

An important class of variables that you can use in procedure definitions
are neither local nor global. These represent the inputs to the procedure.
Parameters or arguments are other names for this class.
Procedure arguments are placeholders for the actual values of data
that you supply when you invoke the procedure, which may have more
than one argument. The following procedure h accepts two quantities, p
and q, and constructs the expression p/q.
> k := proc(p,q)
> p/q;
> end proc:
The arguments to this procedure are p and q. That is, p and q are
placeholders for the actual inputs to the procedure.
> k(103993,33102);
103993
33102
Maple considers floating-point values to be approximations, rather

than exact expressions. Floating-point expressions compute immediately.
> k( 23, 0.56);
41.07142857
In addition to support for exact and floating-point approximate num-

bers and symbols, Maple provides direct support for complex numbers.
By default,
√ Maple uses the capital letter I to represent the imaginary
unit, −1.
> (2 + 3*I)^2;
−5 + 12 I
> k(2 + 3*I, %);
2 3
− I
13 13
> k(1.362, 5*I);
−.2724000000 I
√ Suppose you want to write a procedure which calculates the norm,

a2 + b2 , of a complex number z = a+bi. You can make such a procedure
in several ways. The procedure abnorm takes the real and imaginary parts,
a and b, as separate parameters.
> abnorm := proc(a,b)
> sqrt(a^2+b^2);
> end proc;
abnorm := proc(a, b) sqrt(a2 + b2 ) end proc
Now abnorm can calculate the norm of 2 + 3i.

> abnorm(2, 3);

√
13
You could instead use the Re and Im commands to pick out the real
and imaginary parts, respectively, of a complex number. Hence, you can
also calculate the norm of a complex number in the following manner.
> znorm := proc(z)
> sqrt( Re(z)^2 + Im(z)^2 );
> end proc;
znorm := proc(z) sqrt(<(z)2 + =(z)2 ) end proc
√
The norm of 2 + 3i is still 13.
> znorm( 2+3*I );
√
13
Finally, you can also compute the norm by re-using the abnorm pro-
cedure. The abznorm procedure below uses Re and Im to pass information
to abnorm in the form it expects.
> abznorm := proc(z)
> local r, i;
> r := Re(z);
> i := Im(z);
> abnorm(r, i);
> end proc;
abznorm := proc(z)
local r, i;
r := <(z) ; i := =(z) ; abnorm(r, i)
end proc
Use abznorm to calculate the norm of 2 + 3i.
> abznorm( 2+3*I );
√
13
If you do not specify enough information for Maple to calculate the

norm, abznorm returns a symbolic formula. Here Maple treats x and y
1.2 Basic Programming Constructs • 11
as complex numbers. If they were real numbers, then <(x + i y) would

simplify to x.
> abznorm( x+y*I );
p
<(x + I y)2 + =(x + I y)2
Many Maple commands return unevaluated in such cases. Thus, you

might alter abznorm to return abznorm(x+y*I) in the above example.
Later examples in this book show how to give your own procedures this
behavior.
1.2 Basic Programming Constructs

This section describes the programming constructs you require to get
started with real programming tasks. It covers assignment statements,
for loops and while loops, conditional statements (if statements), and
the use of local and global variables.
The Assignment Statement

Use assignment statements to associate names with computed values.
They have the following form.
variable := value ;
This syntax assigns the name on the left-hand side of := to the com-
puted value on the right-hand side. You have seen this statement used in
many of the earlier examples.
The use of := here is similar to the assignment statement in program-
ming languages, such as Pascal. Other programming languages, such as C
and Fortran, use = for assignments. Maple does not use = for assignments,
since it is such a natural choice for representing mathematical equations.
If you want to write a procedure called plotdiff which plots an
expression f (x) together with its derivative f 0 (x) on the interval [a, b],
you can accomplish this task by computing the derivative of f (x) with
the diff command and then plotting both f (x) and f 0 (x) on the same
interval with the plot command.
> y := x^3 - 2*x + 1;
y := x3 − 2 x + 1
Find the derivative of y with respect to x.

> yp := diff(y, x);
yp := 3 x2 − 2
Plot y and yp together.

> plot( [y, yp], x=-1..1 );
–1 –0.8–0.6–0.4–0.2 0.2 0.4 0.6 0.8 1

x
–1
–2
The following procedure combines this sequence of steps.

> plotdiff := proc(y,x,a,b)
> local yp;
> yp := diff(y,x);
> plot( [y, yp], x=a..b );
> end proc;
plotdiff := proc(y, x, a, b)
local yp;
yp := diff(y, x) ; plot([y, yp], x = a..b)
end proc
The procedure name is plotdiff. It has four parameters: y, the ex-
pression it differentiates; x, the name of the variable it uses to define the
expression; and a and b, the beginning and the end of the interval over
which it generates the plot. The procedure returns a Maple plot object
which you can either display, or use in further plotting routines.
By specifying that yp is a local variable, you ensure that its usage in
the procedure does not clash with any other usage of the variable that
you may have made elsewhere in the current session.
To use the procedure, simply invoke it with appropriate arguments.

Plot cos(t) and its derivative, for t running from 0 to 2π.
> plotdiff( cos(t), t, 0, 2*Pi );
0.5
0 1 2 3 4 5 6
t
–0.5
–1
The for Loop

Use looping constructs, such as the for loop, to repeat similar actions a
number of times. For example, you can calculate the sum of the first five
natural numbers in the following way.
> total := 0;
> total := total + 1;
You may instead perform the same calculations by using a for loop.
> total := 0:
> for i from 1 to 5 do
> total := total + i;
> end do;
total := 1
total := 3
total := 6
total := 10
total := 15
For each cycle through the loop, Maple increments the value of i by
one and checks whether i is greater than 5. If it is not, then Maple executes
the body of the loop again. When the execution of the loop finishes, the
value of total is 15.
> total;
15
The following procedure uses a for loop to calculate the sum of the
first n natural numbers.
> SUM := proc(n)
> local i, total;
> total := 0;
> for i from 1 to n do
> total := total+i;
> end do;
> total;
> end proc:
The purpose of the total statement at the end of SUM is to ensure that
SUM returns the value total. Calculate the sum of the first 100 numbers.
> SUM(100);
5050
The for statement is an important part of the Maple language, but

the language also provides many more succinct and efficient looping con-
structs. For example, the command add.
> add(n, n=1..100);
5050
The Conditional Statement

The loop is one of the two most basic constructs in programming. The
other basic construct is the if or conditional statement . It arises in
many contexts. For example, you can use the if statement to implement
an absolute value function.
º
x if x ≥ 0
|x| =
−x if x < 0.
Below is a first implementation of ABS. Maple executes the if statement
as follows: If x < 0, then Maple calculates −x; otherwise it calculates x. In
either case, the absolute value of x is the last result that Maple computes
and so is the value that ABS returns.
The closing words end if completes the if statement.
> ABS := proc(x)
> if x<0 then
> -x;
> else
> x;
> end if;
> end proc;
ABS := proc(x) if x < 0 then − x else x end if end proc
> ABS(3); ABS(-2.3);
2.3
Returning Unevaluated The ABS procedure above cannot handle non-

numeric input.
> ABS( a );
Error, (in ABS) cannot evaluate boolean: a < 0
The problem is that since Maple knows nothing about a, it cannot

determine whether a is less than zero. In such cases, your procedure should
return unevaluated ; that is, ABS should return ABS(a). To achieve this
result, consider the following example.
> ’ABS’(a);
ABS(a)
The single quotes tell Maple not to evaluate ABS. You can modify the ABS
procedure by using the type(..., numeric) command to test whether
x is a number.
> ABS := proc(x)
> if type(x,numeric) then
> if x<0 then -x else x end if;
> else
> ’ABS’(x);
> end if;
> end proc:
The above ABS procedure contains an example of a nested if statement,
that is, one if statement appearing within another. You need an even
more complicated nested if statement to implement the function


 0 if x ≤ 0
x if 0 < x ≤ 1

hat(x) =

 2 − x if 1 < x ≤ 2
0 if x > 2.

Here is a first version of HAT.

> HAT := proc(x)
> if type(x, numeric) then
> if x<=0 then
> 0;
> else
> if x<=1 then
> x;
> else
> if x<=2 then
> 2-x;
> else
> 0;
> end if;
> end if;
> end if;
> else
> ’HAT’(x);
> end if;
> end proc:
The indentations make it easier to identify which statements belong to
which if conditions.
A better implementation uses the optional elif clause (else if) in the
second-level if statement.
> HAT := proc(x)
> if x<=0 then 0;
> elif x<=1 then x;
> elif x<=2 then 2-x;
> else 0;
> end if;
> else
> ’HAT’(x);
> end if;
> end proc:
You may use as many elif branches as you need.
Symbolic Transformations You can improve the ABS procedure from

the last section even further. Consider the product ab. Since ab is an
unknown, ABS returns unevaluated.
> ABS( a*b );
ABS(a b)
However, the absolute value of a product is the product of the absolute

values.
|ab| → |a||b|
That is, ABS should map over products.
> map( ABS, a*b );
ABS(a) ABS(b)
You can use the type(..., ‘*‘) command to test whether an ex-
pression is a product and use the map command to apply ABS to each
operand of the product.
> ABS := proc(x)
> if x<0 then -x else x end if;
> elif type(x, ‘*‘) then
> map(ABS, x);
> else
> ’ABS’(x);
> end if;
> end proc:
> ABS( a*b );
ABS(a) ABS(b)
This feature is especially useful if some of the factors are numbers.

> ABS( -2*a );
2 ABS(a)
You may want to improve ABS further so that it can calculate the
absolute value of a complex number.
Parameter Type Checking Sometimes when you write a procedure, you

intend it to handle only a certain type of input. Calling the procedure
with a different type of input may not make any sense. You can use type
checking to verify that the inputs to your procedure are of the correct
type. Type checking is especially important for complicated procedures
as it helps you to identify mistakes early .
Consider the original implementation of SUM.
> SUM := proc(n)
> local i, total;
> total := 0;
> total := total+i;
> end do;
> total;
> end proc:
Clearly, n should be an integer. If you try to use the procedure on symbolic
data, it breaks.
> SUM("hello world");
Error, (in SUM) final value in for loop must be numeric
or character
The error message indicates what went wrong inside the for statement
while trying to execute the procedure. The test in the for loop failed
because "hello world" is a string, not a number, and Maple could
not determine whether to execute the loop. The following implemen-
tation of SUM provides a much more informative error message. The
type(...,integer) command determines whether n is an integer.
> SUM := proc(n)
> local i,total;
> if not type(n, integer) then
> error("input must be an integer");
> end if;
> total := 0;
> for i from 1 to n do total := total+i end do;
> total;
> end proc:
Now the error message is more helpful.

Error, (in SUM) input must be an integer
Using type to check inputs is such a common task that Maple provides
a simple means of declaring the type of an argument to a procedure. For
example, you can rewrite the SUM procedure in the following manner. An
informative error message helps you to find and correct a mistake quickly.
> SUM := proc(n::integer)
> local i, total;
> total := 0;
> for i from 1 to n do total := total+i end do;
> total;
> end proc:

Error, invalid input: SUM expects its 1st argument, n,
to be of type integer, but received hello world
Maple understands a large number of types. In addition, you can

combine existing types algebraically to form new types, or you can define
entirely new types. See ?type.
The while Loop

The while loop is an important type of structure. It has the following
structure.
while condition do commands end do;
Maple tests the condition and executes the commands inside the loop
over and over again until the condition fails.
You can use the while loop to write a procedure that divides an inte-
ger n by two as many times as is possible. The iquo and irem commands
calculate the quotient and remainder, respectively, using integer division.
> iquo( 7, 3 );
> irem( 7, 3 );
Thus, you can write a divideby2 procedure in the following manner.

> divideby2 := proc(n::posint)
> local q;
> q := n;
> while irem(q, 2) = 0 do
> q := iquo(q, 2);
> end do;
> q;
> end proc:
Apply divideby2 to 32 and 48.
> divideby2(32);
> divideby2(48);
The while and for loops are both special cases of a more general
repetition statement; see section 4.3.
Modularization
When you write procedures, identifying subtasks and writing these as
separate procedures is a good idea. Doing so makes your procedures easier
to read, and you may be able to reuse some of the subtask procedures in
another application.
Consider the following mathematical problem. Suppose you have a
positive integer, in this case, forty.
> 40;
40
Divide the integer by two, as many times as possible; the divideby2

procedure above does just that for you.
> divideby2( % );
5
Multiply the result by three and add one.

> 3*% + 1;
16
Again, divide by two.

> divideby2( % );
Multiply by three and add one.

> 3*% + 1;
Divide.
> divideby2( % );
The result is 1 again, so from now on you will get 4, 1, 4, 1, . . . .

Mathematicians have conjectured that you always reach the number 1 in
this way, no matter with which positive integer you begin. You can study
this conjecture, known as the 3n + 1 conjecture, by writing a procedure
which calculates how many iterations you need to get to the number 1.
The following procedure makes a single iteration.
> iteration := proc(n::posint)
> local a;
> a := 3*n + 1;
> divideby2( a );
> end proc:
The checkconjecture procedure counts the number of iterations.
> checkconjecture := proc(x::posint)
> local count, n;
> count := 0;
> n := divideby2(x);
> while n>1 do
> n := iteration(n);
> count := count + 1;
> end do;
> count;
> end proc:

You can now check the conjecture for different values of x.
> checkconjecture( 40 );
> checkconjecture( 4387 );
49
You could write checkconjecture as one self-contained procedure

without references to iteration or divideby2. But then, you would have
to use nested while statements, thus making the procedure much harder
to read.
Recursive Procedures
Just as you can write procedures that call other procedures, you can also
write a procedure that calls itself. This is called recursive programming .
As an example, consider the Fibonacci numbers, which are defined in the
following procedure.
fn = fn−1 + fn−2 for n ≥ 2,
where f0 = 0, and f1 = 1. The following procedure calculates fn for any

n.
> Fibonacci := proc(n::nonnegint)
> if n<2 then
> n;
> else
> Fibonacci(n-1)+Fibonacci(n-2);
> end if;
> end proc:
Here is a sequence of the first sixteen Fibonacci numbers.
> seq( Fibonacci(i), i=0..15 );
0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610
The time command tells you the number of seconds a procedure takes
to execute. Fibonacci is not very efficient.
> time( Fibonacci(20) );
.450
The reason is that Fibonacci recalculates the same results over and
over again. To find f20 , it must find f19 and f18 ; to find f19 , it must
find f18 again and f17 ; and so on. One solution to this efficiency problem
is to tell Fibonacci to remember its results. That way, Fibonacci only
has to calculate f18 once. The remember option makes a procedure store
its results in a remember table. Section 2.5 further discusses remember
tables.
> option remember;
> if n<2 then
> n;
> else
> end if;
> end proc:
This version of Fibonacci is much faster.
0.
.133
If you use remember tables indiscriminately, Maple may run out of

memory. You can often rewrite recursive procedures by using a loop, but
recursive procedures are often easier to read. On the other hand, iterative
procedures are more efficient. The procedure below is a loop version of
Fibonacci.
> local temp, fnew, fold, i;
> if n<2 then
> n;
> else
> fold := 0;
> fnew := 1;
> temp := fnew + fold;
> fold := fnew;
> fnew := temp;
> end do;
> fnew;
> end if;
> end proc:
.133
When you write recursive procedures, you must weigh the benefits of
remember tables against their use of memory. Also, you must make sure
that your recursion stops.
The return Statement A Maple procedure by default returns the result

of the last computation within the procedure. You can use the return
statement to override this behavior. In the version of Fibonacci below, if
n < 2 then the procedure returns n and Maple does not execute the rest
of the procedure.
> option remember;
> if n<2 then
> return n;
> end if;
> end proc:
Using the return statement can make your recursive procedures easier
to read; the usually complicated code that handles the general step of the
recursion does not end up inside a nested if statement.
Exercise
1. The Fibonacci numbers satisfy the following recurrence.
F (2n) = 2F (n − 1)F (n) + F (n)2 where n > 1
and
F (2n + 1) = F (n + 1)2 + F (n)2 where n > 1
Use these new relations to write a recursive Maple procedure which
computes the Fibonacci numbers. How much recomputation does this
procedure do?
1.3 Basic Data Structures • 25
1.3 Basic Data Structures

The programs developed so far in this chapter have operated primarily
on a single number or a single formula. More advanced programs often
manipulate more complicated collections of data. A data structure is a
systematic way of organizing data. The organization you choose for your
data can directly affect the style of your programs and how fast they
execute.
Maple has a rich set of built-in data structures. This section will ad-
dress the basic structure of sequences , lists , and sets .
Many Maple commands take sequences, lists, and sets as inputs, and
produce sequences, lists, and sets as outputs. The following problem il-
lustrates how such data structures are useful in solving problems.
Problem: Write a Maple procedure which given n > 0 data values

x1 , x2 , . . . , xn computes their average, where the following equation
gives the average of n numbers.
n
1X
µ= xi .
n
i=1
You can easily represent the data for this problem as a list. nops gives
the total number of entries in a list X, while the ith entry of the list is
denoted X[i].
> X := [1.3, 5.3, 11.2, 2.1, 2.1];
X := [1.3, 5.3, 11.2, 2.1, 2.1]
> nops(X);
> X[2];
5.3
You can add the numbers in a list by using the add command.
> add( i, i=X );
22.0
The procedure average below computes the average of the entries in

a list. It handles empty lists as a special case.
> average := proc(X::list)
> local n, i, total;
> n := nops(X);
> if n=0 then error "empty list" end if;
> total := add(i, i=X);
> total / n;
> end proc:
Using this procedure you can find the average of the list X.
> average(X);
4.400000000
The procedure still works if the list has symbolic entries.

> average( [ a , b , c ] );
1 1 1
a+ b+ c
3 3 3
Exercise
1. Write a Maple procedure called sigma which, given n > 1 data val-
ues, x1 , x2 , . . . , xn , computes their standard deviation. The following
equation gives the standard deviation of n > 1 numbers,
v
u n
u1 X
σ=t (xi − µ)2
n
i=1
where µ is the average of the data values.
You create lists and many other objects in Maple out of more primitive
data structures called sequences . The list X defined previously contains
the following sequence.
> Y := X[];
Y := 1.3, 5.3, 11.2, 2.1, 2.1
You can select elements from a sequence in the same way you select
elements from a list.
> Y[3];
11.2
> Y[2..4];
5.3, 11.2, 2.1
> Y[2..-2];
5.3, 11.2, 2.1
The important difference between sequences and lists is that Maple

flattens a sequence of sequences into a single sequence.
> W := a,b,c;
W := a, b, c
> Y, W, Y;
1.3, 5.3, 11.2, 2.1, 2.1, a, b, c, 1.3, 5.3, 11.2, 2.1, 2.1
In contrast, a list of lists remains just that, a list of lists.

> [ X, [a,b,c], X ];
[[1.3, 5.3, 11.2, 2.1, 2.1], [a, b, c], [1.3, 5.3, 11.2, 2.1, 2.1]]
If you enclose a sequence in a pair of braces, you get a set .

> Z := { Y };
Z := {1.3, 5.3, 11.2, 2.1}
As in mathematics, a set is an unordered collection of distinct objects,

unlike a list which is an ordered sequence of objects. Hence, Z has only
four elements as the nops command demonstrates.
> nops(Z);
You can select elements from a set in the same way you select elements
from a list or a sequence, but the order of the elements in a set is session
dependent. Do not make any assumptions about this order.
You may also use the seq command to build sequences.
> seq( i^2, i=1..5 );
1, 4, 9, 16, 25
> seq( f(i), i=X );
f(1.3), f(5.3), f(11.2), f(2.1), f(2.1)
You can create lists or sets by enclosing a sequence in square brackets

or braces, respectively. The following command creates a list of sets.
> [ seq( { seq( i^j, j=1..3) }, i=-2..2 ) ];
[{−8, −2, 4}, {−1, 1}, {0}, {1}, {2, 4, 8}]
Exercise
1. Write a Maple procedure which, given a list of lists of numerical data,
computes the means of each column of the data.
A MEMBER Procedure
You may want to write a procedure that determines whether a certain
object is an element of a list or a set. The procedure below uses the
return statement discussed in section 1.2.
> MEMBER := proc( a::anything, L::{list, set} )
> local i;
> for i from 1 to nops(L) do
> if a=L[i] then return true end if;
> end do;
> false;
> end proc:
Here 3 is a member of the list.
> MEMBER( 3, [1,2,3,4,5,6] );
true
The type of loop that MEMBER uses occurs so frequently that Maple
has a special version of the for loop for it.
> MEMBER := proc( a::anything, L::{list, set} )
> local i;
> for i in L do
> if a=i then return true end if;
> end do;
> false;
> end proc:
The symbol x is not a member of this set.
> MEMBER( x, {1,2,3,4} );
false
Instead of using your own MEMBER procedure, you can use the built-in
member command.
Exercise
1. Write a Maple procedure called POSITION which returns the position
i of an element x in a list L. That is, POSITION(x,L) should return
an integer i > 0 such that L[i]=x. Return 0 if x is not in the list L.
Binary Search
One of the most basic and well-studied computing problems is that of
searching. A typical problem involves searching a list of words (a dictio-
nary, for example) for a specific word w.
Many possible solutions are available. One approach is to search the
list by comparing each word in turn with w until Maple either finds w or
it reaches the end of the list.
> Search := proc(Dictionary::list(string), w::string)
> local x;
> for x in Dictionary do
> if x=w then return true end if
> end do;
> false
> end proc:
However, if the Dictionary is large, say 50 000 entries, this approach can
take a long time.
You can reduce the execution time required by sorting the Dictionary
before you search it. If you sort the dictionary into ascending order then
you can stop searching as soon as you encounter a word greater than w.
On average, you only have to look halfway through the dictionary.
Binary searching provides an even better approach. Check the word

in the middle of the dictionary. Since you already sorted the dictionary
you can tell whether w is in the first or the second half. Repeat the process
with the appropriate half of the dictionary. The procedure below searches
the dictionary, D, for the word, w, from position, s, to position, f , in D.
The lexorder command determines the lexicographical ordering of two
strings.
> BinarySearch :=
> proc(D::list(string), w::string, s::integer, f::integer)
> local m;
> if s>f then return false end if; # entry was not found.
> m := iquo(s+f+1, 2); # midpoint of D.
> if w=D[m] then
> true;
> elif lexorder(w, D[m]) then
> BinarySearch(D, w, s, m-1);
> else
> BinarySearch(D, w, m+1, f);
> end if;
> end proc:
Here is a short dictionary.
> Dictionary := [ "induna", "ion", "logarithm", "meld" ];
Dictionary := [“induna”, “ion”, “logarithm”, “meld”]
Now search the dictionary for a few words.

> BinarySearch( Dictionary, "hedgehogs", 1, nops(Dictionary) );
false
> BinarySearch( Dictionary, "logarithm", 1, nops(Dictionary) );
true
> BinarySearch( Dictionary, "melodious", 1, nops(Dictionary) );
false
Exercises
1. Can you demonstrate that the BinarySearch procedure always ter-
minates? Suppose the dictionary has n entries. How many words in
the dictionary D does BinarySearch look at in the worst case?
2. Recode BinarySearch to use a while loop instead of calling itself

recursively.
Plotting the Roots of a Polynomial

You can construct lists of any type of object, even lists. A list of two
numbers often represents a point in the plane. The plot command uses
this structure to generate plots of points and lines.
> plot( [ [ 0, 0], [ 1, 2], [-1, 2] ],

> style=point, color=black );
1.5
0.5
–1 –0.5 0.5 1
You can use this approach to write a procedure which plots the com-
plex roots of a polynomial. Consider the polynomial x3 − 1.
> y := x^3-1;
y := x3 − 1
Numeric solutions are sufficient for plotting.

> R := [ fsolve(y=0, x, complex) ];
R := [−.5000000000 − .8660254038 I,
−.5000000000 + .8660254038 I, 1.]
You need to turn this list of complex numbers into a list of points in
the plane. The Re and Im commands pick the real and imaginary parts,
respectively.
> points := map( z -> [Re(z), Im(z)], R );
points := [[−.5000000000, −.8660254038],

[−.5000000000, .8660254038], [1., 0.]]
You can now plot the points.
> plot( points, style=point);
0.8
0.6
0.4
0.2
–0.4 –0.2 0 0.2 0.4 0.6 0.8 1
–0.2
–0.4
–0.6
–0.8
You can automate this technique. The input should be a polynomial

in x with constant coefficients.
> rootplot := proc( p::polynom(constant, x) )
> local R, points;
> R := [ fsolve(p, x, complex) ];
> points := map( z -> [Re(z), Im(z)], R );
> plot( points, style=point, symbol=circle );
> end proc:
Here is a plot of the roots of the polynomial x6 + 3x5 + 5x + 10.
> rootplot( x^6+3*x^5+5*x+10 );
0.5
–3 –2 –1 1
–0.5
–1
The randpoly command generates a random polynomial.

> y := randpoly(x, degree=100);
1.4 Computing with Formulæ • 33
y := 79 x71 + 56 x63 + 49 x44 + 63 x30 + 57 x24 − 59 x18
> rootplot( y );
0.5
–1 –0.5 0 0.5 1
–0.5
–1
When you write procedures, you often have several choices of how to
represent the data with which your procedures work. The choice of data
structure can have great impact on how easy it is to write your procedure
and its resulting efficiency. Section 2.4 describes an example of choosing
a data structure.
1.4 Computing with Formulæ

Maple’s real strength stems from its ability to perform symbolic manip-
ulations. This section demonstrates some of these capabilities through
sample programs for computation with polynomials. While the examples
are specific to polynomials, the techniques and methods apply to more
general formulæ.
In mathematics, a polynomial in the single variable, x, is most easily
recognizable in the expanded form,
n
X
ai xi , where if n > 0, then an 6= 0.
i=0
The ai s are the coefficients . They can be numbers or even expressions in-
volving variables. The crucial point is that each coefficient is independent
of (does not contain) x.
The Height of a Polynomial

The height of a polynomial is the largest magnitude (absolute value) of
the largest coefficient. The procedure below finds the height of a poly-
nomial, p, in the variable x. The degree command finds the degree of a
polynomial and the coeff command extracts specific coefficients from a
polynomial.
> HGHT := proc(p::polynom, x::name)
> local i, c, height;
> height := 0;
> for i from 0 to degree(p, x) do
> c := coeff(p, x, i);
> height := max(height, abs(c));
> end do;
> height;
> end proc:
The height of 32x6 − 48x4 + 18x2 − 1 is 48.
> p := 32*x^6-48*x^4+18*x^2-1;
p := 32 x6 − 48 x4 + 18 x2 − 1
> HGHT(p,x);
48
A significant weakness of the HGHT procedure is its inefficiency with

sparse polynomials; that is, polynomials with few terms relative to their
degree. For example, to find the height of x4321 − 1 the HGHT procedure
has to examine 4322 coefficients.
The coeffs command returns the sequence of coefficients of a poly-
nomial.
> coeffs( p, x );
−1, 32, −48, 18
You cannot map the abs command, or any other command, onto a
sequence. One solution is to turn the sequence into a list or a set.
> S := map( abs, {%} );
S := {1, 18, 32, 48}

The max command, however, works on sequences (but not on lists or

sets), so now you must turn the set into a sequence again.
> max( S[] );
48
The following version of HGHT uses this technique.

> HGHT := proc(p::polynom, x::name)
> local S;
> S := { coeffs(p, x) };
> S := map( abs, S );
> max( S[] );
> end proc:
Try the procedure out on a random polynomial.
> p := randpoly(x, degree=100 );
p := 79 x71 + 56 x63 + 49 x44 + 63 x30 + 57 x24 − 59 x18
> HGHT(p, x);
79
If the polynomial is in expanded form, you can also find its height
in the following manner. You can map a command directly onto a poly-
nomial. The map command applies the command to each term in the
polynomial.
> map( f, p );
f(79 x71 ) + f(56 x63 ) + f(49 x44 ) + f(63 x30 ) + f(57 x24 )
+ f(−59 x18 )
Thus, you can map abs directly onto the polynomial.
> map( abs, p );
79 |x|71 + 56 |x|63 + 49 |x|44 + 63 |x|30 + 57 |x|24 + 59 |x|18
Then use coeffs to find the sequence of coefficients of that polyno-

mial.
> coeffs( % );
79, 56, 49, 63, 57, 59
Finally, find the maximum.

> max( % );
79
Hence, you can calculate the height of a polynomial with this one-
liner.
> p := randpoly(x, degree=50) * randpoly(x, degree=99);
p := (77 x48 + 66 x44 + 54 x37 − 5 x20 + 99 x5 − 61 x3 )

(−47 x57 − 91 x33 − 47 x26 − 61 x25 + 41 x18 − 58 x8 )
> max( coeffs( map(abs, expand(p)) ) );
9214
Exercise
1. Write a pprocedure
Pn that computes the Euclidean norm of a polynomial;
2
that is, i=0 i | .
|a
The Chebyshev Polynomials, Tn (x)

The Chebyshev polynomials, Tn (x), satisfy the following linear recurrence.
Tn (x) = 2xTn−1 (x) − Tn−2 (x), for n ≥ 2.
The first two Chebyshev polynomials are T0 (x) = 1 and T1 (x) = x. This
example is similar to the Fibonacci example in section 1.2. Here is a simple
procedure, T, that computes Tn (x).
> T := proc(n::nonnegint, x::name)
> option remember, system;
> if n=0 then
> return 1;
> elif n=1 then
> return x;
> end if;
> 2*x*T(n-1,x) - T(n-2,x);
> end proc:
Maple does not automatically expand the polynomial.
> T(4,x);
2 x (2 x (2 x2 − 1) − x) − 2 x2 + 1
You can expand the polynomial yourself.

> expand(%);
8 x4 − 8 x2 + 1
You may be tempted to rewrite the procedure so that it expands the

result before returning it. However, this may be a waste of effort since
you do not know whether or not the user of your procedure wants the
Chebyshev polynomial in expanded form. Moreover, since the T procedure
is recursive, it would expand all the intermediate results as well.
Exercise
1. The Fibonacci polynomials, Fn (x), satisfy the linear recurrence
Fn (x) = xFn−1 (x) + Fn−2 (x),
where F0 (x) = 0 and F1 (x) = 1. Write a Maple procedure to compute

and factor Fn (x). Can you see any pattern?
Integration by Parts
Maple’s indefinite integral evaluator is very powerful. This section de-
scribes how you could write your own procedure for integrating formulæ
of the form
p(x)f (x),
where p(x) is a polynomial in x and f (x) is a special function. Here
p(x) = x2 and f (x) = ex .
> int( x^2*exp(x), x );
x2 ex − 2 x ex + 2 ex
As another example, here p(x) = x3 and f (x) = sin−1 (x).

> int( x^3*arcsin(x), x );
1 4 1 3p 3 p 3
x arcsin(x) + x 1 − x2 + x 1 − x2 − arcsin(x)
4 16 32 32
Usually you would use integration by parts to compute integrals of

this form.
> int( u(x)*v(x), x ) = u(x)*int(v(x),x) -
> int( diff(u(x),x) * int(v(x),x), x );
Z Z Z Z
∂
u(x) v(x) dx = u(x) v(x) dx − ( u(x)) v(x) dx dx
∂x
You can verify this formula by differentiating both sides of the equa-
tion.
> diff(%,x);
u(x) v(x) = u(x) v(x)
> evalb(%);
true
Applying integration by parts to the first example yields

Z Z Z Z Z
n x n x n−1
x e dx = x e dx − (nx e dx) dx = x e − n xn−1 ex dx.
x n x
It introduces a new integral, but the degree of x in that new integral is

one smaller than in the old integral. By applying
R the formula repeatedly,
the problem eventually reduces to evaluating ex , which is simply ex .
The following procedure uses integration by parts to calculate the
integral Z
xn ex dx ,
by calling itself recursively until n = 0.

> IntExpMonomial := proc(n::nonnegint, x::name)
> if n=0 then return exp(x) end if;
> x^n*exp(x) - n*IntExpMonomial(n-1, x);
> end proc:
x5 ex dx.
R
IntExpMonomial can calculate
> IntExpMonomial(5, x);
x5 ex − 5 x4 ex + 20 x3 ex − 60 x2 ex + 120 x ex − 120 ex
You can simplify this answer by using the collect command to group
the terms involving exp(x) together.
> collect(%, exp(x));
(x5 − 5 x4 + 20 x3 − 60 x2 + 120 x − 120) ex
You can now write a procedure which calculates p(x)ex dx for any
R
polynomial p. The idea is that integration is linear:

Z Z Z
af (x) + g(x) dx = a f (x) dx + g(x) dx.
The IntExpPolynomial procedure below uses coeff to extract the coef-

ficients of p one at a time.
> IntExpPolynomial := proc(p::polynom, x::name)
> local i, result;
> result := add( coeff(p, x, i)*IntExpMonomial(i, x),
> i=0..degree(p, x) );
> collect(result, exp(x));
> end proc:
(x2 + 1)(1 − 3x)ex dx.
R
Here IntExpPolynomial calculates
> IntExpPolynomial( (x^2+1)*(1-3*x), x );
(24 − 23 x + 10 x2 − 3 x3 ) ex
Exercises
1. Modify the procedure IntExpPolynomial to be more efficient by pro-
cessing only the non-zero coefficients of p(x).
2. The procedure IntExpPolynomial is quadratic in degree. Modify this
procedure again to make it linear in degree.
Computing with Symbolic Parameters

The polynomial 2x5 + 1 is an example of an explicit polynomial in x. All
the elements of the polynomial, except x, are explicit numbers. On the
other hand, polynomials like 3xn + 2, where n is an unspecified positive
integer, or a + x5 , where a is an unknown which is independent of x, are
examples of symbolic polynomials; they contain additional unspecified
symbolic parameters.
The
R procedure IntExpPolynomial in section 1.4 calculates the inte-
x
gral p(x)e dx where p is an explicit polynomial. In its present version
IntExpPolynomial cannot handle symbolic polynomials.
> IntExpPolynomial( a*x^n, x );

Error, invalid input: IntExpPolynomial expects its 1st
argument, p, to be of type polynom, but received a*x^n
You may want to extend IntExpPolynomial so that it can integrate

p(x)ex for symbolic polynomials
R n x p as well. The first problem is that of
finding a formula for x e dx for any natural number n. Often you can
find such a formula by carefully examining the pattern for specific results.
Here are the first few results for explicit values of n.
> IntExpPolynomial(x, x);
(x − 1) ex
> IntExpPolynomial(x^2, x);
(x2 − 2 x + 2) ex
> IntExpPolynomial(x^3, x);
(x3 − 3 x2 + 6 x − 6) ex
With sufficient time and ingenuity you would find the formula
n
(−1)n−i xi
Z X
xn ex dx = n! ex .
i!
i=0
This formula holds only for non-negative integers n. Use the assume
facility to tell Maple that the unknown n has certain properties.
> assume(n, integer);
> additionally(n >= 0);
Note that a simple type check is not sufficient to determine that n is an
integer.
> type(n, integer);
false
You need to use the is command, which is part of the assume facility.
> is(n, integer), is(n >= 0);
true, true
Thus, you can rewrite the IntExpMonomial procedure from section 1.4
in the following manner.
> IntExpMonomial := proc(n::anything, x::name)
> local i;
> if is(n, integer) and is(n >= 0) then
> n! * exp(x) * sum( ( (-1)^(n-i)*xî )/i!, i=0..n );
> else
> error("Expected a non-negative integer but received", n);
> end if;
> end proc:
This version of IntExpMonomial accepts both explicit and symbolic input.
> IntExpMonomial(4, x);
1 2 1 3 1 4
24 ex (1 − x + x − x + x )
2 6 24
In the next example, Maple evaluates the sum in terms of the gamma
function. The tilde (~) on n indicates that n carries an assumption.
> IntExpMonomial(n, x);
n~! ex ((−1)n~ e(−x) + x(n~+1) ((−x)(−1−n~) e(−x) Γ(2 + n~)

− (n~ + 1) (−x)(−1−n~) e(−x) Γ(n~ + 1, −x))/(n~ + 1)!)
You can check the answer by differentiating it with respect to x. The
simplify command reveals xn ex as expected.
> diff(%, x);
n~! ex (
x(n~+1) ((−x)(−1−n~) e(−x) Γ(2 + n~) − %1)
(−1)n~ e(−x) +
(n~ + 1)!
) + n~! ex (−(−1)n~ e(−x) + x(n~+1) (n~ + 1)
((−x)(−1−n~) e(−x) Γ(2 + n~) − %1)/(x (n~ + 1)!)+
(−x)(−1−n~) (−1 − n~) e(−x) Γ(2 + n~)
x(n~+1) (
x
− (−x)(−1−n~) e(−x) Γ(2 + n~)
(n~ + 1) (−x)(−1−n~) (−1 − n~) e(−x) Γ(n~ + 1, −x)
−
x
+ %1 − (n~ + 1) (−x)(−1−n~) e(−x) (−x)n~ ex )/(n~ + 1)!)
%1 := (n~ + 1) (−x)(−1−n~) e(−x) Γ(n~ + 1, −x)
> simplify(%);
ex xn~
Clearly, the use of symbolic constants in this way greatly extends the
power of the system.
Exercise
1. Extend the facility above to compute xn eax+b dx, where n is an
R
integer and a and b are constants. You must handle the case n = −1
separately since Z x
e
dx = −Ei(1, −x) .
x
Use the ispoly command from the Maple library to test for the ex-
pression ax + b which is linear in x.
1.5 Conclusion
This chapter introduced the basics of Maple programming. It first showed
you how to take a few lines of code and turn them into a useful procedure
simply by inserting them between proc() and end proc statements. Then
it introduced local and global variables and how to use them. As well, you
1.5 Conclusion • 43
were introduced to generating error messages, automatic type checking,

recursive procedures, and efficiency.
These concepts should equip you for writing many useful procedures.
In fact, you may put down this book and start writing some procedures
of your own.
The remainder of the book describes procedures in more detail. It
presents a formal introduction to the Maple language (chapter 4) and
the details of procedure definitions (chapter 5). It contains chapters on
special areas of application, including numerical (chapter 8) and graphical
programming (chapter 9). Explore these chapters when you are ready to
continue.
2 Fundamentals
By now, you have no doubt written a number of procedures and found

that Maple’s programming language greatly extends the range of tasks
you can tackle. Chapter 1 introduced a number of simple examples that
are useful as models for creating your own.
When programming in Maple, you may encounter unexpected situa-
tions. For instance, you may develop a sequence of commands which work
reliably and correctly when you execute them interactively, but then no
longer work when you incorporate them into a procedure by encapsulating
them between the proc() and end proc statements.
You are likely to encounter this situation if you write many programs.
Fortunately, the solution is almost always simple. A few fundamental
rules dictate how Maple reads what you type. An understanding of these
basic principles is particularly important within procedures, where you
encounter types of objects with which you may be unfamiliar.
Learning the basics is not difficult, especially if you understand five
particularly important areas:
1. Maple’s evaluation rules;
2. nested procedures;
3. useful details of types: types which modify Maple’s evaluation rules,

structured types, and type matching;
4. data structures: understanding how to make effective use of them in

order to best solve a problem; and
5. remember tables.
This chapter equips you with the fundamentals of Maple program-

ming, thereby allowing you to understand and write nontrivial Maple
code.
45
46 • Chapter 2: Fundamentals
2.1 Evaluation Rules

Maple does not evaluate lines of code within procedures in quite the same
way as it does if you enter those same lines in an interactive session. The
rules for evaluation are demonstrated in this section.
Of course, the evaluation rules within a procedure are different for
good reasons, some which have to do with efficiency. In an interactive
session, Maple evaluates most names and expressions completely. For in-
stance, suppose that you assign a the value b and then assign b the value
c. When you subsequently type a, Maple automatically follows your list
of assignments to determine that the ultimate value of a is c.
> a := b;
a := b
> b := c;
b := c
> a + 1;
c+1
In an interactive session, Maple tirelessly follows your chain of assign-

ments, no matter how long the list. Within a procedure, however, Maple
is not so diligent.
The substitution of assigned values for a name is called evaluation,
and each step in this process is known as an evaluation level. By using
the eval command, you can explicitly ask Maple to perform evaluation
of names to specific levels.
> eval(a, 1);
> eval(a, 2);
If you do not specify a number of levels, Maple evaluates the name to

as many levels as exist.
2.1 Evaluation Rules • 47
> eval(a);
When you enter commands at the prompt, Maple usually evaluates the
names as if you had enclosed each one in an eval(). The main exception
is that evaluation stops whenever evaluating to one more level would turn
the name into one of a table, an array, a procedure, or a module. The
command a + 1 above is almost identical to eval(a) + 1.
In procedures, some rules are different. If you use the previous assign-
ments within a procedure, you may get unexpected results.
> f := proc()
> local a,b;
> a := b;
> b := c;
> a + 1;
> end proc;
f := proc() local a, b; a := b ; b := c ; a + 1 end proc
> f();
b+1
The answer is b + 1 instead of c + 1, because a is a local variable and

Maple evaluates local variables to only one level. The procedure behaves
as if the final line were eval(a,1) + 1. Evaluating local variables fully
is inefficient both in terms of time and memory. To evaluate a variable
fully, Maple may have to follow a long list of assignments, resulting in a
large expression.
The following sections introduce Maple’s evaluation rules systemati-
cally. They discuss what types of variables can exist within a procedure
and the evaluation rules applied to each.
Parameters
Chapter 1 introduced you to local and global variables, but proce-
dures have a more fundamental type of variable: parameters. Param-
eters are variables whose name appears between the parentheses of a
proc()expression. They have a special role within procedures, as Maple
replaces them with arguments when you execute the procedure.
Examine the following procedure which squares its first argument and
assigns the answer to the second argument, which must be a name.
> sqr1 := proc(x::anything, y::name)
> y := x^2;
> end proc;
sqr1 := proc(x::anything , y::name) y := x2 end proc
> sqr1(d, ans);
d2
> ans;
d2
The procedure squares the value of d and assigns the result to the
name ans. Try the procedure again, but this time use the name a which
Maple earlier assigned the value b. Remember to reset ans to a name
first.
> ans := ’ans’;
ans := ans
> sqr1(a, ans);
c2
> ans;
c2
From the answer, Maple clearly remembers that you assigned b to the
name a, and c to the name b. When did this evaluation occur?
To determine when, you must examine the value of x as soon as Maple
enters the procedure. Use the debugger to get Maple to stop just after
entering sqr1.
> stopat(sqr1);
[sqr1 ]
> ans := ’ans’:

> sqr1(a, ans);
sqr1:
1* y := x^2
The value of the formal parameter x is c.

DBG> x
c
sqr1:
1* y := x^2
DBG> cont
c2
> unstopat(sqr1):
In fact, Maple evaluates the arguments before invoking the procedure.

The steps Maple takes are best thought of in the following manner.
When you call a procedure, Maple evaluates the arguments appropriately,
given the context in which the call occurs. For example, if you call sqr1
from inside a procedure, then Maple evaluates a to one level. Thus, in the
procedure g below, Maple evaluates a to b rather than to c.
> g := proc()
> local a,b,ans;
> a := b;
> b := c;
> sqr1(a,ans);
> end proc;
g :=
proc() local a, b, ans ; a := b ; b := c ; sqr1(a, ans ) end proc
> g();
b2
Whether you call a procedure from the interactive level or from inside
a procedure, Maple evaluates the arguments before invoking the proce-
dure. Once Maple evaluates the arguments, it replaces all occurrences
of the procedure’s formal parameters with the actual arguments. Then
Maple invokes the procedure.
Because Maple only evaluates parameters once, you cannot use them
like local variables. The author of procedure cube, below, forgot that
Maple does not re-evaluate parameters.
> cube := proc(x::anything, y::name)
> y := x^3;
> y;
> end proc:
When you call cube as below, Maple does assign ans the value 23 , but
the procedure returns the name ans rather than its value.
> ans := ’ans’;
ans := ans
> cube(2, ans);
ans
> ans;
Maple replaces each y with ans, but Maple does not evaluate these
occurrences of ans again. Thus, the final line of cube returns the name
ans, not the value that Maple assigned to ans.
Use parameters to pass information into the procedure. You may think
of parameters as objects evaluated to zero levels.
Local Variables
Local variables are temporary storage places within a procedure. You can
create local variables by using the local declaration statement at the
beginning of a procedure. If you do not declare whether a variable is
local or global, Maple decides for you. If you make an assignment to a
variable within a procedure then Maple assumes that it should be local.

A local variable is different from any other variable, whether global or
local to another procedure, even if they have the same name. The rules
for determining local variables become a little more involved when nested
procedures are written, but the basic concepts are similar. See Section 2.2
for more details.
Maple only evaluates local variables to one level.
> f := proc()
> local a,b;
> a := b;
> b := c;
> a + 1;
> end proc;
f := proc() local a, b; a := b ; b := c ; a + 1 end proc
When you invoke f, Maple evaluates the a in a+1 one level to b.

> f();
b+1
Maple always uses last name evaluation for tables, arrays, modules,
and procedures. Therefore, if you assign a table, an array, a module, or
a procedure to a local variable, Maple does not evaluate that variable
unless you use eval. Maple creates the local variables of a procedure each
time you call the procedure. Thus, local variables are local to a specific
invocation of a procedure.
If you have not written many programs you might think that one level
evaluation of local variables is a serious limitation, but in fact code which
requires further evaluation of local variables is difficult to understand,
and is unnecessary. Moreover, because Maple does not attempt further
evaluations, it saves many steps, causing procedures to run faster.
Global Variables
Global variables are available from inside any procedure in Maple as well
as at the interactive level. Indeed, any name you use at the interactive
level is a global variable, allowing you to write a procedure which assigns
a value to a variable that is accessible again later from within another
procedure, from within the same procedure, or at the interactive level.
> h := proc()
> global x;
> x := 5;
> end proc:
> h();
> x;
Within procedures, use global variables with caution. The procedure

h assigns a value to the global variable x but it does not leave any warning
in your worksheet. If you then use x thinking that it is an unknown, you
can get unexpected error messages.
> diff( x^2, x);
Error, wrong number (or type) of parameters in function
diff
Moreover, if you write yet another procedure which uses the global
variable x, then the two procedures may use the same x in incompatible
ways.
Whether within a procedure or at the interactive level, Maple always
applies the same evaluation rules to global variables. It evaluates all global
names fully, except when the value of such a variable is a table, an array,
or a procedure, in which case, Maple halts its evaluation at the last name
in the chain of assignments. This evaluation rule is called last name
evaluation.
Hence, Maple evaluates parameters to zero levels, local variables
to one level, and global variables fully, except for last name evaluation.
As with local variables, the rules for determining which variables are
global are fully described in Section 2.2.
Exceptions
This section describes two exceptions of particular note to the rules for
evaluation.
The Ditto Operator The ditto operator , %, which recalls the last result,
is local to procedures but Maple evaluates it fully. When you invoke a
procedure, Maple initializes the local version of % to NULL.
> f := proc()
> local a,b;
> print( "Initially [%] has the value", [%] );
> a := b;
> b := c;
> a + 1;
> print( "Now [%] has the value", [%] );
> end proc:
> f();
“Initially [%] has the value”, []
“Now [%] has the value”, [c + 1]
The same special rules apply to the %% and %%% operators. Using local
variables instead of ditto operators makes your procedures easier to read
and debug.
Environment Variables The variable Digits, which determines the

number of digits that Maple uses when calculating with floating-point
numbers, is an example of an environment variable. Maple evaluates
environment variables in the same manner it evaluates global variables;
that is, Maple evaluates environment variables fully except for last name
evaluation. When a procedure returns, Maple resets all environment vari-
ables to the values they had when you invoked the procedure.
> f := proc()
> print( "Entering f. Digits is", Digits );
> Digits := Digits + 13;
> print( "Adding 13 to Digits yields", Digits );
> end proc:
> g := proc()
> print( "Entering g. Digits is", Digits );
> Digits := 77;
> print( "Calling f from g. Digits is", Digits );
> f();
> print( "Back in g from f. Digits is", Digits );
> end proc:
The default value of Digits is 10.

> Digits;
10
> g();
“Entering g. Digits is”, 10
“Calling f from g. Digits is”, 77

“Entering f. Digits is”, 77
“Adding 13 to Digits yields”, 90
“Back in g from f. Digits is”, 77
When returning from g, Maple resets Digits to 10.

> Digits;
10
See ?environment for a list of environment variables. You can also

make your own environment variables: Maple considers any variable whose
name begins with the four characters _Env to be an environment variable.
2.2 Nested Procedures

You can define a Maple procedure inside another Maple procedure. In-
deed, you may commonly write such procedures without realizing you
are writing nested procedures. In interactive sessions, you are no doubt
familiar with using the map command to apply some operation to the el-
ements of some type of structure. For example, you may want to divide
each element of a list by a number, such as 8.
> lst := [8, 4, 2, 16]:
> map( x->x/8, lst);
1 1
[1, , , 2]
2 4
The map command is also very useful inside a procedure. Consider

another variation on this command which appears in the following proce-
dure. The intent of this new procedure is to divide each element of a list
by the first element of that list.
2.2 Nested Procedures • 55
> nest := proc(x::list)

> local v;
> v := x[1];
> map( y -> y/v, x );
> end proc:
> nest(lst);
1 1
[1, , , 2]
2 4
Maple considers this use of map as an example of nested procedures

and applies its lexical scoping rules, which declare the v within the call
to map as the same v as in the outer procedure, nest.
The following section explains Maple’s scoping rules. You will learn
how Maple decides which variables are local to a procedure and which
are global. Understanding Maple’s evaluation rules for parameters, and
for local and global variables, allows you to make full use of the Maple
language.
Local Versus Global Variables

Usually when you write a procedure, you should explicitly declare which
variables are global and which are local. Declaring the scope of the vari-
ables makes your procedure easier to read and debug. However, sometimes
declaring the variables is not the way to go. In the nest procedure above,
the variable in the map command gets its meaning from the surrounding
procedure. What happens if you define this variable, v, as local to the
invocation of the procedure within map?
> nest2 := proc(x::list)
> local v;
> v := x[1];
> map( proc(y) local v; y/v; end, x );
> end proc:
> nest2(lst);
8 4 2 16
[ , , , ]
v v v v
If you examine nest2 closely, you should be able to determine why it

didn’t work the same as nest. It is obvious that you don’t want to have
the variables declared at all within the inner procedure, so that it can get
it’s proper meaning from the enclosing procedure.
Only two possibilities exist: either a variable is local to a pro-

cedure and certain procedures that are completely within it, or it is
global to the entire Maple session.
The method Maple uses for determining whether a variable is local
or global can be summarized as: The name of the variable is searched
for among the parameters, local declarations, and global declarations
of the procedure, and then among the parameters, local and global
declarations, and implicitly declared local variables of any surrounding
procedure(s), from the inside out. If found, that specifies the binding of
the variable.
If, using the above rule, Maple cannot determine whether a variable
should be global or local, the following default decisions are made for you.
If a variable appears on the left-hand side of an explicit assignment
or as the controlling variable of a for loop, then Maple assumes that
you intend the variable to be local. Otherwise, Maple assumes that the
variable is global to the whole session. In particular, Maple assumes by
default that the variables you only pass as arguments to other procedures,
which may set their values, are global.
The Quick-Sort Algorithm

Sorting algorithms are of particular interest to computer scientists. Even
if you have never formally studied them you can appreciate that many
things need sorting. Sorting a few numbers is quick and easy no matter
what approach you use, but sorting large amounts of data can be very
time consuming; thus, finding efficient methods is important.
The following quick-sort algorithm is a classic algorithm. The key to
understanding this algorithm is to understand the operation of partition-
ing. This involves choosing any one number from the array that you are
about to sort. Then, you reposition the numbers in the array that are less
than the number that you chose to one end of the array and reposition
numbers that are greater to the other end. Lastly, you insert the chosen
number between these two groups.
At the end of the partitioning, you have not yet entirely sorted the
array, because the numbers less or greater than the one you chose may
still be in their original order. This procedure divides the array into two
smaller arrays which are easier to sort than the original larger one. The
partitioning operation has thus made the work of sorting much easier.
Better yet, you can bring the array one step closer in the sorting process
by partitioning each of the two smaller arrays. This operation produces
four smaller arrays. You sort the entire array by repeatedly partitioning
the smaller arrays.
The partition procedure uses an array to store the list because you
can change the elements of an array directly. Thus, you can sort the array
in place and not waste any space generating extra copies.
The quicksort procedure is easier to understand if you look at the
procedure partition in isolation first. This procedure accepts an array
of numbers and two integers. The two integers are element numbers of the
array, indicating the portion of the array to partition. While you could
possibly choose any of the numbers in the array to partition around, this
procedure chooses the last element of the section of the array for that
purpose, namely A[n]. The intentional omission of global and local
statements is to show which variables Maple thinks are local and which
global by default. It is recommended, however, that you not make this
omission in your procedures.
> partition := proc(A::array(1, numeric),
> m::posint, n::posint)
> i := m;
> j := n;
> x := A[j];
> while i<j do
> if A[i]>x then
> A[j] := A[i];
> j := j-1;
> A[i] := A[j];
> else
> i := i+1;
> end if;
> end do;
> A[j] := x;
> eval(A);
> end proc:
Warning, ‘i‘ is implicitly declared local to procedure
‘partition‘
Warning, ‘j‘ is implicitly declared local to procedure
‘partition‘
Warning, ‘x‘ is implicitly declared local to procedure
‘partition‘
Maple declares i, j, and x local because the partition procedure

contains explicit assignments to those variables. partition also assigns
explicitly to A, but A is a parameter, not a local variable. Because you
do not assign to the name eval, Maple makes it the global name which
refers to the eval command.
After partitioning the array a below, all the elements less than 3
precede 3 but they are in no particular order; similarly, the elements
larger than 3 come after 3.
> a := array( [2,4,1,5,3] );
a := [2, 4, 1, 5, 3]
> partition( a, 1, 5);
[2, 1, 3, 5, 4]
The partition procedure modifies its first argument, thus changing

a.
> eval(a);
[2, 1, 3, 5, 4]
The final step in assembling the quick-sort procedure is to insert the

partition procedure within an outer procedure. The outer procedure
first defines the partition subprocedure, then partitions the array. Or-
dinarily, you might want to avoid inserting one procedure within another.
However, you will encounter situations in chapter 3 where you will find it
necessary to nest the procedures. Since the next step is to partition each
of the two subarrays by calling quicksort recursively, partition must
return the location of the element which divides the partition.
> quicksort := proc(A::array(1, numeric),
> m::integer, n::integer)
> local partition, p;
>
> partition := proc(m,n)
> i := m;
> j := n;
> x := A[j];
> while i<j do
> if A[i]>x then
> A[j] := A[i];
> j := j-1;
> A[i] := A[j];
> else
> i := i+1;
> end if;
> end do;
> A[j] := x;
> p := j;
> end proc:
>
> if m<n then # if m>=n there is nothing to do
> p:=partition(m, n);
> quicksort(A, m, p-1);
> quicksort(A, p+1, n);
> end if;
>
> eval(A);
> end proc:
‘partition‘
Warning, ‘j‘ is implicitly declared local to procedure
‘partition‘
Warning, ‘x‘ is implicitly declared local to procedure
‘partition‘
> a := array( [2,4,1,5,3] );
a := [2, 4, 1, 5, 3]
> quicksort( a, 1, 5);
[1, 2, 3, 4, 5]
> eval(a);
[1, 2, 3, 4, 5]
Maple determines that the A and p variables in the partition sub-

procedure are defined by the parameter and local variable (respectively)
from the outer quicksort procedure and everything works as planned.
We could also have passed A as a parameter to the partition subpro-
cedure (as we did when partition was a stand-alone procedure), but
because of the scoping rules, it wasn’t necessary.
Creating a Uniform Random Number Generator

If you want to use Maple to simulate physical experiments, you likely
need a random number generator. The uniform distribution is particu-
larly simple: any real number in a given range is equally likely. Thus, a
uniform random number generator is a procedure that returns a ran-

dom floating-point number within a certain range. This section develops
the procedure, uniform, which creates uniform random number genera-
tors.
The rand command generates a procedure which returns random in-
tegers. For example, rand(4..7) generates a procedure that returns ran-
dom integers between 4 and 7, inclusive.
> f := rand(4..7):
> seq( f(), i=1..20 );
5, 6, 5, 7, 4, 6, 5, 4, 5, 5, 7, 7, 5, 4, 6, 5, 4, 5, 7, 5
The uniform procedure should be similar to rand but should return

floating-point numbers rather than integers. You can use rand to gener-
ate random floating-point numbers between 4 and 7 by multiplying and
dividing by 10^Digits.
> f := rand( 4*10^Digits..7*10^Digits ) / 10^Digits:
> f();
12210706011
2000000000
The procedure f returns fractions rather than floating-point numbers

so you must compose it with evalf; that is, use evalf(f()). Alterna-
tively, you can perform this operation by using Maple’s composition op-
erator, @.
> (evalf @ f)();
6.648630719
The uniform procedure below uses evalf to evaluate the constants in

the range specification, r, to floating-point numbers, the map command to
multiply both endpoints of the range by 10^Digits, and round to round
the results to integers.
> uniform := proc( r::constant..constant )
> local intrange, f;
> intrange := map( x -> round(x*10^Digits), evalf(r) );
> f := rand( intrange );
> (evalf @ eval(f)) / 10^Digits;
> end proc:
You can now generate random floating-point numbers between 4

and 7.
> U := uniform(4..7):
> seq( U(), i=1..20 );
4.559076346, 4.939267370, 5.542851096, 4.260060897,

4.976009937, 5.598293374, 4.547350944,
5.647078832, 5.133877918, 5.249590037,
4.120953928, 6.836344299, 5.374608653,
4.586266491, 5.481365622, 5.384244382,
5.190575456, 5.207535837, 5.553710879,
4.163815544
The uniform procedure suffers from a serious flaw: uniform uses the
current value of Digits to construct intrange; thus, U depends on the
value of Digits when uniform creates it. On the other hand, the evalf
command within U uses the value of Digits that is current when you
invoke U. These two values are not always identical.
> U := uniform( cos(2)..sin(1) ):
> Digits := 15:
> seq( U(), i=1..8 );
.828316845400000, −.328875163100000,
.790988967100000, .624953401700000,
.362773633800000, .679519822000000,
−.0465278542000000, −.291055180800000
The proper design choice here is that U should depend only on the
value of Digits when you invoke U. The version of uniform below ac-
complishes this by placing all the computation inside the procedure that
uniform returns.
> uniform := proc( r::constant..constant )
>
> proc()
> local intrange, f;
> intrange := map( x -> round(x*10^Digits),
> evalf(r) );
> f := rand( intrange );
> evalf( f()/10^Digits );
> end proc;
> end proc:
The r within the inner proc is not declared as local or global, so it

becomes the same r as the parameter to the outer proc.
The procedure that uniform generates is now independent of the value
of Digits at the time you invoke uniform.
> U := uniform( cos(2)..sin(1) ):
> Digits := 15:
> seq( U(), i=1..8 );
.476383408581006, .554836962987261,
.147655743361511, .273247304736175,
.148172828708797, −.258115633420094,
.558246581434993, .518084711267009
This section introduced you to the rules Maple uses to decide which
variables are global or local. You have also seen the principal implications
of these rules. In particular, it introduced you to the tools available for
writing nested procedures.
2.3 Types
Types that Modify Evaluation Rules
Section 2.1 introduces the details of how Maple evaluates different kinds
of variables within a procedure: Maple evaluates global variables fully (ex-
cept for last-name evaluation) and local variables to one level. Maple eval-
uates the arguments to a procedure, depending upon the circumstances,
before invoking the procedure, and then simply substitutes the actual
parameters for the formal parameters within the procedure without any
further evaluation. All these rules seem to imply that nothing within the
procedure in any way affects the evaluation of arguments which occurs
before Maple invokes the procedure. In reality, the exceptions provide
convenient methods for controlling the evaluation of arguments which
make your procedures behave more intuitively. They also prevent eval-
uation which would result in the loss of information you wish available
within your procedure.
Maple uses different evaluation rules for some of its own commands,
for example, the evaln command. You have no doubt used this command
to clear the value of previously defined variables. If this command were
to evaluate its argument normally, it would be of no use for this purpose.
2.3 Types • 63
For example, if you assign x the value π, then Maple evaluates x to π

whenever you use the variable x.
> x := Pi;
x := π
> cos(x);
−1
If Maple behaved the same way when you type evaln(x), then Maple
would pass the value π to evaln, losing all references to the name x.
Therefore, Maple evaluates the argument to evaln in a special way: it
evaluates the argument to a name, not to the value that name may have.
> x := evaln(x);
x := x
> cos(x);
cos(x)
You will find it useful to write your own procedures which exhibit
this behavior. You may want to write a procedure which returns a value
by assigning it to one of the arguments. Section 2.1 describes such a
procedure, sqr1, but each time you call sqr1 you must take care to pass
it an unassigned name.
> sqr1:= proc(x::anything, y::name)
> y := x^2;
> end proc:
This procedure works fine the first time you call it. However, you must
make sure that the second argument is indeed a name; otherwise, an error
results. In the example below, the error occurs because, upon the second
attempt, ans has the value 9.
> ans;
ans
> sqr1(3, ans);
> ans;
> sqr1(4, ans);

Error, invalid input: sqr1 expects its 2nd argument, y,
to be of type name, but received 9
You have two ways around this problem. The first is to use either
single quotes or the evaln command to ensure that Maple passes a name
and not a value. The second is to declare the parameter to be of type
evaln.
Just like the evaln command, declaring a parameter to be of type
evaln causes Maple to evaluate that argument to a name, so you do not
have to worry about evaluation when you use the procedure.
> cube := proc(x::anything, y::evaln)
> y := x^3;
> end proc:
> ans;
> cube(5, ans);
125
> ans;
125
In the above case, Maple passes the name ans to the cube procedure
instead of the value 9.
Using the evaln declaration is generally a good idea. It ensures that
your procedures do what you expect instead of returning cryptic error
messages. However, some Maple programmers like to use the single quotes.
When the call to the procedure is within a procedure itself, the presence
2.3 Types • 65
of the single quotes is a reminder that you are assigning a value to a

parameter. However, if you plan to use your procedure interactively, you
will find using evaln far more convenient.
A second type which modifies Maple’s evaluation rules is uneval.
Where evaln makes Maple evaluate the argument to a name, uneval
leaves the argument unevaluated. This type is useful for two reasons. First,
sometimes you wish to write a procedure which treats a structure as an
object and does not require knowledge of the details. Second, sometimes
expanding the argument within the procedure is useful. You may want
to write a version of the map command which is capable of mapping over
sequences. The standard map command built into Maple is not capable of
this because it evaluates its second argument. If the second argument is
the name of a sequence, Maple evaluates the name to the sequence before
invoking map. Since Maple flattens sequences of sequences, it passes only
the first element of the sequence as the second argument to map and the
other elements become additional arguments.
The smap procedure below uses an uneval declaration to tell Maple
not to evaluate its second argument. Once inside the procedure, the eval
command fully evaluates S. The whattype command returns exprseq if
you pass it a sequence.
> whattype( a, b, c );
exprseq
If S is not a sequence, smap simply calls map. args[3..-1] is the

sequence of arguments to smap after S. If S is a sequence, enclosing it in
square brackets forms a list. You can then map f onto the list and use
the selection operator, [], to turn the resulting list back into a sequence.
> smap := proc( f::anything, S::uneval )
> local s;
> s := eval(S);
> if whattype(s) = ’exprseq’ then
> map( f, [s], args[3..-1] )[];
> else
> map( f, s, args[3..-1] );
> end if;
> end proc:
Now you can map over sequences as well as lists, sets, and other
expressions.
> S := 1,2,3,4;
S := 1, 2, 3, 4
> smap(f, S, x, y);
f(1, x, y), f(2, x, y), f(3, x, y), f(4, x, y)
> smap(f, [a,b,c], x, y);
[f(a, x, y), f(b, x, y), f(c, x, y)]
Both evaln and uneval greatly extend the flexibility of Maple’s pro-
gramming language and the types of procedures you can write.
Structured Types
Sometimes a simple type check, either through declared formal parameters
or explicitly with the type command, does not provide enough informa-
tion. A simple check tells you that 2x is an exponentiation but it does not
distinguish between 2x and x2 .
> type( 2^x, ‘^‘ );
true
> type( x^2, ‘^‘ );
true
To make such distinctions you need structured types . For example,

2 is a constant and x is a name, so 2x has type constant^name but x2
does not.
> type( 2^x, constant^name );
true
> type( x^2, constant^name );
false
Suppose you want to solve a set of equations. Before proceeding you

want to remove any equations that are trivially true, like 4 = 4. Thus,
2.3 Types • 67
you need to write a procedure that accepts a set of equations as input.

The procedure nontrivial below uses automatic type checking to ensure
that the argument is indeed a set of equations.
> nontrivial := proc( S::set( ‘=‘ ) )
> remove( evalb, S );
> end proc:
> nontrivial( { x^2+2*x+1=0, y=y, z=2/x } );
2
{x2 + 2 x + 1 = 0, z = }
x
You can easily extend nontrivial so that it accepts general relations

rather than just equations, and so that it allows both sets and lists of
relations. An expression matches a set of types if it matches one of the
types in the set.
> nontrivial := proc( S::{ set(relation), list(relation) } )
> remove( evalb, S );
> end proc:
> nontrivial( [ 2<=78, 1/x=9 ] );
1
[ = 9]
x
You can extend nontrivial even further: if an element in S is not a

relation but an algebraic expression, f , then nontrivial should treat it
as the equation f = 0.
> nontrivial := proc( S::{ set( {relation, algebraic} ),
> list( {relation, algebraic} ) } )
> local istrivial;
> istrivial := proc(x)
> if type(x, relation) then evalb(x);
> else evalb( x=0 );
> end if;
> end proc;
> remove( istrivial, S );
> end proc:
> nontrivial( [ x^2+2*x+1, 23>2, x=-1, y-y ] );
[x2 + 2 x + 1, x = −1]
Automatic type checking is a very powerful tool. It allows you to do a

large amount of checking for invalid arguments automatically. You should
make using it a habit. Structured types allow checking even when you
design a procedure to accept a variety of inputs, or to rely on a particular

structure in its arguments.
Automatic type checking has two weaknesses. First, if the structure
of the type is complicated, permitting several structures, then the code
for the type checking can become cumbersome. The second is that Maple
does not save any of the information about the structure of the arguments.
It parses and checks them, but then the structure is lost. If you wish to
extract a particular component of the structure you must write more code
to do so.
The complexity of the types is rarely of concern in practice. A pro-
cedure which relies on arguments with a complicated structure is usually
hard to use. The typematch command addresses the duplication of ef-
fort in parsing the arguments. This command provides a more flexible
alternative method of type checking.
Type Matching
Section 1.4 describes the following pair of procedures that implement
indefinite integration of any polynomial multiplied by ex .
> IntExpMonomial := proc(n::nonnegint, x::name)
> if n=0 then return exp(x) end if;
> x^n*exp(x) - n*IntExpMonomial(n-1, x);
> end proc:
> IntExpPolynomial := proc(p::polynom, x::name)
> local i, result;
> end proc:
You may want to modify IntExpPolynomial so that it can also per-

form definite integration. The new version of IntExpPolynomial should
allow its second argument to be a name, in which case IntExpPolynomial
should perform indefinite integration, or the form name =range . You could
use the type command and if statements to do this, but then the pro-
cedure becomes difficult to read.
> IntExpPolynomial := proc(p::polynom, xx::{name, name=range})
> local i, result, x, a, b;
> if type(xx, name) then
> x:=xx;
> else
> x := lhs(xx);
> a := lhs(rhs(xx));
> b := rhs(rhs(xx));
> end if;
2.3 Types • 69
> if type(xx, name) then
> else
> eval(result, x=b) - eval(result, x=a);
> end if;
> end proc:
Using the typematch command makes your procedure much easier to

read. The typematch command not only tests if an expression matches a
certain type, it can also assign variables to pieces of the expression. Below,
typematch checks that expr is of the form name=integer..integer and
it assigns the name to y, the left-hand limit to a, and the right-hand limit
to b.
> expr := myvar=1..6;
expr := myvar = 1..6
> typematch( expr, y::name=a::integer..b::integer );
true
> y, a, b;
myvar , 1, 6
The version of IntExpPolynomial below uses the typematch com-

mand.
> IntExpPolynomial := proc(p::polynom, expr::anything )
> local i, result, x, a, b;
> if not typematch( expr, {x::name,
> x::name=a::anything..b::anything} ) then
> error( "expects a name or name=range but received",
> expr );
> end if;
> if type(expr, name) then
> else
> eval(result, x=b) - eval(result, x=a);
> end if;
> end proc:
Now IntExpPolynomial can perform definite, as well as indefinite,

integrals.
> IntExpPolynomial( x^2+x^5*(1-x), x=1..2 );
−118 e2 + 308 e
> IntExpPolynomial( x^2*(x-1), x);
(−4 x2 + 8 x − 8 + x3 ) ex
2.4 Choosing a Data Structure: Connected

Graphs
When writing programs you have to decide how to represent the data.
Sometimes the choice is straightforward but often it requires considerable
thought and planning. Some choices of data structure may make your
procedures more efficient or easier to write and debug. No doubt you are
familiar with Maple’s many available data structures, such as sequences,
lists, tables, and sets.
This section uses a variety of structures and discusses their advan-
tages. This section also illustrates, by means of an example, the problem
of choosing a data structure.
Suppose you have a number of cities with roads between them. Write a
procedure that determines whether you can travel between any two cities.
You can express this problem in terms of graph theory. Maple has
a networks package that helps you work with graphs and more general
structures. You do not need to understand graph theory or the networks
package to benefit from the examples in this section; these examples pri-
marily use the networks package as a shortcut to the drawing of G,
below.
> with(networks):
Make a new graph G and add a few cities (or vertices , in the termi-
nology of graph theory).
> new(G):
> cities := {Zurich, Rome, Paris, Berlin, Vienna};
2.4 Choosing a Data Structure: Connected Graphs • 71
cities := {Zurich, Rome, Paris , Berlin, Vienna}
> addvertex(cities, G);
Zurich, Rome, Paris , Berlin, Vienna
Add roads between Zurich and each of Paris, Berlin, and Vienna. The
connect command names the roads e1, e2, and e3.
> connect( {Zurich}, {Paris, Berlin, Vienna}, G );
e1 , e2 , e3
Add roads between Rome and Zurich and between Berlin and both
Paris and Vienna.
> connect( {Rome}, {Zurich}, G);
e4
> connect( {Berlin}, {Vienna, Paris}, G);
e5 , e6
Now draw the graph G.

> draw(G);
Paris
Rome
Berlin
Vienna
Zurich
If you look at the drawing above, you can convince yourself that, in
this particular case, you could travel between any two cities. Instead of
visual inspection, you can also use the connectivity command.
> evalb( connectivity(G) > 0 );
true
The data structures that the networks package uses are quite in-
volved, because that package supports more general structures than you
need in this example. The question then is: how would you represent the
cities and roads? Since cities have distinct names and the order of the
cities is irrelevant, you could represent the cities as a set of names.
> vertices(G);
{Zurich, Rome, Paris , Berlin, Vienna}
The networks package assigns distinct names to the roads, so it can

also represent them as set of names.
> edges(G);
{e1 , e2 , e3 , e4 , e5 , e6 }
You can also represent a road as the set consisting of the two cities
the road connects.
> ends(e2, G);
{Zurich, Berlin}
Thus, you can represent the roads as a set of sets.

> roads := map( ends, edges(G), G);
roads := {{Zurich, Rome}, {Zurich, Paris },

{Zurich, Berlin}, {Zurich, Vienna}, {Paris , Berlin},
{Berlin, Vienna}}
Unfortunately, if you want to know which cities are directly connected
to Rome, for example, you have to search through the whole set of roads.
Therefore, representing the data as a set of cities and a set of roads is com-
putationally inefficient for determining whether you can travel between
any two cities.
You can also represent the data as an adjacency matrix : a square

matrix with a row for each city. The (i, j)th entry in the matrix is 1 if
the ith and the jth city have a road between them, and 0 otherwise. The
following is the adjacency matrix for the graph G.
> adjacency(G);
 
0 1 0 1 1
1 0 0 0 1
 
0 0 0 0 1
 
1 0 0 0 1
1 1 1 1 0
The adjacency matrix is an inefficient representation if few roads exist

relative to the number of cities. In that case, the matrix contains many
zeros, representing an overall lack of roads. Also, though each row in the
matrix corresponds to a city, you cannot tell which row corresponds to
which city.
Here is yet another way of representing the cities and roads: Paris has
two roads between it and both Zurich and Berlin; thus, Berlin and Zurich
are the neighbors of Paris.
> neighbors(Paris, G);
{Zurich, Berlin}
You can represent the data as a table of neighbors; one entry should
be in the table for each city.
> T := table( map( v -> (v)=neighbors(v,G), cities ) );
T := table([Zurich = {Rome, Paris , Berlin, Vienna},

Rome = {Zurich}, Paris = {Zurich, Berlin},
Berlin = {Zurich, Paris , Vienna},
Vienna = {Zurich, Berlin}
])
The representation of a system of cities and roads as a table of neigh-
bors is ideally suited to answering the question of whether it is possible
to travel between any two cities. You can begin at one city. The table al-
lows you to efficiently find the neighboring cities to which you can travel.
Similarly, you can find the neighbors of the neighbors, and thus you can
quickly determine how far you can travel.
The connected procedure below determines whether you can travel
between any two cities. It uses the indices command to extract the set
of cities from the table.
> indices(T);
[Zurich], [Rome], [Paris ], [Berlin], [Vienna]
Since the indices command returns a sequence of lists, you must use
the op and map command to generate the set.
> map( op, {%} );
{Zurich, Rome, Paris , Berlin, Vienna}
The connected procedure initially visits the first city, v. Then

connected adds v to the set of cities that it has already visited and v’s
neighbors to the set of cities to which it can travel. As long as connected
can travel to more cities, it will. When connected has no more new cities
to which it can travel, it determines whether it has seen all the cities.
> connected := proc( T::table )
> local canvisit, seen, v, V;
> V := map( op, { indices(T) } );
> seen := {};
> canvisit := { V[1] };
> while canvisit <> {} do
> v := canvisit[1];
> seen := seen union {v};
> canvisit := ( canvisit union T[v] ) minus seen;
> end do;
> evalb( seen = V );
> end proc:
> connected(T);
true
You can add the cities Montreal, Toronto, and Waterloo, and the
highway between them.
> T[Waterloo] := {Toronto};
TWaterloo := {Toronto}
> T[Toronto] := {Waterloo, Montreal};
TToronto := {Waterloo, Montreal }
> T[Montreal] := {Toronto};
TMontreal := {Toronto}
Now you can no longer travel between any two cities; for example,
you cannot travel from Paris to Waterloo.
> connected(T);
false
Exercises
1. The system of cities and roads above splits naturally into two com-
ponents: the Canadian cities and the roads between them, and the
European cities and the roads between them. In each component you
can travel between any two cities but you cannot travel between the
two components. Write a procedure that, given a table of neighbors,
splits the system into such components. You may want to think about
the form in which the procedure should return its result.
2. The connected procedure above cannot handle the empty table of
neighbors.
> connected( table() );
Error, (in connected) invalid subscript selector
Correct this shortcoming.

The importance of this example is not to teach you about networks,
but to emphasize how the choice of data structures suited to the problem
allows you to create an efficient and concise version of the procedure
connected. Sets and tables were the best choices here. The best choice
for a problem that you wish to tackle may be very different. Before writing
code to perform your task, pause to consider which structures best suit
your needs. A good program design begins with choosing structures and
methods which mirror the data and task at hand.
2.5 Remember Tables

Sometimes procedures are designed such that they are called repeatedly
with the same arguments. Each time, Maple has to recompute the same
answer, unless you take advantage of Maple’s concept of remember ta-
bles .
Any Maple procedure can have a remember table. The purpose of
a remember table is to improve the efficiency of a procedure by storing
previous results so that Maple can retrieve them from the table instead
of recomputing them.
A remember table uses the sequence of actual parameters to the pro-
cedure call as the table index, and the results of the procedure calls as
the table values. Whenever you invoke a procedure which has a remem-
ber table, Maple searches the table for an index which is the sequence of
actual parameters. If such an index is found, it returns the corresponding
value in the table as the result of the procedure call. Otherwise, Maple
executes the body of the procedure.
Maple tables are hash tables, so looking up previously computed re-
sults is very fast. The purpose of remember tables is to make use of fast
table lookup in order to avoid recomputing results. Since remember tables
can become large, they are most useful when procedures need the same
results repeatedly and the results are expensive to compute.
The remember Option

Use the remember option to indicate to Maple that it should store the
result of a call to a procedure in a remember table. The Fibonacci pro-
cedure in section 1.2 is an example of a recursive procedure with the
remember option.
> option remember;
> if n<2 then return n end if;
> Fibonacci(n-1) + Fibonacci(n-2);
> end proc:
Section 1.2 demonstrates that the Fibonacci procedure is very slow

without the remember option, since it must compute the lower Fibonacci
numbers many times.
When you ask Fibonacci to calculate the third Fibonacci number, it
adds four entries to its remember table. The remember table is the fourth
operand of a procedure.
> Fibonacci(3);
2.5 Remember Tables • 77
> op(4, eval(Fibonacci));
table([0 = 0, 1 = 1, 2 = 1, 3 = 2])
Adding Entries Explicitly

You can also define entries in procedure remember tables yourself. To do
so, use the following syntax.
f (x ) := result :
Below is another procedure which generates the Fibonacci num-

bers. The fib procedure uses two entries in its remember table, where
Fibonacci uses an if statement.
> fib := proc(n::nonnegint)
> option remember;
> fib(n-1) + fib(n-2);
> end proc:
> fib(0) := 0:
> fib(1) := 1:
You must add entries in the remember table after making the pro-
cedure. The option remember statement does not create the remember
table, but rather asks Maple to automatically add entries to it. The pro-
cedure works without this option, but less efficiently.
You could even write a procedure which chooses which values to add
to its remember table. The following version of fib only adds entries to
its remember table when you call it with an odd-valued argument.
> if type(n,odd) then
> fib(n) := fib(n-1) + fib(n-2);
> else
> fib(n-1) + fib(n-2);
> end if;
> end proc:
> fib(0) := 0:
> fib(1) := 1:
> fib(9);
34
> op(4, eval(fib));
table([0 = 0, 1 = 1, 3 = 2, 5 = 5, 7 = 13, 9 = 34])
As in this case, sometimes you can dramatically improve the efficiency

of a procedure by remembering only some of the values instead of none.
Removing Entries from a Remember Table

You can remove entries from a remember table in the same manner you
remove entries from any other table: assign a table entry to its own name.
The evaln command evaluates an object to its name.
> T := op(4, eval(fib) );
T := table([0 = 0, 1 = 1, 3 = 2, 5 = 5, 7 = 13, 9 = 34])
> T[7] := evaln( T[7] );
T7 := T7
Now the fib procedure’s remember table has only five entries.
> op(4, eval(fib) );
table([0 = 0, 1 = 1, 3 = 2, 5 = 5, 9 = 34])
Maple can also remove remember table entries automatically. If you

give your procedure the system option, then Maple may remove entries in
the procedure’s remember table when Maple performs a garbage collec-
tion. Thus, you should never give the system option to procedures like
fib that rely on entries in its remember table to terminate.
You can remove a procedure’s remember table altogether by substi-
tuting NULL for the procedure’s fourth operand.
> subsop( 4=NULL, eval(Fibonacci) ):
> op(4, eval(Fibonacci));
You should use remember tables only with procedures whose results
depend exclusively on parameters. The procedure below depends on the
value of the environment variable Digits.
> f := proc(x::constant)
> option remember;
> evalf(x);
> end proc:
> f(Pi);
3.141592654
Even if you change the value of Digits, f(Pi) remains unchanged

because Maple retrieves the value from the remember table.
Digits := 44
> f(Pi);
3.141592654
2.6 Conclusion
A thorough understanding of the concepts in this chapter will provide you
with an excellent foundation for understanding Maple’s language. The
time you spend studying this chapter will save you hours puzzling over
trivial problems in subroutines and procedures which appear to behave
erratically. With the knowledge contained here, you should now see the
source of such problems with clarity. Just as you may have done after
finishing chapter 1, you may wish to put this book down for a while and
practice creating more of your own procedures.
Chapter 3 introduces you to more advanced techniques in Maple pro-
gramming. For example, it discusses procedures which return procedures,
procedures which query the user for input, and packages which you can
design yourself.
The remaining chapters of this manual are independent from one an-
other. You can focus on the topics of interest to you, for example, the
Maple debugger or Maple graphics programming. If you wish a more for-
mal presentation of the Maple language, take a look at chapters 4 and 5.
3 Advanced Programming
As you progress in learning the Maple programming language and

tackling more challenging projects, you may discover that you would like
more detailed information. The topics in this chapter are more advanced
than those in previous chapters, and some are difficult to follow without a
sound understanding of Maple’s evaluation rules, scoping rules, and other
principal concepts.
The first two sections in this chapter begin where section 2.2 left off,
using and returning procedures within the same procedure. Armed with
a basic knowledge of Maple’s evaluation rules, you will discover that such
procedures are not difficult to write.
Surprisingly, local variables can exist long after the procedure which
created them has finished. This feature can be particularly useful when
you wish a procedure to return a procedure, but the new procedure needs
a unique place to store information. Maple’s assume facility, for example,
uses such variables. The second section clearly explains and demonstrates
how to use them effectively.
Two special topics make up the remainder of this chapter: interactive
input and extending Maple. Interactive input allows you to write inter-
active procedures, making them more intuitive by querying the user for
missing information. Perhaps you wish to write an interactive tutorial
or a test. You are already aware of the customization power which you
gain through the ability to write procedures; Maple also supplies some
particularly useful mechanisms for modifying or extending Maple’s func-
tionality other than by writing a completely separate group of commands.
In conjunction with the topics you find in the specialized chapters in the
remainder of this book, the topics here will equip you to use Maple to its
fullest.
81
82 • Chapter 3: Advanced Programming
3.1 Procedures Which Return Procedures

Of all the types of procedures you may want to write, procedures which
return procedures are likely to cause the most trouble. Creating these
procedures builds upon the material presented in chapter 2, which cov-
ered procedures within procedures, how Maple evaluates parameters, and
how Maple assigns and evaluates both local and global variables. You
also learned, for example, that an inner procedure recognizes the variable
declarations of an outer procedure.
Some of the standard Maple commands return procedures. For exam-
ple, rand returns a procedure which in turn produces randomly chosen
integers from a specified range. If you use the type=numeric option with
dsolve, it returns a procedure which supplies a numeric estimate of the
solution to a differential equation.
You may wish to incorporate such features into your own programs.
The areas which require your particular attention are those of conveying
values from the outer procedure to the inner procedure, and the use of
local variables to store information unique to a returned procedure. This
section discusses the former. The latter is the topic of the next section,
section 3.2.
Creating a Newton Iteration

Newton’s method is one way of locating the roots of a function. First, you
pick a point on the x-axis that you think might be close to a root. Next,
you find the slope of the curve at the point you picked. Draw the tangent
to the curve at that point and observe where the tangent intersects the
x-axis. For most functions, this second point is closer to the real root than
your initial guess. Thus, to find the root, all you need to do is use the new
point as a new guess and keep drawing tangents and finding new points.
2
1.5
1
0.5
x0 x1
0 1 2 3 4 5 6 7 8
–0.5 x
–1
To find a numerical solution to the equation f (x) = 0, you may use

Newton’s method: guess an approximate solution, x0 ; then use the follow-
3.1 Procedures Which Return Procedures • 83
ing formula, which is the mathematical description of the above process,

to generate better approximations.
f (xk )
xk+1 = xk −
f 0 (xk )
You can implement this algorithm in Maple in a number of ways. The

program below takes a function and creates a new procedure, which takes
an initial guess and, for that particular function, generates the next guess.
Of course, the new procedure will not work for other functions. To find
the roots of a new function, use MakeIteration to generate a new guess-
generating procedure. The unapply command turns an expression into a
procedure.
> MakeIteration := proc( expr::algebraic, x::name )
> local iteration;
> iteration := x - expr/diff(expr, x);
> unapply(iteration, x);
> end proc:
√
Test the procedure on the expression x − 2 x.
> expr := x - 2*sqrt(x);
√
expr := x − 2 x
> Newton := MakeIteration( expr, x);

√
x−2 x
Newton := x → x −
1
1− √
x
It only takes Newton a few iterations to find the solution, x = 4.

> x0 := 2.0;
x0 := 2.0
> to 4 do x0 := Newton(x0); end do;

x0 := 4.828427124
x0 := 4.032533198
x0 := 4.000065353
x0 := 4.000000000
The MakeIteration procedure above expects its first argument to be

an algebraic expression. You can also write a version of MakeIteration
that works on functions. Since MakeIteration below knows that the pa-
rameter f is a procedure, you must use the eval command to evaluate it
fully.
> MakeIteration := proc( f::procedure )
> (x->x) - eval(f) / D(eval(f));
> end proc:
> g := x -> x - cos(x);
g := x → x − cos(x)
> SirIsaac := MakeIteration( g );
x → x − cos(x)
SirIsaac := (x → x) −
x → 1 + sin(x)
Note that SirIsaac does not contain references to the name g; thus,
you can change g without breaking SirIsaac. You can find a good ap-
proximate solution to x − cos(x) = 0 in a few iterations.
> x0 := 1.0;
x0 := 1.0
> to 4 do x0 := SirIsaac(x0) end do;
x0 := .7503638679
x0 := .7391128909
x0 := .7390851334
x0 := .7390851332
3.1 Procedures Which Return Procedures • 85
A Shift Operator
Consider the problem of writing a procedure that takes a function, f , as
input and returns a function, g, such that g(x) = f (x + 1). You can write
such a procedure in the following manner.
> shift := (f::procedure) -> ( x->f(x+1) ):
Try performing a shift on sin(x).
> shift(sin);
x → sin(x + 1)
Maple’s lexical scoping rules declare the f within the inner procedure
to be the same f as the parameter within the outer procedure. Therefore,
the command shift works as written.
The version of shift above works with univariate functions but it
does not work with functions of two or more variables.
> h := (x,y) -> x*y;
h := (x, y) → x y
> hh := shift(h);
hh := x → h(x + 1)
> hh(x,y);
Error, (in h) h uses a 2nd argument, y, which is
missing
If you want shift to work with multivariate functions, you must

rewrite it to deal with the additional parameters. In a procedure, args is
the sequence of actual parameters, and args[2..-1] is the sequence of
actual parameters except the first one; see section 4.4. It follows that the
procedure x->f(x+1,args[2..-1]) passes all its arguments except the
first directly to f .
> shift := (f::procedure) -> ( x->f(x+1, args[2..-1]) ):
> hh := shift(h);
hh := x → h(x + 1, args2..−1 )
> hh(x,y);
(x + 1) y
The function hh depends on h; if you change h, you implicitly change

hh;
> h := (x,y,z) -> y*z^2/x;
y z2
h := (x, y, z) →
x
> hh(x,y,z);
y z2
x+1
3.2 When Local Variables Leave Home

Section 2.2 states that local variables are not only local to a procedure but
also to an invocation of that procedure. Very simply, calling a procedure
creates and uses new local variables each time. If you invoke the same
procedure twice, the local variables it uses the second time are distinct
from those it used the first time.
What may surprise you is that the local variables do not necessar-
ily disappear when the procedure exits. You can write procedures which
return a local variable, either explicitly or implicitly, to the interactive
session, where it may survive indefinitely. You may find these renegade
local variables confusing, particularly since they may have the same name
as some global variables, or even other local variables which another pro-
cedure or a different call to the same procedure created. In fact, you can
create as many distinct variables as you want, all with the same name.
The procedure below creates a new local variable, a, and then returns
this new variable.
3.2 When Local Variables Leave Home • 87
> make_a := proc()

> local a;
> a;
> end proc;
make_a := proc() local a; a end proc
Since a set in Maple contains unique elements, you can easily verify
that each a that make_a returns is unique.
> test := { a, a, a };
test := {a}
> test := test union { make_a() };
test := {a, a}
> test := test union { ’make_a’()$5 };
test := {a, a, a, a, a, a, a}
Obviously, Maple identifies variables by more than their names.

Remember that no matter how many variables you create with the
same name, when you type a name in an interactive session, Maple in-
terprets that name to be of a global variable. Indeed, you can easily find
the global a in the above set test .
> seq( evalb(i=a), i=test);
true, false, false, false, false, false, false
You can use local variables to make Maple print things it would not
ordinarily be able to display. The above set test is an example. Another
example is expressions which Maple would ordinarily simplify automat-
ically. For example, Maple automatically simplifies the expression a + a
to 2a, so displaying the equation a+a = 2a is not easy. You can create the
illusion that Maple is showing you these steps using procedure make_a,
above.
> a + make_a() = 2*a;
a + a = 2a
To Maple, these two variables are distinct, even though they share the
same name.
You cannot easily assign a value to such escapees. Whenever you type
a name in an interactive session, Maple thinks you mean the global vari-
able of that name. While this prevents you from using the assignment
statement, it does not prevent you from using the assignment command.
The trick is to write a Maple expression which extracts the variable you
want. For example, in the equation above, you may extract the local a by
removing the global a from the left-hand side of the equation.
> eqn := %;
eqn := a + a = 2 a
> another_a := remove( x->evalb(x=a), lhs(eqn) );
another _a := a
You may then assign the global name a to this extracted variable and
so verify the equation.
> assign(another_a = a);
> eqn;
2a = 2a
> evalb(%);
true
Should your expression be complicated, you may need a fancier com-

mand to extract the desired variable.
You may have encountered this situation before without realizing it,
when you were using the assume facility and wished to remove an as-
sumption. The assume facility attaches various definitions to the variable
you specify, with one result being that the name subsequently appears
as a local name with an appended tilde. Maple does not understand if
you type the tilde name because no relationship exists with the global
variable name containing a tilde.
> assume(b>0);
> x := b + 1;
x := b~ + 1
> subs( ‘b~‘=c, x);
b~ + 1
When you clear the definition of the named variable the association
between the name and the local name with the tilde is lost, but expressions
created with the local name still contain it.
> b := evaln(b);
b := b
> x;
b~ + 1
If you later wish to reuse your expression, you must either perform a
substitution before removing the assumption or perform some manipula-
tions of your expressions similar to the equation eqn.
Creating the Cartesian Product of a Sequence of Sets

An important use for returning local objects arises when the returned
object is a procedure. When you write a procedure which returns a pro-
cedure, you will often find it useful to have the procedure create a variable
which holds information pertinent only to the returned procedure. This al-
lows different procedures (or different invocations of the same procedure)
to pass information between themselves.
The program introduced in this section uses this idea. When you pass
a sequence of sets to the procedure it constructs a new procedure. The
new procedure returns the next term in the Cartesian product each time
you invoke it. Local variables from the outer procedure are used to keep
track of which term to return next.
The Cartesian product of a sequence of sets is the set of all lists
whose ith entry is an element of the ith set. Thus, the Cartesian product
of {α, β, γ} and {x, y} is
{α, β, γ} × {x, y} = {[α, x], [β, x], [γ, x], [α, y], [β, y], [γ, y]}.
The number of elements in the Cartesian product of a sequence of sets

grows very rapidly as the sequence gets longer or the sets get larger. It
therefore requires a large amount of memory to store all the elements of
the Cartesian product. One way around this is to write a procedure that
returns a new element of the Cartesian product each time you call it. By
calling such a procedure repeatedly you can process every element in the
Cartesian product without ever storing all its elements at once.
The procedure below returns the next element of the Cartesian prod-
uct of the list s of sets. It uses an array, c, of counters to keep track of
which element comes next. For example, c[1]=3 and c[2]=1 correspond
to the third element of the first set and the first element of the second
set.
> s := [ {alpha, beta, gamma}, {x, y} ];
s := [{α, β, γ}, {x, y}]
> c := array( 1..2, [3, 1] );
c := [3, 1]
> [ seq( s[j][c[j]], j=1..2 ) ];
[γ, x]
Before you call the element procedure you must initialize all the coun-
ters to 1, except the first one, which should be 0.
> c := array( [0, 1] );
c := [0, 1]
In element below, nops(s) is the number of sets and nops(s[i])

is the number of elements in the ith set. When you have seen all the
elements, the procedure re-initializes the array of counters and returns
FAIL. Therefore, you can trace through the Cartesian product again by
calling element again.
> element := proc(s::list(set), c::array(1, nonnegint))

> local i, j;
> for i to nops(s) do
> c[i] := c[i] + 1;
> if c[i] <= nops( s[i] ) then
> return [ seq(s[j][c[j]], j=1..nops(s)) ] ;
> end if;
> c[i] := 1;
> end do;
> c[1] := 0;
> FAIL;
> end proc:
> element(s, c); element(s, c); element(s, c);
[α, x]
[β, x]
[γ, x]
> element(s, c); element(s, c); element(s, c);
[α, y]
[β, y]
[γ, y]
> element(s, c);
FAIL
Instead of writing a new procedure for each Cartesian product you

want to study, you can write a procedure, CartesianProduct, that re-
turns such a procedure. CartesianProduct below first creates a list, s,
of its arguments, which should all be sets. Then it initializes the array, c,
of counters and defines the subprocedure element. Finally, the element
subprocedure is invoked inside a proc structure.
> CartesianProduct := proc()
> local s, c, element;
> global S, C;
> s := [args];
> if not type(s, list(set)) then
> error "expected a sequence of sets, but received",
> args ;
> end if;

> c := array( [0, 1$(nops(s)-1)] );
>
> element := proc(s::list(set), c::array(1, nonnegint))
> local i, j;
> for i to nops(s) do
> c[i] := c[i] + 1;
> if c[i] <= nops( s[i] ) then
> return [ seq(s[j][c[j]], j=1..nops(s)) ] ;
> end if;
> c[i] := 1;
> end do;
> c[1] := 0;
> FAIL;
> end proc;
>
> proc()
> element(s, c);
> end proc;
> end proc:
Again, you can find all six elements of {α, β, γ} × {x, y}.
> f := CartesianProduct( {alpha, beta, gamma}, {x,y} );
f := proc() element (s, c) end proc
> to 7 do f() end do;
[α, x]
[β, x]
[γ, x]
[α, y]
[β, y]
[γ, y]
FAIL
You can use CartesianProduct to study several products simultane-

ously.
> g := CartesianProduct( {x, y}, {N, Z, R},
> {56, 23, 68, 92} );
g := proc() element (s, c) end proc
The following are the first few elements of {x, y} × {N, Z, R} ×

{56, 23, 68, 92}.
> to 5 do g() end do;
[x, N, 23]
[y, N, 23]
[x, Z, 23]
[y, Z, 23]
[x, R, 23]
Because the variables s in f and g are local variables to CartesianProduct,

they are not shared by different invocations of CartesianProduct. Sim-
ilarly, the variable c in f and g is also not shared. You can see that the
two arrays of counters are different by invoking f and g a few more times.
> to 5 do f(), g() end do;
[α, x], [y, R, 23]
[β, x], [x, N, 56]

[γ, x], [y, N, 56]
[α, y], [x, Z, 56]
[β, y], [y, Z, 56]
The element procedure in g is also local to CartesianProduct. There-

fore, you can change the value of the global variable element without
breaking g.
> element := 45;
element := 45
> g();
[x, R, 56]
These examples demonstrate not only that local variables can escape
the bounds of the procedures which create them, but that this mechanism
allows you to write procedures which create specialized procedures.
Exercises
1. The procedure that CartesianProduct generates does not work if one
of the sets is empty.
> f := CartesianProduct( {}, {x,y} );
f := proc() element (s, c) end proc
> f();
Error, (in element) invalid subscript selector
Improve the type-checking in CartesianProduct so that it generates

an informative error message in each such case.
2. A partition of a positive integer, n, is a list of positive integers whose

sum is n. The same integer can appear several times in the partition
but the order of the integers in the partition is irrelevant. Thus, the
following are all the partitions of 5:
[1, 1, 1, 1, 1], [1, 1, 1, 2], [1, 1, 3], [1, 2, 2], [1, 4], [2, 3], [5].
Write a procedure that generates a procedure that returns a new

partition of n each time you call it.
3.3 Interactive Input

Normally you pass input to Maple procedures as parameters. Sometimes,
however, you may want to write a procedure that asks the user directly
for input. For example, you could write a procedure that drills students
on some topic; the procedure could generate random problems and verify
the students’ answers. The input may be the value of a certain parameter,
or the answer to a question such as whether a parameter is positive or
not. The two commands in Maple for reading input from the terminal are
the readline command and the readstat command.
3.3 Interactive Input • 95
Reading Strings from the Terminal

The readline command reads one line of text from a file or the keyboard.
You may use the readline command as follows.
readline( filename )
If filename is the special name terminal, then readline reads a line of

text from the keyboard. readline returns the text as a string.
Here is a simple application, prompting the user for an answer to a
question.
> DetermineSign := proc(a::algebraic) local s;
> printf("Is the sign of %a positive? "
> "Answer yes or no:\n",a);
> s := readline(terminal);
> evalb( s="yes" or s = "y" );
> end proc:
> DetermineSign(u-1);
Is the sign of u-1 positive? Answer yes or no:
> y
true
Section 10.5 gives more details on the readline command.
Reading Expressions from the Terminal

You may want to write procedures that require the user to input an
expression rather than a string. The readstat command reads one ex-
pression from the keyboard.
readstat( prompt )
The prompt is an optional string.

> readstat("Enter degree: ");
Enter degree: n-1;
n−1
Notice that the readstat command insists on a terminating semicolon

(or colon). Unlike the readline command, which only reads one line, the
readstat command works like the rest of Maple: it allows you to break
a large expression across multiple lines. Another advantage of using the
readstat command is that if the user makes a mistake in the input,
the readstat command will automatically re-prompt the user for input,
giving the user an opportunity to correct the error.
> readstat("Enter a number: ");
Enter a number: 5^^8;

syntax error, ‘^‘ unexpected:
5^^8;
^
Enter a number: 5^8;
390625
Here is an application of the readstat command for implementing an

interface to the limit command. The procedure does the following: given
the function f (x), assume x is the variable if only one variable is present;
otherwise, ask the user what the variable is, and also ask the user for the
limit point.
> GetLimitInput := proc(f::algebraic)
> local x, a, K;
> # choose all variables in f
> K := select(type, indets(f), name);
>
> if nops(K) = 1 then
> x := K[1];
> else
> x := readstat("Input limit variable: ");
> while not type(x, name) do
> printf("A variable is required: received %a\n", x);
> x := readstat("Please re-input limit variable: ");
> end do;
> end if;
3.3 Interactive Input • 97
> a := readstat("Input limit point: ");

> x = a;
> end proc:
The expression sin(x)/x depends only on one variable, so GetLimitInput

does not ask for any limit variable.
> GetLimitInput( sin(x)/x );
Input limit point: 0;
x=0
Below, the user first tries to use the number 1 as the limit variable.
Since 1 is not a name, GetLimitInput asks for another limit variable.
> GetLimitInput( exp(u*x) );
Input limit variable: 1;

A variable is required: received 1
Please re-input limit variable: x;
Input limit point: infinity;
x=∞
You can specify a number of options to readstat; see section 10.5.
Converting Strings to Expressions

Occasionally, you may need more control over how and when Maple eval-
uates user input to your procedure than the readstat command allows.
In such cases, you can use the readline command to read the input as
a string, and the parse command to convert the string to an expression.
The string must represent a complete expression.
> s := "a*x^2 + 1";
s := “a*x^2 + 1”
> y := parse( s );
y := a x2 + 1
When you parse the string s you get an expression. In this case, you
get a sum.
> type(s, string), type(y, ‘+‘);
true, true
The parse command does not evaluate the expression it returns. You
must use eval to evaluate the expression explicitly. Below, Maple does
not evaluate the variable a to its value, 2, until you explicitly use the eval
command.
> a := 2;
a := 2
> z := parse( s );
z := a x2 + 1
> eval(z);
2 x2 + 1
See section 10.7 for more details about the parse command.
The techniques you have seen in this section are all very simple, but
you can use them to create powerful applications such as Maple tutorials,
procedures that drill students, or interactive lessons.
3.4 Extending Maple

Even though you may find it useful to write your own procedures to per-
form new tasks, sometimes extending the abilities of Maple’s own com-
mands is most beneficial. Many of Maple’s existing commands provide
this service. This section familiarizes you with the most helpful methods,
including making your own types and operators, modifying how Maple
3.4 Extending Maple • 99
displays expressions, and extending the abilities of such useful commands

as simplify and expand.
Defining New Types

If you use a complicated structured type you may find it easier to assign
the structured type to a variable of the form ‘type/name ‘. That way you
only have to write the structure once, thus reducing the risk of errors.
When you have defined the variable ‘type/name ‘, you can use name as
a type.
> ‘type/Variables‘ := {name, list(name), set(name)}:
> type( x, Variables );
true
> type( { x[1], x[2] }, Variables );
true
When the structured type mechanism is not powerful enough, you

can define a new type by assigning a procedure to a variable of the form
‘type/name ‘. When you test whether an expression is of type name,
Maple invokes the procedure ‘type/name ‘ on the expression if such
a procedure exists. Your procedure should return true or false. The
‘type/permutation‘ procedure below determines if p is a permutation
of the first n positive integers. That is, p should contain exactly one copy
of each integer from 1 through n.
> ‘type/permutation‘ := proc(p)
> local i;
> type(p,list) and { op(p) } = { seq(i, i=1..nops(p)) };
> end proc:
> type( [1,5,2,3], permutation );
false
> type( [1,4,2,3], permutation );
true
Your type-testing procedure may have more than one parameter.

When you test if an expression, expr, has type name (parameters ), then
Maple invokes
‘type/name ‘( expr, parameters )
if such a procedure exists. The ‘type/LINEAR‘ procedure below deter-

mines if f is a polynomial in V of degree 1.
> ‘type/LINEAR‘ := proc(f, V::name)
> type( f, polynom(anything, V) ) and degree(f, V) = 1;
> end proc:
> type( a*x+b, LINEAR(x) );
true
> type( x^2, LINEAR(x) );
false
> type( a, LINEAR(x) );
false
Exercises
1. Modify the ‘type/LINEAR‘ procedure so that you can use it to test
if an expression is linear in a set of variables. For example, x + ay + 1
is linear in both x and y, but xy + a + 1 is not.
2. Define the type POLYNOM(X ) which tests if an algebraic expression is

a polynomial in X where X may be a name, a list of names, or a set
of names.
Neutral Operators
Maple understands a number of operators, for example +, *, ^, and, not,
and union. All of these operators have special meaning to Maple: some
represent algebraic operations, such as addition or multiplication; some
represent logical operations; and some represent operations performed on
sets. Maple also has a special class of operators, the neutral operators ,
on which it does not impose any meaning. Instead, Maple allows you
to define the meaning of any neutral operator. The name of a neutral
operator begins with the ampersand character, &. Section 4.4 describes
the naming conventions for neutral operators.
> 7 &^ 8 &^ 9;
(7 &^ 8) &^ 9
> evalb( 7 &^ 8 = 8 &^ 7 );
false
> evalb( (7&^8)&^9 = 7&^(8&^9) );
false
Internally, Maple represents neutral operators as procedure calls.

Thus, 7&^8 is just a convenient way of writing &^(7,8).
> &^(7, 8);
7 &^ 8
Maple uses the infix notation only if your neutral operator has exactly
two arguments.
> &^(4), &^(5, 6), &^(7, 8, 9);
&^(4), 5 &^ 6, &^(7, 8, 9)
You can define the actions of a neutral operator by assigning a pro-

cedure to its name. The example below implements the Hamiltonians by
assigning a neutral operator to a procedure that multiplies two Hamil-
tonians. The next paragraph explains all you need to know about the
Hamiltonians to understand the example.
The Hamiltonians or Quaternions extend the complex numbers in
the same way the complex numbers extend the real numbers. Each Hamil-
tonian has the form a + bi + cj + dk where a, b, c, and d are real numbers.
The special symbols i, j, and k satisfy the following multiplication rules:
i2 = −1, j 2 = −1, k 2 = −1, ij = k, ji = −k, ik = −j, ki = j, jk = i,
and kj = −i.
The ‘&^‘ procedure below uses I, J, and K as the three special sym-
bols. However, I is implemented as the complex imaginary unit in Maple.
Therefore, you should assign another letter to represent the imaginary
unit by using the interface function. See ?interface for more informa-
tion.
> interface(imaginaryunit=j);
You can multiply many types of expressions by using ‘&^‘, making it

convenient to define a new type, Hamiltonian, by assigning a structured
type to the name ‘type/Hamiltonian‘.
> ‘type/Hamiltonian‘ := { ‘+‘, ‘*‘, name, realcons,
> specfunc(anything, ‘&^‘) };
type/Hamiltonian :=
{name, ∗, +, realcons , specfunc(anything , &^)}
The ‘&^‘ procedure multiplies the two Hamiltonians, x and y. If either
x or y is a real number or variable, then their product is the usual product
denoted by * in Maple. If x or y is a sum, ‘&^‘ maps the product onto
the sum; that is, ‘&^‘ applies the distributive laws: x(u + v) = xu + xv
and (u + v)x = ux + vx. If x or y is a product, ‘&^‘ extracts any real
factors. You must take special care to avoid infinite recursion when x
or y is a product that does not contain any real factors. If none of the
multiplication rules apply, ‘&^‘ returns the product unevaluated.
> ‘&^‘ := proc( x::Hamiltonian, y::Hamiltonian )
> local Real, unReal, isReal;
> isReal := z -> evalb( is(z, real) = true );
>
> if isReal(x) or isReal(y) then
> x * y;
>
> elif type(x, ‘+‘) then
> # x is a sum, u+v, so x&^y = u&^y + v&^y.
> map(‘&^‘, x, y);
>
> elif type(y, ‘+‘) then
> # y is a sum, u+v, so x&^y = x&û + x&^v.
> map2(‘&^‘, x, y);
>
> elif type(x, ‘*‘) then
> # Pick out the real factors of x.
> Real, unReal := selectremove(isReal, x);
> # Now x&^y = Real * (unReal&^y)
> if Real=1 then
> if type(y, ‘*‘) then
> Real, unReal := selectremove(isReal, x);
> Real * ’‘&^‘’(x, unReal);
> else
> ’‘&^‘’(x, y);
> end if;

> else
> Real * ‘&^‘(unReal, y);
> end if;
>
> elif type(y, ‘*‘) then
> # Similar to the x-case but easier since
> # x cannot be a product here.
> Real, unReal := selectremove(isReal, y);
> if Real=1 then
> ’‘&^‘’(x, y);
> else
> Real * ‘&^‘(x, unReal);
> end if;
>
> else
> ’‘&^‘’(x,y);
> end if;
> end proc:
You can place all the special multiplication rules for the symbols I,
J, and K in the remember table of ‘&^‘. See section 2.5.
> ‘&^‘(I,I) := -1: ‘&^‘(J,J) := -1: ‘&^‘(K,K) := -1:
> ‘&^‘(I,J) := K: ‘&^‘(J,I) := -K:
> ‘&^‘(I,K) := -J: ‘&^‘(K,I) := J:
> ‘&^‘(J,K) := I: ‘&^‘(K,J) := -I:
Since ‘&^‘ is a neutral operator, you can write products of Hamilto-

nians using &^ as the multiplication symbol.
> (1 + 2*I + 3*J + 4*K) &^ (5 + 3*I - 7*J);
20 + 41 I + 20 J − 3 K
> (5 + 3*I - 7*J) &^ (1 + 2*I + 3*J + 4*K);
20 − 15 I − 4 J + 43 K
> 56 &^ I;
56 I
Below, a is an unknown Hamiltonian until you tell Maple that a is an

unknown real number.
> a &^ J;
a &^ J
> assume(a, real);

> a &^ J;
a~ J
Exercise
1. The inverse of a general Hamiltonian, a + bi + cj + dk, is (a − bi − cj −
dk)/(a2 + b2 + c2 + d2 ). You can demonstrate this fact by assuming
that a, b, c, and d are real and define a general Hamiltonian, h.
> assume(a, real); assume(b, real);
> assume(c, real); assume(d, real);
> h := a + b*I + c*J + d*K;
h := a~ + b~ I + c~ J + d ~ K
By the formula above, the following should be the inverse of h.

> hinv := (a-b*I-c*J-d*K) / (a^2+b^2+c^2+d^2);
a~ − b~ I − c~ J − d ~ K
hinv :=
a~2 + b~2 + c~2 + d ~2
Now all you have to check is that h &^ hinv and hinv &^ h both
simplify to 1.
> h &^ hinv;
a~ (a~ − b~ I − c~ J − d ~ K)
%1
b~ (I a~ + b~ − c~ K + d ~ J)
+
%1
c~ (J a~ + b~ K + c~ − d ~ I)
+
%1
d ~ (K a~ − b~ J + c~ I + d ~)
+
%1
%1 := a~ + b~ + c~2 + d ~2
2 2
> simplify(%);
> hinv &^ h;
a~ (a~ − b~ I − c~ J − d ~ K)
%1
a~ b~ I + b~2 + b~ c~ K − b~ d ~ J
+
%1
a~ c~ J − b~ c~ K + c~2 + c~ d ~ I
+
%1
a~ d ~ K + b~ d ~ J − c~ d ~ I + d ~2
+
%1
%1 := a~ + b~ + c~2 + d ~2
2 2
> simplify(%);
Write a procedure, ‘&/‘, that computes the inverse of a Hamiltonian.

You may want to implement the following rules.
&/( &/x ) = x, &/(x&^y) = (&/y) &^ (&/x),

x &^ (&/x) = 1 = (&/x) &^ x.
Extending Certain Commands

If you introduce your own data structures, Maple cannot know how to ma-
nipulate them. In most cases, you design new data structures because you
want to write special-purpose procedures that manipulate them. However,
sometimes extending the capabilities of one or more of Maple’s built-in
commands is more intuitive. You can extend several Maple commands,
among them expand, simplify, diff, series, and evalf.
Suppose you choose to represent a polynomial an un +an−1 un−1 +· · ·+
a1 u + a0 by using the data structure
POLYNOM( u, a_0, a_1, ..., a_n )
You can then extend the diff command so that you can differentiate
polynomials represented in that way. If you write a procedure with a
name of the form ‘diff/F ‘ then diff invokes it on any unevaluated
calls to F. Specifically, if you use diff to differentiate F (arguments )
with respect to x, then diff invokes ‘diff/F ‘ as follows.
‘diff/F ‘( arguments, x )
The procedure below differentiates a polynomial in u with constant coef-

ficients with respect to x.
> ‘diff/POLYNOM‘ := proc(u)
> local i, s, x;
> x := args[-1];
> s := seq( i*args[i+2], i=1..nargs-3 );
> ’POLYNOM’(u, s) * diff(u, x);
> end proc:
> diff( POLYNOM(x, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1), x );
POLYNOM(x, 1, 2, 3, 4, 5, 6, 7, 8, 9)
> diff( POLYNOM(x*y, 34, 12, 876, 11, 76), x );
POLYNOM(x y, 12, 1752, 33, 304) y
The implementation of the Hamiltonians that section 3.4 describes

does not know that multiplication of Hamiltonians is associative, that
is (xy)z = x(yz). Sometimes, using associativity simplifies a result. Re-
call that I here is not the complex imaginary unit, but rather, one of
the special symbols I, J, and K that are part of the definition of the
Hamiltonians.
> x &^ I &^ J;
(x &^ I) &^ J
> x &^ ( I &^ J );
x &^ K
You can extend the simplify command so that it applies the as-
sociative law to unevaluated products of Hamiltonians. If you write a
procedure with a name of the form ‘simplify/F ‘, then simplify in-
vokes it on any unevaluated function calls to F. Thus, you must write a
procedure ‘simplify/&^‘ that applies the associative law to Hamiltoni-
ans.
The procedure below uses the typematch command to determine if
its argument is of the form (a&^b)&^c and, if so, it picks out the a, b,
and c.
> s := x &^ y &^ z;
s := (x &^ y) &^ z
> typematch( s, ’‘&^‘’( ’‘&^‘’( a::anything, b::anything ),

> c::anything ) );
true
> a, b, c;
x, y, z
You can give the user details about the simplifications your procedure
makes through the userinfo command. The ‘simplify/&^‘ procedure
prints out an informative message if you set infolevel[simplify] or
infolevel[all] to at least 2.
> ‘simplify/&^‘ := proc( x )

> local a, b, c;
> if typematch( x,
> ’‘&^‘’( ’‘&^‘’( a::anything, b::anything ),
> c::anything ) ) then
> userinfo(2, simplify, "applying the associative law");
> a &^ ( b &^ c );
> else
> x;
> end if;
> end proc:
Applying the associative law does make some products of Hamiltoni-

ans simpler.
> x &^ I &^ J &^ K;
((x &^ I) &^ J) &^ K
> simplify(%);
−x
If you set infolevel[simplify] large enough, Maple prints out infor-

mation on what simplify tries in order to make your expression simpler.
> infolevel[simplify] := 5;
infolevel simplify := 5
> w &^ x &^ y &^ z;
((w &^ x) &^ y) &^ z
> simplify(%);
simplify/&^: "applying the associative law"
simplify/&^: "applying the associative law"
w &^ ((x &^ y) &^ z)
The help pages for expand, series, and evalf provide details on how
you may extend those commands. See also section 8.4.
You may employ any or all of the above methods, as you see fit.
Maple’s design affords you the opportunity to customize it to suit your
needs, allowing you great flexibility.
3.5 Conclusion
The topics in this chapter and chapters 1 and 2 form the building blocks
of the programming features in Maple. Although the topics in this chapter
are more specialized than those of earlier chapters, they are still very im-
portant and are among the most useful. In particular, the first two sections
which delve into the workings of procedures which return procedures and
local variables are fundamental as you move on to more advanced pro-
gramming. The later topics, including interactive input and extending
Maple, while not as fundamental, are also extremely beneficial.
The remaining chapters in this book fall into two categories. Chapters
4 and 5 present in a formal manner the structure of the Maple language
and the details of procedures. The other chapters address specific topics,
such as plotting, numerical programming, and the Maple debugger.
4 The Maple Language
This chapter describes the Maple language in detail. The language

definition breaks down into four parts: characters, tokens, syntax (how
you enter commands), and semantics (the meaning Maple gives to the
language). The syntax and semantics are what define a language. Syntax
consists of rules to combine words into sentences; syntax is grammar, and
is purely mechanical. Semantics is the extra information or meaning that
syntax cannot capture, and determines what Maple does when it receives
a command.
Syntax The syntax defines what input constitutes a valid Maple expres-
sion, statement, or procedure. It answers such questions as:
• Do I need the parentheses in x^(y^z)?
• How do I input a string which is longer than a line?
• How can I input the floating-point number 2.3 × 10−3 ?
These are all questions about language syntax. They are concerned
solely with the input of expressions and programs to Maple, not what
Maple does with them.
If the input is not syntactically correct, Maple reports a syntax error.
Consider some interactive examples.
Two adjacent minus signs are not valid.
> --1;
syntax error, ‘-‘ unexpected:

--1;
^
111
112 • Chapter 4: The Maple Language
Maple accepts many kinds of floating-point formats,

> 2.3e-3, 2.3E-03, +0.0023;
.0023, .0023, .0023
but you must place at least one digit between the decimal point and the
exponent suffix.
> 2.e-3;
syntax error, missing operator or

2.e-3
^
The correct way to write this is 2.0e-3.
Semantics The semantics of the language specifies how expressions,

statements, and programs execute, that is, what Maple does with them.
This answers questions such as:
• Does x/2*z equal x/(2*z) or (x/2)*z? What about x/2/z?
• If x has the value 0, what will happen if I compute sin(x)/x?
• Why does computing sin(0)/ sin(0) result in 1 and not in an error?
• What is the value of i after executing the following loop?

> for i from 1 to 5 do print(i^2) end do;
The following is a common mistake. Many users think that x/2*z is

equal to x/(2*z).
> x/2*z, x/(2*z);
1 1x
x z,
2 2 z
4.1 Language Elements • 113
Syntax Errors in Files Maple reports syntax errors which occur when
reading in files and indicates the line number. Write the following program
in a file called integrand.
f:= proc(x)
t:= 1 - x^2
t*sqrt(t)
end proc:
Then read it in to your Maple session by using the read command.

> read integrand;
syntax error, missing operator or

t*sqrt(t)
^
Maple reports an error at the beginning of line 3. There should be a

“;” separating the two calculations, t := 1 - x^2 and t*sqrt(t).
4.1 Language Elements

To simplify the presentation of Maple syntax, consider it in two parts:
first, the language elements and second, the language grammar which
explains how to combine the language elements.
The Character Set

The Maple character set consists of letters, digits, and special characters.
The letters are the 26 lower-case letters
a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t, u, v, w, x, y, z,
and the 26 upper-case letters
A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P, Q, R, S, T, U, V,
W, X, Y, Z.
The 10 digits are
0, 1, 2, 3, 4, 5, 6, 7, 8, 9.
There are also 32 special characters , as shown in Table 4.1. Sections
later in this chapter state the uses of each.
Table 4.1 Special Characters

blank ( left parenthesis
; semicolon ) right parenthesis
: colon [ left bracket
+ plus ] right bracket
- minus { left brace
* asterisk } right brace
/ slash ‘ back quote
^ caret ’ single quote (apostrophe)
! exclamation " double quote
= equal | vertical bar
< less than & ampersand
> greater than _ underscore
@ at sign % percent
$ dollar \ backslash
. period # sharp
, comma ? question mark
Tokens
Maple’s language definition combines characters into tokens. Tokens
consist of keywords (reserved words), programming-language operators,
names, strings, natural integers, and punctuation marks.
Reserved Words Table 4.2 lists the reserved words in Maple. They have
special meanings, and thus you cannot use them as variables in programs.
Many other symbols in Maple have predefined meanings. For example,
mathematical functions such as sin and cos, Maple commands such as
expand and simplify, and type names such as integer and list. How-
ever, you can safely use these commands in Maple programs in certain
contexts. But the reserved words in table 4.2 have a special meanings,
and thus you cannot change them.
Programming-Language Operators Three types of Maple language

operators exist, namely binary, unary, and nullary operators. Tables
4.3 and 4.4 list these operators and their uses. The three nullary oper-
ators, %, %%, and %%% are special Maple names which refer to the three
previously computed expressions.
The ?precedence help page gives the order of precedence of all
programming-language operators.
Table 4.2 Reserved Words

Keywords Purpose
break, next loop control
if, then, elif, else if statement
for, from, in, by, to, for and while loops
while, do
proc, local, global, option, procedures
error, return options, description
export, module, use modules
end ends structure
assuming assume facility
try, catch, finally exception handling
read, save read and save statements
quit, done, stop ending Maple
union, minus, intersect, subset set operators
and, or, not, xor, implies Boolean operators
mod modulus operator
Table 4.3 Programming Binary Operators

Operator Meaning Operator Meaning
+ addition < less than
- subtraction <= less or equal
* multiplication > greater than
/ division >= greater or equal
^ exponentiation <> not equal
$ sequence operator -> arrow operator
@ composition union set union
@@ repeated composition minus set difference
&string neutral operator intersect set intersection
, expression separator :: type declaration,
|| concatenation pattern binding
. decimal point and logical and
.. ellipsis or logical or
mod modulo . non-commutative
:= assignment multiplication
Table 4.4 Programming Unary Operators

Operator Meaning
+ unary plus (prefix)
- unary minus (prefix)
! factorial (postfix)
$ sequence operator (prefix)
not logical not (prefix)
&string neutral operator (prefix)
. decimal point (prefix or postfix)
%integer label (prefix)
Names Maple’s language definition predefines many other tokens, in-

cluding names. For example, mathematical functions like sin and cos, or
commands like expand or simplify, or type names like integer or list
are all examples of names.
The simplest instance of a name consists of letters, digits, and under-
scores, and does not begin with a number. Maple reserves names beginning
with an underscore for internal use only.
Names of the form ~name are allowed for spreadsheet references.
Strings Maple’s language definition also predefines strings. Some simple

strings are "h", "hi", "result" and "Input value1". Generally, enclos-
ing any sequence of characters in double quotes forms a string.
> "The modulus should be prime";
“The modulus should be prime”
> "There were %d values";
“There were %d values”
You should not confuse the double quote character, ", which delimits
a string, with the back quote character, ‘, which forms a symbol or the
single quote, ’, which delays evaluation. A string’s length has no practical
limit in Maple. On most Maple implementations, this means that a string
can contain more than half a million characters.
To make the double quote character appear in a string, type a back-
slash character and a double quote (") where you want the double quote
character to appear.
> "a\"b";
“a\“b”
Similarly, to allow a backslash (escape character) to appear as one of

the characters in a string, type two consecutive backslashes \\.
> "a\\b";
“a\\b”
The special backslash characters mentioned above only count as one

character, as is demonstrated by using the length command.
> length(%);
A reserved word enclosed in double quotes also becomes a valid Maple

string, distinct from its usage as a token.
> "while";
“while”
The enclosing double quotes themselves do not form part of the string.
> length("abcde");
To access individual characters or substrings, strings can be sub-

scripted in much the same way as lists. An integer range provides access.
> S := "This is a string";
S := “This is a string”
> S[6..9];
“is a”
> S[-6..-1];
“string”
As well, iterations can be performed over the characters in a string.

> seq(i,i="over a string");
“o”, “v”, “e”, “r”, “ ”, “a”, “ ”, “s”, “t”, “r”, “i”, “n”, “g”
Integers A natural integer is any sequence of one or more digits. Maple

ignores any leading zeroes.
> 03141592653589793238462643;
3141592653589793238462643
The length limit for integers is system-dependent, but is generally

much larger than users require.
An integer is either a natural integer or a signed integer. Either
+natural or -natural indicates a signed integer.
> -12345678901234567890;
−12345678901234567890
> +12345678901234567890;
12345678901234567890
Token Separators
You can separate tokens by using either white space or punctuation
marks. This tells Maple where one token ends and the next begins.
Blanks, Lines, Comments, and Continuation The white space charac-

ters are space, tab, return, and line-feed. This book uses the terminology
newline to refer to either return or line-feed since the Maple system does
not distinguish between these characters. The terminology blank refers
to either space or tab. The white space characters separate tokens, but
are not themselves tokens.
White space characters cannot normally occur within a token.
> a: = b;
syntax error, ‘=‘ unexpected:

a: = b;
^
You can use white space characters freely between tokens.

> a * x + x*y;
ax + xy
The only instances in which white space can become part of a token
are names and strings, formed by enclosing a sequence of characters in
back quotes and double quotes, respectively. In these cases, the white
space characters are as significant as any other character.
On a line, unless you are in the middle of a string, Maple considers all
characters which follow a sharp character “#” to be part of a comment .
Since white space and newline characters are functionally the same,
you can continue statements from line to line.
> a:= 1 + x +
> x^2;
a := 1 + x + x2
The problem of continuation from one line to the next is less trivial
when long numbers or long strings are involved since these two classes of
tokens are not restricted to a few characters in length. The general mech-
anism in Maple to specify continuation of one line onto the next is as
follows: if the special character backslash, \, immediately precedes a new-
line character, then the parser ignores both the backslash and the newline.
If a backslash occurs in the middle of a line, Maple usually ignores it; see
?backslash for exceptions. You can use this to break up a long sequence
of digits into groups of smaller sequences, to enhance readability.
> "The input should be either a list of\
> variables or a set of variables";
“The input should be either a list of variables or \

a set of variables”
Table 4.5 Maple Punctuation Marks

; semicolon ( left parenthesis
: colon ) right parenthesis
’ single quote [ left bracket
‘ back quote ] right bracket
| vertical bar { left brace
< left angle bracket } right brace
> right angle bracket , comma
> G:= 0.5772156649\0153286060\

> 6512090082\4024310421\5933593992;
G := .5772156649015328606065120900824024\
3104215933593992
Punctuation Marks Table 4.5 lists the punctuation marks .
; and : Use the semicolon and the colon to separate statements. The
distinction between these marks is that a colon during an interactive
session prevents the result of the statement from printing.
> f:=x->x^2;
f := x → x2
> p:=plot(f(x), x=0..10):
’ Enclosing an expression, or part of an expression, in a pair of single

quotes delays evaluation of the expression (subexpression) by one
level. See section 4.4.
> ’’sin’’(Pi);
’sin’(π)
> %;
sin(π)
> %;
‘ To form symbols, use the back quote character.

> limit(f(x), x=0, ‘right‘):
() The left and right parentheses group terms in an expression and group
parameters in a function call.
> (a+b)*c; cos(Pi):
[] Use the left and right square brackets to form indexed (subscripted)
names and to select components from aggregate objects such as arrays,
rtables, tables, and lists. See section 4.4.
> a[1]: L:=[2,3,5,7]: L[3]:
[] and {} Use the left and right square brackets also to form lists, and
the left and right braces to form sets. See section 4.4.
> L:=[2,3,5,2]: S:={2,3,5,2}:
<> and | The left and right angle brackets in conjunction with the ver-
tical bar are used to construct rtable-based Matrices and Vectors.
> <<1,2,3> | <4,5,6>>:
, Use the comma to form a sequence, to separate the arguments of a func-

tion call, and to separate the elements of a list or set. See section 4.4.
> sin(Pi), 0, limit(cos(xi)/xi, xi=infinity):
4.2 Escape Characters

The escape characters are ?, !, #, and \. Their special meanings are
outlined below.
? The question mark character, if it appears as the first non-blank char-
acter on a line, invokes Maple’s help facility. The words following ?
on the same line determine the arguments to the help procedure. Use
either “,” or “/” to separate the words.
! The exclamation mark character, if it appears as the first non-blank
character on a line, passes the remainder of the line as a command
to the host operating system. This facility is not available on all plat-
forms.
# The hash mark character indicates that Maple is to treat the characters
following it on the line as a comment. In other words, Maple ignores
them. They have no effect on any calculation that Maple does.
\ Use the backslash character for continuation of lines and for grouping
of characters within a token. See section 4.1.
4.3 Statements
There are eight types of statements in Maple. They are the
1. assignment statement
2. selection statement
3. repetition statement
4. read statement
5. save statement
6. empty statement
7. quit statement
8. expressions
Section 4.4 discusses expressions at length.
Throughout the remainder of this section, expr stands for any expres-
sion, statseq stands for a sequence of statements separated by semicolons.
4.3 Statements • 123
The Assignment Statement

The syntax of the assignment statement is
name := expr ;
name_1, ..., name_n := expr_1, ..., expr_n ;
This assigns, or sets, the value of the variable name to be the re-
sult of executing the expression expr. Multiple assignments can also be
performed.
Names A name in Maple may be a symbol or an indexed name (sub-

script). Names stand for unknowns in formulæ. They also serve as pro-
gramming variables. A name only becomes a programming variable when
Maple assigns it a value. Otherwise, if Maple does not assign the name a
value, then it remains an unknown.
> 2*y - 1;
2y − 1
> x := 3; x^2 + 1;
x := 3
10
> a[1]^2; a[1] := 3; a[1]^2;
a1 2
a1 := 3
9
> f[Cu] := 1.512;
fCu := 1.512
To define a function, use the arrow notation, ->.

> phi := t -> t^2;
φ := t → t2
Note that the following does not define a function; instead an entry
is created in the remember table for phi. See section 2.5.
> phi(t) := t^2;
φ(t) := t2
Section 5.1 contains more on how to define functions.
Indexed Names Another form of a name in Maple is the indexed name

or subscripted name, which has the form
name [ sequence ]
Note that since an indexed name is itself a valid name, you can add
a succession of subscripts.
> A[1,2];
A1, 2
> A[i,3*j-1];
Ai, 3 j−1
> b[1][1], data[Cu,gold][1];
b11 , data Cu, gold 1
The use of the indexed name A[1,2] does not imply that A is an array,
as in some languages. The statement
> a := A[1,2] + A[2,1] - A[1,1]*A[2,2];
a := A1, 2 + A2, 1 − A1, 1 A2, 2
forms a formula in the four indexed names. (However, if A does evaluate

to an array or table, then A[1,1] refers to the (1, 1) element of the array
or table.)
Table 4.6 Maple Concatenation Operator

v || 5 p || "n" a || (2*i) V || (1..n) r || i || j
The Concatenation Operator Generally, you can form a name by using

the concatenation operator in one of the following three forms.
name || natural
name || string
name || ( expression )
Since a name can appear on the left-hand side, Maple allows a suc-
cession of concatenations. Some examples of the use of the concatenation
operator for name formation are given in Table 4.6.
The concatenation operator is a binary operator which requires a
name or a string as its left operand. Although Maple usually evaluates
expressions from left to right, it evaluates concatenations from right to
left. Maple evaluates the right-most operand, then concatenates to the left
operand. If it evaluates the right operand to an integer, string or name,
then the result of the concatenation is a string or name (depending on the
type of the left-most operand). If it evaluates the right operand to some
other type of object, say a formula, then the result of the operation is an
unevaluated concatenated object.
> p || n;
pn
> "p" || n;
“pn”
> n := 4: p || n;
p4
> p || (2*n+1);
p9
> p || (2*m+1);
p||(2 m + 1)
If the right hand expression is a sequence or a range and the operands

of the range are integers or character strings, then Maple creates a se-
quence of names.
> x || (a, b, 4, 67);
xa, xb, x4 , x67
> x || (1..5);
x1 , x2 , x3 , x4 , x5
> X || ("a".."g");
Xa, Xb, Xc, Xd , Xe, Xf , Xg
If more than one range appears, it composes the extended sequence

of names.
> x || (1..2) || (1..3);
x11 , x12 , x13 , x21 , x22 , x23
Maple never fully evaluates the left-most object, but rather evaluates
it to a name. Concatenations can also be formed with the cat command.
cat( sequence )
Note that all the arguments of the cat command are evaluated nor-
mally (as for any other function call); therefore
> cat( "a", "b", "c" );
“abc”
is equivalent to
> "" || a || b || c;
“abc”
Protected Names Many names in Maple have a predefined meaning,

and you cannot directly assign a value to them. For example, the names
of built-in functions such as sin, the sine function, utility operations such
as degree, which computes the degree of a polynomial, commands such
as diff for differentiation, and type names like integer and list, are
all protected names. When the user attempts to assign to any of these
names, an error occurs.
> list := [1,2];
Error, attempting to assign to ‘list‘ which is
protected
The system protects these names from accidental assignment. It is

possible to assign to these names by first unprotecting them as follows.
> unprotect(sin);
> sin := "a sin indeed";
sin := “a sin indeed”
However, now the areas of Maple that rely on the sine function will
not work properly.
> plot( 1, 0..2*Pi, coords=polar );
Plotting error, empty plot
On the other hand, to write programs in which you want to prevent

a user from assigning to certain names, use the protect command.
> mysqr := x -> x^2;
mysqr := x → x2
> protect( mysqr );

> mysqr := 9;
Error, attempting to assign to ‘mysqr‘ which is
protected
Unassignment: Clearing a Name

When names do not carry assigned values they act as unknowns. When
assigned values, they act as variables. It is often desirable to unassign
(or clear) a name which previously carried an assigned value, so that you
can use the name as an unknown again. The way to do this in Maple is
to assign the name to be itself. Maple understands this to mean clear
the name. The command
evaln(name )
evaluates name to a name (as opposed to evaluating name to its value as

in other function calls). You can thus unassign a name as follows.
> a := evaln(a);
a := a
> i := 4;
i := 4
> a[i] := evaln(a[i]);
a4 := a4
> a || i := evaln(a || i);
a4 := a4
In the special case where name is a string you may also unassign a vari-
able by delaying evaluation of the right-hand side with single quotes (’).
See section 4.4.
> a := ’a’;
a := a
Related Functions You can use the assigned command to test if a name
has an assigned value.
> assigned(a);
false
The assign command assigns a variable.

> assign( a=b );
> assigned(a);
true
> a;
Maple normally evaluates all the arguments of assign. Therefore,

because there was the previous assignment, assign(a=b), Maple assigns
b the value 2 here.
> assign( a=2 );
> b;
One level evaluation of a reveals that a still has the value b.

> eval( a, 1 );
Changing the value of a does not affect the value of b.

> a := 3;
a := 3
> b;
2
Often, applications of the assign command are to a set or list of

equations.
> eqn1 := x + y = 2:
> eqn2 := x - y = 3:
> sol := solve( {eqn1, eqn2}, {x, y} );
−1 5
sol := {y = ,x= }
2 2
Maple assigns the variables x and y according to the set sol of equa-
tions.
> assign(sol);
> x;
5
2
> assigned(x);
true
It is recommended that you not assign values to expressions like f(x).

See section 2.5 for details.
The Selection Statement

The selection or conditional statement has four forms. The syntax of the
first two forms is
if expr then statseq

end if;
if expr then statseq1

else statseq2
end if;
Maple executes the selection statement as follows. It evaluates the ex-

pression in the if clause (expr). If the result is the Boolean value true,
then Maple executes the statement sequence in the then clause. If the
result is the Boolean value false or FAIL, then Maple executes the state-
ments in the else clause.
> x := -2:
> if x<0 then 0 else 1 end if;
The expr must evaluate to one of the Boolean values true, false, or
FAIL; see section 4.4.
> if x then 0 else 1 end if;
Error, invalid boolean expression
Omit the else clause if you do not want to include an alternative

course of action when the condition is false.
> if x>0 then x := x-1 end if;
> x;
−2
The selection statement may be nested, that is, the statement se-
quence in the then clause or else clause may be any statement, including
an if statement.
Compute the sign of a number.
> if x > 1 then 1
> else if x=0 then 0 else -1 end if
> end if;
The following example demonstrates a use of FAIL.

> r := FAIL:
> if r then
> print(1)
> else
> if not r then
> print(0)
> else
> print(-1)
> end if
> end if;
−1
If Maple has many cases to consider, the use of nested if statements

becomes messy and unreadable. Maple provides the following two alter-
natives.

elif expr then statseq
end if;

elif expr then statseq
else statseq
end if;
The elif expr then statseq construct may appear more than
once.
Here you can implement the sign function by using an elif clause.
> x := -2;
x := −2
> if x<0 then -1

> elif x=0 then 0
> else 1
> end if;
−1
In this form, you can view the selection statement as a case statement
with the optional else clause as the default case. For example, if you are
writing a program that accepts a parameter n with four possible values,
0, 1, 2, 3, then you might write
> n := 5;
n := 5
> if n=0 then 0

> elif n=1 then 1/2
> elif n=2 then sqrt(2)/2
> elif n=3 then sqrt(3)/2
> else error "bad argument: %1", n;
> end if;
Error, bad argument: 5
The Repetition Statement

The most general repetition statement in Maple is the for loop. However,
you can replace many loops with more efficient and concise special forms.
See section 4.5.
The for loop has two forms: the for-from loop and the for-in loop.
The for-from Loop A typical for-from loop has the following form.
> for i from 2 to 5 do i^2 end do;
9
16
25
This sequence of results arose as follows. First, Maple assigns i the

value 2. Since 2 is less than 5, Maple executes the statement between the
do and the end do. Then it increments i by 1 to 3, tests again, the loop
executes, and so on until i is (strictly) larger than 5. In this case the final
value of i is
> i;
The syntax of the for-from loop is
for name from expr by expr to expr

while expr do statseq
end do;
You may omit any of the clauses for name , from expr , by expr , to
expr , or while expr . You may omit the sequence of statements statseq.
Except for the for clause, which must always appear first, the other
clauses may appear in any order. If you omit a clause, it has a default
value, which is shown in Table 4.7.
You could also write the previous example as the following.
Table 4.7 Clauses and Their Default Values

Clause Default Value
for dummy variable
from 1
by 1
to infinity
while true
> for i from 2 by 1 to 5 while true do i^2 end do:
If the by clause is negative, the for loop counts downward.

> for i from 5 to 2 by -1 do i^2 end do;
25
16
9
4
To find the first prime number greater than 107 you could write
> for i from 10^7 while not isprime(i) do end do;
Now i is the first prime larger than 107 .

> i;
10000019
Notice that the body of the loop is empty. Maple allows for the empty
statement. Try improving the program by considering only the odd num-
bers.
> for i from 10^7+1 by 2 while not isprime(i) do end do;
> i;
10000019
Here is an example of repeating an action n times. Throw a die five

times.
> die := rand(1..6):

> to 5 do die(); end do;
3
4
6
5
Omitting all clauses produces an infinite loop.
do statseq end do;
This is equivalent to
for name from 1 by 1 to infinity

while true do statseq
end do;
Such a loop statement will loop forever unless the break construct
(see section 4.3) or a return statement (see section 5.5) terminates it, or
if Maple encounters the quit statement, or if an error occurs.
The while Loop The while loop is a for loop with all its clauses omit-
ted except the while clause, that is
while expr do statseq end do;
The expression expr is called the while condition. It must be a

Boolean valued expression, that is, it must evaluate to true, false, or
FAIL. For example,
> x := 256;
x := 256
> while x>1 do x := x/4 end do;

x := 64
x := 16
x := 4
x := 1
The while loop works as follows. First, Maple evaluates the while
condition. If it evaluates to true, Maple executes the body of the loop.
This loop repeats until the while condition evaluates to false or FAIL.
Note that Maple evaluates the while condition before it executes the
body of the loop. An error occurs if the while condition does not evaluate
to one of true, false, or FAIL.
> x := 1/2:
> while x>1 do x := x/2 end do;
> x;
1
2
> while x do x := x/2 end do;

Error, invalid boolean expression
The for-in Loop Suppose you have a list of integers L and want to find
the integers in the list that are at most 7. You could write
> L := [7,2,5,8,7,9];
L := [7, 2, 5, 8, 7, 9]
> for i in L do
> if i <= 7 then print(i) end if;
> end do;
2
5
7
This example cycles through the components of an object. The object,

this time, is a list. But in other examples, the object might be a set, a sum
of terms, a product of factors, or the characters of a string. The syntax
for the for-in loop is
for name in expr

while expr do statseq
end do;
The loop index (the name specified in the for clause of the statement)
takes on the operands of the first expr. See section 4.4 for a description
of the operands associated with each data type. You can test the value
of the index in the optional while clause, and, of course, the value of the
index is available when you execute the statseq. Note that the value of
the index variable name remains assigned at the end of the loop if the
object contains at least one operand.
The break and next Commands Within the Maple language reside
two additional loop control constructs: break and next. When Maple
evaluates the special name break, the result is to exit from the innermost
repetition statement within which it occurs. Execution then proceeds with
the first statement following this repetition statement.
> L := [2, 5, 7, 8, 9];
L := [2, 5, 7, 8, 9]
> for i in L do
> print(i);
> if i=7 then break end if;
> end do;
5
7
When Maple evaluates the special name next, it then proceeds im-
mediately to the next iteration. For example, suppose you want to skip
over the elements in a list that are equal to 7.
> L := [7,2,5,8,7,9];
L := [7, 2, 5, 8, 7, 9]
> for i in L do
> if i=7 then next end if;
> print(i);
> end do;
5
8
9
An error occurs if Maple evaluates the names break or next in a

context other than within a repetition statement.
> next;
Error, break or next not in loop
The read and save Statements

The file system is an important part of Maple. The user may interact with
the file system either explicitly by using the read and save statements, or
implicitly by executing a command that automatically loads information
from a file. For example, the computation of an integral may load many
commands from the Maple library. The read and save statements read
and save Maple data and programs to and from files. See also Chapter 10.
Saving a Maple Session The save statement allows you to save the
values of a sequence of variables. It takes the general form
save nameseq, filename ;
Here nameseq must be a sequence of names of assigned variables.

Maple saves each variable name and its value in an assignment statement
in the file filename. Maple evaluates each argument, except the last one,
to a name. It evaluates the last argument normally.
Clear Maple by using the restart command and assign three new
values.
4.4 Expressions • 139
> restart:
> r0 := x^3:
> r1 := diff(r0,x):
> r2 := diff(r1,x):
The next statement saves r0, r1 and r2 in the ASCII file my_file:
> save r0, r1, r2, "my_file":
This is now the contents of the file my_file.

r0 := x^3;
r1 := 3*x^2;
r2 := 6*x;
The expression filename must evaluate to a name which specifies the
name of a file. You can read it back into Maple later using the read
command.
The read Statement The read statement
read filename ;
reads a file into the Maple session. The filename must evaluate to the
name of the file. The file must be either a Maple internal format file (a
.m file), or a text file.
If the file is a plain text file, then it must contain a sequence of valid
Maple statements, separated by semicolons or colons. The effect of reading
the file is identical to entering the same sequence of statements interac-
tively. The system displays the result of executing each statement that it
reads in from the file.
4.4 Expressions
Expressions are the fundamental entities in the Maple language. The var-
ious types of expressions include constants, names of variables and un-
knowns, formulæ, Boolean expressions, series, and other data structures.
Technically speaking, procedures are also valid expressions since you may
use them wherever an expression is legal. Chapter 5 describes them sep-
arately.
Table 4.8 Primitive Functions

type(f, t ) tests if f is of type t
nops(f ) returns the number of operands of f
op(i, f ) selects the i th operand of f
subsop(i =g, f ) replaces the i th operand of f with g
Expression Trees: Internal Representation

Consider the following formula.
> f := sin(x) + 2*cos(x)^2*sin(x) + 3;
f := sin(x) + 2 cos(x)2 sin(x) + 3
To represent this formula, Maple builds an expression tree.

+
3 * sin
x
sin ^ 2
x
cos 2
The first node of the expression tree labeled “+” is a sum. This in-
dicates the expression’s type. This expression has three branches corre-
sponding to the three terms in the sum. The nodes of each branch tell
you each term’s type in the sum. And so on down the tree until you get
to the leaves of the tree, which are names and integers in this example.
When programming with expressions, you need a way to determine
what type of expression you have, how many operands or branches an
expression has, and a way of selecting those operands. You also need a
way of building new expressions, for example, by replacing one operand
of an expression with a new value. Table 4.8 lists the primitive functions
for doing this.
> type(f, ‘+‘);
true
> type(f, ‘*‘);
false
> nops(f);
> op(1, f);
sin(x)
> subsop(2=0, f);
sin(x) + 3
By determining the type of an expression, the number of operands it

has, and selecting each operand of the expression, you can systematically
work all the way through an expression.
> t := op(2, f);
t := 2 cos(x)2 sin(x)
> type(t, ‘*‘);
true
> nops(t);
> type(op(1,t), integer);
true
> type(op(2,t), ‘^‘);

true
> type(op(3,t), function);
true
The op command has several other useful forms. The first is
op(i ..j, f )
which returns the sequence
op(i, f ), op(i +1, f ), ..., op(j -1, f ), op(j, f )
of operands of f . Another short-form notation is,
op([i, j, k ], f )
which gives the same result as
op(k, op(j, op(i, f )))
The last object in the list may also be a range
op([i, j, k1 ..k2 ], f )
which returns the sequence
op(k1, op(i, f )), op(k1+1, op(i, f )), ...,

op(k2, op(i, f ))
You may want to see the whole sequence of operands of an expression.

You can do this with
op(f )
which is equivalent to op(1..nops(f ),f ). The special operand op(0,f )

generally returns the type of an expression. An exception occurs when f
is a function, in which case it returns the name of the function.
> op(0, f);
> op(1..3, f);
sin(x), 2 cos(x)2 sin(x), 3
> op(0, op(1,f));
sin
> op(0, op(2,f));
> op(0, op(3,f));
Integer
Evaluation and Simplification Consider this example in detail.

> x := Pi/6:
> sin(x) + 2*cos(x)^2*sin(x) + 3;
17
4
What does Maple do when it executes the second command? Maple

first reads and parses the input line. As it is parsing the input line it
builds an expression tree to represent the value
sin(x) + 2 cos(x)2 sin(x) + 3.
Next it evaluates the expression tree, then simplifies the result. Evalu-
ation means substituting values for variables and invoking any functions.
Here x evaluates to π/6. Hence, with these substitutions the expression
is as follows.
sin(π/6) + 2 cos(π/6)2 sin(π/6) + 3
Invoking the sin and cos functions, Maple obtains a new expression
tree, √
1/2 + 2 × (1/2 3)2 × 1/2 + 3.
Table 4.9 Subtypes of Integers

negint negative integer
posint positive integer
nonnegint non-negative integer
nonposint non-positive integer
even even integer
odd odd integer
prime prime integer
Finally, Maple does the arithmetic in this expression tree to obtain

the fraction 17/4. In the following example, evaluation occurs, but no
simplification is possible.
> x := 1;
x := 1
> sin(x) + 2*cos(x)^2*sin(x) + 3;
sin(1) + 2 cos(1)2 sin(1) + 3
We now present in detail every kind of expression, beginning with

the constants. The presentation states how to input the expression, gives
examples of how and where to use the expression, and the action of the
type, nops, op, and subsop commands on the expression.
The numeric constants in Maple are integers, fractions, floating-
point (decimal) numbers, infinity, and undefined. The complex nu-
meric constants are the complex integers (Gaussian integers), complex
rationals, and complex floating-point numbers. The full set of real and
complex numeric constants is exactly what is recognized by type(...,
complex(extended_numeric)).
The Types and Operands of Integers, Strings, Indexed Names,

and Concatenations
The type of an integer is integer. The type command also understands
the subtypes of integers listed in Table 4.9. The op and nops commands
consider an integer to have only one operand, namely, the integer itself.
> x := 23;
x := 23
> op(0, x);
Integer
> op(x);
23
> type(x, prime);
true
The type of a string is string. A string also has only one operand;
the string itself.
> s := "Is this a string?";
s := “Is this a string?”
> type(s, string);
true
> nops(s);
> op(s);
“Is this a string?”
The type of an indexed name is indexed. The operands of an indexed

name are the indices or subscripts and the zeroth operand is the base
name. The type command also understands the composite type name
which Maple defines as either a string or an indexed name.
> x := A[1][2,3];
x := A12, 3
> type(x, indexed);
true
> nops(x);
> op(x);
2, 3
> op(0,x);
A1
> y:=%;
y := A1
> type(y, indexed);
true
> nops(y), op(0,y), op(y);
1, A, 1
The type of an unevaluated concatenation is “||”. This type has two

operands, the left-hand side expression and the right-hand side expres-
sion.
> c := p || (2*m + 1);
c := p||(2 m + 1)
> type(c, ‘||‘);

true
> op(0, c);
||
> nops(c);
> op(c);
p, 2 m + 1
Fractions and Rational Numbers

A fraction is input as
integer /natural
Maple does arithmetic with fractions and integers exactly. Maple al-
ways immediately simplifies a fraction so that the denominator is positive,
and reduces the fraction to lowest terms by canceling out the greatest com-
mon divisor of the numerator and denominator.
> -30/12;
−5
2
If the denominator is 1 after simplification of a fraction, Maple auto-

matically converts it to an integer. The type of a fraction is fraction.
The type command also understands the composite type name rational,
which is an integer or a fraction, that is, a rational number.
> x := 4/6;
2
x :=
3
> type(x,rational);
true
A fraction has two operands, the numerator and denominator. In ad-

dition to the op command, you may use the commands numer and denom
to extract the numerator and denominator of a fraction, respectively.
> op(1,x), op(2,x);
2, 3
> numer(x), denom(x);
2, 3
Floating-Point (Decimal) Numbers

An unsigned float has one of the following six forms:
natural .natural
natural .
.natural
natural exponent
natural .natural exponent
.natural exponent
where the exponent suffix is the letter “e” or “E” followed by a signed
integer with no spaces in the middle. A floating-point number is an
unsigned_float or a signed float (+unsigned_float or -unsigned_float
indicates a signed float).
> 1.2, -2., +.2;
1.2, −2., .2
> 2e2, 1.2E+2, -.2e-2;
200., 120., −.002
Note that
> 1.e2;
syntax error, missing operator or ‘;‘:

1.e2;
^
is not valid, and that spaces are significant.

> .2e -1 <> .2e-1;
−.8 6= .02
The type of a floating-point number is float. The type command

also understands the composite types numeric, which recognizes integer,
fraction and float, and extended_numeric, which recognizes integer,
fraction, float, infinity, and undefined. For information about the
full suite of numeric types and subtypes, see ?numeric_type.
A floating-point number has two parts, the mantissa (or significand) m
and the exponent e, representing the number m × 10e . The decimal point
is taken to be after the rightmost digit of m. To get access to the parts
of a floating-point number, use the Maple commmands SFloatMantissa
and SFloatExponent.
> x := 231.3;
x := 231.3
> SFloatMantissa(x);
2313
> SFloatExponent(x);
−1
The Float command can also be used to construct floating-point num-

bers in Maple:
Float(m, e )
This constructs the floating-point number m × 10e . Again, the mantissa

(or significand) is m, the exponent is e, and the decimal point is to the
right of m.
The mantissa, m, is a Maple integer, and hence it is subject to the
same restrictions in terms of number of digits as any Maple integer, which
is machine dependent, but is always at least 268 435 448. The exponent,
e, is subject to a smaller restriction, which is again machine dependent
but is always at least 2 147 483 646. The exact values of these limits can
be obtained from the Maple_floats command.
You can also input a floating-point number m×10e by simply forming
the literal expression m * 10ê. However this is less efficient, particularly
for large exponents.
Arithmetic with Floating-Point Numbers For arithmetic operations

and the standard mathematical functions, if one of the operands (or argu-
ments) is a floating-point number or evaluates to a floating-point number,
then floating-point arithmetic takes place automatically. The global name
Digits, which has 10 as its default, determines the number of digits which
Maple uses when calculating with floating-point numbers (the number of
digits in the significand).
> x := 2.3: y := 3.7:
> 1 - x/y;
.3783783784
In general, you may use the evalf command to force the evaluation
of a non-floating-point expression to a floating-point expression where
possible.
> x := ln(2);
x := ln(2)
> evalf(x);
.6931471806
An optional second argument to the evalf command specifies the

precision at which Maple is to do this evaluation.
Table 4.10 Types of Complex Numbers

complex(integer) both a and b are integers, possibly 0
complex(rational) both a and b are rationals
complex(float) both a and b are floating-point constants
complex(numeric) any of the above
> evalf[15](x);
.693147180559945
Complex Numerical Constants √

By default, I denotes the complex unit −1 in Maple. In fact, all of the
following are equivalent.
> sqrt(-1), I, (-1)^(1/2);
I, I, I
A complex number a + bi is input as the sum a + b*I or as

Complex(a, b) in Maple. Maple uses a special representation for complex
numeric constants, such as 1.3 + 4.2*I. Note that in an expression such
as x + y*I, where x and y are symbols, Maple does not assume that x is
the real part and y is the imaginary part. Use the commands Re and Im
to select the real and imaginary parts, respectively.
> x := 2+3*I;
x := 2 + 3 I
> Re(x), Im(x);
2, 3
The type of a complex number is complex(numeric). This means

that the real and imaginary parts are of type numeric, that is, integers,
fractions, or floating-point numbers. Other useful type names are listed
in Table 4.10.
Arithmetic with complex numbers is done automatically.
> x := (1 + I); y := 2.0 - I;
x := 1 + I
y := 2.0 − 1. I
> x+y;
3.0 + 0. I
Maple also knows how to evaluate elementary functions and many

special functions over the complex numbers. It does this automatically if
a and b are numeric constants and one of a or b is a decimal number.
> exp(2+3*I), exp(2+3.0*I);
e(2+3 I) , −7.315110095 + 1.042743656 I
If the arguments are not complex floating-point constants, you can

expand the expression in some cases into the form a + bi, where a and b
are real, using the evalc command.
Here the result is not in the form a+bi since a is not of type numeric.
> 1/(a - I);
1
a−I
> evalc(%);
a I
+ 2
a2 +1 a +1
If you prefer to use another letter, say j, for the imaginary unit, use
the interface command as follows.
> interface(imaginaryunit = j);
> solve( {z^2=-1}, {z} );
{z = j}, {z = −j}
The following command reinstates I as the imaginary unit.

> interface(imaginaryunit = I);

> solve( {z^2=-1}, {z} );
{z = I}, {z = −I}
Labels
A label in Maple has the form
%natural
that is, the unary operator % followed by a natural integer. The per-
centage sign takes on double duty, as a label and as the ditto operator,
which represents the result of the last one, two, or three commands.
A label is only valid after Maple’s pretty-printer introduces it. The
purpose is to allow the naming (labeling) of common subexpressions,
which serves to decrease the size of the printed output, making it more
comprehensible. After the pretty-printer introduces it, you may use a label
just like an assigned name in Maple.
> solve( {x^3-y^3=2, x^2+y^2=1}, {x, y} );
1
{y = %1, x = − %1 (−4 %13 − 3 − %12 + 6 %1 + 2 %14 )}
3
%1 := RootOf(3 _Z 2 + 3 − 3 _Z 4 + 2 _Z 6 + 4 _Z 3 )
After you obtain the above printout, the label %1 is an assigned name
and its value is the RootOf expression shown.
> %1;
RootOf(3 _Z 2 + 3 − 3 _Z 4 + 2 _Z 6 + 4 _Z 3 )
Two options are available for adjusting this facility. The option
interface(labelwidth=n )
specifies that Maple should not display expressions less than n characters
wide (approximately) as labels. The default is 20 characters. You may
turn off this facility entirely using
> interface(labelling=false);
Sequences
A sequence is an expression of the form
expression_1, expression_2, ..., expression_n
The comma operator joins expressions into a sequence. It has the

lowest precedence of all operators except assignment. A key property of
sequences is that if any of expression_i themselves are sequences, this
flattens out the result into a single unnested sequence.
> a := A, B, C;
a := A, B, C
> a,b,a;
A, B, C, b, A, B, C
A zero-length sequence is syntactically valid. It arises, for example, in

the context of forming an empty list, an empty set, a function call with
no parameters, or an indexed name with no subscripts. Maple initially
assigns the special name NULL to the zero-length sequence, and you may
use it whenever necessary.
You cannot use the type command to test the type of a sequence, nor
can you use the nops or op commands to count the number of operands
in a sequence or select them. Their use is not possible because a sequence
becomes the arguments to these commands.
> s := x,y,z;
s := x, y, z
The command
> nops(s);
nops
is the same as the command

> nops(x,y,z);

nops
Here the arguments to the nops command are x, y, z, which constitute

too many arguments. If you desire to count the number of operands in a
sequence or select an operand from a sequence, you should first put the
sequence in a list as follows
> nops([s]);
Alternatively, you can use the selection operation discussed in sec-

tion 4.4 to select the operands of a sequence.
Please note that many Maple commands return sequences. You may
wish to put sequences into a list or set data structure. For example, when
the arguments to the solve command are not sets, it returns a sequence
of values if it finds multiple solutions.
> s := solve(x^4-2*x^3-x^2+4*x-2, x);
√ √
s := 1, 1, 2, − 2
The elements of the above sequence are values, not equations, because
you did not use sets in the call to solve. Putting the solutions in a set
removes duplicates.
> s := {s};
√ √
s := {1, 2, − 2}
The seq Command The seq command creates sequences, a key tool for
programming. Section 4.5 describes it in detail. The syntax takes either
of the following general forms.
seq(f, i = a .. b )
seq(f, i = X )
Here f, a, b, and X are expressions and i is a name. In the first form,

the expressions a and b must evaluate to two numerical constants or two
single character strings. The result is the sequence produced by evaluating
f after successively assigning the index i the values a, a+1, . . . , b, (or up

to the last value not exceeding b). If the value a is greater than b then
the result is the NULL sequence.
> seq(i^2,i=1..4);
1, 4, 9, 16
> seq(i,i="d".."g");
“d”, “e”, “f”, “g”
In the second form, seq(f, i =X ), the result is the sequence produced

by evaluating f after successively assigning the index i the operands of the
expression X (or the individual characters, if X is a string). Section 4.4
states the operands of a general expression.
> a := x^3+3*x^2+3*x+1;
a := x3 + 3 x2 + 3 x + 1
> seq(i,i=a);
x3 , 3 x2 , 3 x, 1
> seq(degree(i,x), i=a);
3, 2, 1, 0
> seq(i,i="maple");
“m”, “a”, “p”, “l”, “e”
The $ Operator The sequence operator, $, also forms sequences. The

primary purpose of $ is to represent a symbolic sequence such as x$n as
in the following examples.
> diff(ln(x), x$n);
diff(ln(x), x $ n)
> seq( diff(ln(x), x$n), n=1..5);
1 1 2 6 24
, − 2, 3, − 4, 5
x x x x x
The general syntax of the $ operator is
f $ i = a .. b
f $ n
$ a .. b
where f, a, b, and n are expressions and i must evaluate to a name. In

general, this operator is less efficient than seq and hence the seq command
is preferred for programming.
In the first form, Maple creates a sequence by substituting the values
a, a+1, . . . , b for i in f.
The second form f $n is a shorthand notation for
f $ dummy = 1 .. n
where dummy is a dummy index variable. If the value of n is an integer,

the result of the second form is the sequence consisting of the value of f
repeated n times.
> x$3;
x, x, x
The third form $a..b is a shorthand notation for
dummy $ dummy = a .. b
If the values of a and b are numerical constants, this form is short for
creating a numerical sequence a, a+1, . . . , b (or up to the last value not
exceeding b).
> $0..4;
0, 1, 2, 3, 4
The $ command differs from the seq command in that a and b do

not need to evaluate to numbers. However, when a and b do evaluate to
specific values, seq is more efficient than $. See section 4.5.
Sets and Lists

A set is an expression of the form
{ sequence }
and a list is an expression of the form
[ sequence ]
Note that a sequence may be empty, so {} represents the empty set

and [] the empty list. A set is an unordered sequence of unique ex-
pressions. Maple removes duplicates and reorders the terms in a manner
convenient for internal storage. A list is an ordered sequence of expres-
sions with the order of the expressions specified by the user. Maple retains
duplicate entries in a list.
> {y[1],x,x[1],y[1]};
{x, y1 , x1 }
> [y[1],x,x[1],y[1]];
[y1 , x, x1 , y1 ]
A set is an expression of type set. Similarly, a list is an expression

of type list. The operands in a list or set are the elements in the set or
list. Select the elements of a list or set using either the op command or a
subscript.
> t := [1, x, y, x-y];
t := [1, x, y, x − y]
> op(2,t);
> t[2];
x
Maple’s ordering for sets is the order in which it stores the expressions
in memory. The user should not make assumptions about this ordering.
For example, in a different Maple session, the set above might appear in
the ordering {y[1], x, x[1]}. You can sort elements of a list by using
the sort command.
Selection Operation The selection operation, [], selects components

from an aggregate object. The aggregate objects include tables, arrays,
sequences, lists, and sets. The syntax for the selection operation is
name [ sequence ]
If name evaluates to a table or array, Maple returns the table (array)

entry.
> A := array([w,x,y,z]);
A := [w, x, y, z]
> A[2];
If name evaluates to a list, set, or sequence, and sequence evaluates

to an integer, a range, or NULL, Maple performs a selection operation.
If sequence evaluates to an integer i, then Maple returns the ith
operand of the set, list, or sequence. If sequence evaluates to a range,
then Maple returns a set, list, or sequence containing the operands of the
aggregate object as the range specifies. If sequence evaluates to NULL, then
Maple returns a sequence containing all of the operands of the aggregate
object.
> s := x,y,z:
> L := [s,s];
L := [x, y, z, x, y, z]
> S := {s,s};
S := {z, x, y}
> S[2];
> L[2..3];
[y, z]
> S[];
z, x, y
Negative integers count operands from the right.

> L := [t,u,v,w,x,y,z];
L := [t, u, v, w, x, y, z]
> L[-3];
> L[-3..-2];
[x, y]
You can also use select, remove, and selectremove to select ele-
ments from a list or set. See section 4.5.
Functions
A function call in Maple takes the form
f ( sequence )
Often f will be a name, that is, the name of the function.

> sin(x);
sin(x)
> min(2,3,1);
> g();
g()
> a[1](x);
a1 (x)
Maple executes a function call as follows. First, it evaluates f (typically

yielding a procedure). Next, Maple evaluates the operands of sequence
(the arguments) from left to right. (If any of the arguments evaluate to
a sequence, Maple flattens the sequence of evaluated arguments into one
sequence.) If f evaluated to a procedure, Maple invokes it on the argument
sequence. Chapter 5 discusses this in detail.
> x := 1:
> f(x);
f(1)
> s := 2,3;
s := 2, 3
> f(s,x);
f(2, 3, 1)
> f := g;
f := g
> f(s,x);
g(2, 3, 1)
> g := (a,b,c) -> a+b+c;
g := (a, b, c) → a + b + c
> f(s,x);
A function object’s type is function. The operands are the argu-

ments. The zeroth operand is the name of the function.
> m := min(x,y,x,z);
m := min(1, y, z)
> op(0,m);
min
> op(m);
1, y, z
> type(m,function);
true
> f := n!;
f := n!
> type(f, function);
true
> op(0, f);
factorial
> op(f);
In general, the function name f may be one of the following.
• name
• procedure definition
• integer
• float
• parenthesized algebraic expression
• function
Allowing f to be a procedure definition allows you to write, for exam-

ple
> proc(t) t*(1-t) end proc (t^2);
t2 (1 − t2 )
instead of
> h := proc(t) t*(1-t) end proc;
h := proc(t) t ∗ (1 − t) end proc
> h(t^2);
t2 (1 − t2 )
If f is an integer or a float, Maple treats f as a constant operator.

That is f(x) returns f.
> 2(x);
The following rules define the meaning of a parenthesized algebraic

expression.
> (f + g)(x), (f - g)(x), (-f)(x), (f@g)(x);
f(x) + g(x), f(x) − g(x), −f(x), f(g(x))
@ denotes functional composition; that is, f@g denotes f ◦ g. These

rules together with the previous rule mean that
> (f@g + f^2*g + 1)(x);
f(g(x)) + f(x)2 g(x) + 1
Notice that @@ denotes the corresponding exponentiation. That is,

f@@n denotes f (n) which means f composed with itself n times.
> (f@@3)(x);
(f (3) )(x)
> expand(%);
f(f(f(x)))
Finally, f may be a function, as in

> cos(0);
> f(g)(0);
f(g)(0)
> D(cos)(0);
For more information on how to define a function, see chapter 5.

Table 4.11 The Arithmetic Operators

+ addition
- subtraction
* multiplication
. non-commutative multiplication
/ division
^ exponentiation
The Arithmetic Operators

Table 4.11 contains Maple’s six arithmetic operators. You may use all
these items as binary operators. You may also use the operators + and -
as prefix operators representing unary plus and unary minus.
You can find the types and operands of the arithmetic operations
listed below.
• The type of a sum or difference is +.
• The type of a product or quotient is * and the type of a power is ^.
• The operands of the sum x − y are the terms x and −y.
• The operands of the product xy 2 /z are factors x, y 2 , and z −1 .
• The operands of the power xa are the base x and the exponent a.
> whattype(x-y);
> whattype(x^y);
Arithmetic Maple always computes the five arithmetic operations x + y,

x − y, x × y, x/y, and xn , where n is an integer, if x and y are numbers. If
the operands are floating-point numbers, Maple performs the arithmetic
computation in the floating-point environment.
> 2 + 3, 6/4, 1.2/7, (2 + I)/(2 - 2*I);
3 1 3
5, , .1714285714, + I
2 4 4
> 3^(1.2), I^(1.0 - I);
3.737192819, 0. + 4.810477381 I
The only other simplification done for numerical constants is reduc-

tion of fractional powers of integers and fractions. For integers n, m and
fraction b,
(n/m)b → (nb )/(mb ).
For integers n, q, r, d and fraction b = q + r/d with 0 < r < d,
nb = nq+r/d → nq × nr/d .
> 2^(3/2), (-2)^(7/3);

√
2 2, 4 (−2)(1/3)
Automatic Simplifications Maple automatically does these simplifica-

tions
> x - x, x + x, x + 0, x*x, x/x, x*1, x^0, x^1;
0, 2 x, x, x2 , 1, x, 1, x
for a symbol x, or an arbitrary expression. But these simplifications are

not valid for all x. Some exceptions which Maple catches are
> infinity - infinity;
undefined
> infinity/infinity;
undefined
> 0/0;
Error, numeric exception: division by zero
In the following, a, b, c denote numerical constants and x, y, z denote

general symbolic expressions. Maple understands that addition and multi-
plication are associative and commutative, and so simplifies the following.
ax + bx → (a + b)x
xa × xb → xa+b
a(x + y) → ax + ay
The first two simplifications mean that Maple adds like terms in poly-
nomials automatically. The third means that Maple distributes numerical
constants (integers, fractions, and floats) over sums, but does not do the
same for non-numerical constants.
> 2*x + 3*x, x*y*x^2, 2*(x + y), z*(x + y);
5 x, x3 y, 2 x + 2 y, z (x + y)
The most difficult and controversial simplifications have to do with

simplying powers xy for non-integer exponents y.
Simplification of Repeated Exponentiation In general, Maple does not

do the simplification (xy )z → x(yz) automatically because this procedure
does not always provide an accurate answer. For example, √ letting y = 2
and z = 1/2, the first simplification would imply that x2 = x, which
is not necessarily true. Maple only does the first transformation above if
it is provably correct for all complex x with the possible exception of a
finite number of values, such as, 0 and ∞. Maple does (xa )b → xab if b is
an integer, −1 < a ≤ 1, or x is a positive real constant.
> (x^(3/5))^(1/2), (x^(5/3))^(1/2);
q
x(3/10) , x(5/3)
> (2^(5/3))^(1/2), (x^(-1))^(1/2);

r
1
2(5/6) ,
x
Maple does not simplify ab cb → (ac)b automatically, even if the answer

is correct.
> 2^(1/2)+3^(1/2)+2^(1/2)*3^(1/2);
√ √ √ √
2+ 3+ 2 3
√ √ √
The reason is that combining 2 3 to 6 would introduce a third
unique square root. Calculating with roots is, in general, difficult and
expensive, so Maple is careful not to create new roots. You may use the
combine command to combine roots if you desire.
Non-Commutative Multiplication
The non-commutative multiplication operator &* acts as an inert operator
(for example, the neutral operators described in section 4.4), but the
parser understands its binding strength to be equivalent to the binding
strength of * and /.
The evalm command in the Maple Library interprets &* as the table-
based matrix multiplication operator.1 The evalm command also under-
stands the form &*() as a generic matrix identity.
> with(LinearAlgebra):
> A := matrix(2,2,[a,b,c,d]);
´ µ
a b
A :=
c d
> evalm( A &* &*() );

´ µ
a b
c d
> B := matrix(2,2,[e,f,g,h]);
´ µ
e f
B :=
g h
> evalm( A &* B - B &* A );

´ µ
bg − cf af + bh − eb − f d
ce + dg − ga − hc cf − bg
1
The Maple library interprets . (dot) as the rtable-based Matrix multiplication
operator.
The Composition Operators

The composition operators are @ and @@. The @ operator represents func-
tion composition, that is, f@g in Maple denotes f ◦ g.
> (f@g)(x);
f(g(x))
> (sin@cos)(Pi/2);
The @@ operator is the corresponding exponentiation operator repre-

senting repeated functional composition, that is, f (n) is denoted f@@n in
Maple.
> (f@@2)(x);
(f (2) )(x)
> expand(%);
f(f(x))
> (D@@n)(f);
(D(n) )(f )
Usually f n (x) denotes composition; for example, Dn denotes the dif-

ferential operator composed n times. Also sin−1 (x) denotes the inverse
of the sine function, that is, composition to the power −1. But, some-
times mathematicians use f n (x) to denote ordinary powering, for exam-
ple, sin2 (x) is the square of sine of x. Maple always uses f n (x) to denote
repeated composition, and f (x)n to denote powering.
> sin(x)^2, (sin@@2)(x), sin(x)^(-1), (sin@@(-1))(x);
1
sin(x)2 , (sin(2) )(x), , arcsin(x)
sin(x)
The Ditto Operators

The value of the nullary operator, %, is the most recent expression. The
first and second expressions preceding the most recent are the values of
the nullary operators %% and %%%, respectively. The most common use of
these operators is in an interactive Maple session where they refer to the
previously computed results. The sequence of expressions defining these
three nullary operators is the last three non-NULL values generated in the
Maple session.
The Factorial Operator

Maple uses the unary operator ! as a postfix operator which denotes the
factorial function of its operand n. The input n ! is shorthand for the
functional form factorial(n ).
> 0!, 5!;
1, 120
For negative integers, the factorial returns an error.

> (-2)!;
For floating-point n, generalized factorial function values n ! are cal-

culated by using GAMMA(n +1).
> 2.5!;
3.323350970
The type of an unevaluated factorial is !. Note that in Maple, n!! does

not denote the double factorial function. It denotes repeated factorial,
n!! = (n!)!.
> 3!!;
720
The mod Operator

The mod operator evaluates an expression modulo m, for a non-zero integer
m. That is, Maple writes a mod m as a mod m. Maple uses one of two
representations for an integer modulo m.
• In the positive representation, integer mod m is an integer between

zero and m -1, inclusive. The following assignment selects the positive
representation explicitly.
> ‘mod‘ := modp;
This is the default representation.
• In the symmetric representation, integer mod m is an integer be-

tween -floor((abs(m )-1)/2) and floor(abs(m )/2). The following
assignment selects the symmetric representation.
> ‘mod‘ := mods;
Notice that you need back quotes around mod since it is a reserved word.
You may invoke the commands modp and mods directly if you desire.
For example
> modp(9,5), mods(9,5);
4, −1
The mod operator understands the inert operator &^ for powering.
That is, i&^j mod m calculates ij mod m. Instead of first computing the
integer ij , which may be too large to compute, and then reducing modulo
m, Maple computes the power using binary powering with remainder.
> 2^(2^100) mod 5;
Error, numeric exception: overflow
> 2 &^ (2^100) mod 5;
The first operand of the mod operator may be a general expression.

Maple evaluates the expression over the ring of integers modulo m. For
polynomials, this means that it reduces rational coefficients modulo m.
The mod operator knows many functions for polynomial and matrix arith-
metic over finite rings and fields. For example, Factor for polynomial
factorization, and Nullspace for matrix null-space.
> 1/2 mod 5;
> 9*x^2 + x/2 + 13 mod 5;
4 x2 + 3 x + 3
> Factor(4*x^2 + 3*x + 3) mod 5;
4 (x + 4) (x + 3)
Do not confuse, for example, the commands factor and Factor. The
former evaluates immediately; the latter is an inert command which Maple
does not evaluate until you make the call to mod.
The mod command also knows how to compute over a Galois field
GF (pk ), that is, the finite field with pk elements. See the ?mod online
documentation for a list of the commands that mod knows, and for further
examples.
The Neutral Operators

Maple possesses a user-defined or neutral operators facility. Form a
neutral operator symbol by using the ampersand character “&” followed by
one or more characters. The two varieties of &-names depend on whether
the sequence of characters is alphanumeric or non-alphanumeric:
• Any Maple name not requiring back quotes, preceded by the & char-
acter; for example, &wedge.
• The & character followed by one or more non-alphanumeric characters;

for example, &+ or &++.
The following characters cannot appear in an &-name after the initial &:
& | ( ) [ ] { } ; : ’ ‘ # \ %
as well as newline and blank characters.
Maple singles out the particular neutral operator symbol &* as a spe-
cial token representing the non-commutative multiplication operator. The
special property of &* is that the parser understands its binding strength
to be equivalent to Maple’s other multiplication operators. All other
neutral operators have binding strength greater than the standard al-
gebraic operators. See ?precedence for the order of precedence of all
programming-language operators. See section 4.4 which describes where
to use &* in Maple.
You can use neutral operators as unary prefix operators, infix binary
operators, or function calls. In any of these cases, they generate function
calls with the name of the function being that of the neutral operator.
(In the usual pretty-printing mode, these particular function calls are
printed in binary operator format when exactly two operands exist and in
unary operator format when exactly one operand exists, but the internal
representation is an unevaluated function.) For example,
> a &~ b &~ c;
(a &~ b) &~ c
> op(%);
a &~ b, c
> op(0,%%);
&~
Maple imposes no semantics on the neutral operators. The user may

define the operator to have a meaning by assigning the name to a Maple
procedure. You can define manipulations on expressions containing such
operators via Maple’s interface to user-defined procedures for various
standard library functions, including simplify, diff, combine, series,
evalf, and many others. See section 3.4.
Relations and Logical Operators

You can form new types of expressions from ordinary algebraic expressions
by using the relational operators <, >, <=, >=, =, and <>. The semantics
of these operators is dependent on whether they occur in an algebraic
context or in a Boolean context.
In an algebraic context, the relational operators are simply place hold-
ers for forming equations or inequalities. Maple fully supports addition
of equations or inequalities and multiplication of an equation or inequal-

ity by an algebraic expression. In the case of adding or subtracting two
equations, for example, Maple applies the addition or subtraction to each
side of the equations, thus yielding a new equation. In the case of multi-
plying an equation by an expression, Maple distributes the multiplication
to each side of the equation. You may perform similar operations with
inequalities.
> e := x + 3*y = z;
e := x + 3 y = z
> 2*e;
2x + 6y = 2z
The type of an equation is = or equation. An equation has two

operands, the left-hand side, and the right-hand side. You can use the
commands lhs and rhs to select the operands of an equation instead of
op.
> op(0,e);
> lhs(e);
x + 3y
The type command also understands the types <>, <, and <=. Maple
automatically converts inequalities involving > or >= to < and <=, respec-
tively. All the relational types have two operands.
> e := a > b;
e := b < a
> op(e);
b, a
In a Boolean context, Maple evaluates a relation to the value true

or the value false. A Boolean context includes the condition in an if
statement and the condition in the while clause of a loop. You may also
use the evalb command to evaluate a relation in a Boolean context.
In the case of the operators <, <=, >, and >=, the difference of the
operands must evaluate to a numeric constant and Maple compares this
constant with zero.
> if 2<3 then "less" else "not less" end if;
“less”
In the case of the relations = and <>, the operands may be arbitrary
expressions (algebraic or non-algebraic). This equality test for expressions
deals only with syntactic equality of the Maple representations of the
expressions, which is not the same as mathematical equivalence.
> evalb( x + y = y + x );
true
> evalb( x^2 - y^2 = (x - y)*(x + y) );
false
In the latter example, applying the expand command results in an

equation which evaluates to true.
> evalb( x^2 - y^2 = expand( (x - y)*(x + y) ) );
true
You may use the is command instead of evalb to evaluate relations

in a Boolean context. The is command tries much harder than evalb to
determine whether relations are true.
> is( x^2 - y^2 = (x - y)*(x + y) );
true
> is( 3<Pi );

true
The Logical Operators Generally, you can form an expression by using

the logical operators and, or, and not, where the first two are binary op-
erators and the third is a unary (prefix) operator. An expression contain-
ing one or more logical operators is automatically evaluated in a Boolean
context.
> 2>3 or not 5>1;
false
The precedence of the logical operators and, or, and not is analo-
gous to multiplication, addition, and exponentiation, respectively. Here
no parentheses are necessary.
> (a and b) or ((not c) and d);
a and b or not c and d
The type names for the logical operators and, or, and not are and,
or, and not, respectively. The first two have two operands, the latter one
operand.
> b := x and y or z;
b := x and y or z
> whattype(b);
or
> op(b);
x and y, z
Among operators of the same precedence, the evaluation of Boolean

expressions involving the logical operators and and or proceeds from left
to right and terminates as soon as Maple can determine the truth of the
whole expression. Consider the evaluation of the following.
a and b and c
If the result of evaluating a is false, you know that the result of the
whole Boolean expression will be false, regardless of what b and c evaluate
to. These evaluation rules are commonly known as McCarthy evaluation
rules. They are quite crucial for programming. Consider the following
statement
if x <> 0 and f(x)/x > 1 then ... end if;
If Maple always evaluated both operands of the and clause, then when
x is 0, evaluation would result in a division by zero error. The advantage
of the above code is that Maple will attempt to check the second condition
only when x 6= 0.
Boolean Expressions In general, a Boolean context requires a Boolean

expression. Use the Boolean constants true, false, and FAIL, the rela-
tional operators and the logical operators for forming Boolean expres-
sions. The type command understands the name boolean to include all
of these.
The evaluation of Boolean expressions in Maple uses the following
three-valued logic. In addition to the special names true and false,
Maple also understands the special name FAIL. Maple sometimes uses
the value FAIL as the value that a procedure returns when it is unable to
completely solve a problem. In other words, you can view it as the value
don’t know.
> is(sin(1),positive);
true
> is(a-1,positive);
FAIL
In the context of the Boolean clause in an if statement or a while

statement, Maple determines the branching of the program by treating
the value FAIL the same as the value false. Without three valued logic,
whenever you use the is command you would have to test for FAIL sep-
arately. You would write
Table 4.12 Truth Tables

and false true FAIL or false true FAIL
false false false false false false true FAIL
true false true FAIL true true true true
FAIL false FAIL FAIL FAIL FAIL true FAIL
not false true FAIL

true false FAIL
if is(a - 1, positive) = true then ...
The three valued logic allows you to write
if is(a - 1, positive) then ...
The evaluation of a Boolean expression yields true, false, or FAIL

according to table 4.12.
Note that three-valued logic leads to asymmetry in the use of if state-
ments and while statements. For example, the following two statements
are not equivalent.
if condition then statseq_1 else statseq_2 end if;

if not condition then statseq_2 else statseq_1 end if;
Depending on the desired action in the case where condition has the value
FAIL, either the first or the second of these two if statements may be
correct for a particular context.
Tables
The table data type in Maple is a special data type for representing data
in tables. Create a table either explicitly via the table command or im-
plicitly by assignment to an indexed name. For example, the statements
> a := table([(Cu,1) = 64]);
a := table([(Cu, 1) = 64])
> a[Cu,1] := 64;

aCu, 1 := 64
have the same effect. They both create a table object with one compo-
nent. The purpose of a table is to allow fast access to data with
> a[Cu,1];
64
The type of a table object is table. The first operand is the indexing
function. The second operand is a list of the components. Note that tables
(and arrays, which are a special case of a table) have special evaluation
rules; in order to access the table (or array) object, you must first apply
the eval command.
> op(0,eval(a));
table
Table a has no indexing function and only one entry.

> op(1,eval(a));
> op(2,eval(a));
[(Cu, 1) = 64]
The array data type in Maple is a specialization of the table data

type. An array is a table with specified dimensions, with each dimension
an integer range. Create an array via the array command call.
> A := array(symmetric, 1..2, 1..2, [(1,1) = 3]);
´ µ
3 A1, 2
A :=
A1, 2 A2, 2
> A[1,2] := 4;
A1, 2 := 4
> print(A);
´ µ
3 4
4 A2, 2
The ranges 1..2,1..2 specify two dimensions and bounds for the
integers. You may include entries in the array command or insert them
explicitly as shown. You may leave entries unassigned. In this example,
the (2, 2) entry is unassigned.
> op(0,eval(A));
array
As for tables, the first operand is the indexing function (if any).
> op(1,eval(A));
symmetric
The second operand is the sequence of ranges.

> op(2,eval(A));
1..2, 1..2
The third operand is a list of entries.

> op(3, eval(A));
[(1, 1) = 3, (1, 2) = 4]
The example above displays only two entries in the array A since
Maple knows the (2, 1) entry implicitly through the indexing function.
Series
The series data type in Maple represents an expression as a truncated
power series with respect to a specified indeterminate, expanded about a
particular point. Although you cannot input a series directly into Maple
as an expression, you can create a series data type with the taylor or
series commands which have the following syntax
taylor( f, x =a, n )
taylor( f, x )
series( f, x =a, n )
series( f, x )
If you do not specify the expansion point, it is by default x = 0. If you

do not specify the order n, it is the value of the global variable Order,
which by default is 6.
> s := series( exp(x), x=0, 4 );
1 2 1 3
s := 1 + x + x + x + O(x4 )
2 6
The type name for the series data type is series.

> type(s, series);
true
The zeroth operand is the expression x − a where x denotes the spec-

ified indeterminate and a denotes the particular point of expansion.
> op(0, s);
The odd (first, third, . . . ) operands are the coefficients of the series
and the even operands are the corresponding integer exponents.
> op(s);
1 1
1, 0, 1, 1, , 2, , 3, O(1), 4
2 6
The coefficients may be general expressions but Maple restricts the

exponents to word-size integers on the host computer, with a typical
limit of nine or ten digits, ordered from least to greatest. Usually, the
final pair of operands in the series data type are the special order symbol
O(1) and the integer n which indicates the order of truncation.
The print routine displays the final pair of operands by using the
notation O(xn ) rather than more directly as O(1)xn , where x is op(0,s).
If Maple knows that the series is exact then it will not contain an order
term. An example of this occurs when you apply the series command
to a polynomial whose degree is less than the truncation degree for the
series. A very special case is the zero series, which Maple immediately
simplifies to the integer zero.
The series data structure represents generalized power series, which
include Laurent series with finite principal parts. More generally, Maple
allows the series coefficients to depend on x provided their growth is less
than polynomial in x. O(1) represents such a coefficient, rather than an
arbitrary constant. An example of a non-standard generalized power series
is
> series( x^x, x=0, 3 );
1
1 + ln(x) x + ln(x)2 x2 + O(x3 )
2
Maple can compute more general series expansions than the series
data type supports. The Puisseux series is such an example. In these
cases, the series command does not return a series data type, it returns
a general algebraic expression.
> s := series( sqrt(sin(x)), x );
√ 1 (5/2) 1
s := x− x + x(9/2) + O(x(13/2) )
12 1440
> type(s, series);
false
> type(s, ‘+‘);
true
Ranges
You often need to specify a range of numbers. For example, when you
want to integrate a function over a range. In Maple, use the ellipsis oper-
ation to form ranges.
expression_1 .. expression_2
Specify the operator “..” using two consecutive periods. The ellipsis
operator simply acts as a place holder in the same manner as using the
relational operators in an algebraic context, primarily as a notational
tool. A range has type “..” or range. A range has two operands, the
left-limit and the right-limit, which you can access with the lhs and rhs
commands.
> r:=3..7;
r := 3..7
> op(0,r);
..
> lhs(r);
A typical application of ranges occurs in Maple’s int, sum, and

product commands. Interpret the operands of the ellipsis to mean the
lower and upper limits of integration, summation, or products, respec-
tively.
> int( f(x), x=a..b );
Z b
f(x) dx
a
You can use the range construct, with Maple’s built-in command op,
to extract a sequence of operands from an expression. The notation
op(a..b, c)
is equivalent to
seq(op(i,c),i=a..b)
For example,
> a := [ u, v, w, x, y, z ];
a := [u, v, w, x, y, z]
> op(2..5,a);
v, w, x, y
You can also use the range construct in combination with the con-
catenation operator to form a sequence as follows.
> x || (1..5);
x1 , x2 , x3 , x4 , x5
See section 4.3.
Unevaluated Expressions
Maple normally evaluates all expressions, but sometimes you need to tell
Maple to delay evaluating an expression.
An expression enclosed in a pair of single quotes
’expression ’
is called an unevaluated expression. For example, the statements

> a := 1; x := a + b;
a := 1
x := 1 + b
assign the value 1 + b to the name x, while the statements

> a := 1; x := ’a’ + b;
a := 1
x := a + b
assign the value a + b to the name x if b has no value.

The effect of evaluating a quoted expression is to strip off one level of
quotes, so in some cases nested levels of quotes are very useful. Note the
distinction between evaluation and simplification in the statement
> x := ’2 + 3’;
x := 5
which assigns the value 5 to the name x even though this expression con-
tains quotes. The evaluator simply strips off the quotes, but the simplifier
transforms the expression 2 + 3 into the constant 5.
The result of evaluating an expression with two levels of quotes is an
expression of type uneval. This expression has only one operand, namely
the expression inside the outermost pair of quotes.
> op(’’x - 2’’);
x−2
> whattype(’’x - 2’’);
uneval
A special case of unevaluation arises when a name, which Maple may

have assigned a value, needs unassigning so that in the future the name
simply stands for itself. You can accomplish this by assigning the quoted
name to itself.
> x := ’x’;
x := x
Now x stands for itself as if Maple had never assigned it a value.

Another special case of unevaluation arises in the function call
’f’(sequence )
Suppose the arguments evaluate to the sequence a. Since the result of

evaluating 0 f 0 is not a procedure, Maple returns the unevaluated function
call f (a).
> ’’sin’’(Pi);
’sin’(π)
> %;
sin(π)
> %;
You will find this facility useful when writing procedures which im-
plement simplification rules. See section 3.4.
Constants
Maple has a general concept of symbolic constants , and initially assigns
the global variable constants the following expression sequence of names
> constants;
false, γ, ∞, true, Catalan, FAIL, π
implying that Maple understands these particular names to be of type

constant. The user may define additional names (specifically, they must
be the simplest type of names called strings —see section 4.3) to be con-
stants by redefining the value of this global variable.
> type(g,constant);
false
> constants := constants, g;
constants := false, γ, ∞, true, Catalan, FAIL, π, g

> type(g,constant);
true
Generally, a Maple expression is of type constant if it is of type

complex(numeric), or one of the initially-known constants, or an uneval-
uated function with all arguments of type constant, or a sum, product,
or power with all operands of type constant. For example, the following
expressions are of type constant: 2, sin(1), f(2,3), exp(gamma), 4+Pi,
3+I, 2*gamma/Pi^(1/2)
Structured Types
Sometimes a simple type check does not give enough information. For
example, the command
> type( x^2, ‘^‘ );
true
tells you that x^2 is an exponentiation but it does not tell you whether
or not the exponent is, say, an integer. In such cases, you need structured
types .
> type( x^2, nameînteger );
true
Since x is a name and 2 is an integer, the command returns true.

The square root of x does not have this type.
> type( x^(1/2), nameînteger );
false
The expression (x+1)^2 does not have type nameînteger, because

x+1 is not a name.
> type( (x+1)^2, nameînteger );
false
The type anything matches any expression.

> type( (x+1)^2, anythingînteger );
true
An expression matches a set of types if the expression matches one of

the types in the set.
> type( 1, {integer, name} );
true
> type( x, {integer, name} );
true
The type set(type ) matches a set of elements of type type.

> type( {1,2,3,4}, set(integer) );
true
> type( {x,2,3,y}, set( {integer, name} ) );
true
Similarly, the type list(type ) matches a list of elements of type

type.
> type( [ 2..3, 5..7 ], list(range) );
true
Note that e2 is not of type anything^2.

> exp(2);
e2
> type( %, anything^2 );
false
The reason is that e2 is simply the pretty-printed version of exp(2).

> type( exp(2), ’exp’(integer) );
true
You should use single quotes (’) around Maple commands in type
expressions to delay evaluation.
> type( int(f(x), x), int(anything, anything) );
Error, testing against an invalid type
Here Maple evaluated int(anything, anything) and got

> int(anything, anything);
1
anything 2
2
which is not a valid type. If you put single quotes around the int com-
mand, the type checking works as intended.
> type( int(f(x), x), ’int’(anything, anything) );
true
The type specfunc(type, f ) matches the function f with zero or

more arguments of type type.
> type( exp(x), specfunc(name, exp) );
true
> type( f(), specfunc(name, f) );
true
The type function(type ) matches any function with zero or more

arguments of type type.
> type( f(1,2,3), function(integer) );
true
> type( f(1,x,Pi), function( {integer, name} ) );
true
You can also test the number (and types) of arguments. The type
anyfunc(t1, ..., tn ) matches any function with n arguments of the
listed types.
> type( f(1,x), anyfunc(integer, name) );
true
> type( f(x,1), anyfunc(integer, name) );
false
Another useful variation is to use the And, Or, and Not type construc-
tors to create Boolean combinations of types.
> type(Pi, ’And( constant, numeric)’);
false
> type(Pi, ’And( constant, Not(numeric))’);
true
See ?type,structured for more information on structured types or

?type,definition on how to define your own types.
4.5 Useful Looping Constructs

Section 4.3 describes the for loop and while loop. Many common kinds
of loops appear so often that Maple provides special purpose commands
for them. These commands help to make writing programs simpler and
more efficient. They are the “bread and butter” commands in the Maple
language. You can group the eight loop-like commands in Maple into three
categories
4.5 Useful Looping Constructs • 191
1. map, select, remove, selectremove
2. zip
3. seq, add, mul
The map, select, remove, and selectremove Commands

The map command applies a function to every element of an aggregate
object. The simplest form of the map command is
map( f, x )
where f is a function and x is an expression. The map command re-

places each operand x _i of the expression x with f (x _i).2
> map( f, [a,b,c] );
[f(a), f(b), f(c)]
For example, if you have a list of integers, create a list of their absolute
values and of their squares by using the map command.
> L := [ -1, 2, -3, -4, 5 ];
L := [−1, 2, −3, −4, 5]
> map(abs,L);
[1, 2, 3, 4, 5]
> map(x->x^2,L);
[1, 4, 9, 16, 25]
The general syntax of the map command is
2
Exception: for an rtable, table or array, Maple applies the function to the entries
of the table or array, and not to the operands or indices.
map( f, x, y1, ..., yn )
where f is a function, x is any expression, and y1, . . . , yn are expressions.

The action of map is to replace each operand x _i of x by f (x _i, y1,
..., yn ).
> map( f, [a,b,c], x, y );
[f(a, x, y), f(b, x, y), f(c, x, y)]
> L := [ seq(xî, i=0..5) ];
L := [1, x, x2 , x3 , x4 , x5 ]
> map( (x,y)->x^2+y, L, 1);
[2, x2 + 1, x4 + 1, x6 + 1, x8 + 1, x10 + 1]
The select, remove, and selectremove commands have the same

syntax as the map command and they work in a similar way. The simplest
forms are
select( f, x )
remove( f, x )
selectremove( f, x )
where f is a Boolean-valued function and x is an expression which must

be one of a sum, product, list, set, function, or indexed name.
The select command selects the operands of x which satisfy the
Boolean-valued function f, creating a new object of the same type as x.
Maple discards those operands for which f does not return true.
The remove command does the opposite of select. It removes the
operands of x that satisfy f.
> X := [seq(i,i=1..10)];
X := [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]
> select(isprime,X);
[2, 3, 5, 7]
> remove(isprime,X);
[1, 4, 6, 8, 9, 10]
> selectremove(isprime,X);
[2, 3, 5, 7], [1, 4, 6, 8, 9, 10]
The general forms of the select and remove commands are
select( f, x, y1, ..., yn )

remove( f, x, y1, ..., yn )
selectremove( f, x, y1, ..., yn )
where f is a function, x is a sum, product, list, set, function or indexed

name, and y1, . . . , yn are expressions. As with the general form of the
map command the expressions y1, . . . , yn are passed to the function f.
> X := {2, sin(1), exp(2*x), x^(1/2)};
√
X := {2, sin(1), e(2 x) , x}
> select(type, X, function);
{sin(1), e(2 x) }
> remove(type, X, constant);

√
{e(2 x) , x}
> X := 2*x*y^2 - 3*y^4*z + 3*z*w + 2*y^3 - z^2*w*y;
X := 2 x y 2 − 3 y 4 z + 3 z w + 2 y 3 − z 2 w y
> select(has, X, z);
−3 y 4 z + 3 z w − z 2 w y
> remove( x -> degree(x)>3, X );
2 x y2 + 3 z w + 2 y3
The zip Command

Use the zip command to merge two lists or vectors. The zip command
has two forms
zip(f, u, v )
zip(f, u, v, d )
where f is a binary function, u and v are both lists or vectors, and d is a

value. The action of zip is for each pair of operands u _i, v _i, to create
a new list or vector out of f (u _i,v _i). The following is an example of
the action of zip.
> zip( (x,y)->x || y, [a,b,c,d,e,f], [1,2,3,4,5,6] );
[a1 , b2 , c3 , d4 , e5 , f6 ]
If the lists or vectors are not the same length, the length of the result
depends on whether you provide d. If you do not specify d, the length will
be the length of the smaller of u and v.
> zip( (x,y)->x+y, [a,b,c,d,e,f], [1,2,3] );
[a + 1, b + 2, c + 3]
If you do specify d, the length of the result of the zip command will be
the length of the longer list (or vector) and Maple uses d for the missing
value(s).
> zip( (x,y)->x+y, [a,b,c,d,e,f], [1,2,3], xi );
[a + 1, b + 2, c + 3, d + ξ, e + ξ, f + ξ]
Note that Maple does not pass the extra argument, xi, to the function
f as it does with the map command.
The seq, add, and mul Commands

The seq, add, and mul commands form sequences, sums, and products,
respectively. Use the following syntax with these commands.
seq(f, i = a ..b )
add(f, i = a ..b )
mul(f, i = a ..b )
where f, a, and b are expressions and i is a name. The expressions a and

b must evaluate to numerical constants (except in the case of seq, where
they may be single character strings).
The result of seq is the sequence that Maple produces by evaluating
f after successively assigning the index name i the values a, a +1, . . . , b,
(or up to the last value not exceeding b). The result of add is the sum of
the same sequence, and the result of mul is the product of the sequence.
If the value a is greater than b, the result is the NULL sequence, 0, and 1,
respectively.
> seq(i^2,i=1..4);
1, 4, 9, 16
> mul(i^2,i=1..4);
576
> add(x[i], i=1..4);
x1 + x2 + x3 + x4
> mul(i^2, i = 4..1);
> seq(i, i = 4.123 .. 6.1);
4.123, 5.123
You can also use the commands seq, add, and mul with the following
syntax.
seq(f, i = X )
add(f, i = X )
mul(f, i = X )
where f is an expression, X is an expression (or string for seq), and i is

a name.
The result of seq in this form is the sequence that Maple produces by
evaluating f after successively assigning the operands of the expression X
(or the characters of string X ) to the index i. The result of add is the sum
of the same sequence, and the result of mul is the product of the same
sequence.
> a := x^3 + 3*x^2 + 3*x + 1;
a := x3 + 3 x2 + 3 x + 1
> seq(degree(i,x), i=a);
3, 2, 1, 0
> seq(i, i="square");
“s”, “q”, “u”, “a”, “r”, “e”
> add(degree(i,x), i=a);
> a := [23,-42,11,-3];
a := [23, −42, 11, −3]
> mul(abs(i),i=a);
31878
> add(i^2,i=a);
2423
4.6 Substitution • 197
seq, add, and mul Versus $, sum, and product Note that the dollar
operator, $, and the sum and product commands are very similar to the
seq, mul, and add commands. However, they differ in an important way.
The index variable i and the end points a and b do not need to be integers.
For example
> x[k] $ k=1..n;
xk $ (k = 1..n)
The design of these commands is for symbolic sequences, sums, and

products. As with the int (integration) command, the index variable k
is a global variable to which you must not assign a value.
When should you use $, sum, and product? versus seq, add, mul?
When you are computing a symbolic sum or product. For example,
if the end points are unknowns, then clearly you must use $, sum, and
product. When you are computing an explicit finite sequence, sum, or
product, that is, you know that the range points a and b are integers,
then use seq, add, or mul. These latter commands are more efficient than
their symbolic counterparts $, sum, and product.
4.6 Substitution
The subs command does a syntactic substitution. It replaces subex-
pressions in an expression with a new value; the subexpressions must be
operands in the sense of the op command.
> expr := x^3 + 3*x + 1;
expr := x3 + 3 x + 1
> subs(x=y, expr);
y3 + 3 y + 1
> subs(x=2, expr);
15
The syntax of the subs command is
subs( s, expr )
where s is either an equation, a list, or set of equations. Maple traverses

the expression expr and compares each operand in expr with the left-hand
side(s) of the equation(s) s. If an operand is equal to a left-hand side of
an equation in s, then subs replaces the operand with the right-hand side
of the equation. If s is a list or set of equations, then Maple makes the
substitutions indicated by the equations simultaneously.
> f := x*y^2;
f := x y 2
> subs( {y=z, x=y, z=w}, f );
y z2
The general syntax of the subs command is
subs( s1, s2, ..., sn, expr )
where s1, s2, . . . , sn are equations or sets or lists of equations, n > 0,

and expr is an expression. This is equivalent to the following sequence of
substitutions.
subs( sn, ..., subs( s2, subs( s1, expr ) ) )
Thus, subs substitutes according to the given equations from left to right.
Notice the difference between the previous example and the following one.
> subs( y=z, x=y, z=w, f );
y w2
Maple does not evaluate the result of a substitution.

> subs( x=0, sin(x) + x^2 );
sin(0)
4.6 Substitution • 199
If you want to combine the acts of substitution and evaluation, use

the two-parameter version of the eval command instead of subs.
> eval(sin(x) + x^2, x=0);
Substitution only compares operands in the expression tree of expr

with the left-hand side of an equation.
> subs(a*b=d, a*b*c);
abc
The substitution did not result in d*c as intended because the

operands of the product a*b*c are a, b, c. That is, the products a*b,
b*c, and a*c do not appear explicitly as operands in the expression a*b*c;
consequently, subs does not see them.
The easiest way to make such substitutions is to solve the equation
for one unknown and substitute for that unknown, that is
> subs(a=d/b, a*b*c);
dc
You cannot always do this, and you may find that it does not always
produce the results you expect. The algsubs routine provides a more
powerful substitution facility.
> algsubs(a*b=d, a*b*c);
dc
Note also that operands of a rational power xn/d are x and n/d.
Although in the following example
> subs( x^(1/2)=y, a/x^(1/2) );
a
√
x
√
it looks as though the output has a x in it, the operands of this ex-
pression are a and x−1/2 . Think of the division as a negative power in
a product, that is, a × x−1/2 . Because the operands of x−1/2 are x and
−1/2, subs does not see x1/2 in x−1/2 . The solution is to substitute for
the negative power x−1/2 .
> subs( x^(-1/2)=1/y, a/x^(1/2) );
a
y
The reader should refer to the online help information under ?algsubs
for more details. Note that the algsubs command, as powerful as it is, is
also much more computationally expensive than the subs command.
4.7 Conclusion
This chapter discusses the elements of Maple’s language. Maple breaks
your input into its smallest meaningful parts, called tokens. Its language
statements include assignments, conditional, looping, and reading from
and saving to files. Many types of expressions exist within Maple, and
using its expression trees tells you of the type and operands in an expres-
sion. You have seen the efficient looping constructs map, zip, and seq,
and how to make substitutions.
5 Procedures
The proc command defines procedures in Maple. This chapter de-

scribes the syntax and semantics of the proc command in the same man-
ner as chapter 4 describes the rest of the Maple programming language.
This chapter explains the concepts of local and global variables and how
Maple passes arguments to procedures. The chapter also provides exer-
cises to help extend your understanding of Maple procedures.
5.1 Procedure Definitions

A Maple procedure definition has the following general syntax.
proc( P )
local L ;
global G ;
options O ;
description D ;
B
end proc
Here, B is a sequence of statements forming the body of the procedure.

The formal parameters, P, along with the local, global, options, and
description clauses are all optional.
The following is a simple Maple procedure definition. It has two for-
mal parameters , x and y, no local, global, options, or description
clauses, and only one statement in the body of the procedure.
201
202 • Chapter 5: Procedures
> proc(x,y)
> x^2 + y^2
> end proc;
proc(x, y) x2 + y 2 end proc
You can give a name to a procedure as to any other Maple object.

> F := proc(x,y) x^2 + y^2 end proc;
F := proc(x, y) x2 + y 2 end proc
You can then execute (invoke) it by using the function call,
F ( A )
When Maple executes the statements of the body of a procedure, it re-

places the formal parameters, P, with the actual parameters, A, from the
function call. Note that Maple evaluates the actual parameters, A, before
substituting them for the formal parameters, P.
Normally, the result a procedure returns after executing is the value
of the last executed statement from the body of the procedure.
> F(2,3);
13
Mapping Notation
You can also define simple one-line procedures by using an alternate syn-
tax borrowed from algebra.
( P ) -> B
The sequence, P, of formal parameters may be empty and the body, B,

of the procedure must be a single expression or an if statement.
> F := (x,y) -> x^2 + y^2;
F := (x, y) → x2 + y 2
If your procedure involves only one parameter, then you may omit the
parentheses around the formal parameter.
5.1 Procedure Definitions • 203
> G := n -> if n<0 then 0 else 1 end if;
G := proc(n)
option operator , arrow ;
if n < 0 then 0 else 1 end if
end proc
> G(9), G(-2);
1, 0
The intended use for the arrow notation is solely for simple one-line
function definitions. It does not provide a mechanism for specifying local
or global variables, or options.
Unnamed Procedures and Their Combinations

Procedure definitions are valid Maple expressions. You can create, ma-
nipulate, and invoke all of them without assigning to a name.
> (x) -> x^2;
x → x2
You invoke an unnamed procedure in the following manner.

> ( x -> x^2 )( t );
t2
> proc(x,y) x^2 + y^2 end proc(u,v);
u2 + v 2
A common use of unnamed procedures occurs in conjunction with the

map command.
> map( x -> x^2, [1,2,3,4] );
[1, 4, 9, 16]
You can add procedures together, or, if appropriate, you can process
them by using commands, such as the differential operator , D.
> D(x -> x^2);
x → 2x
> F := D(exp + 2*ln);
1
F := exp + 2 (a → )
a
You can apply the result, F, directly to arguments.
Procedure Simplification
When you create a procedure, Maple does not evaluate the procedure but
it does simplify the body of the procedure.
> proc(x) local t;
> t := x*x*x + 0*2;
> if true then sqrt(t); else t^2 end if;
> end proc;
proc(x) local t; t := x3 ; sqrt(t) end proc
Maple simplifies procedures with the operator option even further.

> x -> 3/4;
3
4
> (x,y,z) -> h(x,y,z);
Procedure simplification is a simple form of program optimization.
5.2 Parameter Passing

Consider what happens when Maple evaluates a function or procedure.
5.2 Parameter Passing • 205
F ( ArgumentSequence )
First, Maple evaluates F. Then it evaluates the ArgumentSequence. If

any of the arguments evaluate to a sequence, Maple flattens the result-
ing sequence of sequences into a single sequence, the sequence of actual
parameters . Suppose F evaluates to a procedure.
proc( FormalParameters )
body
end proc
Maple then executes the statements in the body of the procedure, sub-
stituting the actual parameters for the formal parameters.
Consider the following example.
> s := a,b: t := c:
> F := proc(x,y,z) x + y + z end proc:
> F(s,t);
a+b+c
Here, s,t is the argument sequence, a,b,c is the actual parameter

sequence and x,y,z is the formal parameter sequence.
The number of actual parameters, n, may differ from the number of
formal parameters. If too few actual parameters exist, then an error occurs
if (and only if) a missing parameter is actually used during the execution
of the procedure body. Maple ignores extra parameters.
> f := proc(x,y,z) if x>y then x else z end if end proc:
> f(1,2,3,4);
> f(1,2);
Error, (in f) f uses a 3rd argument, z, which is
missing
> f(2,1);
2
Declared Parameters
You may write procedures that only work for certain types of input. Use
declared formal parameters so that when you use the procedure with the
wrong types of input Maple issues an informative standard error message.
A type declaration has the following syntax.
parameter :: type
Maple knows many types of expressions; see ?type.

When invoking the procedure, before executing the body of the pro-
cedure, Maple tests the types of the actual parameters from left to right.
Any of these tests may generate an error message. If no type error occurs,
the procedure executes.
> MAX := proc(x::numeric, y::numeric)
> if x>y then x else y end if
> end proc:
> MAX(Pi,3);
Error, invalid input: MAX expects its 1st argument, x,
to be of type numeric, but received Pi
You can also use declared parameters with the operator option.
> G := (n::even) -> n! * (n/2)!;
1
G := n::even → n! ( n)!
2
> G(6);
4320
> G(5);
Error, invalid input: G expects its 1st argument, n, to
be of type even, but received 5
If you do not declare the type of a parameter, it can have any type.
Thus, proc(x ) is equivalent to proc(x ::anything). If that is what you
intend, you should use the latter form to inform other users that you
intend your procedure to work for any input.
5.2 Parameter Passing • 207
The Sequence of Arguments

You do not need to supply names for the formal parameters. You can
access the entire sequence of actual arguments from within the procedure,
by using the name args . The following procedure simply builds a list of
its arguments.
> f := proc() [args] end proc;
f := proc() [args] end proc
> f(a,b,c);
[a, b, c]
> f(c);
[c]
> f();
[]
The ith argument is simply args[i]. Thus, the following two pro-
cedures are equivalent, provided you call them with at least two actual
parameters of type numeric.
> MAX := proc(x::numeric,y::numeric)
> if x > y then x else y end if;
> end proc;
MAX := proc(x::numeric, y::numeric)

if y < x then x else y end if
end proc
> MAX := proc()
> if args[1] > args[2] then args[1] else args[2] end if;
> end proc;
MAX := proc()
if args2 < args1 then args1 else args2 end if
end proc
The nargs command provides the total number of actual parameters.

This allows you to easily write a procedure, MAX, which finds the maximum
of any number of arguments.
> MAX := proc()
> local i,m;
> if nargs = 0 then return -infinity end if;
> m := args[1];
> for i from 2 to nargs do
> if args[i] > m then m := args[i] end if;
> end do;
> m;
> end proc:
The maximum of the three values 2/3, 1/2, and 4/7 is
> MAX(2/3, 1/2, 4/7);
2
3
5.3 Local and Global Variables

Variables inside a procedure are either local to that procedure or global.
Variables outside procedures are global. Maple considers local variables
in different procedure invocations to be different variables, even if they
have the same name. Thus, a procedure can change the value of a local
variable without affecting variables of the same name in other procedures
or a global variable of the same name. You should always declare which
variables are local and which are global in the following manner.
local L1, L2, ..., Ln ;

global G1, G2, ..., Gm ;
In the procedure below, i and m are local variables.

> MAX := proc()
> local i,m;
> m := args[1];
> end do;
> m;
> end proc:
5.3 Local and Global Variables • 209
In the case of nested procedures, where one procedure is defined within

the body of another, variables can also acquire local or global declaration
from procedures which enclose them. See Section 2.2 for details and ex-
amples of nested procedures.
If no declaration is made of whether a variable is local or global, Maple
decides. A variable is automatically made local in both of these cases:
• It appears on the left-hand side of an assignment statement. For ex-

ample, A in A := y or A[1] := y.
• It appears as the index variable in a for loop, or in a seq, add, or

mul command.
If neither of these two rules applies, the variable is a global variable.

> MAX := proc()
> m := args[1];
> end do;
> m;
> end proc:
Warning, ‘m‘ is implicitly declared local to procedure
‘MAX‘
‘MAX‘
Maple declares m local because it appears on the left-hand side of the

assignment m:=args[1], and i local because it is the index variable of a
for loop.
Do not rely on this facility to declare local variables. Declare all
your local variables explicitly. Rely instead on the warning messages to
help you identify variables that you have misspelled or have forgotten to
declare.
The newname procedure below creates the next unused name in the
sequence C1, C2, . . . . The name that newname creates is a global variable
since neither of the two rules above apply to cat(C,N).
> newname := proc()
> global N;
> N := N+1;
> while not assigned(cat(C,N)) do
> N := N+1;
> end do;
> cat(C,N);
> end proc:
> N := 0;
N := 0
The newname procedure does not take any arguments.

> newname() * sin(x) + newname() * cos(x);
C1 sin(x) + C2 cos(x)
Assigning values to global variables inside procedures is generally a

poor idea. Any change of the value of a global variable affects all uses of
the variable, even any of which you were unaware. Thus, you should only
use this technique judiciously.
Evaluation of Local Variables

Local variables are special in another very important way. During the
execution of a procedure body, they evaluate exactly one level. Maple
evaluates global variables fully, even inside a procedure.
This section should help to clarify this concept. Consider the following
examples.
> f := x + y;
f := x + y
> x := z^2/ y;
z2
x :=
y
> z := y^3 + 3;
z := y 3 + 3
The normal full recursive evaluation yields

> f;
5.3 Local and Global Variables • 211
(y 3 + 3)2
+y
y
You can control the actual level of evaluation by using eval . Using
the following sequence of commands, you can evaluate to one level, two
levels, and three levels.
> eval(f,1);
x+y
> eval(f,2);
z2
+y
y
> eval(f,3);
(y 3 + 3)2
+y
y
The notion of the use of one-level evaluation 1 is important for effi-

ciency. It has very little effect on the behavior of programs because you
tend to write code in an organized sequential fashion. In the rare case
where a procedure body requires a full-recursive evaluation of a local
variable, you may use the eval command.
> F := proc()
> local x, y, z;
> x := y^2; y := z; z := 3;
> eval(x)
> end proc:
> F();
Without the call to eval, the answer would be y 2 .

You can still use local variables as unknowns just like global variables.
For example, in the following procedure, the local variable x does not have
1
Such a concept of evaluation does not occur in traditional programming languages.
However, here, you may assign to a variable a formula involving other variables, which
in turn you may assign values and so on.
an assigned value. The procedure uses it as the variable in the polynomial

xn − 1.
> RootsOfUnity := proc(n)
> local x;
> [solve( x^n - 1=0, x )];
> end proc:
> RootsOfUnity(5);
1 1√ 1 √ p √
[1, − + 5 + I 2 5 + 5,
4 4 4
1 1√ 1 √ p √ 1 1√ 1 √ p √
− − 5 + I 2 5 − 5, − − 5 − I 2 5 − 5,
4 4 4 4 4 4
1 1√ 1 √ p √
− + 5 − I 2 5 + 5]
4 4 4
5.4 Procedure Options and the Description Field

Options
A procedure may have one or more options. You may specify options by
using the options clause of a procedure definition.
options O1, O2, ..., Om ;
You may use any symbol as an option but the following options have
special meanings.
The remember and system Options When you invoke a procedure with
the remember option, Maple stores the result of the invocation in the
remember table associated with the procedure. Whenever you invoke
the procedure, Maple checks whether you have previously called the pro-
cedure with the same parameters. If so, Maple retrieves the previously
calculated result from the remember table rather than executing the pro-
cedure again.
> option remember;
> fib(n-1) + fib(n-2);
> end proc;
5.4 Procedure Options and the Description Field • 213
fib := proc(n::nonnegint )
option remember ;
fib(n − 1) + fib(n − 2)
end proc
You may place entries in the remember table of a procedure by direct
assignment; this method also works for procedures without the remember
option.
> fib(0) := 0;
fib(0) := 0
> fib(1) := 1;
fib(1) := 1
The following is the fib procedure’s remember table.
table([0 = 0, 1 = 1])
Since fib has the remember option, invoking it places new values in
its remember table.
> fib(9);
34
Below is the new remember table.
table([0 = 0, 1 = 1, 2 = 1, 3 = 2, 4 = 3, 5 = 5, 6 = 8, 7 = 13,
8 = 21,
9 = 34
])
The use of remember tables can drastically improve the efficiency of
recursively defined procedures.
The system option allows Maple to remove entries from a proce-
dure’s remember table. Such selective amnesia occurs during garbage
collection, an important part of Maple’s memory management scheme.
See section 2.5 for more details and examples of remember tables.
The operator and arrow Options The operator option allows Maple
to make additional simplifications to the procedure, and the arrow option
indicates that the pretty-printer should display the procedure by using the
arrow notation.
> proc(x)
> option operator, arrow;
> x^2;
> end proc;
x → x2
Section 5.1 describes procedures by using the arrow notation.
The Copyright Option Maple considers any option that begins with
the word Copyright to be a Copyright option. Maple does not print
the body of a procedure with a Copyright option unless the interface
variable verboseproc is at least 2.
> f := proc(expr::anything, x::name)
> option ‘Copyright (c) 1684 by G. W. Leibnitz. All rights reserved‘;
> Diff(expr, x);
> end proc;
f := proc(expr ::anything , x::name) . . . end proc
The builtin Option Maple has two main classes of procedures: those
which are part of the Maple kernel, and those which the Maple language
itself defines. The builtin option indicates the kernel procedures. You
can see this when you fully evaluate a built-in procedure.
> eval(type);
proc() option builtin; 268 end proc
Each built-in procedure is uniquely identified by a number. Of course,

you cannot create built-in procedures of your own.
The Description Field

The last part of the procedure header is the description field. It must
appear after any local clause, global clause, or options clause, and
before the body of the procedure. It takes the following form.
5.5 The Value Returned by a Procedure • 215
description symbol ;
The description field has no effect on the execution of the procedure.

Its use is for documentation purposes. Unlike a comment, which Maple
discards when you read in a procedure, the description field provides a
way to attach a one line comment to a procedure.
> f := proc(x)
> description ‘computes the square of x‘;
> x^2; # compute x^2
> end proc:
> print(f);
proc(x)
description ‘computes the square of x ‘;
x2
end proc
Also, Maple prints the description field even if it does not print the
body of a procedure due to a Copyright option.
> f := proc(x)
> option ‘Copyrighted ?‘;
> description ‘computes the square of x‘;
> x^2; # compute x^2
> end proc:
> print(f);
proc(x)
description ‘computes the square of x ‘
...
end proc
5.5 The Value Returned by a Procedure

When you invoke a procedure, the value that Maple returns is normally
the value of the last statement in the statement sequence of the body of
the procedure. Three other types of returns from procedures are a return
through a parameter, an explicit return, and an error return.
Assigning Values to Parameters

Sometimes you may want to write a procedure that returns a value
through a parameter. Consider writing a Boolean procedure, MEMBER,
which determines whether a list L contains an expression x. Moreover,

if you call MEMBER with a third argument, p, then MEMBER should assign
the position of x in L to p.
> MEMBER := proc(x::anything, L::list, p::evaln) local i;
> for i to nops(L) do
> if x=L[i] then
> if nargs>2 then p := i end if;
> return true
> end if;
> end do;
> false
> end proc:
If you call MEMBER with two arguments, then nargs is two, so the body of
MEMBER does not refer to the formal parameter, p. Therefore, Maple does
not complain about a missing parameter.
> MEMBER( x, [a,b,c,d] );
false
If you call MEMBER with three arguments, then the type declaration
p::evaln ensures that Maple evaluates the third actual parameter to a
name2 rather than by using full evaluation.
> q := 78;
q := 78
> MEMBER( c, [a,b,c,d], q );
true
> q;
Maple evaluates parameters only once. This means that you cannot
use formal parameters freely like local variables within a procedure body.
Once you have made an assignment to a parameter you should not
2
If the third parameter has not been declared as evaln, then you should enclose the
name q in single quotes (’q’) to ensure that the name and not the value of q is passed
to the procedure.
refer to that parameter again. The only legitimate purpose for assigning
to a parameter is so that on return from the procedure the corresponding
actual parameter has an assigned value. The following procedure assigns
the value −13 to its parameter, then returns the name of that parameter.
> f := proc(x::evaln)
> x := -13;
> x;
> end proc:
> f(q);
The value of q is now −13.

> q;
−13
The count procedure below is a more complicated illustration of this

phenomenon. count should determine whether a product of factors, p,
contains an expression, x. If p contains x, then count should return the
number of factors that contain x in the third parameter, n.
> count := proc(p::‘*‘, x::name, n::evaln)
> local f;
> n := 0;
> for f in p do
> if has(f,x) then n := n+1 end if;
> end do;
> evalb( n>0 );
> end proc:
The count procedure does not work as intended.
> count(2*x^2*exp(x)*y, x, m);
−m < 0
The value of the formal parameter n inside the procedure is always m,

the actual parameter that Maple determines once and for all when you
invoke the procedure. Thus, when execution reaches the evalb statement,
the value of n is the name m, and not the value of m. Worse yet, the n:=n+1
statement assigns to m the name m+1, as you can see if you evaluate m
one level.
> eval(m, 1);
m+1
The m in the above result also has the value m+1.

> eval(m, 2);
m+2
Thus, if you were to evaluate m fully, Maple would enter an infinite

loop.
A general solution to this type of problem is to use local variables and
to view the assignment to a parameter as an operation which takes place
just before returning from the procedure.
> count := proc(p::‘*‘, x::name, n::evaln)
> local f, m;
> m := 0;
> for f in p do
> if has(f,x) then m := m + 1 fi;
> od;
> n := m;
> evalb( m>0 );
> end:
The new version of count works as intended.
> count(2*x^2*exp(x)*y, x, m);
true
> m;
Explicit Returns
An explicit return occurs when you invoke the return statement, which
has the following syntax.
return sequence
The return statement causes an immediate return from the procedure

and the value of the sequence becomes the value of the procedure invo-
cation.
For example, the following procedure computes the first position i of
a value x in a list of values L. If x is not in the list L, the procedure
returns 0.
> POSITION := proc(x::anything, L::list)
> local i;
> for i to nops(L) do
> if x=L[i] then return i end if;
> end do;
> 0;
> end proc:
In most applications of the return statement, it returns only a single

expression. Returning a sequence, however, including the empty sequence,
is quite legitimate. For example, the GCD procedure below computes the
greatest common divisor g of two integers a and b. It returns the sequence
g, a/g, b/g. GCD must treat the case a = b = 0 separately because that
makes g zero.
> GCD := proc(a::integer, b::integer)
> local g;
> if a=0 and b=0 then return 0,0,0 end if;
> g := igcd(a,b);
> g, iquo(a,g), iquo(b,g);
> end proc:
> GCD(0,0);
0, 0, 0
> GCD(12,8);
4, 3, 2
Of course, instead of returning a sequence, you may also return a list

or a set of values.
Error Returns
An error return occurs when you raise an exception by invoking the
error statement, which has the following syntax:
error msgString
error msgString, msgParams
The msgString is a string value which is independent of any parame-

ters that are to be part of the message (for instance, the string complaining
about an unassigned variable should not mention the variable by name).
In the msgString, numbered parameters are used as placeholders
for actual values. For example, the error "f has a 2nd argument,
x, which is missing" is specified by the msgString "%1 has a %-2
argument, %3, which is missing", and the msgParams f, 2, and x.
Each numbered parameter consists of the percent symbol, "%", option-
ally followed by a minus sign, "-", followed by one or more digits, making
up a single-digit integer n. At message display time, the nth msgParam
will be substituted for the numbered parameter. A numbered parameter
of the form %n will display the nth msgParam in lineprinted notation (i.e.,
as lprint would display it). A numbered parameter of the form %-n will
display the nth msgParam, assumed to be an integer, in ordinal form. For
example, the value 2 is displayed as "2nd". The special parameter, %0,
will display all the msgParams, separated by a comma and space.
The msgParams are one or more arbitrary Maple objects that will be
substituted into numbered parameter locations in the msgString in the
event that the exception is ever printed as an error message.
The error statement will evaluate its arguments, and then create
an exception object, which is an expression sequence with the following
elements:
• The name of the procedure in which the exception was raised, or the
constant 0 if the exception was raised at the top-level.
• The msgString.
• The msgParams, if any.
The created exception object is assigned to the global variable

lastexception as an expression sequence. The actual arguments to the
error statement are also assigned to lasterror for compatibility with
older versions of Maple.
The error statement normally causes an immediate exit from the
current procedure to the Maple session. Maple prints an error message of
the form:
Error, (in procName ) msgText
Here msgText is the text of the error message, which is constructed

from the msgString and optional msgParams of the error statement. The
procedure in which the error occurred is given by procName. If the pro-
cedure does not have a name, procName is "unknown". If the error occurs
at the top-level, outside of any procedure, the entire "(in procName)"
part is omitted from the message.
A common use of the error statement is to check that the actual
parameters to a procedure are of the correct type, but parameter decla-
rations are not sufficient for the job. The pairup procedure below takes
a list L of the form [x1 , y1 , x2 , y2 , . . . , xn , yn ] as input, and creates from it
a listlist of the form [[x1 , y1 ], [x2 , y2 ], . . . , [xn , yn ]]. A simple type check
cannot determine if the list L has an even number of elements, so you
need to check that explicitly.
> pairup := proc(L::list)
> local i, n;
> n := nops(L);
> if irem(n,2) = 1 then
> error "list must have an even number of entries, but had %1", n
> end if;
> [seq( [L[2*i-1],L[2*i]], i=1..n/2 )]
> end proc:
> pairup([1, 2, 3, 4, 5]);
Error, (in pairup) list must have an even number of
entries, but had 5
> pairup([1, 2, 3, 4, 5, 6]);
[[1, 2], [3, 4], [5, 6]]
Trapping Exceptions
You can trap exceptions using the try statement. The syntax for the try
statement is:
try tryStatSeq
catch catchStrings : catchStatSeq
end try
try tryStatSeq
finally finalStatSeq
end try
When execution enters a try statement, the tryStatSeq is executed. If

no exceptions occur during the execution of tryStatSeq, execution contin-
ues with finalStatSeq if a finally clause was provided. After that, or if
no finally clause was provided, execution continues with the statement
after the end try.
If an exception does occur during the execution of tryStatSeq, execu-
tion of tryStatSeq terminates immediately. The exception object corre-
sponding to the exception is compared against each catchString in turn
until a match is found. Any number of catch clauses can be provided,
and each can have any number of catchStrings, separated by commas. A
catch clause can also have no catch string.
If a matching catch clause is found, or the catch clause contains no
catchStrings, the catchStatSeq of that catch clause is executed, and the
exception is considered to have been caught. If no matching catch clause
is found, the exception is considered not-caught, and is re-raised outside
the try construct.
A catchStatSeq can contain an error statement with no arguments,
which also re-raises the exception. When an exception is re-raised, a new
exception object is created that records the current procedure name, and
the message and parameters from the original exception.
Under normal circumstances, the finalStatSeq of the finally clause,
if there is one, is always executed before control leaves the try statement.3
This is true even if a catchStatSeq re-raises the exception, raises a new
one, or executes a return, break, or next statement.
Under certain abnormal circumstances, the finalStatSeq is not exe-
cuted:
• If an exception is raised in a catch clause and this exception is caught

by the debugger and the user exits the debugger, the user’s command
to stop execution overrides everything.
3
This is true whether or not an exception occurs, and if one does occur, whether
or not it is caught, and if it is caught, whether or not another exception occurs in the
catch clause.
• If one of the following untrappable exceptions occurs, the exception

is not caught, and the finalStatSeq is not executed:
1. Computation timed out (this can only be caught by timelimit,

which raises a "time expired" exception, which can be caught).
2. Computation interrupted (i.e., user pressed Ctrl-C, Break, or equiv-
alent).
3. Internal system error (i.e., which indicates a bug in Maple itself).
4. ASSERT or local variable type assertion failure (assertion failures are
not trappable because they indicate a coding error, not an algorith-
mic failure).
5. Stack overflow (when that happens, there’s generally not enough
stack space to do anything like run cleanup code).
If an exception occurs during the execution of a catchStatSeq or the

finalStatSeq, it is treated in the same way as if it occurred outside the
try...end statement entirely.
When looking for a matching catch clause, the following definition of
"matching" is used:
• Neither the exception object nor the catchStrings are evaluated (the
exception object will already have been evaluated by the error state-
ment that produced it).
• The catchStrings are considered to be prefixes of the exception ob-

ject’s synmsgString. If a catchString has n characters, only the first n
characters of the msgString need match the catchString. This allows
one to define classes of exceptions.
• A missing catchString will match any exception.
• The "result" of a try statement (i.e., the value that % would return
if evaluated immediately after execution of the try statement) is the
result of the last statement executed within the try statement.
A given catchString (or a catch clause without one) can appear only
once in a try...end construct.
A very useful application of the try and error statements is to abort
an expensive computation as quickly and cleanly as possible. For example,
suppose you are trying to compute an integral by using one of several
methods, and in the middle of the first method, you determine that it
will not succeed. You would like to abort that method and go on to try
another method. The code that tries the different methods might look
like this:
> try
> result := MethodA(f,x)
> catch "FAIL":
> result := MethodB(f,x)
> end try;
MethodA can abort its computation at any time by executing the state-
ment error "FAIL". The catch clause will catch that exception, and pro-
ceed to try MethodB. If any other error occurs during the execution of
MethodA, or if an error occurs during the execution of MethodB, it will not
be caught.
Another useful application of the try statement is to make sure certain
resources are freed when you are done with them, regardless of whether
or not anything went wrong while you were using them. For example, you
may wish to use the facilities of the I/O library (see Chapter 1) to read
the lines of a file and process them in some way:
> f := fopen("myfile",TEXT,READ):
> try
> line := readline(f);
> while line < 0 do
> ProcessContentsOfLine(line);
> line := readline(f)
> end do
> finally
> fclose(f)
> end try;
In this example, if any exception occurs while reading or processing

the lines of the file, it will not be caught, because there is no catch clause.
However, the fclose(f) will be executed before execution leaves the try
statement, regardless of whether or not there was an exception.
The final example makes use of both catch and finally clauses. In
this example, we are writing to a file instead of reading from one.
> f := fopen("myfile",TEXT,WRITE):
> try
> for i to 100 do
> fprintf(f,"Result %d is %q\n",i,ComputeSomething(i))
> end do
> catch:
> fprintf(f,"Something went wrong: %q\n",lastexception);
> error
> finally
> fclose(f)
> end try;
If any exception occurs, we catch it with a catch clause with no

catchString, and write the exception object into the file. We then re-raise
the exception, by executing an error statement with no msgString. In all
cases, we close the file by executing fclose(f) in the finally clause.
Returning Unevaluated
Maple often uses a particular form of return as a fail return, in the sense
that it cannot carry out the computation and so returns the unevaluated
function invocation as the result. The procedure MAX, below, calculates
the maximum of two numbers, x and y.
> MAX := proc(x,y) if x>y then x else y end if end proc:
The above version of MAX is unacceptable for a symbolic computation

system because it insists on its arguments being numerical values so that
Maple can determine if x > y.
> MAX(3.2, 2);
3.2
> MAX(x, 2*y);

Error, (in MAX) cannot evaluate boolean: 2*y-x < 0
The absence of symbolic capabilities in MAX causes problems when you

try to plot expressions involving MAX.
> plot( MAX(x, 1/x), x=1/2..2 );
Error, (in MAX) cannot evaluate boolean: 1/x-x < 0
The error occurs because Maple evaluates MAX(x, 1/x) before invok-
ing the plot command.
The solution is to make MAX return unevaluated when its parame-
ters, x and y, are not numeric. That is, in such cases MAX should return
’MAX’(x,y).
> MAX := proc(x, y)
> if type(x, numeric) and type(y, numeric) then
> if x>y then x else y end if;
> else
> ’MAX’(x,y);
> end if;
> end proc:
The new version of MAX handles both numeric and non-numeric input.
> MAX(3.2, 2);
3.2
> MAX(x, 2*y);
MAX(x, 2 y)
> plot( MAX(x, 1/x), x=1/2..2 );
1.8
1.6
1.4
1.2
1
0.6 0.8 1 1.2 1.4 1.6 1.8 2
x
You can improve MAX so that it can find the maximum of any number
of arguments. Inside a procedure, args is the sequence of actual parame-
ters, nargs is the number of actual parameters, and procname is the name
of the procedure.
> MAX := proc()
> local m, i;
> m := -infinity;
> for i in (args) do
> if not type(i, numeric) then
> return ’procname’(args);
> end if;
> if i>m then m := i end if;
> end do;
> m;
> end proc:
> MAX(3,1,4);
> MAX(3,x,1,4);
5.6 The Procedure Object • 227
MAX(3, x, 1, 4)
The sin function and the int integration command follow the same
model as the MAX procedure above. If Maple can compute the result, it
returns it; otherwise, sin and int return unevaluated.
Exercise
1. Improve the MAX procedure above so that MAX(3,x,1,4) returns
MAX(x,4); that is, the procedure returns the maximum numerical
value along with all non-numerical values.
5.6 The Procedure Object

This section describes the procedure object, its type and operands, its
special evaluation rule, and how to save it to a file and retrieve it again.
Last Name Evaluation

Maple evaluates ordinary expressions in a full recursive evaluation
mode. All future references to a name that you assign a value return
the computed value instead of the name.
> f := g;
f := g
> g := h;
g := h
> h := x^2;
h := x2
Now f evaluates to x2 .
> f;
x2
Names of procedures, modules, arrays, and tables are exceptions. For

such names, Maple uses a last name evaluation model. This model avoids
printing all the details forming the procedure definition.
> F := G;
F := G
> G := H;
G := H
> H := proc(x) x^2 end proc;
H := proc(x) x2 end proc
Now F evaluates to H because H is the last name before the actual

procedure.
> F;
You can use the eval command to evaluate a procedure fully.

> eval(F);
proc(x) x2 end proc
See also section 2.1.
The Type and Operands of a Procedure

Maple recognizes all procedures (including those created by using the
mapping notation) as being of type procedure, as are any names that
you give to procedures.
> type(F,name);
true
> type(F,procedure);
true
> type(F,name(procedure));
true
> type(eval(F),procedure);
true
Thus, you can use the following test to ensure that F is the name of
a procedure.
> if type(F, name(procedure)) then ... end if
A procedure has seven operands:
1. The sequence of formal parameters.
2. The sequence of local variables.
3. The sequence of options.
4. The remember table.
5. The description string.
6. The sequence of global variables.
7. The lexical table.
As an example of the structure of a procedure, consider the following.

> f := proc(x::name, n::posint)
> local i;
> global y;
> option Copyright;
> description "a summation";
> sum( x[i] + y[i], i=1..n );
> end proc:
Place an entry in the procedure’s remember table.

> f(t,3) := 12;
f(t, 3) := 12
You can see the various parts of f below.

The name of the procedure:
> f;
The procedure itself:

> eval(f);
proc(x::name, n::posint )
description “a summation”
...
end proc
The formal parameters:
> op(1, eval(f));
x::name, n::posint
The local variables:

> op(2, eval(f));
The options:
> op(3, eval(f));
Copyright
The remember table:

> op(4, eval(f));
table([(t, 3) = 12])
The description:
> op(5, eval(f));
“a summation”
The global variables:

> op(6, eval(f));
The body of a procedure is not one of its operands, so you cannot

gain access to the body with the op command. If you need to manipulate
the body of a procedure, see ?codegen.
Saving and Retrieving Procedures

While you develop a new procedure, you can save your work by saving
the whole worksheet. When you have finished, save the procedure.
> CMAX := proc(x::complex(numeric), y::complex(numeric))
> if abs(x)>abs(y) then
> x;
> else
> y;
> end if;
> end proc:
Use the save command to save procedures in the same manner you save
any other Maple object.
> save CMAX, "CMAX.mws":
The read command retrieves the objects stored in a .m file.

> read "CMAX.mws":
Some Maple users prefer to write Maple procedures with their favorite
text editor. You can also use the read command to read in data from such
files. Maple executes each line in the file as if you had typed it directly
into your session.
If you make a number of related procedures, you may want to save
them as a Maple package. Making a package allows you to load the pro-
cedures using the with command. See section 6.3.
5.7 Explorations
The purpose of the exercises in this section is to deepen your understand-
ing of how Maple procedures work. In some cases you may wish to study
the on-line help pages for the various Maple commands that you will need.
Exercises √
1. Implement the function f (x) = ( 1 − x2 )3 − 1, first as a procedure,
then by using the mapping notation. Compute f (1/2) and f (0.5) and
comment on the different results. Use the D operator to compute f 0 ,
and then compute f 0 (0).
2. Write a procedure, SPLIT, which on input of a product f and a variable

x returns a list of two values. The first item in the list should be
the product of the factors in f that are independent of x, and the
second item should be the product of the factors that have an x in
them. Hint: You may want to use the has, select, remove, and
selectremove commands.
3. The following program tries to compute 1 − x|a| .

> f := proc(a::integer, x::anything)
> if a<0 then a := -a end if;
> 1-xâ;
> end proc:
What is wrong with this procedure? You may want to use the Maple
debugger to isolate the error. See chapter 7.
4. ab/g gives the least common multiple of two integers, a and b, where
g is the greatest common divisor of a and b. For example, the least
common multiple of 4 and 6 is 12. Write a Maple procedure, LCM,
which takes as input n > 0 integers a1 , a2 , . . . , an and computes their
least common multiple. By convention, the least common multiple of
zero and any other number is zero.
5. The following recurrence relation defines the Chebyshev polynomials

of the first kind, Tn (x).
T0 (x) = 1, T1 (x) = x, Tn (x) = 2xTn−1 (x) − Tn−2 (x)
The following procedure computes Tn (x) in a loop for any given inte-
ger n.
> T := proc(n::integer, x)
> local t1, tn, t;
> t1 := 1; tn := x;
> t := expand(2*x*tn - t1);
> t1 := tn; tn := t;
> end do;
> tn;
> end proc:
The procedure has several errors. Which variables should have been
declared local? What happens if n is zero or negative? Identify and
correct all errors, using the Maple debugger where appropriate. Mod-
ify the procedure so that it returns unevaluated if n is a symbolic
value.
5.8 Conclusion
In this chapter, you have seen the details of the proc command. You
have learned the finer points of the options at your disposal when defin-
ing procedures. You have learned about functional operators, unnamed
procedures, and procedure simplification.
In addition, you have reviewed Maple’s evaluation rules which chap-
ter 2 introduced. For example, Maple generally evaluates local variables
to one level and global variables fully. Maple evaluates the arguments to a
procedure at the time you invoke it. How they are evaluated depends upon
the environment in which the call occurs, and in some cases, the types
specified within the procedure definition. Once evaluated, Maple substi-
tutes the values into the procedure and then executes it. Maple does no
further evaluation on the values which it substituted, unless you specifi-
cally use a command such as eval. This rule makes it impractical to use
parameters to store temporary results, as you would use local variables.
This chapter extended the discussion of type declarations, which were
introduced in chapters 1 and 2. Type declarations are particularly useful
as a means of stating the intended purpose of your procedures and as a
convenient means of supplying error messages to any user who might call
them with inappropriate values.
This chapter concludes the formal review of the Maple language which
began in chapter 4. The remaining chapters deal with specific areas of
Maple programming. For example, chapter 7 discusses the Maple debug-
ger, chapter 8 introduces you to the details of numerical programming
in Maple, and chapter 9 shows how to extend Maple’s extensive plotting

facilities to suit your needs.
6 Programming with
Modules
In the same way that procedures allow you to abstract a sequence

of commands typed to the Maple interpreter, “modules” allow you to
abstract collections of related procedures and data.
This chapter describes Maple’s module system. Modules are a type of
Maple expression, like numbers, equations, and procedures, that enable
you to write generic algorithms, create packages, or simply use Pascal-
style records in your programs.
Modules can be used a number of ways. In this chapter, we describe
four broad categories of application for them.
1. Encapsulation
2. Packages
3. Modeling Objects
4. Generic Programming
Encapsulation refers to the provision of a guarantee that an ab-

straction is used only according to its specified interface. This pro-
vides the developer with the ability to write significant software systems
that are transportable and reusable and that offer clean, well-defined
user interfaces. This makes your code more maintainable and easier to
understand—important properties for large software systems.
Packages are a vehicle for bundling together collections of related
Maple procedures to address computations in some well-defined problem
domain. Much of the functionality of the standard Maple library resides
in packages.
Objects are easily represented using modules. An object is a repre-
sentation, in software, of something that has both state and behavior.
235
236 • Chapter 6: Programming with Modules
You compute with objects by sending them “messages”, to which they

respond by performing services.
Generic programs are written without knowledge of how the objects
upon which they operate are represented. A generic program will work
with any object that honors the “contract” against which the program is
written, regardless of how the object satisfes that contract.
These are four very practical software engineering concepts, but they
are all just different facets of the same underlying, abstract Maple lan-
guage technology. We have devoted less space to the abstract language
feature than to its practical implications.
Before we go into too many details about modules, it is helpful to
examine a small example module to get an idea what they are all about.
A First Example Here is an example of a simple, but nontrivial mod-

ule. When Maple evaluates the right-hand side of the assignment to
TempGenerator, it creates a module using the module definition that
begins with module()... and ends with end module.
> TempGenerator := module()
> description "generator for temporary symbols";
> export gentemp;
> local count;
>
> count := 0;
> gentemp := proc()
> count := 1 + count;
> ‘tools/gensym‘( T || count )
> end proc;
> end module;
TempGenerator := module()
local count ;
export gentemp;
description “generator for temporary symbols”;
end module
The module definition that appears above resembles a procedure defi-
nition. The main differences visible here are the use of the keyword mod-
ule instead of proc (and the corresponding terminator), and the export
declaration following the description string.
We could do something quite similar using only procedures.
> TempGeneratorProc := proc()
> description "generator for temporary symbols";
> local count, gentemp;
• 237
> count := 0;
> gentemp := proc()
> ‘tools/gensym‘( T || count )
> end proc;
> eval( gentemp, 1 )
> end proc:
We can assign the procedure returned by TempGeneratorProc, and then
use it to generate temporary symbols.
> f := TempGeneratorProc();
f := proc()
count := 1 + count ; ‘tools/gensym‘(T ||count )
end proc
> f();
T1
> f();
T2
The module TempGenerator and the procedure TempGeneratorProc

are similar. Understanding this similarity will take you a long way towards
understanding how modules work.
In the procedure version, the local variable gentemp is assigned a
little procedure that references another local variable count; the value of
gentemp is returned by the procedure to its caller. The module version of
the generator does much the same thing, except that its gentemp variable
is declared as an export, not a local, and there is no explicit return. In
fact, in both versions of the generator, the variables count and gentemp
are local variables. The significant difference here is that, in the module
version, one of those local variables is exported . This means that it is
made available outside the scope in which it was created. Special syntax
is used for accessing exported local variables. For example, to call the
exported variable gentemp of the module, you can type
> TempGenerator:-gentemp();
T1
using the member selection operator :-. In a sense, a module definition

“returns” a data structure (a module) that contains all of its exported
local variables.
A second mechanism exists for conveniently accessing module exports
— the use statement.
> use TempGenerator in
> gentemp();
> gentemp();
> gentemp();
> end use;
T2
T3
T4
Within the body of a use statement, the exported local variables of

the module that appears after the use keyword can be accessed directly,
without the need for the member selection operator :-.
About This Chapter

This chapter provides many examples of the use of modules to help you
understand the ideas presented. Some examples are very simple, designed
to illustrate a very specific point. Others are more substantial, and can be
used in your own programs. Many of the nontrivial examples are available
as Maple source code in the samples directory of your Maple installation.
You can load any of these into your private Maple library and experiment
with them. You are encouraged to modify, extend, and improve on these
code samples, and to use them in your own work.
6.1 Syntax and Semantics

The syntax of module definitions is very similar to that of procedures,
given in Chapter 5. Here is an example of a very simple module definition.
> module()
> export e1;
> local a, b;
>
> a := 2;
> b := 3;
6.1 Syntax and Semantics • 239
> e1 := x -> a^x/b^x;

> end module:
Evaluating this expression results in a module with one “export”, e1, and
two local variables, a and b.
A template for a module definition looks like:
module()
local L ;
export E ;
global G ;
options O ;
description D ;
B
end module
The simplest valid module definition is

> module() end;
module() end module
This module definition has no exported variables, no locals, no refer-

ences, and no global variables. It does not even have a body of statements.
The module to which this evaluates is not very useful.
Every module definition begins with the keyword module, followed
by an empty pair of parentheses. Following that is an optional declaration
section and the module body. The keyword combination end module (or
just end) terminates a module definition.
The Module Body

The body of a module definition consists of zero or more Maple state-
ments. The body is executed when the module definition is evaluated,
producing a module as a result. Typically, a module body consists of a
number of assignment statements that give values to the exported names
of the module. It may also include assignments to local variables, and
may, in fact, perform arbitrary computations. The body of a module may
not contain a break or next statement outside a loop, but may contain
a return statement. The effect of executing a return statement is to ter-
minate the execution of the body of the module definition at the point at
which the return statement occurs.
Module Parameters
Module definitions begin with the Maple keyword module, followed by an
(empty) pair of parentheses. This is similar to the parentheses that follow
the proc keyword in a procedure definition. Unlike procedures, however,
module definitions do not have explicit parameters. This is because, unlike
procedures (which result from the evaluation of procedure definitions),
modules are not “called” (or “invoked”) with arguments.
Instead, every module definition has an implicit parameter called
thismodule. Within the body of a module definition, this special name
evaluates to the module in which it occurs. This allows you to refer to
a module within its own definition (before the result of evaluating it has
been assigned to any name).
You have seen implicit parameters before; all procedure definitions
may reference the implicit parameters procname, args, and nargs. The
args and nargs currently have no meaning for modules, and the difference
between thismodule and procname is that procname evaluates to a
name, while thismodule evaluates to the module expression itself. This
is because the “invocation” phase of evaluating a module definition is
part of its normal evaluation, and it occurs immediately. Procedures, on
the other hand, are not invoked until called with arguments. Normally,
at least one name for a procedure is known by the time it is called; this
is not the case for modules.
Named Modules
An optional symbol may appear after the module keyword in a module
definition. Modules created with this variant on the syntax are called
named modules. Semantically, named modules are nearly identical to
normal modules, but the exported members of named modules are printed
differently, in a way that allows the module from which it was exported
to be identified visually.
> NormalModule := module() export e; end;
NormalModule := module() export e; end module
> NormalModule:-e;
> module NamedModule() export e; end module;

module NamedModule () export e; end module
> NamedModule:-e;
NamedModule : −e
When the definition of a named module is evaluated, the name (which

appears immediately after the module keyword) is assigned the module
as its value, and the name is protected. Therefore, a named module can,
ordinarily, be created only once. For example, an attempt to execute the
named module definition yields an error.
> module NamedModule() export e; end module;
Error, (in NamedModule) attempting to assign to
‘NamedModule‘ which is protected
Executing the normal module definition again creates a new in-

stance of the module, but does not result in an error. (It simply reassigns
the variable NormalModule to the new module instance.)
> NormalModule := module() export e; end;
NormalModule := module() export e; end module
It is also important (if you expect sensible results) that you never assign
a named module to another variable.
> SomeName := eval( NamedModule );
SomeName :=
module NamedModule () export e; end module
> SomeName:-e;
NamedModule : −e
Exports of named modules are printed using the distinguished name

that was given the module when it was created, regardless of any other
names that you may refer to it by.
Whether a module has a name also affects the reporting of errors
that occur during its evaluation. When the second attempt to evaluate
the named module definition above failed, the error message reported the
location of the error by name. By contrast, when an error occurs during

the evaluation of a normal module definition, the name unknown is used
instead.
> NormalModule := module() export e; error "oops"; end;
Error, (in unknown) oops
This differs from the way error reporting works with procedures.
You cannot report the name of a normal module (where, by “the name”,
we mean the name of the variable to which the module is assigned), be-
cause the evaluation of the right side of an assignment occurs before the
assignment to the name takes place. So the error occurs before any asso-
ciation between a variable and the module has occurred.
Declarations
The declarations section of the module must appear immediately after
the parentheses. All statements in the declarations section are optional,
but at most one of each kind may appear. Most module declarations are
the same as those for procedures.
Description Strings Many modules constitute a nontrivial body of soft-

ware. When you write a module, you should provide a brief description
so that users who encounter it will be able to recognize what it is or what
it does. Use the description keyword to do this, just as you would in a
procedure definition.
> Hello := module()
> description "my first module";
> export say;
> say := proc()
> print( "HELLO WORLD" )
> end proc;
> end module:
When the module is printed, its description string is displayed.
> eval( Hello );
module()
export say ;
description “my first module”;
end module
The export declaration that appears in this example is explained
later in this chapter.
Global Variables Global variables referenced within a module definition

should be declared with the global declaration. Following the keyword
global is a sequence of one or more symbols. These symbols are bound
to their global instances. In certain cases you must declare a name as a
global to prevent implicit scoping rules from making it a local variable.
> Hello := module()
> export say;
> global message;
> say := proc()
> message := "HELLO WORLD!"
> end proc;
> end module:
> message;
message
> Hello:-say();
“HELLO WORLD!”
> message;
“HELLO WORLD!”
Local Variables You can refer to variables that are local to the module
definition by using the local declaration. Its format is exactly the same
as for procedures. Here is a variant on our Hello module above which
makes (gratuitous) use of a local variable.
> Hello := module()
> local loc;
> export say;
> loc := "HELLO WORLD!";
> say := proc()
> print( loc )
> end proc;
> end module:
Local variables are not visible outside the definition of the module
in which they occur. They are “private” to the module, and are exactly
analogous to local variables of procedures.
A local variable foo in a module (or procedure) is a distinct object
from a global variable with the same name foo. Local variables are nor-
mally “short-lived” variables; the normal lifetime of a local variable is
the execution time of the body of code (a module or procedure body) to

which it is local. (Local variables may persist once execution of the scope
in which they occur has completed, but they are normally inaccessable
and will eventually be recycled by Maple’s automatic storage management
system.)
Exported Local Variables

Procedures and modules both support local variables. Only modules sup-
port exported local variables, often referred to simply as “exports”.
Module exports are declared using the export declaration. It begins
with the keyword export, after which follows a (nonempty) sequence of
symbols. A name is never exported “implicitly”; exports must be de-
clared.
The result of evaluating a module definition is a module. You can
view a module as a collection of its exports, which are also referred to as
“members” of the module. These are simply names that may (but need
not) be assigned values. You can establish initial values for the exports
by assigning to them in the body of the module definition.
The word “export” is short for “exported local variable”. In most
respects, a module export is a local variable (such as those declared via
the local declaration.) The crucial difference is that you can access the
exported local variables of a module after it has been created.
To access an export of a module, use the :- member selection opera-
tor. Its general syntax is:
modexpr :- membername
Here, modexpr must be an expression that evaluates to a module, and

membername must be the name of an export of the module to which
modexpr evaluates. Anything else signals an exception. You cannot ac-
cess local variables of an instantiated module by using this syntax.
Local variables of a procedure are created when the procedure is called
(or invoked). Normally, the locals persist only during the execution of the
statements that form the body of the procedure. Sometimes, however,
local variables persist beyond the procedure activation that instantiated
them. For example:
> gen := proc()
> local s, p;
> s := 2;
> p := x -> s * x;
> p
> end proc:
> g := gen();
g := p
> g( 3 );
The local variable s of gen persists after gen has returned. It is “cap-
tured” in the closure of the procedure p, whose name is returned by gen.
Thus, both local variables p and s of gen “escape”, but in different ways.
The local name p is accessible because it is the assigned value of the global
variable g. However, there is no way to refer to s once gen has returned.
No Maple syntax exists for that purpose. What the member selection op-
erator :- does is provide a syntax for referencing certain local variables
of modules–those declared as exports.
Our most recent Hello example has one export, named say. In this
case, say is assigned a procedure. To call it, you can type
> Hello:-say();
“HELLO WORLD!”
Whereas the following expression raises an exception, because the

name noSuchModule is not assigned a module expression.
> noSuchModule:-e;
Error, ‘noSuchModule‘ does not evaluate to a module
Here, a module expression is assigned to the name m, and the

member selection expression m:-e evaluates to the value of the exported
variable e of m.
> m := module() export e; e := 2 end module:
> m:-e;
Since m does not export a variable named noSuchExport, the following

expression raises an exception.
> m:-noSuchExport;
Error, module does not export ‘noSuchExport‘
Important! Suppose that we construct a module that exports an unas-

signed name.
> m := module() export e; end:
References to the exported name e in m evaluate to the name e.
> m:-e;
Note, however, that this is a local name e, not the global instance of
the name.
> evalb( e = m:-e );
false
The first e in the expression above refers to the global e, while the
expression m:-e evaluates to the e that is local to the module m. This
means that you can have a special version of sin, for instance, that is
private to your module, and that assigning to it will not affect the value
of the global name sin.
The Procedure exports You can determine the names of the exports
of a module by using the procedure exports.
> exports( Hello );
say
> exports( VerifyImplementation );

exports
This returns the global instances of the export names.

> exports( m );
> evalb( % = e );
true
You can also obtain the local instances of those names by passing the
option instance.
> exports( m, ’instance’ );
> evalb( % = e );
false
> evalb( %% = m:-e );
true
For this reason, you cannot have the same name declared both as a
local and an export.
> module() export e; local e; end;
Error, export and local ‘e‘ have the same name
(The declared exports and locals really form a partition of the

names that are local to a module.)
The Procedure member You have already seen the built-in procedure
member that is used to test for membership in a set or list.
> member( 4, { 1, 2, 3 } );
false
This procedure can be used for membership tests in modules as well.

> member( say, Hello );
true
> member( cry, Hello );
false
The first argument is a (global) name whose membership is to be

tested, and the second argument is a module. It returns the value true if
the module has an export whose name is the same as the first argument.
The procedure member also has a three argument form that can be
used with lists to determine the (first) position at which an item occurs.
> member( b, [ a, b, c ], ’pos’ );
true
The name pos is now assigned the value 2 because b occurs at the
second position of the list [ a, b, c].
> pos;
When used with modules, the third argument is assigned the local
instance of the name whose membership is being tested, provided that
the return value is true.
> member( say, Hello, ’which’ );
true
> which;
say
> eval( which );
proc() print(loc) end proc
If the return value from member is false, then the name remains
unassigned (or maintains its previously assigned value).
> unassign( ’which’ ):
> member( cry, Hello, ’which’ );
false
> eval( which );

which
Module Options
As with procedures, a module definition may declare options. The op-
tions that Maple recognizes as being meaningful for modules are differ-
ent from those for procedures. Only the options trace, package, and
‘Copyright...‘ are meaningful (and have the same meaning) for both
procedures and modules.
A special module option load takes the form load = thunk, where
thunk is the name of an exported or local module member. thunk must
be a procedure that is invoked when the (instantiated) module is first
created or read from a Maple repository. This option may be used for any
per-session initialization required.
Implicit Scoping Rules

The bindings of names that appear within a module definition are deter-
mined when the module definition is simplified. Module definitions are
subject to the same implicit scoping rules that procedure definitions are.
Under no circumstances is a name ever “implicitly” determined to be ex-
ported by a module; implicitly scoped names can resolve only to locals or
globals.
Lexical Scoping Rules

Module definitions, along with procedure definitions, obey standard lex-
ical scoping rules. Modules may be nested, in the sense that a module
may have any of its exports assigned to a module whose definition occurs
within the body of the outer module.
Here is a simple example of a submodule.
> m := module()
> export s;
> s := module()
> export e;
> e := proc()
> print( "HELLO WORLD!" )
> end proc;
> end module
> end module:
The global name m is assigned a module that exports the name s. Within
the body of m, the export s is assigned a module that exports the name
e. We call s a submodule of m. The Shapes package, described later,
illustrates a nontrivial use of submodules.
Modules and procedures can be mutually nested to an arbitrary depth.

The rules for the “visibility” of local variables (including exported locals
of modules) and procedure parameters are exactly the same as the rules
for nested procedures.
Parameterized Modules We have already remarked that modules do

not take explicit parameters. It could be useful, however, to be able to
write a “generic” module that could then be specialized by providing one
or more parameters.
For example, here is a module for arithmetic modulo 6.
> z6 := module()
> export add, mul;
> add := ( a, b ) -> a + b mod 6;
> mul := ( a, b ) -> a * b mod 6;
> end module:
> z6:-add( 5, 4 );
> z6:-mul( 2, 3 );
It would be nice to write a generic module for arithmetic modulo any

positive integer n, and then specialize it for each integer that you want
to use. This is already possible because of the standard lexical scoping
rules. So to do this, write a constructor procedure for the module that
accepts the value of n as an argument. Here is a generic version of the z6
example above.
> MakeZn := proc( n::posint )
> module()
> export add, mul;
> add := ( a, b ) -> a + b mod n;
> mul := ( a, b ) -> a * b mod n;
> end module
> end proc:
To generate a module that does arithmetic modulo 7, simply call the
constructor MakeZn with the number 7 as its argument.
> z7 := MakeZn( 7 );
z7 := module() export add , mul ; end module

> z7:-add( 3, 4 );
Modules and Types

Two Maple types are associated with modules. First, the name module
is a type name. Naturally, an expression is of type module only if it is a
module. When used as a type name, the name module must be enclosed
in name quotes (‘).
> type( module() end, ’‘module‘’ );
true
> type( LinearAlgebra, ’‘module‘’ );
true
Secondly, a type called moduledefinition identifies expressions that

are module definitions. In the first example above, the module definition
> module() end:
was evaluated before being passed to type, so the expression that was
tested was not the definition, but the module to which it evaluates. We
must use unevaluation quotes (’) to delay the evaluation of a module
definition.
> type( ’module() end’, ’moduledefinition’ );
true
Other important type tests satisfied by modules are the types atomic
and last_name_eval.
> type( module() end, ’atomic’ );
true
The procedure map has no effect on modules; they pass through un-
changed.
> map( print, module() export a, b, c; end );
module() export a, b, c; end module
Modules also suffer last name evaluation rules.

> m := module() end:
> m;
> type( m, ’last_name_eval’ );
true
Although type module is a surface type, it acts also as a structured

type. Parameters passed as arguments to the unevaluated name module
are taken to be the names of exports. For example, the module
> m := module() export a, b; end:
has type structured module type ‘module‘( a, b ):

> type( m, ’‘module‘( a, b )’ );
true
It also has type type ‘module‘( a )

> type( m, ’‘module‘( a )’ );
true
because any module that exports symbols a and b is a module that

exports the symbol a.
Example: A Symbolic Differentiator

In this section we will illustrate the various module concepts discussed
so far by writing a simple symbolic differentiator. Maple already pro-
vides a built-in differentiator diff, so we will call our differentia-
tor differentiate. Its (final) implementation is hidden in a module
DiffImpl, which holds all the local states for the program. Much of
the code for the differentiator is designed to implement either a stan-

dard “rule” (such as the rule that the derivative of a sum is the sum of
the derivatives of the summands), or special case rules for mathematical
functions such as sin and exp. Our simple differentiator will handle only
real valued functions of a single real variable.
We will show several steps in the development of the module, from
a very simple “first try” to the final, fully functional program. The final
form of the differentiator is a good illustration of a very common Maple
“design pattern”. This pattern arises when you have a single top-level
routine that has to dispatch to a number of subroutines to handle special
cases using special purpose algorithms.
The First Attempt Initially, we will present the differentiator as an

ordinary procedure, not yet involving modules. The first version of the
differentiator is shown below.
> differentiate := proc( expr, var )
> local a, b;
>
> if type( expr, ’constant’ ) then
> 0
> elif expr = var then
> 1
> elif type( expr, ’‘+‘’ ) then
> map( procname, args )
> elif type( expr, ’‘^‘’ ) then
> a, b := op( expr );
> if a = var and not has( b, var ) then
> b * a ^ ( b - 1 )
> else
> ’procname( args )’
> end if
> elif type( expr, ’‘*‘’ ) then
> a, b := op( 1, expr ), subsop( 1 = 1, expr );
> procname( a, var ) * b + a * procname( b, var )
> else
> end if
> end proc:
Trivial cases are handled first: The derivative of a constant expression is
equal to 0, and the derivative of the variable with respect to which we are
differentiating is equal to 1. The additivity of the derivative operator is
expressed by mapping the procedure over sums, using the idiom
> map( procname, args );
This is commonly used to map a procedure over its first argument,

passing along all the remaining arguments. Only the simple case of powers
of the differentiation variable is handled so far, provided also that the

power is independent of the differentiation variable. The product rule for
derivatives is expressed by splitting expressions of type product into two
“pieces”:
• the first factor in the product, and
• the product of all the remaining factors.
This is achieved by the double assignment of

so the input expression expr is expressed as expr = a * b. The stan-

dard technique of returning “unevaluated” is used so that computation
may proceed symbolically on expressions that the differentiator is unable
to do anything meaningful with.
This first version is quite simple, but it is already able to handle
polynomials with numeric coefficients.
> differentiate( 2 - x + x^2 + 3*x^9, x );
−1 + 2 x + 27 x8
However, it fails on expressions containing calls to standard mathe-

matical functions.
> differentiate( sin( x ), x );
differentiate(sin(x), x)
It is also unable to deal successfully with symbolic coefficients.

> differentiate( a*x^2 + b*x + c, x );
differentiate(a, x) x2 + 2 a x + differentiate(b, x) x + b
+ differentiate(c, x)
Adding Missing Functionality To add the missing functionality, we

can add a case for expressions of type function.

> local a, b;
>
> if not has( expr, var ) then
> 0
> 1
> a, b := op( expr );
> if not has( b, var ) then
> b * a ^ ( b - 1 ) * procname( a, var )
> else
> end if
> elif type( expr, ’function’ ) and nops( expr ) = 1 then
> # functions of a single variable; chain rule
> b := op( 0, expr ); # the name of the function
> a := op( 1, expr ); # the argument
> if b = ’sin’ then
> cos( a ) * procname( a, var )
> elif b = ’cos’ then
> -sin( a ) * procname( a, var )
> elif b = ’exp’ then
> exp( a ) * procname( a, var )
> elif b = ’ln’ then
> ( 1 / a ) * procname( a, var )
> else
> end if
> else
> end if
> end proc:
This uses the chain rule to compute the derivatives of calls to “known”
functions.
> differentiate( sin( x ) + cos( exp( x ) ), x );
cos(x) − sin(ex ) ex
> differentiate( sin( x^2 ) + cos( x^2 ), x );
2 cos(x2 ) x − 2 sin(x2 ) x
> differentiate( sin( x )^2 + cos( x )^3, x );

2 sin(x) cos(x) − 3 cos(x)2 sin(x)
At the same time, we have also improved the handling of expressions

independent of the variable of differentiation.
> differentiate( a*x^2 + b*x + c, x );
2ax + b
This is effected by using the expression has( expr, var ) instead

of the weaker test type( expr, ’constant’ ). The “power rule” now
handles more than just powers of var.
> differentiate( sin( x )^2, x );
2 sin(x) cos(x)
However, adding new functions to our differentiator is tedious and

error-prone, and the job of handling the chain rule has to be repeated for
each function “known” to it.
Introducing a Function Table A better way to deal with the many

functions that we should eventually want to add is to store them, and the
rules used for their differentiation, in a table.
> local a, b, functab;
>
> functab := table();
> functab[ ’sin’ ] := ’cos’;
> functab[ ’cos’ ] := x -> -sin( x );
> functab[ ’exp’ ] := exp;
> functab[ ’ln’ ] := x -> 1 / x;
>
> 0
> 1
> a, b := op( expr );
> if a = var and not has( b, var ) then
> b * a ^ ( b - 1 ) * procname( a, var )
> else
> end if
> # functions of a single variable; chain rule
> b := op( 0, expr ); # the name of the function
> a := op( 1, expr ); # the argument
> if assigned( functab[ b ] ) then
> # This is a ‘‘known’’ function
> functab[ b ]( a ) * procname( a, var )
> else
> # This function is not known; return unevaluated
> end if
> else
> end if
> end proc:
This not only simplifies the code used for the function case, but also
makes it very easy to add new functions.
Unfortunately, this implementation suffers from a serious drawback.
It is not extensible. The known functions are hardcoded as part of the
procedure definition for differentiate.
New functions cannot be added without editing this source code.
A second problem relates to performance. A complete implementation
would require a table of dozens or hundreds of functions. That large table
would have to be created and initialized each time differentiate was
invoked.
Encapsulation and Extensibility One way to fix both problems is

to make the table of functions a global variable. However, using global
variables can be dangerous, because they pollute the user namespace and
are subject to unwanted inspection and tampering.
A better solution is to put the differentiate procedure, along with
its table of functions, into a module. The table is then initialized only
once–when the module is created–and can be saved to a Maple repository
along with the rest of the module by using a savelib call. By making the
table a local variable of the module, we prevent users from modifying the
table, or otherwise inspecting it in unwanted ways.
This does not prevent us from making the differentiator user-extensible,
however. We can add an access procedure addFunc that allows users to
add their own rules for differentiating new functions. For instance, we can
use the call
> addFunc( ’cos’, x -> -sin(x) );
to add the derivative of the sin function. The export addFunc of the
DiffImpl module is a procedure that requires two arguments. The first
is the name of a function whose derivative is to be made known to the
differentiator. The second is a Maple procedure of one argument that
expresses the derivative of the function being added.
With this strategy in mind, we will create a module DiffImpl, with
principal export differentiate. At the same time, we will also take the
opportunity to make the basic differentiation rules extensible.
Here is the complete source code for the differentiator with these im-
provements.
> DiffImpl := module()
> description "a symbolic differentiator";
> local functab, ruletab, diffPower;
> export differentiate, addFunc, addRule, rule;
>
> addFunc := proc( fname::symbol, impl )
> functab[ fname ] := impl
> end proc;
>
> addRule := proc( T, impl )
> if type( T, ’{ set, list }’ ) then
> elif type( T, ’And( name, type )’ ) then
> ruletab[ T ] := impl
> else
> error "expecting a type name, but got %1", T
> end if
> end proc;
>
> rule := proc( T )
> if type( T, ’And( name, type )’ ) then
> if assigned( ruletab[ T ] ) then
> eval( ruletab[ T ], 1 )
> else
> error "no rule for expressions of type %1", T
> end if
> else
> error "expecting a type symbol, but got %1", T
> end if
> end proc;
>
> local a, b, e;
> 0
> 1
> e := op( 0, expr );
> a := op( expr );

> if assigned( functab[ e ] ) then
> functab[ e ]( a ) * procname( a, var )
> else
> end if
> else
> b := whattype( expr );
> if assigned( ruletab[ b ] ) then
> ruletab[ b ]( expr, var )
> else
> end if
> end if
> end proc;
>
> addRule( ’{list,set,tabular}’,
> () -> map( differentiate, args ) );
> addRule( ’‘+‘’,
> addRule( ’‘*‘’,
> (expr,var) ->
> op(1,expr)*differentiate(subsop(1=1,expr),var)
> + differentiate(op(1,expr),var)*subsop(1=1,expr) );
> diffPower := proc( expr, var )
> local b, e;
> Assert( type( expr, ’‘^‘’ ) );
> b, e := op( expr );
> if has( e, var ) then
> expr * ( differentiate( e, var ) * ln( b )
> + e * differentiate( b, var ) / b )
> else # simpler formula
> e * b^(e - 1) * differentiate( b, var )
> end if;
> end proc;
> addRule( ’‘^‘’, eval( diffPower ) );
>
> addFunc( ’sin’, cos );
> addFunc( ’cos’, x -> -sin(x) );
> addFunc( ’exp’, exp );
> addFunc( ’ln’, x -> 1/x );
> # ... etc.
>
> end module:
> differentiate := DiffImpl:-differentiate:
To give our set of rules for nonfunctional expressions a similar kind of
extensibility, we will store those rules in a table as well. The table will be
indexed by the primary (or basic) type name for the expression type, as
given by the Maple procedure whattype.
> whattype( a + 2 );
> whattype( a / b );
> whattype( a^sqrt(2) );
> whattype( [ f( x ), g( x ) ] );
list
A “rule” is expressed by a procedure of two arguments, expr and

var, in which expr is the expression to be differentiated, and var is the
variable of differentiation. For instance, we would like the differentiator
to handle items such as sets and lists by differentiating their individual
components. So we add the rule
> addRule( ’{ list, set, tabular }’, () -> map( differentiate,
> args ) );
The first version of our differentiator dealt with sums by mapping

itself over the sum expression. In the new scheme, this is expressed by the
statement
> addRule( ’‘+‘’, () -> map( differentiate, args ) );
in the module body. Now, the advantage of using this scheme is that,
not only can the author of the differentiator extend the system, but so
can users of the system. Having instantiated the module DiffImpl, any
user can add rules or new functions, simply by issuing appropriate calls
to addRule and addFunc.
The differentiator cannot handle the procedure tan
> differentiate( tan( x )/exp( x ), x );
tan(x) differentiate(tan(x), x)
− +
ex ex
so we will add it to the database of known functions.

> DiffImpl:-addFunc( ’tan’, x -> 1 + tan(x)^2 );
x → 1 + tan(x)2
> differentiate( tan( x )/exp( x ), x );
tan(x) 1 + tan(x)2
− +
ex ex
Similarly, there is not yet any rule for handling equations and other
relations.
> differentiate( y( x ) = sin( x^2 ) - cos( x^3 ), x );
differentiate(y(x) = sin(x2 ) − cos(x3 ), x)
> DiffImpl:-addRule( ’{ ‘=‘, ‘<‘, ‘<=‘ }’,

{() → map(differentiate, args)}
> differentiate( y( x ) = sin( x^2 ) - cos( x^3 ), x );
differentiate(y(x), x) = 2 cos(x2 ) x + 3 sin(x3 ) x2
The Extension Mechanism is “Module AwareÔ Take care not

to confuse the extension mechanism proposed above for our differentia-
tor with the extension mechanism used by the built-in Maple command
diff. diff uses a traditional string concatenation mechanism for adding
knowledge of the derivatives of functions, and all its rules are built-in,
so they cannot be extended. For instance, to add a new function F to
Maple’s built-in diff command, you can define a procedure ‘diff/F‘
that computes the derivative of F.
By contrast, the extension mechanism used in our differentiate
example is “module aware”. Suppose that you want to add knowledge
of the derivative of some top-level function F. You can simply issue a
command, such as
> DiffImpl:-addFunc( ’F’, x -> sin( x ) + cos( x ) );
x → sin(x) + cos(x)
(We are supposing that, in fact, the derivative of F( x ) is sin( x

) + cos( x ).) Now suppose that you have defined a module with some
special functions, one of which is also called F.
> SpecFuncs := module()
> export F; # etc.
> # definition of F() and others
> end module:
You can now add this new F to the known functions.
> DiffImpl:-addFunc( SpecFuncs:-F, x -> exp( 2 * x ) );
x → e(2 x)
> differentiate( F( x ), x );
sin(x) + cos(x)
> use SpecFuncs in

> differentiate( F( x ), x );
> end use;
e(2 x)
With the traditional mechanism, this does not work.

> ‘diff/‘ || F := x -> sin( x ) + cos( x );
diff /F := x → sin(x) + cos(x)
> diff( F( x ), x );
sin(x) + cos(x)
> use SpecFuncs in

> ‘diff/‘ || F := x -> exp( 2 * x );
> diff( F( x ), x );
> end use;
6.2 Records • 263
diff /F := x → e(2 x)
e(2 x)
The definition for the global F has been lost.

> diff( F( 2 * x ), x );
e(4 x)
(We use a different argument to diff to avoid recalling the answer

from its remember table.) The traditional mechanism fails because it relies
on the external representation of names, and not upon their bindings,
so each attempt to define an extension to diff in fact added a definition
for the derivative of all functions whose names are spelled "F".
A commented version of the differentiator module is available in the
sample source code of your Maple installation. The implementation shown
in the text has been somewhat simplified.
6.2 Records
The simplest way in which to use modules is as Pascal-style records (or
“structures”, as in C and C++). A record is a data structure that has
some number of named “slots” or “fields”. In Maple, these slots can be
assigned arbitrary values.
Instantiating Records To create a record, use the Record constructor.

It takes the slot names as arguments.
> rec := Record( ’a’, ’b’, ’c’ );
rec :=
module() export a, b, c; option record ; end module
The name rec is now assigned a record with slots named a, b, and c.
These are the slot names for the record rec. You can access, and assign,
these slots by using the expressions rec:-a, rec:-b, and rec:-c.
> rec:-a := 2;
a := 2
> rec:-a;
If not assigned, the record slot evaluates to the local instance of the
slot name.
> rec:-b;
> evalb( % = b );
false
The usefulness of this is that the entire record may be passed around
as an aggregate data structure.
The record constructor accepts initializers for record slots. That is,
you may specify an initial value for any among the slots of a newly created
record by passing an equation with the slot name on the left side and the
initial value on the right.
> r := Record( ’a’ = 2, ’b’ = sqrt( 3 ) );
r := module() export a, b; option record ; end module
> r:-b;
√
3
Record Types Expressions created with the Record constructor are of

type record.
> type( rec, ’record’ );
true
This is a structured type that works the same way that type ‘module‘
does but recognizes records specifically.
6.2 Records • 265
Using Records to Represent Quaternions Records are useful for

implementing simple aggregate data structures for which named access
to slots is wanted. For example, four real numbers can be combined to
form a quaternion, and we can represent this using a record structure, as
follows.
> MakeQuaternion := proc( a, b, c, d )
> Record( ’re’ = a, ’i’ = b, ’j’ = c, ’k’ = d )
> end proc:
> z := MakeQuaternion( 2, 3, 2, sqrt( 5 ) );
z := module()
export re, i, j, k;
option record ;
end module
√
In this example, z represents the quaternion 2 + 3i + 2j + 5k (where
i, j, and k are the nonreal quaternion basis units). The quaternion records
can now be manipulated as single quantities. The following procedure ac-
cepts a quaternion record as its sole argument and computes the Euclidean
length of the quaternion that the record represents.
> qnorm := proc( q )
> use re = q:-re, i = q:-i, j = q:-j, k = q:-k in
> sqrt( re * re + i * i + j * j + k * k )
> end use
> end proc:
> qnorm( z );
√
22
A Maple type for quaternions can be introduced as a structured record

type.
> ‘type/quaternion‘ := ’record( re, i, j, k )’:
> type( z, ’quaternion’ );
true
6.3 Packages
Modules are ideal for writing Maple packages. They provide facilities that
are better equipped for large software projects than are the older table-
and procedure-based methods.
What Is a Package?
A package is a collection of procedures, and other data, that can be treated
as a whole. Packages typically gather a number of procedures that enable
you to perform computations in some well-defined problem domain. Pack-
ages may contain data other than procedures, and may even contain other
packages (“subpackages”).
Packages in the Standard Library A number of packages are

shipped with the standard Maple library. For example, the group,
numtheory, codegen, and LinearAlgebra packages are all provided with
Maple, along with several dozen others. The group package provides pro-
cedures that allow you to compute with groups that have a finite represen-
tation in terms of permutations, or of generators and defining relations.
The LinearAlgebra package has a large number of procedures available
for computational linear algebra.
Table-Based Packages Many packages are implemented as tables.

The essential idea underlying this implementation scheme is that the
name of a package routine is used as the index into a table of proce-
dures. The table itself is the concrete representation of the package.
Use Modules for New Packages Modules are the new implementa-
tion vehicle for packages. A module represents a package by its exported
names. The exported names can be assigned arbitrary Maple expressions,
typically procedures, and these names form the package.
Package Exports Some of the data in a package is normally made

accessible to the user as an “export” of the package. For packages imple-
mented as modules, the package exports are the same as the exports of
the underlying module. For packages implemented as tables, the package
exports are the names used to index the underlying table.
Accessing the exports of a package is a fundamental operation that is
supported by all packages. If P is a Maple package, and e is one among
its exports, you can access e by using the fully qualified reference P[ e ].
If it happens that P is a module, then you can also use the syntax P:-e.
6.3 Packages • 267
These methods of accessing the exports of a module are normally used

when programming with a package.
Note that :- is a left-associative operator. If S is a submodule of a
module P, and the name e is exported by S, then the notation P:-S:-e is
parsed as (P:-S):-e, and so it refers to the instance of e local to S. This
fact is important to reference members of subpackages. For example,
> m := Matrix(2,2,[[1-x,2-x],[3-x,4-x]],
> ’datatype’ = ’polynom(integer)’ );
´ µ
1−x 2−x
m :=
3−x 4−x
> LinearAlgebra:-LA_Main:-Norm( m, 1, conjugate = false );
max(|x − 1| + |x − 3| , |x − 2| + |x − 4|)
calls the procedure Norm in the subpackage LA_Main of the LinearAlgebra

package. (You cannot use indexed notation for this.)
Using Packages Interactively For interactive use, it is inconvenient

to have to type fully qualified references to all the exports of a package. To
ease this burden, the Maple procedure with is provided for the interactive
management of package namespaces. Using with, you can globally impose
the exported names of a package. This allows you to access the package
exports, without typing the package prefix, by making the names of the
exports visible at the top-level of your Maple session. For example, to use
the numtheory package, you can issue the command
> with( numtheory );
Warning, the protected name order has been redefined
and unprotected
[GIgcd , bigomega, cfrac, cfracpol , cyclotomic, divisors ,

factorEQ , factorset , fermat , imagunit , index ,
integral _basis , invcfrac, invphi , issqrfree , jacobi ,
kronecker , λ, legendre, mcombine, mersenne,
minkowski , mipolys, mlog , mobius , mroot , msqrt ,
nearestp, nthconver , nthdenom, nthnumer , nthpow ,
order , pdexpand , φ, π, pprimroot , primroot , quadres ,
rootsunity , safeprime, σ, sq2factor , sum2sqr , τ, thue]
The effect of this command is to make the names exported by tne
numtheory package (a list of which is returned by the call to with) avail-
able temporarily as top-level Maple commands.
> cfrac( ( 1 + x )^k, x, 5, ’subdiagonal’, ’simregular’ );
1
kx
1−
1 (k + 1) x
1+
2 1 (k − 1) x
1−
6 1 (k + 2) x
1+
6 1 + ...
In this section, we are concerned primarily with how to write Maple

packages by using modules. The following subsections present several ex-
amples that illustrate how to do this.
Example: The LinkedList Package

The first example package is a small package called LinkedList. This
example illustrates the basic structure of a package implemented by using
modules.
Background Linked lists are a basic data structure used in programs

for many different purposes. There are many different kinds of linked lists,
with variations on the basic idea intended to deal with various issues
involving performance and functionality. The example package shown in
this subsection provides a few operations on the simplest possible form of
linked lists.
The “links” in a linked list are formed from a very simple data struc-
tured called a “pair”. A pair is essentially a container with space for
exactly two elements. Pairs can be modeled by fixed length records with
two slots. When used to implement linked lists, the first slot holds the
data for the list entry, and the second slot stores a pointer to the next
pair in the list.
The LinkedList package implements an abstract data definition for
the pair data structure, and adds some higher level operations on pairs
to effect the list abstraction. A linked list is effectively represented by its
first pair.
The “pair” abstract data structure is very simple. It consists of a
constructor “pair”, and two accessors called “head” and “tail” that satisfy
the algebraic specification
p = pair(head(p), tail(p))
for each pair p. In addition, there is a distinguished “pair” nil, satisfying

this algebraic relation, that is unequal to any other pair, and satisfies
head(nil) = nil, tail(nil) = nil.
Note that linked lists are quite different from Maple’s builtin list struc-
tures, which are really immutable arrays. Linked lists are best suited for
applications in which you want to incrementally build up the list from its
members.1
Package Implementation The LinkedList package is implemented

as a module containing the primitive operations on pairs, and higher level
operations that implement the list abstraction.
> macro( _PAIR = ‘‘ ): # for nice printing
> LinkedList := module()
> description "routines for simple linked lists";
> export
> nil,
> nullp,
> pair,
> head,
> tail,
> list,
> length,
> member,
> reverse,
> append,
> map;
> local
1
Lisp programmers will recognise the pair, head and tail operations as the more
traditional operations known as “consÔ, “carÔ and “cdrÔ.
> setup,
> cleanup,
> map1,
> reverse1,
> _PAIR;
> option
> package,
> load = setup,
> unload = cleanup;
>
> setup := proc()
> global ‘type/Pair‘, ‘type/LinkedList‘;
> ‘type/Pair‘ := ’{ _PAIR( anything, anything ),
> identical( nil ) }’;
> ‘type/LinkedList‘ := proc( expr )
> if expr = nil then
> true
> elif type( expr, Pair ) then
> type( tail( expr ), ’LinkedList’ )
> else
> false
> end if
> end proc;
> userinfo( 1, ’LinkedList’,
> "new types ‘Pair’ and ‘LinkedList’ defined" );
> NULL
> end proc;
>
> cleanup := proc()
> global ‘type/Pair‘, ‘type/LinkedList‘;
> userinfo( 1, ’LinkedList’,
> "cleaning up global types" );
> ‘type/Pair‘ := evaln( ‘type/Pair‘ );
> ‘type/LinkedList‘ := evaln( ‘type/LinkedList‘ );
> NULL
> end proc;
>
> pair := ( a, b )
> -> setattribute( ’_PAIR’( a, b ), ’inert’ );
> head := ( c::Pair )
> -> ‘if‘( c = nil, nil, op( 1, c ) );
> tail := ( c::Pair )
> -> ‘if‘( c = nil, nil, op( 2, c ) );
> nullp := ( pair )
> -> evalb( pair = nil );
>
> list := proc()
> local a, L;
> L := nil;
> for a in args do
> L := pair( a, L )
> end do
> end proc;
>
> length := proc( lst )

> if nullp( lst ) then
> 0
> else
> 1 + length( tail( lst ) )
> end if
> end proc;
>
> member := proc( item, lst )
> false
> elif item = head( lst ) then
> true
> else
> procname( item, tail( lst ) )
> end if
> end proc;
>
> map := proc( p, lst )
> nil
> else
> pair( p( head( lst ) ),
> procname( p, tail( lst ) ) )
> end if
> end proc;
>
> append := proc( lst1, lst2 )
> if nullp( lst1 ) then
> lst2
> else
> pair( head( lst1 ),
> procname( tail( lst1 ), lst2 ) )
> end if
> end proc;
>
> reverse1 := proc( sofar, todo )
> if nullp( todo ) then
> sofar
> else
> procname( pair( head( todo ), sofar ),
> tail( todo ) )
> end if
> end proc;
>
> reverse := lst -> reverse1( nil, lst );
>
> setup();
>
> end module:
Normally, a package definition like this would be entered into a Maple
source file using a text editor, or in a worksheet using Maple’s GUI inter-
face. In either case, the definition would then be followed by a call to the
savelib procedure using the name of the module as its sole argument:
> savelib( ’LinkedList’ );
Evaluating the savelib call saves the module to the first repository
found in the global variable libname, or the repository named with the
global variable savelibname, if it is defined. (At least one of these must
be defined.) You should always ensure that your standard Maple
library is write-protected to avoid saving your own expressions
in it. If you accidentally save something to the standard Maple library,
you may need to restore the original from the media on which you obtained
your copy of Maple.
The package exports are listed as the exports of the module. A few
local variables are used to implement the package. The local procedures
map1 and reverse1 are part of the package implementation that is not
available to users of the package. They are visible only within the module
definition. This allows the package author to make improvements to the
package without disturbing any code that uses it. If the local procedures
reverse1 and map1 were exported (thus, available to users), it would be
difficult for the author to replace these routines without breaking existing
code that relies upon them.
The package includes two special (local) procedures, setup and
cleanup. These are executed, respectively, when the module is first read
from a repository, and when the package is either garbage collected or
when Maple is about to exit.
Using the Package The package exports can always be accessed by

using the long form of their names.
> LinkedList:-pair( a, b );
(a, b)
For consistency with the older table-based package implementations,

an indexed notation may also be used.
> LinkedList[ ’pair’ ]( a, b );
(a, b)
This form requires that the index (in this case, the symbol pair) be
protected from evaluation, and the notation does not extend to packages
with nested subpackages.
To access the package exports interactively, use the with command.
> with( LinkedList );
Warning, the protected names length, map and member
have been redefined and unprotected
[append , head , length, list , map, member , nil , nullp, pair ,

reverse, tail ]
Note that, since some of the package exports shadow global procedures
with the same name, with issues warnings. These warnings are normal and
merely serve to remind you that these names will now refer to expressions
other than the expressions to which they referred prior to issuing the
command. Once the exports of the package LinkedList have been bound,
you can call them as you would global Maple routines with those names.
Note that you can still access the global version of member, for example,
by using the syntax :-member.
> use LinkedList in
> member( a, %% );
> :-member( a, [ a, b, c, d ] )
> end use;
true
true
This is one of the principal advantages of using modules and binding,

rather than assignment, to implement packages.
Lists are either built up incrementally using the pair export of the
package, or by calling the list export.
> L := nil:
> L := pair( i, L )
> end do;
L := (1, nil )
L := (2, (1, nil ))

L := (3, (2, (1, nil )))
L := (4, (3, (2, (1, nil ))))
L := (5, (4, (3, (2, (1, nil )))))
L := (6, (5, (4, (3, (2, (1, nil ))))))
L := (7, (6, (5, (4, (3, (2, (1, nil )))))))
L := (8, (7, (6, (5, (4, (3, (2, (1, nil ))))))))
L := (9, (8, (7, (6, (5, (4, (3, (2, (1, nil )))))))))
L := (10, (9, (8, (7, (6, (5, (4, (3, (2, (1, nil ))))))))))
> length( L );
10
> member( 3, L );
true
> member( 100, L );
false
> reverse( L );
(1, (2, (3, (4, (5, (6, (7, (8, (9, (10, nil ))))))))))
> map( x -> x^2, L );
(100, (81, (64, (49, (36, (25, (16, (9, (4, (1, nil ))))))))))
> member( 100, % );
true
> L2 := list( a, b, c, d );
L2 := (d, (c, (b, (a, nil ))))
> map( sin, L2 );
(sin(d), (sin(c), (sin(b), (sin(a), nil ))))
> eval( L2, { a = 1, b = 2, c = 3, d = 4 } );
(4, (3, (2, (1, nil ))))
> map( evalf[ 10 ], % );
(4., (3., (2., (1., nil ))))
The Maple source code for this package is available in the ‘samples’
directory of your Maple installation as the file ‘samples/ch06/ll.mpl’.
The source code in the ‘samples’ directory may differ slightly from that
shown in this book.
Example: A Code Coverage Profiling Package

Our next example is a package called coverage. It instruments procedures
and modules for coverage profiling. Besides serving as an example of a
small package, it also illustrates some of the ways in which modules can
be manipulated.
Design When you have written some Maple code, you will want to
write tests that exercise each part of the program to ensure that it works
correctly, and that it continues to work when it, or other programs on
which it depends, change over time. It is important to be able to determine
whether each statement in a procedure is executed by some test case.
The traceproc option of the Maple command debugopts provides that
capability. It takes the name p of a procedure, using the syntax
debugopts( ’traceproc’ = p );
and instruments the procedure assigned to p for coverage profiling. Here

is an example.
> p := proc( x )
> if x < 0 then
> 2 * x
> else
> 1 + 2 * x
> end if
> end proc:
> debugopts( ’traceproc’ = p ):
Once the procedure has been instrumented, then each time it is executed,
profiling information at the statement level is stored. To view the profiling
information, use the procedure showstat.
> p( 2 );
> showstat( p );
p := proc(x)
|Calls Seconds Words|
PROC | 1 0.000 12|
1 | 1 0.000 12| if x < 0 then
2 | 0 0.000 0| 2*x
else
3 | 1 0.000 0| 1+2*x
end if
end proc
The display shows that only one branch of the if statement that
forms the body of p was taken so far. This is because only a non-negative
argument has been supplied as an argument to p. To get complete cover-
age, a negative argument must also be supplied.
> p( -1 );
−2
> showstat( p );
p := proc(x)
PROC | 2 0.000 24|
1 | 2 0.000 24| if x < 0 then
2 | 1 0.000 0| 2*x
else
3 | 1 0.000 0| 1+2*x
end if
end proc
Now the display shows that each statement in the body of p has
been reached.
The profiling information is stored in an rtable assigned to a name
that is formed by concatenating the name TRACE with the name of the
procedure (the one used in the call to debugopts), separated by a /
character.
> eval( ‘TRACE/p‘ );
 
2 0 24
2 0 24 
 
1 0 0
1 0 0
The little package illustrated in this section helps to extend this func-
tionality to modules, and acts as a front end to the debugopts with the
traceproc option.
The coverage package has two exports: profile and covered. Two
private procedures, rprofile and traced, are used as subroutines. They
are stored in local variables of the underlying module of the package.
The Package Source Here is the source code for the package.
> coverage := module()
> description "a package of utilities for "
> "code coverage profiling";
> option package;
> export profile, covered;
> local rprofile, traced;
>
> # Instrument a procedure or module
> # for coverage profiling. Return the
> # number of procedures instrumented.
> profile := proc()
> local arg;
> add( rprofile( arg ), arg = [ args ] )

> end proc;
>
> rprofile := proc( s::name )
> local e;
> if type( s, ’procedure’ ) then
> debugopts( ’traceproc’ = s );
> 1
> elif type( s, ’‘module‘’ ) then
> add( procname( e ),
> e = select( type,
> [ exports( s, ’instance’ ) ],
> ’{ ‘module‘, procedure }’ ) )
> else
> error "only procedures and modules can be profiled"
> end if
> end proc;
>
> # Subroutine to recognize an rtable that
> # is used to store profiling information,
> # based on its name.
> traced := proc( s )
> evalb( substring( convert( s, ’string’ ),
> 1 .. 6 ) = "TRACE/" )
> end proc;
>
> # Determine which procedures have
> # coverage information.
> covered := proc()
> local S;
> S := [ anames( ’rtable’ ) ];
> S := select( traced, S );
> if nargs > 0 and args[ 1 ] = ’nonzero’ then
> S := select( s -> evalb( s[1,1] <> 0 ), S )
> elif nargs > 0 then
> error "optional argument is the name nonzero"
> end if;
> map( parse, map( substring,
> map( convert, S, ’string’ ), 7 .. -1 ) )
> end proc;
> end module:
How the Package Works The export profile is an interface to the

package’s principal facility: instrumenting procedures and modules for
coverage profiling. It returns the number of procedures instrumented, and
calls the private subroutine rprofile to do most of the work.
The procedure rprofile accepts a name s as an argument. If s is the
name of a procedure, rprofile simply calls debugopts to instrument the
procedure assigned to that name. Otherwise, if s is the name of a module,
rprofile picks out any exports of the module that are procedures or
modules and calls itself recursively to instrument them. If the parameter

s is assigned a value of any other type, then an exception is raised.
The expression [ exports( s, ’instance’ ) ] evaluates to a list
of all the exported variables of the module that are assigned to s. It is
important to pass the instance option to exports, because when those
names are passed to rprofile in a recursive call, rprofile must test the
type of their assigned values. This list contains all the module exports,
so those that are of type procedure, or of type module, are picked out by
using a call to select. The recursion is effected in the call to add, which
sums the return values of all the recursive calls to rprofile.
The exported procedure covered is used to determine which pro-
cedures have been instrumented and called, with profiling information
stored. One possible design would store this information in a private ta-
ble in the coverage package. With this design, covered could simply
query that internal table for the names of the procedures that have been
instrumented and that have profiling information stored. However, a user
may have instrumented the procedure “manually” by calling debugopts
directly, or historical profiling data may have been read from a Maple
repository. Therefore, we have adopted a design that queries the system
directly, without regard to how a procedure came to be instrumented
initially.
The procedure covered queries Maple for all the names currently
assigned values of type rtable, using the Maple command anames (“as-
signed names”). Names beginning with the character string "TRACE/" are
selected, by using the subroutine traced, as there may be other rtables in
the system. If the nonzero option is passed to covered, then only those
which have actually been called are chosen. The final statement
> map( parse, map( substring,
> map( convert, S, ’string’ ),
> 7 .. -1 ) )
first converts the names to strings, then removes the "TRACE/" prefix
by forming the substring from the seventh position to the end of the string,
and finally calls parse on each string to convert it to the procedure for
which profiling data is stored.
Using the Package As with all packages, you can access the coverage
package interactively by using the with command.
> with( coverage );
Warning, the protected name profile has been redefined
and unprotected
[covered , profile]
A list of the package exports is returned. Alternatively, the package ex-

ports can always be accessed by using the long forms coverage:-profile
and coverage:-covered.
Suppose that we want to test the procedure copy (chosen because it
is short). This procedure produces a new copy of a table, array, or rtable.
Now that the coverage package has been globally imposed by using with,
we can simply call
> profile( copy );
The return value of 1 indicates that, as expected, one procedure was

instrumented. Next we call copy with a few arguments (output sup-
pressed):
> copy( table() ):
> copy( array( 1 .. 3 ) ):
Using covered, we see that copy has indeed had its profiling information
stored.
> covered( ’nonzero’ );
[copy ]
From the output of showstat,

> showstat( copy );
copy := proc(A)
PROC | 2 0.000 640|
1 | 2 0.000 640| if type(A,rtable) then
2 | 0 0.000 0| return rtable(rtable_indfns(A),
rtable_dims(A),A,rtable_options(A),
readonly = false)
elif type(A,{array, table}) then
3 | 2 0.000 582| if type(A,name) then
4 | 0 0.000 0| return map(proc () args end proc,
eval(A))
else
5 | 2 0.000 574| return map(proc () args end proc,A)
end if
else
6 | 0 0.000 0| return A
end if
end proc
we see that we have missed the rtable case (statement 2), so we

add a test for that.
> copy( rtable() ):
> showstat( copy );
copy := proc(A)
PROC | 3 0.000 828|
readonly = false)
eval(A))
else
end if
else
6 | 0 0.000 0| return A
end if
end proc
Statement 4 is still missing. This statement can be reached by

assigning an array or table to a name and by calling copy with that name
as argument.
> t := table():
> copy( t ):
> showstat( copy );
copy := proc(A)
PROC | 4 0.000 1127|
readonly = false)
eval(A))
else

end if
else
6 | 0 0.000 0| return A
end if
end proc
The only missing case now is the one in which the argument to
copy is something other than an rtable, array, or table.
> copy( 2 ):
> showstat( copy );
copy := proc(A)
PROC | 5 0.000 1153|
readonly = false)
eval(A))
else
end if
else
6 | 1 0.000 10| return A
end if
end proc
The final output shows that every statement has been reached by
our test cases. This functionality is very useful for interactively developing
unit tests for Maple programs.
The source presented here for the coverage package has been simpli-
fied for presentation in printed form. The full source code is available in
the samples directory of your Maple installation.
Example: The Shapes Package

Modules permit the construction of packages with hierarchical structure.
This cannot be done with table-based implementations of packages.
This section presents a detailed look at how you can organize your
source code for a (potentially) large package that has nontrivial substruc-
ture. A package Shapes is described, and the details of its design and
implementation are provided. We also give some “hints” related to source

code organization.
The mathematical functionality of this package is trivial. It provides
the means to compute areas and circumferences of various planar figures,
which are called “shapes”.
Only portions of the source code for this package are shown here. The
fully commented source code can be found in the samples directory of
your Maple installation.
Source Code Organization The Shapes package is organized into sev-

eral source files:
shapes.mpl
point.mpl
segment.mpl
circle.mpl
square.mpl
triangle.mpl
To avoid platform-specific differences, a flat directory structure is used.

(All the source files reside in the same directory or folder.)
shapes.mpl ShapesPackage
point.mpl
segment.mpl
circle.mpl
square.mpl
triangle.mpl
To define the module that implements this package, we use the Maple
preprocessor to include the remaining source files at the appropriate point
in the “master” source file shapes.mpl. A number of $include directives
are included in shapes.mpl, such as
$include "point.mpl"
$include "segment.mpl"
...
Splitting a large project into a number of source files makes it easier
to manage, and allows several developers to work on a project simultane-
ously. The source file is divided into shape-specific functionality. Most of
the functionality for points, for instance, is implemented by source code
stored in the file point.mpl.
Package Architecture The Shapes package is structured as a module

with a number of exported procedures. Individual submodules provide
shape-specific functionality for each shape type supported by the package.
Each of these shape-specific submodules is stored in its own source file;

these are the files included into the main package source file, shapes.mpl.
The shape-specific submodules are submodules of another submod-
ule, called Shapes. That is, the package module is called Shapes; it has
a submodule, also called Shapes; and the module Shapes:-Shapes con-
tains one submodule for each kind of shape supported. This submodule
hierarchy is illustrated below.
Shapes
Shapes
point triangle
segment square
circle
ShapesPackageSubmoduleStructure
The result of preprocessing the main file shapes.mpl produces a module

whose source has the following general outline.
Shapes := module()
export make, area, circumference;
local Shapes, circum_table;
Shapes := module()
export point, segment, circle, square, triangle;
point := module() ... end;
segment := module() ... end;
.....
end module;
make := proc() ... end;
area := proc() ... end;
circum_table := table(); ...
circumference := proc() ... end;
end module:
The Package API The Shapes package exports the following routines:
1. make
2. area
3. circumference
The exported procedure make is a constructor for shapes. It is used
to create a shape expression from the input data. For example, points are
constructed from their x and y coordinates.
> org := make( ’point’, 0, 0 );
org := make(point , 0, 0)
A circle is constructed from its center and radius.

> circ := make( ’circle’, org, 2 );
circ := make(circle, make(point , 0, 0), 2)
In each case, the name of the shape is passed as the first argument,
to tell make what kind of shape to return.
To compute the area of a shape, call the exported procedure area
with the shape as its argument.
> area( circ );
area(make(circle, make(point , 0, 0), 2))
Similarly, the exported procedure circumference computes the cir-

cumference of a given shape.
> circumference( circ );
circumference(make(circle, make(point , 0, 0), 2))
Shape Representation Shapes are represented as unevaluated func-

tion calls. The arguments to the call are the instance-specific data for
the shape. For example, a point with coordinates (2, 3) is represented
by the unevaluated function call POINT( 2, 3 ). Some instance data
may be shapes themselves. For instance, a segment is represented, us-
ing its endpoints, as an unevaluated function call of the form SEGMENT(
start_point, end_point ). The start and end points of the segment
can be obtained by calls to the point constructor.
Procedure Dispatch The Shapes package illustrates three types of pro-

cedure dispatching:
1. dispatching on submodule exports
2. conditional dispatching
3. table-based dispatching
1. Dispatching on Submodule Exports The procedure makei, which

is exported from the Shapes package, uses its submodule Shapes:-Shapes
to effect procedure dispatching. To test whether a method for a given
shape is available, the procedure make tests whether there is a submodule
by that name in the Shapes:-Shapes submodule. If no such submod-
ule is found, an exception is raised. Otherwise, the export make from
the submodule is passed the arguments that were given to the top-level
Shapes:-make procedure. Here is the make source code.
> make := proc( what::symbol )
> description "constructor for shapes";
> local ctor, # the shape constructor,
> # if found
> theShape; # the submodule for the
> # kind of shape requested
>
> if not member( what, Shapes, ’theShape’ ) then
> error "shape ‘%1’ not available", what

> end if;
> if member( ’:-make’, theShape, ’ctor’ ) then
> ctor( args[ 2 .. nargs ] )
> else
> error "no constructor provided for "
> "shape %1", what
> end if
> end proc:
The first argument to make is a symbol that denotes the kind of

shape to construct (point, circle, triangle). This symbol is used as
an “index” into the Shapes:-Shapes submodule. The first statement
uses member to test whether the symbol passed in the parameter what
is exported by the submodule Shapes:-Shapes. If it is not found, an ap-
propriate diagnostic is issued, and an exception raised. If member returns
the value true, then its third argument, the local variable theShape, is
assigned the export found in the submodule.
For example, if what is the symbol circle, then the local variable
theShape is assigned the submodule Shapes:-Shapes:-circle that im-
plements operations on circles. The same idea is used to pick out the
shape-specific constructor; it is the value assigned to the local variable
ctor upon a true value being returned from the second call to member.
Any remaining arguments are taken to be data used to construct the
shape. These are passed on to the make export in a shape-specific sub-
module, if found, and are not checked further at this level. This design
keeps the knowledge of each kind of shape localized to the submodule
responsible for it.
2. Conditional Dispatching The procedure area uses a simple condi-

tional dispatching mechanism. The “tag” of the input shape is extracted
and is used in direct comparisons with hard-coded values to determine
which shape-specific area subroutine to call to perform the area compu-
tation.
> area := proc( shape )
> description "compute the area of a shape";
> local tag;
>
> if not type( shape, ’function’ ) then
> error "expecting a shape expression, "
> "but got %1", shape
> end if;
>
> # Extract the "tag" information from the shape
> tag := op( 0, shape );
>
> # Dispatch on the "tag" value

> if tag = ’:-POINT’ then
> Shapes:-point:-area( shape )
> elif tag = ’:-SEGMENT’ then
> Shapes:-segment:-area( shape )
> elif tag = ’:-CIRCLE’ then
> Shapes:-circle:-area( shape )
> elif tag = ’:-SQUARE’ then
> Shapes:-square:-area( shape )
> elif tag = ’:-TRIANGLE’ then
> Shapes:-triangle:-area( shape )
> else
> error "not a recognized shape: %1", tag
> end if
> end proc:
3. Table-based Dispatching The third dispatch method illustrated in

the Shapes package is table-based. This technique is used by the exported
procedure circumference, which references the table circum_table to
look up the appropriate routine to call. This table is built simply by
assigning its entries in the body of the Shapes package.
> circum_table := table();
> circum_table[ ’POINT’ ] := Shapes:-point:-circumference;
> circum_table[ ’SEGMENT’ ] := Shapes:-segment:-circumference;
> circum_table[ ’CIRCLE’ ] := Shapes:-circle:-circumference;
> circum_table[ ’SQUARE’ ] := Shapes:-square:-circumference;
> circum_table[ ’TRIANGLE’ ] := Shapes:-triangle:-circumference;
The source code for the procedure circumference follows.

> circumference := proc( shape )
> description "compute the circumference of a "
> "shape expression";
> if not type( shape, ’function’ ) then
> error "expecting a shape, but got %1", shape
> end if;
> if assigned( circum_table[ op( 0, shape ) ] ) then
> circum_table[ op( 0, shape ) ]( shape )
> else
> error "no circumference method available "
> "for shape %1. Supported shapes "
> "are: %2", tag,
> sprintf( "%q", op( ALL_SHAPES ) )
> end if
> end proc:
Minimal checking is done to ensure that the input has the right struc-
ture. If an entry is found in the table circum_table for the shape “tag”
(as with the area routine), then the corresponding procedure is called
with the given shape as argument. (The shape must be passed as an ar-
gument, so that the shape-specific subroutine can extract the instance
data from it.) Otherwise, a diagnostic is issued, and an exception raised.
Shape-specific Submodules As already noted, each shape is imple-

mented in a shape-specific submodule. The set of exports of each module
varies, but each supports at least the “required” exports make, area, and
circumference. Particular shapes may support other operations. Only
two of these submodules are described here. You can see the source for
the other submodules in the sample source code.
The point Submodule The submodule that implements points is

fairly simple. In fact, it makes no reference to any lexically scoped vari-
ables in any of its parent modules (either Shapes or Shapes:-Shapes).
> point := module()
> description "support routines for points";
> export make, area, circumference, xcoord, ycoord;
> option package;
>
> make := ( x, y ) -> ’POINT’( x, y );
> area := () -> 0;
> circumference := () -> 0;
> xcoord := p -> op( 1, p );
> ycoord := p -> op( 2, p );
> end module:
Since the area and circumference of a point are both 0, these proce-
dures are trivial to implement. In addition to the “required” exports, the
point submodule also exports two utility routines, xcoord and ycoord
for retrieving the x and y coordinates of a point. Providing these makes
it possible for clients of this submodule to use it without knowing any-
thing about the concrete representation of points. This makes it easier to
change the representation later should that be required.
Within this submodule, the names make, area, and circumference
shadow the names with the same external representation at the top-level
Shapes module.
The circle Submodule circle submodule provides the circle-specific

subroutines for the Shapes package.
> circle := module()
> export make, center, radius, diameter,
> area, circumference;
> option package;
>
> make := proc( cntrPt, radius )
6.4 The use Statement • 291
> ’CIRCLE’( cntrPt, radius )

> end proc;
>
> center := circ -> op( 1, circ );
> radius := circ -> op( 2, circ );
> diameter := circ -> 2 * radius( circ );
> circumference := circ -> Pi * diameter( circ );
> area := circ -> Pi * radius( circ )^2;
> end module:
Again, a few “extra” routines are provided beyond those required at

the top-level of the Shapes package. The exported procedure radius is
used to define some of the other routines. It could have been made local
to this submodule.
6.4 The use Statement

The use statement is formally unrelated to modules, but is expressly
designed to complement them and to make programming with modules
easier in some circumstances.
Syntax and Semantics The keyword use introduces the use state-
ment, which has the following syntax template:
use env in
body
end use;
Here, env is an expression sequence of binding equations, each of

which is either a module or an equation whose left-hand side is a symbol;
and body is a sequence of Maple statements. The right-hand side of a
binding equation may be any Maple expression.
Executing a use statement executes the body of the statement. Each
occurrence of a name that appears on the left-hand side of any of the
binding equations is replaced by the right-hand side of the corresponding
equation.
For example,
> use f = sin, g = cos in
> f( x )^2 + g( x )^2
> end use;
sin(x)2 + cos(x)2
Characteristics of use The use statement can be nested.

> use f = sin in
> use g = cos in
> simplify( f(x)^2 + g(x)^2 )
> end use
> end use;
When nested use statements are encountered, the name bindings es-
tablished by the inner use statement take precedence over those of the
outer one.
> use a = 2, b = 3 in
> use a = 3 in a + b end
> end use;
In this example, the inner binding of the name a to the value 3 takes
precedence, so the value of the expression a + b (and hence of the entire
statement) is the number 6. The inner binding of a to 3 has an effect
only within the body of the inner use statement. Once the execution has
exited the inner use statement, the binding of a to 2 is restored.
> use a = 2, b = 3 in
> # here a is bound to 2 and b to 3
> use a = 3 in
> # here, b is still bound to 3, but a is bound to 3
> a + b
> end use;
> # binding of a to 2 is restored
> a + b
> end use;
The use statement is unique in the Maple language because it is fully

resolved during automatic simplification. It is not possible to evaluate
a use statement. (Recall that Maple uses a modified “read-eval-print”
loop, which actually involves the four stages: parsing (reading), automatic
simplification, evaluation and printing.)
To see how this works, consider an example in which the use statement
appears inside a procedure.
> f := proc( a, b )
> use x = a + b, y = a - b in
> x * y
> end use
> end proc;
f := proc(a, b) (a + b) ∗ (a − b) end proc
Note that the body of the procedure f contains no use statement. Dur-
ing automatic simplification, the use statement that formed the body of f
was “expanded”, yielding the expression that involves only the parameters
a and b.
Modules and use Statements As a special case, a module m may ap-

pear in the binding sequence of a use statement. The module is regarded
as an abbreviation for the sequence of equations a = m:-a, b = m:-b,
. . . , where a,b,... are the exports of the module m.
For example,
> m := module() export a, b; a := 2; b := 3; end:
> use m in a + b end;
This is useful for programming with packages.

> m := Matrix( 4, 4, [[ 26, 0, 0, 30 ],
> [ 0, -41, -90, 0],
> [ 0, -7, -56, 0 ],
> [ 0, 0, 0, 0]] );
 
26 0 0 30
 0 −41 −90 0 
m := 
 0 −7 −56 0 

0 0 0 0
> use LinearAlgebra in

> Determinant( m );
> Rank( m );
> CharacteristicPolynomial( m, ’lambda’ )
> end use;
3
(26 − λ) (−1666 λ − 97 λ2 − λ3 )
Please note that a name that appears in a binding list for a use
statement that is intended to be a module must evaluate to a module at
the time the use statement is simplified. This is necessary because the
simplification of the use statement must be able to determine the exports
of the module. In particular, the following attempt to pass a module as a
parameter to a procedure does not work, and yields an error during the
simplification of the procedure.
> proc( m, a, b )
> use m in e( a, b ) end
> end proc;
Error, no bindings were specified or implied
The correct way to use a module as a parameter is to specify the

names to be bound explicitly, such as in this example:
> proc( m, a, b )
> use e = m:-e in a + b end
> end proc;
proc(m, a, b) a + b end proc
This is necessary because, until the procedure is called with a module

expression as first argument, Maple does not know whether the e refers
to a module export or to something else (such as a global name). To
expand the use statement, this must be known at the time the procedure
is simplified.
Operator Rebinding
An additional feature of the use statement is that it allows most infix and
prefix operators in the Maple language to be rebound. This is not really
the same thing as the “operator overloading” found in some programming
languages (such as C++), because the rebinding occurs during automatic

simplification in Maple.
If an operator name appears on the left-hand side of a binding equa-
tion for a use statement (consequently, if it is an exported name of a
module that is bound via use), then the corresponding operator expres-
sions in the body of the use statement are transformed into function calls.
For example:
> use ‘+‘ = F in a + b end;
F(a, b)
> m := module()
> export ‘*‘, ‘+‘;
> ‘+‘ := ( a, b ) -> a + b - 1;
> ‘*‘ := ( a, b ) -> a / b;
> end module:
> s * ( s + t );
s (s + t)
> use m in s * ( s + t ) end;

s
s+t−1
The operators that can be rebound are summarized in the following

table.
Operator Arity Position Description

Arithmetic Operators
+ binary infix addition
* binary infix multiplication
. binary infix multiplication
^ binary infix exponentiation
- unary prefix negation
/ unary prefix inversion (reciprocal)
Logical Operators
and binary infix logical and
or binary infix logical or
not unary prefix logical negation
Relational Operators
< binary infix less than
<= binary infix less than or equal
> binary infix greater than
>= binary infix greater than or equal
= binary infix equality
<> binary infix not equal
Other Operators
@ binary infix composition
@@ binary infix power composition
! unary postfix factorial
Please note that the operators - and / are treated as unary opera-
tors (that represent negation and inversion, respectively). Subtraction is
represented internally in Maple by composing addition and negation: a
- b = a + (-b). Similarly for division. Therefore, it is not necessary to
override the binary infix operators - and /.
Note also that an expression such as a + b + c + d is treated as
though it were parenthesized as ((a + b) + c) + d, so that each + op-
erator is binary. For example,
> use ‘+‘ = F in
> a + b + c + d;
> a + ( ( b + c ) + d )
> end use;
F(F(F(a, b), c), d)
F(a, F(F(b, c), d))

6.5 Modeling Objects • 297
6.5 Modeling Objects

A principle application of modules is modeling objects. An object is some-
thing that has both state and behavior. Many programming languages
provide support for programming with objects. Some of these are called
"object-oriented"; popular examples include Smalltalk, CLOS, Java, and
C++.
Maple is not an object-oriented programming language, but it does
support programming with objects. In Maple, an object can be repre-
sented by a module. The state of the object (module) is stored in the
local and exported data variables. The behavior of the object is repre-
sented by procedures assigned to the exported variables. Since, in Maple,
procedures stand on an equal footing with all other values in the lan-
guage, this distinction between state and behavior is somewhat artificial
and exists only as a convention.
The essential idea behind programming with objects is that the “ob-
jects” carry their behavior around with them. Clients of an object can
elicit behavior by sending the object “messages”. Objects respond to these
messages by performing some prescribed computation that is determined
by both the recipient of the message (the object) and the message itself
(which may be parameterized by other arguments). This is in contrast to
non-object-oriented approaches to programming, in which the objects in
a software system merely contain static data and serve as inputs and out-
puts of the algorithms, which are represented separately from the objects
by procedures or other routines.
Objects and Constructors Objects are usually created by invoking a

“constructor”. A constructor is a procedure that builds the object expres-
sion from some (possibly empty) set of inputs. Maple uses constructors
for a number of its native expression types. For example, the procedure
table is a constructor for Maple tables, and series is a constructor for
Maple series expressions. Here we are interested in constructors for objects
represented by modules.
A constructor must be used to create objects that have no input
syntax (such as series and tables, in Maple), but may also be used for
expressions that do have an input syntax (the Float constructor is an
example of the latter case). Therefore, most user-defined objects must be
created by using a constructor. So most of our object examples will be
defined by specifying a constructor for the object.
Example: a Complex Number Constructor A simple example of

an object is the following representation of a complex number.
> MakeComplex := proc( real, imag )
> if nargs <> 2 then
> error "real and imaginary parts are required"
> end if;
> module()
> description "a complex number";
> local real_part, imag_part;
> export re, im, abs, arg;
>
> real_part, imag_part := real, imag;
> re := () -> real_part;
> im := () -> imag_part;
> abs := () -> sqrt( re()^2 + im()^2 );
> arg := () -> arctan( im(), re() );
> end module
> end proc:
To create the complex number 1 + i, we call the constructor as follows.
> z := MakeComplex( 1, 1 );
z := module()
local real _part , imag _part ;
export re, im, abs, arg ;
description “a complex number”;
end module
The procedure MakeComplex is a constructor for complex number ob-
jects. The value returned by the procedure is the instantiation of the
module whose definition appears in the body of MakeComplex.
The local state of the complex number is represented by the local
variables of the module, real_part and imag_part. The behavior is rep-
resented by the exported procedures re, im, abs, and arg.
The exports of a module that represents an object are sometimes
viewed also as messages . Objects respond to these messages by exhibiting
the behavior that the messages elicit.
> z:-re(), z:-im();
1, 1
> z:-abs();
√
2
> z:-arg();
1
π
4
For instance, the expression z:-abs() is viewed as sending the abs

“message” to the complex number object z. The object responds by com-
puting its absolute value.
Note that each time the procedure MakeComplex is invoked, a new
module is created using the module definition that is visible within the
procedure body. Thus, complex numbers created by different calls to the
constructor are distinct, even if the arguments real and imag are the
same. Whether a constructor should produce distinct objects for the same
input (instance) data depends on the nature of the objects being modeled.
For complex number objects, we would likely want to have the same
object produced for multiple calls with the same inputs. This can be
easily achieved by using the remember option in the constructor.
Effect of Immutable Local States The MakeComplex constructor

above represented the local state of complex number objects by using two
local variables real_part and imag_part. For many object constructors,
some or all of the local state of the object is expected to be immutable. In
these cases, local variables do not have to be allocated in the module to
store the local state of the object. The state can instead be represented
by the parameters to the constructor, which are visible within the module
because of Maple’s lexical scoping rules. Using this idea, the constructor
above can be simplified as follows.
> MakeComplex := proc( real, imag )
> if nargs <> 2 then
> error "real and imaginary parts are required"
> end if;
> module()
> description "a complex number";
> export re, im, abs, arg;
>
> re := () -> real;
> im := () -> imag;
> abs := () -> sqrt( real^2 + imag^2 );
> arg := () -> arctan( imag, real );
> end module
> end proc:
Example: Priority Queues

A very useful data structure that can be implemented in an object-
oriented way with modules is the priority queue. A priority queue is
a container data structure that admits the following operations:
• test for an empty priority queue
• insert a prioritized item into a priority queue
• return (non-destructively) the highest-priority item on the priority

queue
• delete the highest priority item from a priority queue
Design An object representation of priority queues will have the fol-

lowing methods.
empty test for an empty priority queue
top return the highest-priority item
insert insert a prioritized item
delete remove (and return) the highest priority item
This representation leads directly to the following Maple type, which
can be used to identify priority queues.
> ‘type/PriorityQueue‘ := ’‘module‘( empty, top, insert,
> delete )’:
Constructor Implementation We can implement priority queues as

Maple objects satisfying this interface by writing a constructor for the
objects.
> PriorityQueue := proc( priority::procedure )
> description "priority queue constructor";
> local largs, lnargs;
>
> lnargs := nargs;
> if lnargs > 1 then
> largs := [ args[ 2 .. -1 ] ]
> else
> largs := []
> end if;
>
> module()
> description "a priority queue";
> export empty, top, insert,

> size, delete, init;
> local heap, nitems,
> bubbleup, bubbledown;
>
> nitems := 0;
> heap := table();
>
> bubbleup := proc( child::posint )
> local parent;
> parent := iquo( child, 2 );
> if child > 1
> and priority( heap[ child ] ) > priority( heap[
> parent ] ) then
> heap[ parent ], heap[ child ] := heap[ child ],
> heap[ parent ];
> procname( parent ) # recurse
> end if
> end proc;
>
> bubbledown := proc( parent::posint )
> local child;
> child := 2 * parent;
> if child < nitems
> and priority( heap[ 1 + child ] ) > priority(
> heap[ child ] ) then
> child := 1 + child
> end if;
> if child <= nitems
> and priority( heap[ parent ] ) < priority( heap[
> child ] ) then
> heap[ parent ], heap[ child ] := heap[ child ],
> heap[ parent ];
> procname( child ) # recurse (new parent)
> end if
> end proc;
>
> # Initialize the priority queue.
> init := proc()
> heap := table();
> nitems := 0
> end proc;
>
> # Test whether the priority queue is empty.
> empty := () -> evalb( nitems < 1 );
>
> # Return the number of items on the priority queue.
> size := () -> nitems;
>
> # Query the highest priority item.
> top := proc()
> if empty() then
> error "priority queue is empty"
> else
> heap[ 1 ]
> end if
> end proc;
>
> # Delete the highest priority item from the
> # priority queue.
> delete := proc()
> local val;
> val := heap[ 1 ]; # val := top()
> # move bottom to the top
> heap[ 1 ] := heap[ nitems ];
> # allow expression to be collected
> heap[ nitems ] := evaln( heap[ nitems ] );
> # decrement the bottom of heap counter
> nitems := nitems - 1;
> # heapify the array
> bubbledown( 1 );
> # return the value
> val
> end proc;
>
> # Insert an item into the priority queue.
> insert := proc( v )
> if nargs > 1 then
> op( map( procname, [ args ] ) )
> else
> nitems := 1 + nitems;
> heap[ nitems ] := v;
> bubbleup( nitems )
> end if
> end proc;
>
> # Insert any intially specified items.
> if lnargs > 1 then
> insert( op( largs ) )
> end if
> end module
> end proc:
The constructor takes a Maple procedure priority as its argument. For
each expression that may be placed on the queue, this procedure should
return a numeric measure of its “priority”. Items on the queue are main-
tained in a prioritized order so that the highest priority items are removed
first.
In this sample computation with a priority queue, we use the Maple
built-in procedure length as the “priority” of an expression. Here, the
randomly generated expressions are all polynomials.
> pq := PriorityQueue( x -> length( x ) );
pq := module()
local heap, nitems , bubbleup, bubbledown;
export empty , top, insert , size, delete, init ;
description “a priority queue”;
end module
> pq:-insert( randpoly( x ) );
> end do:
> while not pq:-empty() do
> pq:-delete()
> end do;
−85 x5 − 55 x4 − 37 x3 − 35 x2 + 97 x + 50
−99 x5 − 85 x4 − 86 x3 + 30 x2 + 80 x + 72
−53 x5 + 85 x4 + 49 x3 + 78 x2 + 17 x + 72
79 x5 + 56 x4 + 49 x3 + 63 x2 + 57 x − 59
−86 x5 + 23 x4 − 84 x3 + 19 x2 − 50 x + 88
−50 x5 − 12 x4 − 18 x3 + 31 x2 − 26 x − 62
−58 x5 − 90 x4 + 53 x3 − x2 + 94 x + 83
77 x5 + 66 x4 + 54 x3 − 5 x2 + 99 x − 61
45 x5 − 8 x4 − 93 x3 + 92 x2 + 43 x − 62
x5 − 47 x4 − 91 x3 − 47 x2 − 61 x + 41
Priority Queue Usage Priority queues can be used to implement a

heapsort algorithm.
> HeapSort := proc( L::list(numeric) )
> local pq, t, count;
> pq := PriorityQueue( x -> -x, op( L ) );
> t := array( 1 .. nops( L ) );
> count := 0;
> while not pq:-empty() do
> t[ count ] := pq:-delete()
> end do;
> ASSERT( count = nops( L ) );
> [ seq( t[ count ], count = 1 .. nops( L ) ) ]
> end proc:
> r := rand(100):
> L := [ seq( r(), i = 1 .. 20 ) ]:

> HeapSort( L );
[7, 7, 15, 25, 27, 27, 28, 29, 42, 51, 52, 55, 62, 74, 82,
88, 94, 97, 97, 98]
The fully commented source code for the Priority Queue constructor
is available in the sample source code of your Maple installation.
An Object-oriented Shapes Package

In this subsection, we will demonstrate an object-oriented approach to the
Shapes package described earlier. The earlier revision of the package used
unevaluated function calls as the concrete representation of shapes. In this
section we will demonstrate how to put together a package that offers the
same functionality, but which represents shapes as “objects”. Each shape
will use a module as its concrete representation. The “package” itself does
not export the area and circumference features of the traditional style
package, because these features are available as part of each shape object.
Instead, the package is merely a collection of constructors for the various
kinds of shapes. You could use the object representation at a lower level,
and present exactly the same interface as the first Shapes package, but
we have not done so here, choosing instead to make the object-oriented
nature of shape expressions more apparent to the user.
The point Constructor Points are quite simple shapes, so the corre-
sponding constructor is similarly simple.
> point := proc( x, y )
> module()
> export area, circumference, xcoord, ycoord;
> xcoord := () -> x;
> ycoord := () -> y;
> area := () -> 0;
> end module
> end proc:
The module returned by this constructor uses the lexically scoped

parameters x and y, representing the abscissa and ordinate of the point.
These values are part of the local state, or instance data, of each point
constructed. These points are captured in the closures of the exported
methods, so that variables local to the module in which to store these
values are not necessary.
The segment Constructor Segments are represented using the start

and end points of the segment. These are the points returned by the point
constructor.
> segment := proc( pt1, pt2 )
> module()
> export area,
> circumference,
> length,
> start_point,
> end_point;
> local mymidpoint;
>
> start_point := () -> pt1;
> end_point := () -> pt2;
> area := () -> 0;
> length := proc()
> local x, y;
> x := pt1:-xcoord() - pt2:-xcoord();
> y := pt1:-ycoord() - pt2:-ycoord();
> sqrt( x^2 + y^2 )
> end proc;
> midpoint := proc()
> local x, y;
> if assigned( mymidpoint ) then
> mymidpoint
> else
> y := (pt1:-ycoord() + pt2:-ycoord())/2;
> x := (pt1:-xcoord() + pt2:-xcoord())/2;
> point( x, y )
> end if
> end proc;
> end module
> end proc:
The segment objects implement methods in addition to the required

area and circumference methods. Apart from the trivial syntax methods
start_point and end_point, there are methods for computing the length
of a segment and its midpoint.
The circle Constructor Circles are represented by using the center

and radius of the circle as instance data.
> circle := proc( ctr, rad )
> module()
> export area, circumference, diameter,
> center, centre, radius;
> radius := () -> rad;
> center := () -> ctr;
> centre := eval( center ); # UK spelling
> diameter := () -> 2 * radius();
> circumference := () -> Pi * diameter();

> area := () -> Pi * rad * rad;
> end module
> end proc:
Again, the lexically scoped parameters ctr and rad encode the in-
stance data of the circle object.
The remainder of the object oriented version of the Shapes package
can be read in the sample source code file ShapeObj.mpl.
6.6 Interfaces and Implementations

Generic programming is a programming style and a software engineering
methodology for software reuse. In this sense, many Maple builtin oper-
ations are generic. The addition operator + is able to compute sums of
integers, rational numbers, complex numbers, polynomials, special func-
tions, and so on. It is not necessary for + to know how an expression is
represented to do its job. (The automatic simplifier, on the other hand,
knows a very great deal about how Maple expressions are represented.)
As with any dynamically typed language, Maple provides much genericity
without you having to do anything special. Most built-in Maple operations
(including many standard library routines) are naturally polymorphic in
that they are able to perform successfully with a large variety of data
formats.
Genericity as Good Software Engineering Practice On any large

project, it is important to write reusable code; that is, code that can
perform a well-defined function in a wide variety of situations. Generic
programs do not rely on the details of how their inputs are represented.
They are able to perform their function on any inputs that satisfy a
specified set of constraints. Normally, these constraints are described in
terms of the behavior of the inputs in response to various “messages”
rather than on their physical representation or the storage layout of their
concrete representation. This behavior is sometimes called a “contract”.
The idea is that an object makes a “promise” to behave a certain way,
and anything which uses that object knows that the “promised” behavior
can be relied upon. Generic programs rely only on the “promises” made
by an object’s “contract”. They do not rely on knowledge of how an
object is implemented. So, generic software separates “interfaces” from
implementations.
6.6 Interfaces and Implementations • 307
With this discussion, we are finally able to see the real meaning be-
hind the distinction between local and exported variables in a module.
A module’s exports are part of its “promise” to those who would use it.
Whatever is expressed through its local variables is the business only of
the module, and is not to be relied upon, or even known, by clients of the
module. (Client access is, in fact, the only technical difference between
module locals and exports.)
Before the introduction of the module system, Maple’s “design by
contract” was enforced only by convention. Maple routines whose names
had to be enclosed in name quotes (‘) were considered “private”, and not
for client use. But this was only a convention. Moreover, it was necessary
to use global variables to communicate information and state between the
various routines that made up a subsystem (such as solve or assume).
Now, using modules, it is possible to design software systems that enforce
their contracts by a mechanism embedded in the Maple language itself.
Interfaces
The “contracts” discussed above are represented formally in Maple by an
interface. An interface is a special kind of structured type. It has the
form
‘module‘( symseq );
in which symseq is a sequence of symbols or of typed symbols (ex-

pressions of the form symbol::type). For example, an interface for a ring
might be written as
> ‘type/ring‘ := ’‘module‘( ‘+‘, ‘*‘, ‘-‘, zero, one )’:
while that for an (additive) abelian group could take the form
> ‘type/abgroup‘ := ’‘module‘( ‘+‘, ‘-‘, zero )’:
These symbols are the ones that clients are “allowed” to access as
module exports.
A module is said to satisfy, or to implement, an interface if it is of
the type defined by the interface.
> z5 := module()
> description "the integers modulo 5";
> export ‘+‘, ‘*‘, ‘-‘, zero, one;
> ‘+‘ := (a,b) -> a+b mod 5;
> ‘*‘ := (a,b) -> a*b mod 5;
> ‘-‘ := s -> 5-s mod 5;
> zero := 0;
> one := 1;
> end module:

> type( z5, ’ring’ );
true
A module may satisfy more than one interface.

> type( z5, ’abgroup’ );
true
Interfaces are an abstraction that form part of Maple’s type system.

They provide a form of constrained polymorphism. Not every Maple type
is an interface; only those that have the form just described are. We can
define a Maple type (that, as it happens, is not itself an interface) to
describe interfaces.
> ‘type/interface‘ := ’specfunc( {symbol,symbol::type},
> ‘module‘ )’:
This is a structured type that describes expressions that are themselves
structured types that have the form of an unevaluated function call with
operator the symbol ‘module‘ and all arguments of type symbol, or of
type symbol::type. In the two examples at the start of this section, the
types type/ring and type/abgroup are the interface expressions, and the
names ring and abgroup are the respective names of those interfaces.
A Package for Manipulating Interfaces Interfaces are sufficiently im-

portant that it is worthwhile to develop a package for manipulating them.
The package is small enough that it can be reproduced here, in full, but
it is also available in the samples/ directory of your Maple installation.
> Interface := module()
> description "a package for manipulating interfaces";
> global ‘type/interface‘;
> export define, # define an interface
> extend, # extend an interface
> extends, # test for an extension
> equivalent,# test equivalence
> savelib, # save an interface
> satisfies; # test whether a module satisfies
> # an interface
> local gassign, # assign to a global variable
> totype, # convert from interface name to type
> toset, # convert from interface name to a set
> setup; # install ‘type/interface‘ globally
> option load = setup;
>
> # Define a global type for interfaces.

> # This assignment takes care of installing the type
> # in the Maple session in which this module definition
> # is evaluated. Calling ‘setup()’ ensures that this also
> # happens when the instantiated module is read from a
> # repository.
> ‘type/interface‘
> := ’specfunc( {symbol, ‘::‘}, ‘module‘ )’;
>
> # Ensure that ‘type/interface‘ is defined. This thunk is
> # called when the instantiated ‘Interface’ module is read
> # from a Maple repository.
> setup := proc()
> global ‘type/interface‘;
> ‘type/interface‘
> := ’specfunc( {symbol, ‘::‘}, ‘module‘ )’;
> NULL # quiet return
> end proc;
>
> # Assign to the global instance of a name
> gassign := proc( nom::symbol, val )
> option inline;
> eval( subs( _X = nom,
> proc()
> global _X;
> _X := val
> end ) )()
> end proc;
>
> # Convert an interface name to the corresponding type.
> totype := ( ifc::symbol ) -> ( ‘type/‘ || ifc );
>
> # Convert an interface name to a set of symbols.
> toset := ( ifc::symbol ) -> { op( ( ‘type/‘ || ifc ) ) };
>
> # Install a new interface into the type system.
> define := proc( ifc )
> description "define an interface";
> if map( type, {args}, ’symbol’ ) <> { true } then
> error "arguments must all be symbols"
> end if;
> gassign( ‘type/‘ || ifc,
> ’‘module‘’( args[ 2 .. nargs ] ) );
> ifc # return the interface name
> end proc;
>
> # Implement subtyping.
> extend := proc( new, old )
> description "extend an existing inteface";
> if map( type, {args}, ’symbol’ ) <> { true } then
> error "arguments must all be symbols"
> end if;
> if not type( totype( old ), ’interface’ ) then
> error "cannot find an interface named %1", old
> end if;

> define( new, op( totype( old ) ), args[3..nargs] )
> end proc;
>
> # Test whether ifc2 is an extension of ifc1.
> extends := proc( ifc1, ifc2 )
> description "test whether the second interface "
> "extends the first";
> local t1, t2;
> t1, t2 := op( map( totype, [ ifc1, ifc2 ] ) );
> if not type( [t1,t2], ’[interface,interface]’ ) then
> if not type( t1, ’interface’ ) then
> error "arguments must be interface names, "
> "but got %1", ifc1
> else
> error "arguments must be interface names, "
> end if
> end if;
> toset( ifc1 ) subset toset( ifc2 )
> end proc;
>
> # Save an interface to the repository.
> savelib := proc()
> description "save a named interface to a "
> "repository";
> local ifc;
> for ifc in map( totype, [ args ] ) do
> if not type( ifc, ’interface’ ) then
> error "arguments must be interfaces, "
> "but got %1", ifc
> end if;
> :-savelib( totype( ifc ) )
> end do
> end proc;
>
> # Test whether a module satisfies an interface.
> # This is simply an alternative to a call
> # to ‘type()’.
> satisfies := proc( m, ifc )
> description "test whether a module satisfies an interface";
> if not type( totype( ifc ), ’interface’ ) then
> error "second argument must be an interface name, "
> "but got %1", ifc
> end if;
> type( m, ifc )
> end proc;
>
> # Test whether two interfaces are equivalent.
> # Since unevaluated function calls compare
> # differently if their arguments are in a
> # different order, we convert them to sets first,
> # and then test for equality.
> equivalent := proc( ifc1, ifc2 )
> description "test whether two interfaces "

> "are equivalent";
> local t1, t2;
> t1, t2 := totype( ifc1 ), totype( ifc2 );
> if not type( t1, ’interface’ ) then
> error "expecting an interface name, "
> elif not type( t2, ’interface’ ) then
> error "expecting an interface name, "
> end if;
> evalb( { op( t1 ) } = { op( t2 ) } )
> end proc;
> end module:
This little package implements the interface abstraction. It allows you

to manipulate interfaces without having to worry about how they fit into
Maple’s type system.
> with( Interface );
Warning, the protected names define and savelib have
been redefined and unprotected
[define, equivalent , extend , extends , satisfies , savelib]
> define( ’abgroup’, ’‘+‘’, ’‘-‘’, ’zero’ );
abgroup
> type( ‘type/abgroup‘, ’interface’ );
true
> type( z5, ’abgroup’ );
true
> satisfies( z5, ’abgroup’ );
true
> extend( ’ring’, ’abgroup’, ’‘*‘’, ’one’ );

ring
> type( ‘type/ring‘, ’interface’ );
true
> extends( abgroup, ring );
true
> satisfies( z5, ’ring’ );
true
> type( z5, ’ring’ );
true
The load= Option Besides providing a nice abstraction of the inter-

face concept in Maple, this package also serves to illustrate a module
feature not demonstrated earlier. This is the load=thunk option. In the
Interface package, this option is used in a fairly typical way. The dec-
laration
option load = setup;
that appears in the module definition instructs Maple that, when the
instantiated module is read from a repository, it is to call the procedure
setup. The procedure named must be a local or an exported local of the
module. The local procedure setup in this module simply ensures that
the global variable type/interface is assigned an appropriate value. This
assignment is also made in the body of the module so that the assignment
is also executed in the session in which the module is instantiated. This
was done for illustrative purposes. A better scheme would simply have
invoked setup at some point in the body of the module definition.
Example: Generic Graph Algorithms

We use (simple) graph algorithms as an example of generic programming
with a computer science orientation.
Mathematical Description A directed graph may be thought of as

an object that consists of a set V of vertices and a set E ⊆ V × V of
ordered pairs of vertices, called “edges”. Graphs may be visualized by
diagrams like the following.
a
b

e

c d

f
This diagram represents a graph with vertex set V = {a, b, c, d, e, f }, and
edge set E = {(a, b), (a, c), (b, d), (c, f ), (f, d), (b, e), (d, e), (c, b), (c, d)}.
Software Models Graphs may be represented in software in a vari-

ety of ways. The choice of storage mechanism depends on the expected
applications of the graph. Three possibilities for representing graphs in
software are:
1. store the set V of vertices and the set E of edges explicitly;
2. store the “adjacency matrix” of the graph;
3. store, for each vertex of the graph, the set of all its neighbours.
(The adjacency matrix is a square matrix whose rows are columns are
indexed by the vertices of the graph; the (i, j)-entry is equal to 1 if there
is an edge from i to j, and is equal to 0 otherwise.) We would like to be
able to write software that can manipulate a graph regardless of which of
the above (or other) representations is chosen.
Designing a Graph Interface To demonstrate how this can be

achieved, let us consider graphs as objects that implement the follow-
ing methods:
vertices returns the set of vertices of the graph
edges returns the set of edges of the graph
addedge allows one to add a new edge to a graph
order returns the number of vertices of the graph
size returns the number of edges of the graph
Then, we can represent the abstract interface of a graph by a Maple

type.
> ‘type/Graph‘ := ’‘module‘( vertices, edges, addedge, order,
> size )’:
We say that an object implements the Graph interface if it is of type
Graph.
Computing Vertex Degrees Generically If we have an object that

implements this interface, then we can write generic code based on that
interface. For example, we can write the following procedure to compute
the in-degree and out-degree of a vertex of a given graph.
> vdeg := proc( G::Graph, v )
> local vs, vt;
> description "compute the in- and out-degrees "
> "of a vertex in a graph";
> if member( v, G:-vertices() ) then
> vs := select( e -> evalb( v = e:-source() ),
> G:-edges() );
> vt := select( e -> evalb( v = e:-target() ),
> G:-edges() );
> nops( vs ), nops( vt )
> else
> 0, 0
> end if
> end proc:
We could write this procedure even though we have, as yet, no idea how
graphs are going to be implemented. Although this is but a tiny example,
this capability is very important when you are designing a large software
system.
Edge Object Representation We are tacitly assuming that edges

will also be represented as objects that implement, at least, the interface
‘module‘( source, target ), which provides methods for extracting
the source and target vertices from an edge. Writing a constructor Edge
for edges is easy.
> Edge := proc( src, targ )
> module()
> local the_source, the_target;
> export source, target, setsource, settarget;
> the_source := src;
> the_target := targ;
> source := () -> the_source;
> target := () -> the_target;
> setsource := proc( v )
> the_source := v
> end proc;
> settarget := proc( v )

> the_target := v
> end proc;
> end module
> end proc:
First Graph Constructor At first, we might choose to adopt a graph

representation that is simple to implement. Here is a graph constructor
that produces graphs represented by storing the vertex and edge sets
explicitly as part of the state of a module.
> Graph1 := proc()
> local vertex_set, edge_set;
> description "graph constructor";
>
> edge_set := { args };
> if map( type, edge_set, ’[ anything, anything ]’ )
> <> { true } then
> error "graph must be specified by a sequence of edges"
> end if;
> if map( nops, edge_set ) <> { 2 } then
> error "each edge must be specified "
> "as a [ source, target ] pair"
> end if;
> vertex_set := map( op, edge_set );
> edge_set := map( Edge@op, edge_set );
> module()
> export order, size,
> vertices, edges,
> addedge; # required exports
> vertices := () -> vertex_set;
> edges := () -> edge_set;
> addedge := proc( src, targ )
> edge_set := { Edge( src, targ ) }
> union edge_set;
> vertex_set := { src, targ }
> union vertex_set;
> NULL
> end proc;
> order := () -> nops( vertices() );
> size := () -> nops( edges() );
> end module
> end proc:
If we now create a small graph using this constructor
> g1 := Graph1( [ a, b ], [ a, c ], [ b, c ] ):
> type( g1, ’Graph’ );
true
we can use the routine vdeg with the graph g1, since graphs produced
by Graph1 implement the Graph interface.
> vdeg( g1, a );
2, 0
> vdeg( g1, b );
1, 1
> vdeg( g1, c );
0, 2
The important feature of the procedure vdeg is its genericity. It can

be used with any implementation of graphs that implements the Graph
interface specified above.
Second Graph Constructor Here is another, different implementa-

tion of the Graph interface. The graph is represented by using a table N
in which the neighbors of each vertex are stored.
> Graph2 := proc()
> local vertex_set, edge_set;
> description "graph constructor";
>
> edge_set := { args };
> vertex_set := map( op, edge_set );
> if map( type, edge_set, ’list’ ) <> { true } then
> error "graph must be specified by a sequence of edges"
> end if;
> if map( nops, edge_set ) <> { 2 } then
> error "each edge must be specified "
> "as a [ source, target ] pair"
> end if;
> module()
> export order, size,
> vertices, edges,
> addedge;
> local N, e, v, n, edge_pairs;
> N := table();
> edge_pairs := () -> { seq(
> seq( [ v, n ], n = N[ v ] ),
> v = map( op, { indices( N ) } )
> ) };
> vertices := () -> map( op, edge_pairs() );
> edges := () -> map( Edge@op, edge_pairs() );

> addedge := proc( src, targ )
> if assigned( N[ src ] )
> and not member( targ, N[ src ] ) then
> N[ src ] := { op( N[ src ] ), targ }
> else
> N[ src ] := { targ };
> end if;
> NULL
> end proc;
> order := () -> nops( vertices() );
> size := () -> nops( edges() );
> for e in edge_set do
> addedge( op( 1, e ), op( 2, e ) )
> end do
> end module
> end proc:
A graph returned by the constructor Graph2 also satisfies the Graph in-
terface.
> g2 := Graph2( [ a, b ], [ a, c ], [ b, c ] ):
> type( g2, ’Graph’ );
true
Because of this, the generic procedure vdeg works equally well with
it.
> vdeg( g2, a );
2, 0
> vdeg( g2, b );
1, 1
> vdeg( g2, c );
0, 2
Generic Computation of Adjacency Matrices Another example of

a procedure generic over the Graph interface is the following routine for
computing the adjacency matrix of a graph.
> AdjacencyMatrix := proc( g::Graph )

> local a, # the adjacency matrix; returned
> n, # the order of the graph g
> V, # the vertex set of the graph
> E, # the edge set of the graph
> row, # row index for matrix
> col, # column index for matrix
> e; # induction variable for loop
>
> n := g:-order();
> a := Matrix( n, n, ’storage’ = ’sparse’ );
> V := sort( convert( g:-vertices(), ’list’ ) );
> E := g:-edges();
> for e in E do
> if not member( e:-source(), V, ’row’ )
> or not member( e:-target(), V, ’col’ ) then
> error "inconsistent graph structure detected"
> end if;
> a[ row, col ] := 1
> end do;
> a
> end proc:
> AdjacencyMatrix( g1 );
 
0 1 1
0 0 1
0 0 0
> AdjacencyMatrix( g2 );
 
0 1 1
0 0 1
0 0 0
Example: Quotient Fields

As an example of generic programming, we will discuss a generic quotient
field (or “field of fractions”) construction algorithm.
Mathematical Description Given an integral domain D, its quotient

field is (up to isomorphism) the unique field k, paired with a nonzero
ring homomorphism η : D −→ k, with the property that, for any nonzero
ring homomorphism ϕ : D −→ F , in which F is a field, is a unique ring
homomorphism σ for which the diagram
k
σ
η F
ϕ
D
commutes. Because a nonzero ring homomorphism into a field must be
injective, this says that every field F that contains D as a subring must
also contain an isomorphic copy of k.
Concretely, the quotient field of an integral domain D can be thought
of as the set of “reduced fractions” n/d, with n, d ∈ D. A formal con-
struction can be produced by defining an equivalence relation on the set
D × (D \ {0}), according to which two pairs (n1, d1) and (n2, d2) are
equivalent only if,
n1 · d2 = n2 · d1.
A representative from each equivalence class is chosen to represent the
field element defined by that class. This understanding guides the com-
puter representation of the quotient field.
Unit Normal Representatives If R is a commutative ring with mul-

tiplicative identity, then
U (R) × R −→ R : (u, r) → u · r
is a natural action of the group U (R) of units of R on R. Each orbit of this
action has a representative called the unit normal representative of the
class, and we will suppose the existence of an effective mapping R −→ R
that selects the unit normal representative of each class. For instance, for
the ring Z of integers, the group U (Z) of units is the set {1, −1}, the orbits
are the sets {n, −n} for n ∈ Z \ {0} together with {0}, and we take the
unit normal representative to be the positive member of each orbit, or 0
for the orbit {0}. (Thus, the unit normal mapping simply computes the
sign and absolute value of an integer.) The unit normal mapping on the
ring k[T ] of polynomials in an indeterminate T over a field k is
1
p(T ) → · p(T ),
lc(p(T ))
in which lc(p(T )) denotes the leading coefficient of the polynomial p(T ).
(The group of units in k[T ] is the set k ∗ = k \ {0}, of nonzero members of
k, and each orbit of k[T ] under the action of k ∗ contains an unique monic
polynomial that is its representative.)
Designing the Ring Interfaces The first step in representing these

ideas in software is to devise an interface that describes the rings we are
to work with. We will suppose that our rings are equipped with the basic
ring operations, as well as several methods that implement the kind of
computations we want to do.
> ‘type/Ring‘ := ’‘module‘(
> ‘+‘::procedure,
> ‘*‘::procedure,
> ‘-‘::procedure,
> iszero::procedure,
> isone::procedure,
> zero, one
> )’:
This interface corresponds quite naturally with a formal mathematical
characterization of the ring as a tuple
hS, +, ∗, 0, 1i
that satisfies a number of properties, and to which we have added some

computational capabilities. We have added unary negation (-) because of
the way operator overrides work in Maple. (In a more tightly integrated
system, we might also specify the number and types of arguments to each
of the procedures.)
For the kind of computations that we want to do, we will need a
slightly richer structure.
> ‘type/GcdRing‘ := ’‘module‘(
> quo::procedure,
> rem::procedure,
> gcd::procedure,
> unormal::procedure,
> isone::procedure,
> zero, one
> )’:
This interface extends the Ring interface defined previously. Note that
nothing in the signature enforces any ring-theoretic properties (such as
being an integral domain, or having unique factorization). It merely spec-
ifies the admissible operations. Since we want to be able to compute with
infinite rings (and even large finite ones), we avoid requiring an enumer-
ation of the elements of the ring, but focus entirely on the effectively
computable operations that the ring must support.
Representing the ring Z of Integers One of the simplest examples of

a ring that supports the computations we require is the ring Z of integers
in its native Maple representation.
> MapleIntegers := module()
> description "the ring of integers";
> export ‘+‘, ‘*‘, ‘-‘,
> gcd, unormal, iszero,
> isone, zero, one, rem, quo;
> ‘+‘ := ( a, b ) -> a + b;
> ‘*‘ := ( a, b ) -> a * b;
> ‘-‘ := i -> -i;
> quo := ( a, b ) -> :-iquo( a, b );
> rem := ( a, b ) -> :-irem( a, b );
> gcd := ( a, b ) -> :-igcd( a, b );
> unormal := proc( i::integer )
> if i < 0 then
> -1, -i
> else
> 1, i # includes 0
> end if
> end proc;
> iszero := i -> evalb( i = 0 );
> isone := i -> evalb( i = 1 );
> zero := 0;
> one := 1;
> end module:
This is a software representation of the ring of integers. The unit normal
mapping is represented by the exported procedure unormal. It returns an
expression sequence of length two, whose first member is a unit, and whose
second member is the unit normal form of its argument. The product of
the output values yields the input ring element. The other methods just
invoke the corresponding, built-in Maple operations.
> type( MapleIntegers, ’Ring’ );
true
> type( MapleIntegers, ’GcdRing’ );
true
An Interface for Fields Our quotient field constructor produces a

field. An interface that describes fields differs from the one for integral
domains by the absence of a gcd method (since they are trivial) and the
addition of the (unary) / operator that computes inverses. The methods
rem and quo are also not included in the signature for fields, because
those too are trivial in a field. We do include two new methods: make for
constructing field elements from their numerators and denominators, and
embed, the natural embedding of the integral domain D into its field k of
fractions. Additionally, the two methods numer and denom allow the user
to extract the components of a fraction.
> ‘type/Field‘ := ’‘module‘(
> ‘/‘::procedure,
> normal::procedure,
> isone::procedure,
> zero, one,
> make::procedure,
> embed::procedure
> )’:
Naturally, the ring Z of integers is not a field.
> type( MapleIntegers, ’Field’ );
false
Fields produced by the quotient field constructor will satisfy this in-
terface.
The Quotient Field Functor Here is the generic constructor for quo-
tient fields.
> QuotientField := proc( R::GcdRing )
> description "quotient field functor";
> module()
> description "a quotient field";
> export ‘+‘, ‘*‘, ‘-‘, ‘/‘,
> zero, one,
> iszero, isone,
> make,
> numer, denom,
> normal, embed;
> make := proc( n, d )
> local u, nd;
> if R:-iszero( d ) then
> error "division by zero"
> end if;
> u, nd := R:-unormal( d );
> ’FRACTION’( u*n, nd )
> end proc;
> embed := d -> make( d, R:-one );
> numer := f -> op( 1, f );
> denom := f -> op( 2, f );

> zero := embed( R:-zero );
> one := embed( R:-one );
> iszero := f -> evalb( normal( f ) = zero );
> isone := f -> evalb( normal( f ) = one );
> normal := proc( f )
> local g, a, b;
> g := R:-gcd( numer( f ), denom( f ) );
> if R:-isone( g ) then
> f
> else
> a := R:-quo( numer( f ), g );
> b := R:-quo( denom( f ), g );
> make( a, b )
> end if
> end proc;
> ‘-‘ := f -> normal( R:-‘-‘( numer( f ) ), denom( f ) );
> ‘/‘ := f -> normal( make( denom( f ), numer( f ) ) );
> ‘+‘ := proc( a, b )
> use ‘+‘ = R:-‘+‘, ‘*‘ = R:-‘*‘ in
> normal( make( numer( a ) * denom( b )
> + denom( a ) * numer( b ),
> denom( a ) * denom( b ) ) )
> end use
> end proc;
> ‘*‘ := proc( a, b )
> use ‘*‘ = R:-‘*‘ in
> normal( make( numer( a ) * numer( b ),
> denom( a ) * denom( b ) ) )
> end use
> end proc;
> end module
> end proc:
Most of the exported routines are quite straightforward. The fraction
constructor make accepts two members of the ring R as arguments and
returns the constructed fraction, which we represent by an unevaluated
function call of the form
FRACTION( numerator, denominator )
The exported procedure embed is the canonical embedding η of the inte-

gral domain into its quotient field, described previously. This makes the
constructor functorial. The arithmetic operators are simple implementa-
tions of the familiar rules for fraction arithmetic:
a c ad + bc
+ =
b d bd
a c ac
× =
b d bd
° a ±−1 b
=
b a
°a± −a
− =
b b
After applying these simple formulae, the result is normalized by using
a call to the local routine normal (not :-normal). normal does most of
the interesting work in the ring generated by this constructor. It uses the
manifestation of the division algorithm in the ring R via the exported
procedures quo and gcd to reduce each fraction to the “lowest terms”.
Together, the fraction constructor make and the method normal ensure
that field elements are represented by the normal form representative of
the equivalence class that is the field element. Preventing division by zero,
and forcing denominators to be unit normal representatives, is performed
by make. Ensuring that fractions are reduced to “lowest terms” is handled
by normal.
The most important property of the QuotientField functor is that
it is generic. It relies solely on the GcdRing interface to do its job. No
knowledge of the concrete representation of the input integral domain R
(other than that it is a module that satisfies the required interface) is used
anywhere in the construction. Therefore, it will work with any implemen-
tation of the GcdRing interface that implements the correct semantics for
its public operations and that satisfies the abstract constraint that it be
a software representation of an integral domain. (The latter constraint is
required to ensure that the arithmetic operations are well defined.)
Constructing the Rationals as the Quotient Field of Z To con-

struct the quotient ring of the ring MapleIntegers defined previously, we
will proceed as follows.
> FF := QuotientField( MapleIntegers );
FF := module()
export‘ + ‘, ‘ ∗ ‘, ‘ − ‘, ‘/‘, zero, one, iszero, isone, make,
numer , denom, normal , embed ;
description “a quotient field”;
end module
> type( FF, ’Field’ );
true
> a := FF:-make( 2, 3 );
a := FRACTION(2, 3)
> b := FF:-make( 2, 4 );
b := FRACTION(2, 4)
> use FF in
> a + b;
> a * b;
> a / b
> end use;
FRACTION(7, 6)
FRACTION(1, 3)
FRACTION(4, 3)
The Quotient Field of the Polynomial Ring Q[T ] To illustrate the

genericity of this constructor, we will construct the field Q[T ] of rational
functions in a single indeterminate T from a concrete representation of
Maple rational polynomials.
> MaplePoly := module()
> description "the ring of rational polynomials";
> export ‘+‘, ‘*‘, ‘-‘,
> zero, one,
> iszero, isone,
> gcd, unormal,
> quo, rem;
> ‘+‘ := ( a, b ) -> expand( a + b );
> ‘*‘ := ( a, b ) -> expand( a * b );
> ‘-‘ := p -> -p;
> gcd := ( a, b ) -> :-gcd( a, b );
> unormal := proc( p )
> local lc;
> if iszero( p ) then
> one, zero
> else
> use lc = lcoeff( p ) in
> lc, :-normal( p / lc )
> end use
> end if
> end proc;
> iszero := p -> Testzero( p );
> isone := p -> Testzero( p - 1 );
> zero := 0;
> one := 1;
> rem := ( a, b ) -> :-rem( a, b );

> quo := ( a, b ) -> :-quo( a, b );
> end module:
The unormal method produces the leading coefficient and monic associate
of a given polynomial in Q[T ]. The remaining exports simply capture
built-in Maple operations on univariate rational polynomials.
> RR := QuotientField( MaplePoly );
RR := module()
export‘ + ‘, ‘ ∗ ‘, ‘ − ‘, ‘/‘, zero, one, iszero, isone, make,
numer , denom, normal , embed ;
description “a quotient field”;
end module
> type( RR, ’Field’ );
true
To make printed fractions more readable, we introduce the following

extension to the print command.
> ‘print/FRACTION‘ := ( n, d ) -> sort( n ) / sort( d ):
Finally, we will construct a few examples, and test the arithmetic.
> a := RR:-make( randpoly( ’T’, ’degree’ = 4, ’terms’ = 3 ),
> randpoly( ’T’, ’degree’ = 4, ’terms’ = 3 ) );
−2072 T 2 − 1960 T + 5432

a :=
7 9
T3 + T2 +
8 8
> b := RR:-make( randpoly( ’T’, ’degree’ = 4, ’terms’ = 3 ),

> randpoly( ’T’, ’degree’ = 4, ’terms’ = 3 ) );
−2790 T 3 + 496 T 2 + 5766

b :=
77 33
T2 − T−
62 31
> use RR in
> a + b,
> a * b,
> a / b
> end use;
7781 5 401827 3
(−2790 T 6 − T − 1638 T 4 + T
4 124
¾
1943715 2 144452 87333
+ T − T+ ) (
124 31 124
91 4 1067 3 6 2 693 297
T5 − T − T + T − T− ), (
248 496 31 496 248
5780880 T 5 + 4440688 T 4 − 16127440 T 3 − 9252880 T 2
− 11301360 T + 31320912)/(
91 4 1067 3 6 2 693 297
T5 − T − T + T − T− ), (
248 496 31 496 248
5780880 T 4 − 1711080 T 3 − 28100520 T 2 + 13000680 T
+ 16133040)/(
251 5 7 4 113 3 241 2 93
T6 + T − T − T − T − )
360 45 120 120 40
Example: A Generic Group Implementation
In this section, we illustrate how to develop a moderately complex software
system based on the use of features of Maple’s module system. Generic
programming is at the heart of the design. Only a fraction of the complete
system from which the examples are taken is shown. The examples that
follow comprise a system for computing with finite groups. Recall that a
group is a set of objects together with an associative binary operation,
for which there is an unique two-sided identity element, and with respect
to which each member of the underlying set possesses an unique inverse.
Examples of groups include systems of numbers, using addition, closed
sets of invertible matrices (all of the same size, with a common ground
field) using multiplication (“linear groups”), closed sets of permutations
(bijective mappings on a set) using composition (“permutation groups”),
and groups of points on elliptic curves. We are concerned here only with
finite groups.
An Interface for Finite Groups First, we must decide how to represent

the generic group interface. This is, in large measure, determined by the
use to which the group objects will be put. Once again, our design takes a
group to be a repository of data and computational services that we may
query or invoke.
The Group signature that we will use in our examples describes a com-
putational model of abstract groups that supports the following methods.
id the group identity

‘.‘ the binary operation on the group
mul n-ary version of ‘.‘
inv unary inversion operation
pow computes integral powers of group elements
eq tests whether two group elements are equal
member tests membership in the group and in sets
gens a generating set for the group
order returns the order of the group
elements returns an enumeration of the group’s members
> ‘type/Group‘ := ’‘module‘(
> id, ‘.‘, mul, inv,
> eq, member,
> gens,
> order, elements
> )’:
A corresponding constructor for groups is easily written using the Record
constructor introduced earlier. For the examples in this section, we will
not need to introduce any default methods.
> Group := proc()
> Record( op( ‘type/Group‘ ) );
> end proc:
This constructor does very little work on its own. It relies on more spe-
cialized constructors to establish useful values or defaults for the methods
exported.
We can begin to write generic algorithms using this interface immedi-
ately. A few simple examples are these routines for computing conjugates
and commutators of group elements. The conjugate of a group member
a by a group member b is b−1 ab. This routine computes the conjugate of
an element a by an element b in a group G.
> Conjugate := proc( G, a, b )
> description "compute the conjugate of a "
> "group element by another";
> use ‘/‘ = G:-inv, ‘.‘ = G:-‘.‘ in
> b^(-1) . a . b
> end use
> end proc:
Since the group operations ‘.‘ and inv in a generic group remain unas-
signed, the following computation is done symbolically.
> Conjugate( Group(), ’x’, ’y’ );
(inv(y)) . x . y
Similarly, we can compute the commutator [a, b] = a(−1) b(−1) ab,

generically, as follows.
> Commutator := proc( G, a, b )
> description "compute the commutator of "
> "two group elements";
> use ‘/‘ = G:-inv, mul = G:-mul in
> mul( inv( a ), inv( b ), a, b )
> end use
> end proc:
Again, this computation is done symbolically, so the group operations
return unevaluated.
> Commutator( Group(), ’x’, ’y’ );
mul(inv(x), inv(y), x, y)
The ability to write algorithms generic over a given interface is im-

portant for the management of large software projects involving many
developers. Here, one developer can be assigned the task of implementing
particular group constructors along with the attendant arithmetic, while
another developer can begin coding generic routines like those above. The
two developers can work independently, provided each ensures that their
work conforms to some agreed-upon interface and semantics.
Permutation Groups Before attempting to develop any complicated al-

gorithms, it is helpful to have available a few constructors for specific
kinds of groups. These can then be used to validate generic algorithms in
specific instances. For this reason, we develop a straight-forward imple-
mentation of permutation groups.
Permutations are represented using Maple lists. For example, the list
[2,1,3] represents the permutation that maps 1 → 2, maps 2 → 1,
and leaves 3 fixed. (In cycle notation, this is written as the transposition
(12).) The constructor takes a positive integer as its first argument, indi-
cating the degree of the permutation group. The remaining arguments are
expected to be permutations (represented as lists) of the stated degree.
These are used to form the generating set of the group returned by the
constructor.
> PermutationGroup := proc( deg::posint )
> description "permutation group constructor";
> local G, gens;
> gens := { args[ 2 .. -1 ] };
> G := Group();
> G:-id := [ $ 1 .. deg ];
> G:-‘.‘ := proc( a, b )
> local i;
> [ seq( b[ i ], i = a ) ]
> end proc;
> G:-mul := () -> foldl( G:-‘.‘, G:-id, args );
> G:-inv := proc( g )
> local i, a;
> a := array( 1 .. deg );
> for i from 1 to deg do
> a[ g[ i ] ] := i
> end do;
> [ seq( a[ i ], i = 1 .. deg ) ]
> end proc;
> G:-member := proc( g, S, pos::name )
> if nargs = 1 then
> type( g, ’list( posint )’ )
> and { op( g ) } = { $ 1 .. deg }
> else
> :-member( args )
> end if
> end proc;
> G:-eq := ( a, b ) -> evalb( a = b );
> G:-gens := gens;
> eval( G, 1 )
> end proc:
For example, to construct the group h(12), (123)i in the symmetric group
S4 , we use the PermutationGroup constructor as follows.
> G := PermutationGroup( 4, { [2,1,3,4], [2,3,1,4] } );
G := module()
export
id , ‘.‘, mul , inv , eq , member , gens , order , elements ;
option record ;
end module
We can now call upon the “services” provided by the methods ex-
ported by the instantiated group G to compute with its elements.
> use G in
> inv( [ 2,1,3,4 ] ) . [2,3,1,4];
> end use;
[3, 2, 1, 4]
It is useful to provide more specialized permutation group con-

structors for special kinds of groups. Using the general constructor
PermutationGroup, and over-riding some of the exported methods, we
can define several of these specialized constructors as follows.
The full symmetric group Sn on the n points {1, 2, 3, . . . , n} is pro-

duced by specifying a particular set of generators for a given degree (which
must be specified as an argument to the constructor).
> Symmetric := proc( n::posint )
> description "symmetric group constructor";
> if n < 2 then
> error "argument must be an integer larger than 1"
> elif n = 2 then
> PermutationGroup( 2, [2,1] );
> else
> PermutationGroup( n, [2,1,$3..n], [$2..n,1] );
> end if
> end proc:
This uses the fact that Sn is the two-generator group
Sn = h(12), (123 · · · n)i,
for any integers n ≥ 3.

A second special case worth considering is the class of dihedral groups.
We can think of these as the groups of symmetries of regular plane poly-
gons. The symmetry group of the regular n-gon is the dihedral group of
degree n and order 2n; it is denoted by Dn .
D6
(12)
(123456)
We’ll use the following little utility for reversing a list.

> lreverse := proc( L::list )
> description "reverse a list";
> option inline;
> [ seq( L[ -i ], i = 1 .. nops( L ) ) ]
> end proc:
> Dihedral := proc( n::posint )
> description "dihedral group constructor";
> local a, b, D;
> if n = 2 or n = 3 then
> return Symmetric( n )

> end if;
> a := [ $ 2 .. n, 1 ];
> b := [ 1, op( lreverse( [ $ 2 .. n ] ) ) ];
> D := PermutationGroup( n, { a, b } );
> D:-order := () -> 2*n;
> eval( D, 1 )
> end proc:
1. Use the fact that the alternating group An of degree n >= 3 is

generated by the set {(123), (234), (345), . . . , (n − 2, n − 1, n)} of 3-
cycles to write a constructor Alternating for this class of groups.
Dimino’s Algorithm Dimino’s algorithm is used to compute a complete

enumeration of the elements of a finite group, given a generating set for
the group. Suppose that we are given a generating set {g1 , g2 , . . . , gn } for
a finite group G. The idea behind Dimino’s algorithm is to enumerate,
successively, the elements of each of the subgroups
Gk = hg1 , g2 , . . . , gk i
of G, which form a chain
hg1 i = G1 ≤ G2 ≤ · · · ≤ Gk ≤ · · · ≤ Gn = G.
These elements can be enumerated by forming products of the genera-

tors g1 , g2 , . . . , gn in all possible ways, until all the elements of G have
been found. Dimino’s algorithm does this in a careful way, so as to avoid
computing unnecessary products.
We’ll use the following utility routine to determine the entries assigned
to a table. It can be used when we are certain no entry is a non-NULL ex-
pression sequence. Since it is sufficiently simple, it is defined with option
inline;.
> Entries := proc( T )
> description "return a set of simple table entries";
> option inline;
> map( op, { entries( T ) } )
> end proc:
Here is the code for Dimino’s algorithm.

> Dimino := proc( G::Group )
> description "enumerate the elements of a finite group";
> local s, g, ord, elements, i, j, prev_ord, rep_pos,
> elt, addElt, gens;
>
> if nargs > 1 then

> gens := args[ 2 ]
> else
> gens := G:-gens
> end if;
>
> if not type( gens, ’{ set, list }’ ) then
> error "no generating set specified"
> end if;
>
> if nops( gens ) = 0 then
> # trivial group
> return { G:-id }
> end if;
>
> addElt := proc( h )
> ord := 1 + ord;
> elements[ ord ] := h
> end proc;
>
> elements := table();
> ord := 0;
> addElt( G:-id );
>
> # Handle the first cyclic subgroup
> s := gens[ 1 ];
> g := s;
> while not G:-eq( g, G:-id ) do
> addElt( g );
> g := G:-‘.‘( g, s )
> end do;
> userinfo( 1, ’Dimino’, "finished first cycle; order is:", ord );
>
> for i from 2 to nops( gens ) do
> userinfo( 1, ’Dimino’, "Adding generator number:", i );
> s := gens[ i ];
> if not G:-member( s, Entries( elements ) ) then
> prev_ord := ord;
> addElt( s );
> for j from 2 to prev_ord do
> addElt( G:-‘.‘( elements[ j ], s ) )
> end do;
> rep_pos := 1 + prev_ord;
> do
> for s in gens[ 1 .. i ] do
> elt := G:-mul( elements[ rep_pos ], s );
> if not G:-member( elt, Entries( elements ) ) then
> addElt( elt );
> for j from 2 to prev_ord do
> addElt( G:-‘.‘( elements[ j ], elt ) )
> end do
> end if
> end do;
> rep_pos := rep_pos + prev_ord;
> if rep_pos > ord then

> break
> end if
> end do
> end if
> end do;
> Entries( elements )
> end proc:
The coding of this algorithm is generic. The exported members of the
group object G are used to effect computations within the procedure.
Even comparisons of equality use the export eq instead of Maple’s builtin
comparison operator ‘=‘. (The need for this will be seen below.)
Using the Symmetric constructor defined above, we can compute the
elements of the symmetric group S4 , using Dimino’s algorithm, as follows.
> G := Symmetric( 4 );
G := module()
export
option record ;
end module
> Dimino( G );
{[2, 1, 3, 4], [2, 3, 1, 4], [1, 2, 3, 4], [3, 2, 1, 4],

[2, 3, 4, 1], [3, 2, 4, 1], [1, 3, 4, 2], [3, 1, 4, 2],
[3, 4, 1, 2], [4, 3, 1, 2], [2, 4, 1, 3], [4, 2, 1, 3],
[3, 4, 2, 1], [4, 3, 2, 1], [4, 1, 2, 3], [1, 4, 2, 3],
[3, 1, 2, 4], [1, 3, 2, 4], [4, 1, 3, 2], [1, 4, 3, 2],
[4, 2, 3, 1], [2, 4, 3, 1], [1, 2, 4, 3], [2, 1, 4, 3]}
Anticipating later developments, we have coded the procedure Dimino
to accept a second, optional argument that specifies an alternate set of
generators to use. Thus, we could compute the same set using the set
{(12), (23), . . . , (n − 1, n)} of transpositions instead.
> Dimino( G, { [2,1,3,4], [1,3,2,4], [1,2,4,3] } );
{[2, 1, 3, 4], [2, 3, 1, 4], [1, 2, 3, 4], [3, 2, 1, 4],

[2, 3, 4, 1], [3, 2, 4, 1], [1, 3, 4, 2], [3, 1, 4, 2],
[3, 4, 1, 2], [4, 3, 1, 2], [2, 4, 1, 3], [4, 2, 1, 3],
[3, 4, 2, 1], [4, 3, 2, 1], [4, 1, 2, 3], [1, 4, 2, 3],
[3, 1, 2, 4], [1, 3, 2, 4], [4, 1, 3, 2], [1, 4, 3, 2],
[4, 2, 3, 1], [2, 4, 3, 1], [1, 2, 4, 3], [2, 1, 4, 3]}
We still need to pass the group object G for Dimino to access its
operations.
Dimino’s algorithm is a useful “fallback” algorithm, but many fi-
nite groups of interest can be enumerated more efficiently using specific
knowledge of their structure. For “small” examples, the implementation
presented here suffices, but a well-optimized implementation that takes
advantage of fast arithmetic for group elements would be required for
serious use.
Representing Subgroups A subset of a group that forms a group in its

own right (using the operations inherited from the group, by restriction)
is called a subgroup. For example, the 3-member set {(123), (132), (1)}
is a subgroup of the full symmetric group S3 of degree 3 (which has
6 members). There are a number of approaches we could take to the
representation of subgroups. One way is to represent a subgroup H of a
“known” group G by specifying a generating set for H and copying the
computational services from the representation of G to the representation
of H. Thus, the Maple representations G and H of G and H would both
be of type Group.
Instead, we shall adopt a different approach that is better suited to im-
plicit representations of subgroups. This design can be extended to allow
implicit representations of subgroups that we need not be able to compute
with directly. The idea is to represent a subgroup by a simpler structure
that maintains a link to its parent group and an indication of how it is
defined in terms of its parent group. Thus, a subgroup will be represented
by a module with an export parent that is assigned the group in which
the subgroup is contained. A second export has a name depending upon
the way in which the subgroup is defined. One way to define a subgroup
in terms of its parent is to specify a generating set. Subgroups defined in
this way will be represented by a module having the export gens of type
set. A second way to define a subgroup is by a “property”. For example,
the center of a group is defined by the property that all its members com-
mute with every element of the group (or, equivalently, that each among
its members commutes with all the generators of the parent group). We
can ask that properties be specified by requiring a procedure that tests
for membership in the subgroup. Thus, subgroups can be described by
either the following interfaces.
parent the parent group
test a membership test (a procedure)
gens a set of generators
Only one of the methods test and gens need be present. A Maple im-
plementation of this interface is as follows.
> ‘type/SubGroup‘ := ’{
> ‘module‘( parent::Group, gens::set ),
> ‘module‘( parent::Group, test::procedure )
> }’:
The SubGroup constructor must dispatch on the type of its second argu-
ment to determine which kind of record to create to model the subgroup.
> SubGroup := proc( G::{Group,SubGroup}, how::{set,procedure} )
> description "subgroup constructor";
> local S;
> if type( how, ’procedure’ ) then
> S:= Record( ’parent’, ’test’ = eval( how, 1 ) )
> else
> S := Record( ’parent’, ’gens’ = how )
> end if;
> S:-parent := G;
> eval( S, 1 )
> end proc:
For example, the center of the symmetric group S3 can be defined as
follows.
> S3 := Symmetric( 3 ):
> Z := SubGroup( S3, proc( z )
> local g;
> use S3 in
> for g in gens do
> if not eq( mul( inv( g ), inv( z ), g ), z ) then
> return false
> end if
> end do;
> end use;
> true
> end proc );
Z := module()
export parent , test ;
option record ;
end module
> Z:-test( [2,1,3] );
false
> Z:-test( [2,3,1] );
false
> Z:-test( [1,2,3] );
true
Similarly, we can write a constructor for the centralizer of an element

in a group.
> Centralizer := proc( G, g )
> SubGroup( G, proc( s )
> use ‘.‘ = G:-‘.‘, ‘=‘ = G:-eq in
> s . g = g . s
> end use end proc )
> end proc:
Generic Interfaces Dimino’s algorithm is fairly expensive. For many

classes of groups, better alternatives to enumerating group elements ex-
ist. We will take advantage of the opportunity to use them, relying on
Dimino’s algorithm only as a “last resort”. The advantage of Dimino’s
algorithm is that it works for any finite group. To provide a clean and
uniform interface to the enumeration functionality, we develop a fron-
tend procedure that takes care of hiding the details of how we go about
choosing the best available algorithm.
> GroupElements := proc( G )
> description "enumerate the elements of a finite group";
> if type( G, ’Group’ ) then
> if type( G:-elements, ’set’ ) then
> G:-elements
> elif type( G:-elements, ’procedure’ ) then
> G:-procedure()
> else
> G:-elements := Dimino( G )
> end if
> else
> ’procname’( args )
> end if
> end proc:
Several elements of the design allow us to take advantage of structural

knowledge to improve efficiency. This routine first checks whether the
export elements of its input group is of type set. If it is, then it is
taken to be a stored enumeration of the group elements and is simply
returned. Otherwise, if the export elements is a procedure, then it is
taken to be a (perhaps specialized) routine for computing the requested
enumeration. Finally, Dimino’s algorithm is used as a “last resort” if no
better alternative is provided. As a simple optimisation, the result of
Dimino’s algorithm is stored as a new value for the elements export so
that it need only be computed once.
Providing the GroupElements interface shields the user from having
to know what the available alternatives are and how to use them. An ad-
ditional benefit of the design is that it allows us to change the algorithm
selection criteria at any time (to correct software faults, or make func-
tional or performance improvements). Code using this interface need not
be modified, provided that the routine continues to honor its “contract”.
Enumerating Elements in Subgroups Once the elements of the parent

group are known, the members of the subgroup can be computed using a
call to the builtin Maple command select.
> select( C:-test, Dimino( G ) );
How best to enumerate the elements of a subgroup depends upon how

it is defined and what is known about the parent group. The procedure
SubGroupElements that follows takes a subgroup as argument and at-
tempts to find the “best” way to compute the elements of the subgroup
from among the available methods.
> SubGroupElements := proc( S )
> description "enumerate the elements of "
> "a subgroup of a group";
> local P;
> P := S;
> while type( P, ’SubGroup’ ) do
> P := P:-parent
> end do;
> if type( P, ’Group’ ) then
> if member( :-test, S ) then
> select( S:-test, GroupElements( P ) )
> else
> ASSERT( member( :-gens, S ) );
> Dimino( P, S:-gens )
> end if
> else
> end if
> end proc:

G := module()
export
option record ;
end module
> SubGroupElements( Centralizer( G, [ 1, 3, 2, 4 ] ) );
{[1, 2, 3, 4], [4, 3, 2, 1], [1, 3, 2, 4], [4, 2, 3, 1]}
With SubGroupElements implemented, it is a good idea to extend

GroupElements to accept subgroups also, thus providing a common in-
terface.
> GroupElements := proc( G )
> description "enumerate the elements of a "
> "group or subgroup";
> if type( G, ’SubGroup’ ) then
> SubGroupElements( G )
> elif type( G, ’Group’ ) then
> if type( G:-elements, ’set’ ) then
> G:-elements
> elif type( G:-elements, ’procedure’ ) then
> G:-elements()
> else
> G:-elements := Dimino( G )
> end if
> else
> end if
> end proc:
Computing the Order of a Group Since we have the capability of enu-

merating all of a group’s elements, it is always possible to determine its
order. (Note that this is rarely the best way to do this, however.) In many
cases, it is possible to provide much better ways to compute the order of
a group. For instance, the symmetric group of degree n has order equal
to n!, so its order export could be redefined to compute this number
instead.
A generic interface to computing group orders, in the same spirit as
GroupElements may be written as follows.
> GroupOrder := proc( G )

> description "compute the order of a finite group";
> if type( G, ’SubGroup’ ) then
> nops( GroupElements( G ) )
> elif type( G, ’Group’ ) then
> if type( G:-order, ’posint’ ) then
> G:-order
> elif type( G:-elements, ’set’ ) then
> G:-order := nops( G:-elements )
> elif type( G:-order, ’procedure’ ) then
> G:-order()
> else
> nops( GroupElements( G ) )
> end if
> else
> end if
> end proc:
As with GroupElements, this routine checks the possible “shortcuts”
that might be available for a group, beginning with those that are
likely to involve the least computation and progressing through more
and more costly alternatives. Only as a last resort does the procedure
call GroupElements to compute a full enumeration of the group elements
only to return their number.
G := module()
export
option record ;
end module
> C := Centralizer( G, [ 1, 3, 2, 4 ] );
C := module()
export parent , test ;
option record ;
end module
> GroupOrder( G );
24
> GroupOrder( C );
Note that, when the argument G is neither a group nor a subgroup, the
procedure GroupElements returns unevaluated. This allows us to extend
other Maple operations, such as expand, combine or simplify to be effec-
tive on algebraic expressions involving unevaluated calls to GroupOrder.
Matrix Groups So far, all our groups have been permutation groups
returned by one of the constructors presented above. If we are to have
any confidence in the genericity of the code we have developed, we must
test it on some other kinds of groups. A good source for examples of finite
groups are the finite groups of exact matrices.
Equality and Membership Tests for Matrices Because distinct

matrices with equal entries compare differently using Maple’s equality
comparison operator ‘=‘, it is necessary to implement a specialized test
for membership in a set. For example, consider the matrices
> A := Matrix( [[1,0],[0,1]] );
´ µ
1 0
A :=
0 1
> B := Matrix( [[2,3],[3,4]] );

´ µ
2 3
B :=
3 4
> C := Matrix( [[1,0],[0,1]] );

´ µ
1 0
C :=
0 1
Both A and C have the same entries, and represent mathematically

identical objects. However, because matrices are mutable data structures
(necessary for efficiency in matrix computations), they are distinct as
Maple objects. Thus, for instance, we see that:
> member( A, { B, C } );
false
To deal with this property of the data structures we are using, we

need to implement a generic version of the Maple command member.
This routine, gmember accepts arguments after its first that are like those
required by member, but must be passed an additional, first argument test
that is an equality test to use. We’ll use this utility in our implementation
of the matrix group constructor below.
> gmember := proc( test, g::anything, S::{set,list}, pos::name )
> description "a generic membership predicate";
> local i;
> if type( test, ’procedure’ ) then
> for i from 1 to nops( S ) do
> if test( g, S[ i ] ) then
> if nargs > 3 then
> pos := i
> end if;
> return true
> end if
> end do;
> false
> elif test = ’‘=‘’ then
> # use the standard membership test
> :-member( args[ 2 .. -1 ] )
> else
> end if
> end proc:
The builtin procedure Equal package provides an equality predicate that
is suitable for use with matrices.
> gmember( LinearAlgebra:-Equal, A, { B, C } );
true
The MatrixGroup Constructor Except for the member export, most

the the exported methods for matrix groups simply delegate to the ap-
propriate routine in the LinearAlgebra package. The MatrixGroup con-
structor takes the degree n of the matrix group as its first argument and,
if given more than one argument, takes the remaining ones to be matrices
that form a set of generators for the group.
> MatrixGroup := proc( n::posint )
> description "matrix group constructor";
> local matgens, G;
> use LinearAlgebra in
> matgens := { args[ 2 .. -1 ] };

> G := Record(
> ’id’ = Matrix( n, n, ( i, j ) -> ‘if‘( i = j, 1, 0 ) ),
> ’‘.‘’ = ( ( a, b ) -> MatrixMatrixMultiply( a, b ) ),
> ’mul’ = ( () -> foldl( G:-‘.‘, G:-id, args ) ),
> ’inv’ = ( m -> MatrixInverse( m ) ),
> ’gens’ = matgens,
> ’eq’ = ( ( a, b ) -> Equal( a, b ) ),
> ’member’ = proc( g, S, pos::name )
> local i, s;
> if nargs = 1 then
> if type( g, ’Matrix( square )’ ) then
> evalb( Determinant( g ) <> 0 )
> else
> false
> end if
> else
> gmember( G:-eq, args )
> end if
> end proc,
> ’order’, ’elements’ );
>
> if nargs = 1 then
> G:-order := 1;
> G:-elements := { G:-id }
> end if
> end use;
> eval( G, 1 )
> end proc:
Here, we use the matrix group constructor to generate a dihedral matrix
group of order 12.
> theta := Pi / 3;
1
θ := π
3
> a := Matrix( 2, 2, [[ 0, 1 ], [ 1, 0 ]] );
´ µ
0 1
a :=
1 0
> b := Matrix( 2, 2,
> [[cos(theta),sin(theta)],
> [-sin(theta),cos(theta)]] );
 1 1√ 
3
2 2
b := 
 
1 √ 1

− 3
2 2
> B := MatrixGroup( 2, a, b );
B := module()
export
id , ‘.‘, mul , inv , gens , eq , member , order , elements ;
option record ;
end module
> GroupElements( B );
 1 1√   1√ 1   1√ 1 
3 − 3 3
2 2   2 2  2 2

,  ,  ,
 
1√ 1 1 1√ 1 1√

− −

3 3 3
2 2 2 2 2 2
 −1 1√   1 1√   1√ −1  ´
3 − 3 − 3 µ
2 2   2 2   2 2  0 1
,  √ ,  , ,

1√ −1 1 1 −1 1√ 1 0

− 3 3 3
2 2 2 2 2 2
 1√ −1   −1 1√  
´ µ 3 − 3 ´ µ ´ µ
1 0 2 2   2 2 −1 0 0 −1
, ,  √ , ,

0 1 −1 1√ 1 −1 0 −1 −1 0 
− 3 3
2 2 2 2
Direct Products To enrich the supply of example groups that we can
work with, we’ll develop a constructor for the direct product of (two)
groups. (Extending the constructor to handle any finite number of groups
is straight-forward, but complicates the exposition unnecessarily.) Direct
products are very important in the study of finite groups because all
finitely generated abelian groups possess an unique “factorisation” as a
direct product of cyclic groups. (In the abelian theory, direct products are
often referred to as direct sums.)
The direct product of two groups A and B is the group G whose
elements are all pairs (a, b), with a ∈ A and b ∈ B. The group product in
G is defined by (a1 , b1 )·(a2 , b2 ) = (a1 ·a2 , b1 ·b2 ). The inverse of an element
(a, b) is the pair (a−1 , b−1 ). All the operations are defined component-wise.
We can represent the elements (a, b) of the direct product by two-element

lists. Here is the constructor DirectProduct.
> DirectProduct := proc( A::Group, B::Group )
> description "direct product constructor";
> local G, a, b;
> if type( A, ’Group’ ) and type( B, ’Group’ ) then
> G := Group();
> G:-id := [ A:-id, B:-id ];
> G:-‘.‘ := ( u, v ) -> [ A:-‘.‘( u[1], v[1] ),
> B:-‘.‘( u[2], v[2] ) ];
> G:-mul := () -> foldl( G:-‘.‘, G:-id, args );
> G:-inv := v -> [ A:-inv( v[ 1 ] ),
> B:-inv( v[ 2 ] ) ];
> G:-gens := [ seq( seq( [ a, b ],
> a = A:-gens ), b = B:-gens ) ];
> G:-eq := ( u, v ) -> A:-eq( u[ 1 ], v[ 1 ] )
> and B:-eq( u[ 2 ], v[ 2 ] );
> G:-order := () -> GroupOrder( A ) * GroupOrder( B );
> G:-member := proc( g, S, pos::name )
> if nargs = 1 then
> A:-member( g[ 1 ] )
> and B:-member( g[ 2 ] )
> else
> gmember( G:-eq, args )
> end if
> end proc;
> G:-elements := () -> [ seq( seq( [ a, b ],
> a = GroupElements( A ) ), b = GroupElements( B ) ) ];
> eval( G, 1 )
> else
> end if
> end proc:
Most of the group methods are quite straightforward, but note that we
take the opportunity to use the known group structure to reduce the com-
plexity of some computations such as those for the order and elements
exports.
> A := Symmetric( 3 ):
> G := DirectProduct( A, B ):
> GroupOrder( G );
72
> nops( GroupElements( G ) );
72
Homomorphisms In all algebraic theories, homomorphisms play a key

role. A group homomorphism is a mapping from a group to another
(possibly the same) group which commutes with the group operations.
That is, a map ϕ : A −→ B of groups A and B is a homomorphism if
ϕ(ab) = ϕ(a)ϕ(b), for all a and b in A. A homomorphism is determined
uniquely by its effect on a generating set for its domain, so to define a
homomorphism, it is enough to specify the images of each among a set of
generators for the domain.
We’ll use the following interface for homomorphisms.
domain the domain of the homomorphism

codomain the codomain of the homomorphism
genmap the mapping of the generators of the domain into the codomain
This lead directly to a fairly simple constructor for homomorphism ob-
jects.
> ‘type/Homomorphism‘ := ’‘module‘( domain, codomain, genmap )’:
> Homomorphism := proc( A::Group, B::Group, p::procedure )
> description "homomorphism constructor";
> Record( ’domain’ = A, ’codomain’ = B, ’genmap’ = p )
> end proc:
The image of a group homomorphism ϕ : A −→ B is the subset ϕ(A) of
B consisting of all elements of B having the form ϕ(a), for some element
a in A. It is a subgroup of B. The various design choices we’ve made along
the way lead to a very simple formulation for computing or representing
images of homomorphisms.
> HomImage := proc( hom::Homomorphism )
> description "compute the image of a homomorphism";
> SubGroup( hom:-codomain,
> map( hom:-genmap, hom:-domain:-gens ) )
> end proc:
As an example computation, we compute the image of a homomorphism
from the symmetric group S4 onto a two-element matrix group generated
by the reflection
> Matrix( [ [ 0, 1 ], [ 1, 0 ] ] );
´ µ
0 1
1 0
First, define the groups.

> A := Symmetric( 4 ):
> B := MatrixGroup( 2, Matrix( [[0,1],[1,0]] ) ):
We can define a mapping from the generators of A to the group B by
inserting the images of the generators into a procedure’s remember table.
> h( [2,1,3,4] ) := Matrix( [[0,1],[1,0]] ):
> h( [2,3,4,1] ) := Matrix( [[1,0],[0,1]] ):
This defines a Maple procedure h that performs the indicated mapping
and returns unevaluated for any other arguments.
> eval( h );
proc() option remember ; ’procname(args)’ end proc
Now we use A, B and h to construct the homomorphism object.

> hom := Homomorphism( A, B, h );
hom := module()
export domain, codomain, genmap;
option record ;
end module
> type( hom, ’Homomorphism’ );
true
We can now use the machinery developed earlier in this example to

compute the order of the image of this homomorphism.
> GroupOrder( HomImage( hom ) );
Thus, we see that the homomorphism is surjective (as expected). We

can compute the elements explicitly.
> GroupElements( B );
´ µ ´ µ
0 1 1 0
{ , }
1 0 0 1
> GroupElements( HomImage( hom ) );

´ µ ´ µ
1 0 0 1
{ , }
0 1 1 0
1. An automorphism α of a group G is called inner if there is an

element a in G for which α(g) = a−1 ga, for all g in G. Write a
constructor for inner automorphisms of groups.
Summary The power of generic programming is that we need only im-

plement computation in quotient fields or groups once — in the construc-
tors and generic procedures. The functor QuotientField and the various
generic group constructors and procedures are parameterized by the com-
putational domains upon which their computed values depend. We view
rings, fields, groups and subgroups as collections of computational capabil-
ities, which we use to construct new instances with derived computational
capabilities. The ability to override default methods (which may not be
efficient, but at least, are always present) with better methods that take
advantage of specific structural information allows for efficient computa-
tion without sacrificing generality. This leads to a powerful paradigm for
software reuse, and is the principal motivation underlying Maple’s module
system.
6.7 Conclusion
This chapter introduced the concept of Maple modules. It described the
structure and flexibility of modules.
Encapsulation and generic programming with modules allow you to
write code that can be reused, transported, and easily maintained. By
collecting procedures into a module called a package, you can organize
your procedures into distinct sets of related functions. You can also use
modules to implement objects in Maple.
The descriptions in this chapter are complemented by numerous ex-
amples to help you learn the syntax and semantics of modules and provide
you with modules that can be customized and used in your own work.
7 Debugging Maple
Programs
New programs, whether developed in Maple or any other language, often

work incorrectly. Problems that occur in the execution of a program are
usually due to syntax errors introduced during implementation, or logic
errors in the design of the algorithm. Most errors are subtle and hard
to find by visual inspection of the program alone. Maple provides error
detection commands and a debugger to help you find these errors.
The Maple debugger lets you stop execution within a Maple proce-
dure, inspect and modify the values of local and global variables, and
continue execution, either to completion, or one statement or block at a
time. You can stop execution when Maple reaches a particular statement,
when it assigns some value to a particular local or global variable, or
when a particular error occurs. This facility lets you investigate the inner
workings of a program to determine why it is not doing what you expect.
Alternatively, Maple provides a variety of commands to help you find
errors in procedures. Among these are commands to trace procedure exe-
cution, check assertions, raise exceptions and trap errors, and verify pro-
cedure semantics and syntax.
Even when a program is working correctly, you may want to ana-
lyze its performance to try to improve its efficiency. Maple commands
are available to analyze the time and memory consumption involved in
running the program.
7.1 A Tutorial Example

The Maple debugger is a tool that you can use to detect errors in your
procedures. Using this facility, you can follow the step-by-step execution
349
350 • Chapter 7: Debugging Maple Programs
of your program to determine why it is not returning the results that you
expect.
This section illustrates how to use the Maple debugger as a tool for
debugging a Maple procedure. The debugger commands are introduced
and described as they are applied. Additional information about the com-
mands is provided in Section 7.2.
The following procedure, sieve, is used as a case study. It implements
the Sieve of Eratosthenes : given a parameter n, return a count of the
prime numbers less than n, inclusive. To debug the sieve procedure, we
use breakpoints and watchpoints, which cause Maple to stop the execution
of the procedure.
> sieve := proc(n::integer)
> local i, k, flags, count,twicei;
> count := 0;
> flags[i] := true
> end do;
> if flags[i] then
> twicei := 2*i;
> for k from twicei by i to n do
> flags[k] = false;
> end do;
> count := count+l
> end if;
> end do;
> count;
> end proc:
Numbering the Procedure Statements I

To use the Maple debugger, you must enter a variety of debugger com-
mands. Many of these debugger commands must refer to statements
within the procedures that you are debugging. Statement numbers al-
low such references. The showstat command displays a Maple procedure
along with numbers preceeding each line that begins a new statement.
> showstat(sieve);
sieve := proc(n::integer)
local i, k, flags, count, twicei;
1 count := 0;
2 for i from 2 to n do
3 flags[i] := true
end do;
5 if flags[i] then
7.1 A Tutorial Example • 351
6 twicei := 2*i;
7 for k from twicei by i to n do
8 flags[k] = false
end do;
9 count := count+l
end if
end do;
10 count
end proc
Note that the numbers preceeding each line differ from line numbers
that may display in a text editor. For example, keywords that end a state-
ment (such as end do and end if) are not considered separate commands
and are therefore not numbered.
Invoking the Debugger I

To invoke the Maple debugger you must start the execution of a procedure,
and the execution must be made to stop within the procedure. To execute
a Maple procedure, call it by using a Maple command at the top-level,
or call it from another procedure. The simplest way to cause execution to
stop within the procedure, is to set a breakpoint in the procedure.
Setting a breakpoint Use the stopat command to set a breakpoint in

procedure sieve.
> stopat(sieve);
[sieve]
This command sets a breakpoint before the first statement in proce-

dure sieve. When you subsequently execute sieve, Maple stops before
executing the first statement. When execution does stop, the debugger
prompt appears (DBG>).1
The following example demonstrates an initial execution of sieve.
> sieve(10);
sieve:
1* count := 0;
1
If a procedure has a remember table, you may have to execute a restart command
before issuing a second or subsequent stopat command. For more information about
remember tables, see ?remember.
Preceding the debugger prompt are several pieces of information.
• The previously computed result. (This particular execution stopped

at the first statement before making any computations, so no result
appears.)
• The name of the procedure in which execution has stopped is sieve.
• Execution is stopped before statement number 1. An asterisk (*) fol-

lows this statement number to indicate that a breakpoint was set
before the statement.
At the debugger prompt, you can evaluate Maple expressions, and

invoke debugger commands. Maple evaluates expressions in the context
of the stopped procedure. You have access to exactly the same procedure
parameters, and local, global, and environment variables, as the stopped
procedure. For example, since sieve was called with parameter value 10,
the formal parameter n has the value 10.
DBG> n
10
sieve:
1* count := 0;
Notice that for each expression that Maple evaluates, it displays:
• the result of the expression,
• the name of the stopped procedure,
• the statement number where the procedure stopped followed by the

statement, and
• a new debugger prompt.
Note: To remove a breakpoint from a procedure, use the unstopat

command.
Controlling Execution of a Procedure during Debugging I

Debugger commands control the execution of the procedure once the de-
bugger is active. Some commonly used debugger commands are next,
step, list, into, outfrom, and cont.
The next command executes the next statement at the current nesting
level. After the statement is executed, control is returned to the debugger
as though a breakpoint had been set. If the statement is a control struc-
ture (an if statement or a loop), the debugger executes any statements
within the control structure that it would normally execute. It stops ex-
ecution before the next statement after the control structure. Similarly,
if the statement contains calls to procedures, the debugger executes these
procedure calls in their entirety before execution stops.
DBG> next
0
sieve:
...
end do;
The 0 in the first line of the output represents the result of the exe-
cuted statement—that is, the result of count := 0. A “*” does not appear
next to the statement number because there is no breakpoint set immedi-
ately before statement 2. The debugger does not show the body of the for
loop, which itself consists of statements with their own statement num-
bers, unless execution actually stops within its body. Maple represents
the body of compound statements by ellipses (...).
Executing the next command again results in the following output.
DBG> next
true
sieve:
...
end do;
Execution now stops before statement 4. Statement 3 (the body of

the previous for loop) is at a deeper nesting level. Therefore, the loop
is executed n-1 times. The debugger displays the last result computed in
the loop (the assignment of the value true to flags[10]).
If you want to step into a nested control structure (such as an if
statement or for loop) or a procedure call, use the step debugger com-
mand.
DBG> step
true
sieve:
5 if flags[i] then
...
end if
DBG> step
true
sieve:
6 twicei := 2*i;
If you use the step debugger command when the next statement to
execute is not a deeper structured statement, it has the same effect as
the next debugger command.
DBG> step
4
sieve:
...
end do;
At any time during the debugging process, you can use the showstat
debugger command to display the current status of the debugging process.
DBG> showstat
1* count := 0;
3 flags[i] := true
end do;
5 if flags[i] then
6 twicei := 2*i;
7 ! for k from twicei by i to n do
8 flags[k] = false
end do;
9 count := count+l
end if
end do;
10 count
end proc
Notice that the debugger prompt is displayed to indicate that you

are still working inside the Maple debugger. The asterisk (*) marks the
unconditional breakpoint. An exclamation point (!) that follows a state-
ment number (see line 7) indicates the statement at which the procedure
is stopped.
To continue the debugging process, issue another debugger command.
For example, you can use into or step to enter the innermost loop. The
into debugger command is a compromise between the next and step
commands. Execution stops at the next statement within the current
procedure regardless of whether it is at the current nesting level or within
the body of a control structure (an if statement or a loop). In other words,
the into command steps into nested statements, but not into procedure
calls.
DBG> into
4
sieve:
8 flags[k] = false
A debugger command that is related to showstat is the list com-

mand. It displays the previous five statements, the current statement, and
the next statement, to quickly provide some idea of where the procedure
has stopped.
DBG> list
...
3 flags[i] := true
end do;
5 if flags[i] then
6 twicei := 2*i;
8 ! flags[k] = false
end do;
9 count := count+l
end if
end do;
...
end proc
You can use the outfrom debugger command to finish execution at

the current nesting level or deeper. Execution of the procedure is stopped
once a statement at a shallower nesting level is reached, that is, after a
loop terminates, a branch of an if statement executes, or the current
procedure call returns.
DBG> outfrom
true = false
sieve:
9 count := count+l
DBG> outfrom
l
sieve:
5 if flags[i] then
...
end if
The cont debugger command continues execution, until either the

procedure terminates normally or it encounters another breakpoint.
DBG> cont
9l
You can now see that the procedure does not give the expected output.
Although you may find the reason obvious from the previous debugger
command examples, in most cases it is not easy to find procedure errors.
Therefore, pretend not to recognize the problem, and continue to use the
debugger. First, use the unstopat command to remove the breakpoint
from sieve.
> unstopat(sieve);
[]
Invoking the Debugger II

The procedure sieve keeps track of the changing result in the variable
count. Therefore, a logical place to look during debugging is wherever
Maple modifies count. The easiest way to do this is by using a watchpoint,
which invokes the debugger whenever Maple modifies a watched variable.
Setting a watchpoint Use the stopwhen command to set watchpoints.

In this case, you want to stop execution whenever Maple modifies the
variable count in the procedure sieve.
> stopwhen([sieve,count]);
[[sieve, count ]]
The stopwhen command returns a list of all the variables currently

being watched.
Execute the sieve procedure again.
> sieve(10);
count := 0
sieve:
...
end do;
Execution stops because Maple has modified count, and the debugger
displays the assignment statement count := 0. As in the case of break-
points, the debugger then displays the name of the procedure and the next
statement to be executed in the procedure. Note that execution stops af-
ter Maple has assigned a value to count.
This first assignment to count is correct. Use the cont debugger com-
mand to continue execution of the procedure.
DBG> cont
count := l
sieve:
5 if flags[i] then
...
end if
If you do not look carefully, this also looks correct. Assume that noth-
ing is wrong and continue execution.
DBG> cont
count := 2*l
sieve:
5 if flags[i] then
...
end if
This output is suspicious because Maple should have simplified 2*1.

Notice that it has printed 2*l (two times the letter l) instead. By studying
the source text for the procedure, you can see that the letter “l” was
typed instead of the number “1”. Since the source of the error has been
discovered, there is no reason to continue the execution of the procedure.

Use the quit debugger command to exit the debugger and then use the
unstopwhen command to remove the watchpoint from the procedure.
DBG> quit
Warning, computation interrupted

> unstopwhen();
[]
After correcting the source text for sieve, issue a restart command,
read the corrected version of sieve back into Maple, and execute the
procedure again.
> sieve(10);
9l
This result is still incorrect. There are four primes less than 10, namely
2, 3, 5, and 7. Therefore, invoke the debugger once more, stepping into the
innermost parts of the procedure to investigate. Since you do not want to
start at the beginning of the procedure, set the breakpoint at statement
6.
> stopat(sieve,6);
[sieve]
> sieve(10);
true
sieve:
6* twicei := 2*i;
DBG> step
4
sieve:
...
end do;
DBG> step
4
sieve:
8 flags[k] = false
DBG> step
true = false
sieve:
8 flags[k] = false
The last step reveals the error. The previously computed result should
have been false (from the assignment of flags[k] to the value false),
but instead true = false was returned. An equation was used instead
of an assignment. Therefore, Maple did not set flags[k] to false.
Once again, exit the debugger and correct the source text.
DBG> quit

The following code represents the corrected procedure.
> sieve := proc(n::integer)
> local i, k, flags, count,twicei;
> count := 0;
> flags[i] := true
> end do;
> if flags[i] then
> twicei := 2*i;
> for k from twicei by i to n do
> flags[k] := false;
7.2 Maple Debugger Commands • 361
> end do;

> count := count+1
> end if;
> end do;
> count;
> end proc:
Execute procedure sieve again to test the corrections.

> sieve(10);
The sieve procedure now returns the correct result.
7.2 Maple Debugger Commands

This section provides additional details about the commands used in the
tutorial in Section 7.1 and a description of other debugger commands.
Numbering the Procedure Statements II

The showstat command is called by using the following syntax. The
procedureName parameter is optional.
showstat( procedureName );
If showstat is called with no arguments, all procedures that contain

breakpoints are displayed.
You can also use the showstat command to display a single statement
or a range of statements by using the following syntax.
showstat( procedureName, number );

showstat( procedureName, range );
In these cases, the statements that are not displayed are represented
by ellipses (...). The procedure name, its parameters, and its local and
global variables are always displayed.
> f := proc(x)
> if x <= 2 then
> print(x);
> end if;
> print(-x)
> end proc:
> showstat(f, 2..3);
f := proc(x)
...
2 print(x)
end if;
3 print(-x)
end proc
Invoking the Debugger III

This section provides additional information about breakpoints and
watchpoints.
Setting breakpoints The stopat command is called by using the follow-

ing syntax where procedureName is the name of the procedure in which
to set the breakpoint, statementNumber is the line number of the state-
ment in the procedure before which the breakpoint is set, and condition
is a Boolean expression which must be true for execution to stop. The
statementNumber and condition arguments are optional.
stopat( procedureName, statementNumber, condition );
This condition argument can refer to any global variable, local vari-
able, or parameter of the procedure. These conditional breakpoints are
marked by a question mark (?) if showstat is used to display the proce-
dure.
Since the stopat command sets the breakpoint before the specified
statement, when Maple encounters a breakpoint, execution stops and
Maple engages the debugger before the statement. This means that
it is not possible to set a breakpoint after the last statement in a
statement sequence—that is, at the end of a loop body, an if state-
ment body, or a procedure.
If two identical procedures exist, depending on how you created them,
they may share breakpoints. If you entered the procedures individually,
with identical procedure bodies, then they do not share breakpoints. If
you created a procedure by assigning it to the body of another procedure,
then their breakpoints are shared.
> f := proc(x) x^2 end proc:

> g := proc(x) x^2 end proc:
> h := op(g):
> stopat(g);
[g, h]
> showstat();
g := proc(x)
1* x^2
end proc
h := proc(x)
1* x^2
end proc
Removing breakpoints The unstopat command is called by using the

following syntax where procedureName is the name of the procedure that
contains the breakpoint, and statementNumber is the line number of the
statement where the breakpoint is set. The statementNumber parameter
is optional.
unstopat( procedureName, statementNumber );
If statementNumber is omitted in the call to unstopat, then all

breakpoints in procedure procedureName are cleared.
Setting explicit breakpoints You can set an explicit breakpoint by in-

serting a call to the DEBUG command in the source text of a procedure. The
DEBUG command is called by using the following syntax. The argument
parameter is optional.
DEBUG( argument );
If no argument is included in the DEBUG command, execution in the

procedure stops at the statement following the location of the DEBUG
command and the debugger is invoked.2
2
The showstat command does not mark explicit breakpoints with a “*Ô or a “?Ô.
> f := proc(x,y) local a;

> a:=x^2;
> DEBUG();
> a:=y^2;
> end proc:
> showstat(f);
f := proc(x, y)
local a;
1 a := x^2;
2 DEBUG();
3 a := y^2
end proc
> f(2,3);
4
f:
3 a := y^2
DBG> quit

If the argument of the DEBUG command is a Boolean expression, then
execution stops only if the Boolean expression evaluates to true. If the
Boolean expression evaluates to false or FAIL, then the DEBUG command
is ignored.
> f := proc(x,y) local a;
> a:=x^2;
> DEBUG(a<1);
> a:=y^2;
> DEBUG(a>1);
> print(a);
> end proc:
> f(2,3);
9
f:
5 print(a)
DBG> quit

If the argument of the DEBUG command is anything but a Boolean
expression, then the debugger prints the value of the argument instead of
the last result when execution stops at the following statement.
> f := proc(x)
> x^2;
> DEBUG("This is my breakpoint. The current value of x is:", x);
> x^3
> end proc:
> f(2);
"This is my breakpoint. The current value of x is:"
2
f:
3 x^3
Removing explicit breakpoints The unstopat command cannot remove

explicit breakpoints. You must remove breakpoints that were set using
DEBUG by editing the source text for the procedure.
DBG> unstopat
[f, g, h]
f:
3 x^3
DBG> showstat
f := proc(x)
1 x^2;
2 DEBUG("This is my breakpoint. The current value of x is:",x);
3 ! x^3
end proc
DBG> quit

Note: If you display the contents of a procedure by using print (or
lprint) and the procedure contains a breakpoint that was set by using
stopat, the breakpoint appears as a call to DEBUG.
> stopat(f);
[f, g, h]
> print(f);
proc(x) DEBUG() ; x2 end proc
Setting watchpoints The stopwhen command is called by using one of

the following.
stopwhen( globalVariableName );
stopwhen( [procedureName, variableName ] );
The first form specifies that the debugger should be invoked whenever
the global variable globalVariableName is changed. Maple environment
variables, such as Digits, can also be monitored by using this method.
> stopwhen(Digits);
[Digits ]
The second form invokes the debugger whenever the (local or global)
variable variableName is changed in the procedure procedureName.
When stopwhen is called in either form or with no arguments, Maple
returns a list of the watchpoints that are currently set.
Execution stops after Maple has already assigned a value to the
watched variable. The debugger displays an assignment statement in-
stead of the last computed result (which would be the right-hand side
of the assignment statement).
Clearing watchpoints The syntax to call unstopwhen is the same as

that for stopwhen. Similar to stopwhen, unstopwhen returns a list of all
(remaining) watchpoints.
If no arguments are entered in the call to unstopwhen, then all watch-
points are cleared.
Setting watchpoints on specified errors You can use an error watch-

point to invoke the debugger whenever Maple returns a specified error
message. When a watched error occurs, execution of the procedure stops
and the debugger displays the statement in which the error occurred.
Error watchpoints are set by using the stoperror command. The
stoperror command is called by using the following syntax
stoperror( "errorMessage" );
where errorMessage is a string or a symbol that represents the er-

ror message returned from a procedure. A list of the currently set error
watchpoints is returned.
If no argument is entered in the call to stoperror, the current list of
watchpoints is returned.
> stoperror();
[]
> stoperror( "numeric exception: division by zero" );
[“numeric exception: division by zero”]

> stoperror();
If the special name ‘all‘ is used instead of a specific error message

as the parameter to the stoperror command, execution of a procedure
stops whenever any error that would not be trapped occurs.
Errors trapped by a traperror construct (try...catch statement)
do not generate an error message and so stoperror cannot be used to
catch them. For more information about the try...catch structure, see
Trapping errors on page 390. If the special name ‘traperror‘ is used
instead of a specific error message as the parameter to the stoperror
command, execution of a procedure stops whenever any error that is
trapped occurs. If the errorMessage parameter is entered in the form
traperror["message"] to stoperror, the debugger is invoked only if
the error specified by "message" is trapped.
When execution of a procedure stops due to an error which causes
an exception, continued execution is not possible. Any of the execution
control commands, such as next or step (see Controlling Execution of
a Procedure during Debugging I and II in Section 7.1 and this section,
respectively), process the error as if the debugger had not intervened. For
example, consider the following two procedures. The first procedure, f,
calculates 1/x. The other procedure, g, calls f but traps the "division
by zero" error that occurs when x = 0.
> f := proc(x) 1/x end:
> g := proc(x) local r;
> try
> f(x);
> catch:
> infinity;
> end try;
> end proc:
If procedure g is executed at x=9, the reciprocal is returned.

> g(9);
1
9
At x=0, as expected, infinity is returned.

> g(0);
The stoperror command stops execution when you call f directly.

> stoperror("numeric exception: division by zero");
> f(0);
f:
1 1/x
DBG> cont
Error, (in f) numeric exception: division by zero
The call to f from g is inside a traperror (try...catch statement),

so the "division by zero" error does not invoke the debugger.
> g(0);
Instead, try to use stoperror(traperror).

> unstoperror( "numeric exception: division by zero" );
[]
> stoperror( ‘traperror‘ );
[traperror ]
This time Maple will not stop at the error in f.

> f(0);
However, Maple invokes the debugger when the trapped error occurs.
> g(0);
f:
1 1/x
DBG> step
g:
3 infinity
DBG> step
In the case that a particular error message is specified in the form

traperror["message"], the debugger is invoked only if the error specified
by "message" is trapped.
Clearing watchpoints on specified errors Error watchpoints are cleared

by using the top-level unstoperror command. The syntax to call unstoperror
is the same as for stoperror. Like stoperror, unstoperror returns a list
of all (remaining) error watchpoints.
If no argument is entered in the call to unstoperror, then all error
watchpoints are cleared.
> unstoperror();
[ ]
Controlling Execution of a Procedure during Debugging II

Once the execution of a procedure is stopped and the debugger is invoked,
you can examine the values of variables or perform other experiments (see
Changing the State of a Procedure during Debugging ). After you have
examined the state of the procedure, you can cause execution to continue
by using a number of different debugger commands.
The most commonly used debugger commands are cont, next, step,
into, outfrom, return, and quit.
The return debugger command causes execution of the currently ac-
tive procedure call to complete. Execution stops at the first statement
after the current procedure.
The other commands were described in the tutorial in Section 7.1.
For more information about these and other debugger commands, see
?debugger.
Changing the State of a Procedure during Debugging

When a breakpoint or watchpoint stops the execution of a procedure, the
Maple debugger is invoked. In the debugger mode, you can examine the
state of the global variables, local variables, and parameters of the stopped
procedure. You can also determine where execution stopped, evaluate
expressions, and examine procedures.
While in the debugger mode, you can evaluate any Maple expres-
sion and perform assignments to local and global variables. To evaluate
an expression, type the expression at the debugger prompt. To perform
assignments to variables, use the standard Maple assignment statement.
> stopat(f);
[f ]
> f(10);
f:
1* x^2
DBG> sin(3.0)
.1411200081
f:
1* x^2
DBG> cont
100
The debugger evaluates any variable names that you use in the ex-
pression in the context of the stopped procedure. Names of parameters
or local variables evaluate to their current values within the procedure.
Names of global variables evaluate to their current values. Environment
variables, such as Digits, evaluate to their values in the stopped proce-
dure’s environment.
If an expression corresponds to a debugger command (for example,
your procedure has a local variable named step), you can still evaluate
it by enclosing it in parentheses.
> f := proc(step) local i;
> for i to 10 by step do
> i^2
> end do;
> end proc:
> stopat(f,2);
[f ]
> f(3);
f:
2* i^2
DBG> step
1
f:
2* i^2
DBG> (step)
3
f:
2* i^2
DBG> quit

While execution is stopped, you can modify local and global vari-
ables by using the assignment operator (:=). The following example sets
a breakpoint in the loop only when the index variable is equal to 5.
> sumn := proc(n) local i, sum;
> sum := 0;
> for i to n do
> sum := sum + i
> end do;
> end proc:
> showstat(sumn);
sumn := proc(n)
local i, sum;
1 sum := 0;
2 for i to n do
3 sum := sum+i
end do
end proc
> stopat(sumn,3,i=5);
[sumn]
> sumn(10);
10
sumn:
3? sum := sum+i
Reset the index to 3 so that the breakpoint is encountered again.

DBG> i := 3
sumn:
3? sum := sum+i
DBG> cont
17
sumn:
3? sum := sum+i
Maple has added the numbers 1, 2, 3, 4, 3, and 4 and returned 17 as

the result. Continuing the execution of the procedure, the numbers 5, 6,
7, 8, 9, and 10 are added and 62 is returned as the result.
DBG> cont
62
Examining the State of a Procedure during Debugging

There are two debugger commands available that return information
about the state of the procedure execution. The list debugger command
shows you the location within a procedure where execution stopped, and
the where debugger command shows you the stack of procedure activa-
tions.
The list debugger command has the following syntax.
list procedureName statementNumber[..statNumber]
The list debugger command is similar to showstat, except in the

case where you do not specify any arguments. If no arguments are in-
cluded in the call to list, only the five previous statements, the current
statement, and the next statement to be executed are displayed. This pro-
vides some context in the stopped procedure. In other words, it indicates
the static position where execution stopped.
The where debugger command shows you the stack of procedure ac-
tivations. Starting from the top-level, it shows you the statement that
is executing and the parameters it passed to the called procedure. The
where debugger command repeats this for each level of procedure call
until it reaches the current statement in the current procedure. In other
words, it indicates the dynamic position where execution stopped. The
where command has the following syntax.
where numLevels
To illustrate these commands, consider the following example. The

procedure check calls the sumn procedure from the previous example.
> check := proc(i) local p, a, b;
> p := ithprime(i);
> a := sumn(p);
> b := p*(p+1)/2;
> evalb( a=b );
> end proc:
There is a (conditional) breakpoint in sumn.

> showstat(sumn);
sumn := proc(n)
local i, sum;
1 sum := 0;
2 for i to n do
3? sum := sum+i
end do
end proc
When check calls sumn, the breakpoint invokes the debugger.

> check(9);
10
sumn:
3? sum := sum+i
The where debugger command reveals that:
• check was invoked from the top-level with argument 9,

• check called sumn with argument 23, and
• execution stopped at statement number 3 in sumn.

DBG> where
TopLevel: check(9)
[9]
check: a := sumn(p)
[23]
sumn:
3? sum := sum+i
DBG> cont
true
The next example illustrates the use of where in a recursive function.

> fact := proc(x)
> if x <= 1 then
> 1
> else
> x * fact(x-1)
> end if;
> end proc:
> showstat(fact);
fact := proc(x)
1 if x <= 1 then
2 1
else
3 x*fact(x-1)
end if
end proc
> stopat(fact,2);
[fact ]
> fact(5);
fact:
2* 1
DBG> where
TopLevel: fact(5)
[5]
fact: x*fact(x-1)
[4]
fact: x*fact(x-1)
[3]
fact: x*fact(x-1)
[2]
fact: x*fact(x-1)
[1]
fact:
2* 1
If you are not interested in the entire history of the nested procedure
calls, then use the numLevels parameter in the call to where to print out
only a certain number of levels.
DBG> where 3
fact: x*fact(x-1)
[2]
fact: x*fact(x-1)
[1]
fact:
2* 1
DBG> quit

The showstop command (and the showstop debugger command) dis-
plays a report of all currently set breakpoints, watchpoints, and error
watchpoints. Outside the debugger at the top-level, the showstop com-
mand has the following syntax.
showstop();
The next example illustrates the use of showstop.

> f := proc(x) local y;
> if x < 2 then
> y := x;
> print(y^2);
> end if;
> print(-x);
> x^3;
> end proc:
Set some breakpoints.

> stopat(f):
> stopat(f,2):
> stopat(int);
[f, int ]
Set some watchpoints.

> stopwhen(f,y):
> stopwhen(Digits);
[[f, y], Digits ]
Set an error watchpoint.

> stoperror( "numeric exception: division by zero" );
The showstop command reports all the breakpoints and watchpoints.

> showstop();
Breakpoints in:
f
int
Watched variables:
y in procedure f
Digits
Watched errors:
"numeric exception: division by zero"
Using Top-Level Commands at the Debugger Prompt

The showstat, stopat, unstopat, stopwhen, unstopwhen, stoperror,
and showstop commands can be used at the debugger prompt. The fol-
lowing list describes the syntax rules for top-level commands used at the
debugger prompt.
• Do not enclose the arguments of the command in parentheses.
• Do not separate the arguments of the command with commas. The

arguments must be separated by a space character.
• Do not use colons or semicolons to end statements.
• The procedure name is not required by any command. Commands

that use a procedure name assume the currently stopped procedure if
one is not specified.
• For the stoperror command, the quotation marks ("") are not re-
quired.
Except for these rules, the debugger prompt call for each command
is of the same form and takes the same arguments as the corresponding
top-level command call.
Restrictions
At the debugger prompt, the only permissible Maple statements are de-
bugger commands, expressions, and assignments. The debugger does not
permit statements such as if, while, for, read, and save. However, you
can use ‘if‘ to simulate an if statement, and seq to simulate a loop.
The debugger cannot set breakpoints in, or step into, built-in kernel
routines, such as diff and has. These routines are implemented in C
and compiled into the Maple kernel. Debugging information about these
routines is not accessible to Maple since the routines deal with objects at
a level lower than the debugger can access.
Finally, if a procedure contains two identical statements that are ex-
pressions, the debugger cannot determine with certainty the statement at
which execution stopped. If this situation occurs, you can still use the de-
bugger and execution can continue. The debugger merely issues a warning
that the displayed statement number may be incorrect.3
7.3 Detecting Errors

This section describes some simple commands that you can use for de-
tecting errors in procedures that are written in Maple. If you are not
successful in finding the error by using these commands, you can use the
Maple debugger, which is discussed in Sections 7.1 and 7.2, to display the
stepwise execution of a procedure.
Tracing a Procedure
The simplest tools available for error detection in Maple are the printlevel
global variable, and the trace and tracelast commands. These facili-
ties enable you to trace the execution of both user-defined and Maple
library procedures. However, they differ in the type of information that
is returned about a procedure.
The printlevel variable is used to control how much information is
displayed when a program is executed. By assigning a large integer value
to printlevel, you can monitor the execution of statements to selected
levels of nesting within procedures. The default value of printlevel is
1. Larger, positive integer values cause the display of more intermediate
steps in a computation. Negative integer values suppress the display of
information.
The printlevel global variable is set by using the following syntax,
where n is the level to which Maple commands are evaluated.
printlevel := n ;
To determine what value of n to use, remember that statements within

a particular procedure are recognized in levels that are determined by the
nesting of conditional or repetition statements, and by the nesting of pro-
cedures. Each loop or if condition increases the evaluation level by 1, and
each procedure call increases the evaluation level by 5. Alternatively, you
can use a sufficiently large value of n to ensure that all levels are traced.
3
This problem occurs because Maple stores all identical expressions as a single
occurrence of the expression, and the debugger has no way to determine at which
invocation execution stopped.
7.3 Detecting Errors • 381
For example, printlevel := 1000 displays information in procedures up

to 200 levels deep.
> f := proc(x) local y; y := x^2; g(y) / 4; end proc;
f := proc(x) local y; y := x2 ; 1/4 ∗ g(y) end proc
> g := proc(x) local z; z := x^2; z * 2; end proc;
g := proc(x) local z; z := x2 ; 2 ∗ z end proc
> f(3);
81
2
> printlevel := 5;
printlevel := 5
> f(3);
{--> enter f, args = 3
y := 9
81
2
<-- exit f (now at top level) = 81/2}

> printlevel := 10;
printlevel := 10
> f(3);
y := 9
{--> enter g, args = 9
z := 81
162
<-- exit g (now in f) = 162}

81
2

81
2
The amount of information that is displayed depends on whether the

call to the procedure was terminated with a colon or a semicolon. If a
colon is used, only entry and exit points of the procedure are printed. If
a semicolon is used, the results of the statements are also printed.
To reset the value of the printlevel variable, reassign its value to 1.
> printlevel := 1;
printlevel := 1
By assigning a large value to printlevel, the trace of all subsequent

Maple procedure calls is displayed. To display the trace of specific pro-
cedures, you can use the trace command. The trace command has the
following syntax, where arguments is one or more procedure names.
trace(arguments );
The trace command returns an expression sequence containing the

names of the traced procedures. To begin tracing, you must call the pro-
cedure.
> trace(f,g);
f, g
> f(3):

<-- exit g (now in f) = 162}
> f(3);
y := 9
z := 81
162
<-- exit g (now in f) = 162}

81
2

81
2
Like printlevel, the amount of information that is displayed during

tracing when trace is used depends on whether the call to the procedure
was terminated with a colon or a semicolon. If a colon is used, only entry
and exit points of the procedure are printed. If a semicolon is used, the
results of the statements are also printed.
To turn off the tracing of specific procedures, use the untrace com-
mand.4
> untrace(f,g);
f, g
> f(3);
81
2
If a procedure returns an error message, you can use the tracelast

command to determine the last statement executed and the values of vari-
ables at the time of the error. The tracelast command has the following
syntax.
tracelast;
When a procedure returns an error message, the following information

is returned from a call to tracelast.
• The first line displays which procedure was called and what parameter
was used.
• The second line displays the # symbol, the procedure name with the
line number of the statement that was executed, and the statement
that was executed.
• Finally, if there are any local variables in the procedure, they are
displayed with their corresponding values.
4
You can use debug and undebug as alternate names for trace and untrace.
> f := proc(x) local i,j,k;

> i := x;
> j = x^2;
> seq(k, k=i..j);
> end proc;
f := proc(x)
local i, j, k;
i := x ; j = x2 ; seq(k, k = i..j)
end proc
> f(2,3);
Error, (in f) unable to execute seq
> tracelast;
f called with arguments: 2, 3

#(f2,3): seq(k,k = i .. j)
Error, (in f) unable to execute seq
locals defined as: i = 2, j = j, k = k

You can find the error in this procedure by studying the results of the
tracelast command—the assignment to the local variable j incorrectly
used an equal sign (=) instead of an assignment operator (:=).
The information provided by tracelast can become unavailable
whenever Maple does a garbage collection. Therefore, it is advisable to
use tracelast immediately after an error occurs. 5
Using Assertions
An assertion is a statement about a procedure that you “assert” to be
true. You can include assertions in your procedure to guarantee pre- and
post-conditions, and loop invariants during execution by using the ASSERT
command. You can also use assertions to guarantee the value returned by
a procedure or the value of local variables inside a procedure. The ASSERT
command has the following syntax.
5
For more information about garbage collection in Maple, see ?gc.
ASSERT(condition, message );
If condition evaluates to false, an error is generated and message

is printed. If the first argument evaluates to true, ASSERT returns NULL.
To check assertions, you must turn assertion checking on prior to
executing a procedure that contains an ASSERT command. To query the
current state of assertion checking, and to turn assertion checking on and
off, you must use the kernelopts command.6
The default state for assertion checking is false.
> kernelopts(ASSERT); #query the current state
false
If you enter a kernelopts command to turn assertion checking on,

kernelopts returns its previous value.
> kernelopts(ASSERT=true);
false
At any time during the Maple session, you can confirm whether as-
sertion checking is on by entering the following command.
> kernelopts(ASSERT);
true
If assertion checking is on and a procedure that contains an ASSERT

statement is executed , the condition represented by the ASSERT statement
is checked.
> f := proc(x,y) local i,j;
> i:=0;
> j:=0;
> while (i <> x) do
> ASSERT(i > 0,‘invalid index‘);
> j := j + y;
> i := i + 1;
> end do;
> j;
> end proc;
6
For more information about kernelopts, see ?kernelopts.
f := proc(x, y)
local i, j;
i := 0 ;
j := 0 ;
while i 6= x do
ASSERT(0 < i, ‘invalid index ‘) ; j := j + y ; i := i + 1
end do;
j
end proc
> f(2,3);
Error, (in f) assertion failed, invalid index
Use the kernelopts command again to turn assertion checking off.

(Again, kernelopts returns its previous value.) When assertion checking
is off, the overhead of processing an ASSERT statement within a procedure
is negligible.
> kernelopts(ASSERT=false);
true
Related to assertions are Maple warning messages. The WARNING com-

mand causes a specified warning, preceded by the string "Warning,", to
display. The WARNING command has the following syntax.
WARNING( msgString, msgParam1, msgParam2, ...);
The msgString parameter is the text of the warning message and

msgParami are optional parameters to substitute into msgString, if any.
> f := proc(x)
> if x < 0 then
> WARNING("the result is complex")
> end if;
> sqrt(x)
> end proc;
f := proc(x)
if x < 0 then WARNING(“the result is complex”) end if ;
sqrt(x)
end proc
> f(-2);
Warning, the result is complex
√
I 2
You can turn the WARNING command off by using interface(warnlevel=0).

In this case, the warning is not displayed and the call to WARNING has no
effect.
> interface(warnlevel=0);
> f(-2);
√
I 2
Handling Exceptions
An exception is an event that occurs during the execution of a procedure
that disrupts the normal flow of instructions. Many kinds of errors can
cause exceptions—for example, attempting to read from a file that doesn’t
exist. Maple has two mechanisms available when such situations arise:
• the error statement to raise an exception, and
• the try...catch...finally block to handle exceptions.
Raising exceptions The error statement raises an exception. Execution

of the current statement sequence is interrupted, and the block and proce-
dure call stack is popped until either an exception handler is encountered,
or execution returns to the top-level (in which case the exception becomes
an error). The error statement has the following syntax.
error msgString, msgParam1, msgParam2, ...
The msgString parameter is a string that gives the text of the error
message. It can contain numbered parameters of the form %n or %-n, where
n is an integer. These numbered parameters are used as placeholders for
actual values. In the event that the exception is ever printed as an error
message, the actual values are specified by the msgParams.
For example, the error message "f has a 2nd argument, x, which
is missing" is specified by the following error statement.
error "%1 has a %-2 argument, %3, which is missing", f, 2, x
A numbered parameter of the form %n displays the nth msgParam
in line-printed notation (i.e., as lprint would display it). A numbered
parameter of the form %-n displays the nth msgParam, assumed to be
an integer, in ordinal form. For example, the %-2 in the error statement
above is displayed as “2nd”. The special parameter %0 displays all the
msgParams, separated by a comma and a space.
The error statement evaluates its arguments, and then creates an
exception object which is an expression sequence with the following ele-
ments.
• The name of the procedure in which the exception was raised, or the
constant 0 if the exception was raised at the top-level.
• The msgString.
• The msgParams, if any.
The created exception object is assigned to the global variable

lastexception as an expression sequence.7
The error statement normally causes an immediate exit from the
current procedure to the Maple session. Maple prints an error message of
the following form.
Error, (in procName ) msgText
In this case, msgText is the text of the error message (which is con-
structed from the msgString and optional msgParams of the error state-
ment), and procName is the procedure in which the error occurred. If
7
The actual arguments to the error statement are also assigned to lasterror for
compatibility with older versions of Maple. For more information, see ?traperror.
the procedure does not have a name, procName is displayed as unknown.

If the error occurs at the top-level, outside of any procedure, the (in
procName) part of the message is omitted.
The error statement is commonly used when parameter declarations
are not sufficient to check that the actual parameters to a procedure are of
the correct type. The following pairup procedure takes a list L of the form
[x1 , y1 , x2 , y2 , . . . , xn , yn ] as input, and creates from it a list of the form
[[x1 , y1 ], [x2 , y2 ], . . . , [xn , yn ]]. A simple type check cannot determine if list
L has an even number of elements, so you need to check this explicitly by
using an error statement.
> pairup := proc(L::list)
> local i, n;
> n := nops(L);
> if irem(n,2) = 1 then
> error "list must have an even number of "
> "entries, but had %1", n
> end if;
> [seq( [L[2*i-1],L[2*i]], i=1..n/2 )]
> end proc:
> pairup([1, 2, 3, 4, 5]);

Error, (in pairup) list must have an even number of
entries, but had 5
> pairup([1, 2, 3, 4, 5, 6]);
[[1, 2], [3, 4], [5, 6]]
Trapping errors The try statement is a mechanism for executing pro-

cedure statements in a controlled environment so that if an error occurs,
it does not immediately terminate the procedure. The try statement has
the following syntax (the finally clause is optional).
try tryStatSeq
finally finalStatSeq
end try
If procedure execution enters a try...catch block, the tryStatSeq

is executed. If no exceptions occur during the execution of tryStatSeq,
procedure execution continues with the statement after end try.
If procedure execution enters a try...catch...finally block, the

tryStatSeq is executed. If no exceptions occur during the execution of
tryStatSeq, the finalStatSeq in the finally clause is executed. Exe-
cution then continues with the statement after end try.
If an exception does occur during the execution of tryStatSeq, exe-
cution of tryStatSeq terminates immediately. The exception object cor-
responding to the exception is compared against each catchString. Any
number of catch clauses can be provided, and each can have any num-
ber of catchStrings, separated by commas. Alternatively, a catch clause
need not have a catch string. Any given catchString (or a catch clause
without one) can appear only once in a try...end try construct.
If a matching catch clause is found, or the catch clause contains no
catchStrings, the catchStatSeq of that catch clause is executed, and
the exception is considered to have been caught. If no matching catch
clause is found, the exception is considered not caught, and is re-raised
outside the try block.
When Maple is looking for a matching catch clause, the following
definition of “matching” is used.
• Neither the exception object nor the catchStrings are evaluated (the
exception object has already been evaluated by the error statement
that produced it).
• The catchStrings are considered to be prefixes of the exception ob-

ject’s msgString. If a catchString has n characters, only the first
n characters of the msgString need match the catchString. This
permits the definition of classes of exceptions.
• A missing catchString will match any exception.
• The “result” of a try statement (the value that % returns if it is

evaluated immediately after execution of the try statement) is the
result of the last statement executed within the try statement.
A catchStatSeq can contain an error statement with no arguments,

which also re-raises the exception. When an exception is re-raised, a new
exception object is created that records the current procedure name, and
the message and parameters from the original exception.
Under normal circumstances, the finalStatSeq of the finally
clause, if there is one, is always executed before control leaves the try
statement. This is true in the case that an exception occurs, independent
of whether it is caught or whether another exception occurs in the catch
clause.
This is true even if a catchStatSeq re-raises the exception, raises a

new one, or executes a return, break, or next statement.
Under certain abnormal circumstances, the finalStatSeq is not ex-
ecuted:
• If an exception is raised in a catch clause and this exception is caught

by the debugger and the user exits the debugger, the user’s command
to stop execution overrides everything.
• If one of the following untrappable exceptions occurs, the exception

is not caught, and finalStatSeq is not executed:
1. Computation timed out. (This can only be caught by timelimit,

which raises a “time expired” exception that can be caught.)
2. Computation interrupted. (In other words, the user pressed Ctrl+C,
Break, or equivalent.)
3. Internal system error. (This indicates a bug in Maple itself.)
4. ASSERT or local variable type assertion failure. (Assertion failures
are not trappable because they indicate a coding error, not an al-
gorithmic failure.)
5. Stack overflow. (If a stack overflow occurs, there is generally not
enough stack space to do anything such as running cleanup code.)
If an exception occurs during the execution of a catchStatSeq or the

finalStatSeq, it is treated in the same way as if it occurred outside the
try...end try statement.
A useful application of the try and error statements is to abort an
expensive computation as quickly and cleanly as possible. For example,
suppose that you are trying to compute an integral by using one of several
methods, and in the middle of the first method, you determine that it will
not succeed. You would like to abort that method and go on to try another
method. The following code implements this example.
> try
> result := MethodA(f,x)
> catch "FAIL":
> result := MethodB(f,x)
> end try:
MethodA can abort its computation at any time by executing the state-
ment error "FAIL". The catch clause will catch that exception, and pro-
ceed to try MethodB. If any other error occurs during the execution of
MethodA, or if an error occurs during the execution of MethodB, it is not

caught.
Another useful application of the try statement is to ensure that
certain resources are freed when you are done with them, regardless of
whether or not anything went wrong while you were using them. For
example, you can use the following code to access Maple’s I/O facilities
to read the lines of a file and process them in some way.
> f := fopen("myfile",TEXT,READ):
> try
> line := readline(f);
> while line < 0 do
> ProcessContentsOfLine(line);
> line := readline(f)
> end do
> finally
> fclose(f)
> end try:
In this example, if any exception occurs while reading or processing

the lines of the file, it is not caught because there is no catch clause. How-
ever, fclose(f) is executed before execution leaves the try statement,
regardless of whether or not there was an exception.
The next example uses both catch and finally clauses to write to a
file instead of reading from one.
> f := fopen("myfile",TEXT,WRITE):
> try
> for i to 100 do
> fprintf(f,"Result %d is %q\n",i,ComputeSomething(i))
> end do
> catch:
> fprintf(f,"Something went wrong: %q\n",lastexception);
> error
> finally
> fclose(f)
> end try:
If any exception occurs, it is caught with the catch clause that has no
catchString, and the exception object is written into the file. The excep-
tion is re-raised by executing the error statement with no msgString. In
all cases, the file is closed by executing fclose(f) in the finally clause.
Checking Syntax
Maple’s maplemint command generates a list of semantic errors for a spec-
ified procedure, if any. The semantic errors for which maplemint checks
include parameter name conflicts, local and global variable name conflicts,
unused variable declarations, and unreachable code. The maplemint com-

mand has the following syntax.
maplemint( procedureName );
In the case where the specified procedure is free of semantic errors,

maplemint returns NULL.
> f := proc() local a,i; global c;

> print(i);
> if (a=5) then
> a:=6;
> return true;
> print(‘test‘);
> end if;
> end do;
> end do;
> end proc;
> maplemint(f);
This code is unreachable:
print(test)
These global variables were declared, but never used:
c
These local variables were used before they were assigned
a value:
a
These variables were used as the same loop variable for
nested loops:
i
Similar to maplemint, Maple also has an external program utility

called mint. The mint program is called from outside Maple and it is used
to check both semantic and syntax errors in an external Maple source file.
For more information about mint, see ?mint.
7.4 Conclusion
This chapter surveyed a variety of Maple commands that are available to
help you find errors in procedures. In particular, the Maple debugger was
presented as a tool that you can use to find and correct errors.
8 Numerical Programming
in Maple
Representation and manipulation of expressions in symbolic mode, that

is, in terms of variables, functions, and exact constants, is a powerful
feature of the Maple system. However, practical scientific computation
also demands floating-point calculations which represent quantities by
approximate numerical values. Typically, numerical computations are
used for one of three reasons.
First, not all problems have analytical or symbolic solutions. For ex-
ample, of the many partial differential equations known, only a small
subset have known closed-form solutions. But, you can usually find nu-
merical solutions.
Second, the analytic answer that Maple returns to your problem may
be very large or complex. You are not likely to do calculations by hand
which involve rational numbers containing many digits or equations with
hundreds of terms, but Maple does not mind such expressions. To un-
derstand big expressions, sometimes it helps to compute a floating-point
approximation.
Third, you may not always need an exact answer. Computing an ana-
lytic answer of infinite precision is not necessary when your only interest
is in an approximation. This situation typically arises in plotting. Calcu-
lating the points in the graph too accurately is wasteful because normal
plotting devices are not capable of displaying ten digits of resolution.
While the rest of this book primarily shows Maple’s powerful sym-
bolic methods, the focus of this chapter is on how to perform floating-
point calculations in Maple. You will quickly discover that Maple has
some extraordinary capabilities in this regard. You have your choice of
software floating-point calculations of arbitrary precision or hardware
floating-point arithmetic. The former is unaffected, save for speed, by
395
396 • Chapter 8: Numerical Programming in Maple
the machine you are using. The latter is determined by the architecture
of your computer, but offers the advantage of exceptional speed.
8.1 The Basics of evalf

The evalf command is the primary tool in Maple for performing floating-
point calculations. It causes Maple to evaluate in software floating-point
mode. Maple’s software floating-point arithmetic (see section 8.3) has an
n-digit machine floating-point model as its basis, but allows computations
at arbitrary precision. The environment variable Digits , which has an
initial setting of 10, determines the default number of digits for calcula-
tions.
> evalf(Pi);
3.141592654
You may alter the number of digits either by changing the value of
Digits, or by specifying the number as an index to evalf. Note that
when you specify the number of digits as an index to evalf, the default,
Digits, remains unchanged.
> Digits := 20:
> evalf(Pi);
3.1415926535897932385
> evalf[200](Pi);
3.1415926535897932384626433832795028841\
97169399375105820974944592307816406286\
20899862803482534211706798214808651328\
23066470938446095505822317253594081284\
81117450284102701938521105559644622948\
9549303820
> evalf(sqrt(2));
1.4142135623730950488
8.1 The Basics of evalf • 397
> Digits := 10:
The number of digits you specify is the number of decimal digits

that Maple uses during calculations. Specifying a larger number of digits
is likely to give you a more accurate answer, and the maximum value of
Digits is sufficiently large enough to be considered infinite for practical
purposes. Unlike most hardware implementations of floating-point arith-
metic, Maple stores and performs software operations on floating-point
numbers in base 10.
Because all floating-point computations are carried out in finite pre-
cision, with intermediate results generally being rounded to Digits
precision, it is possible for such round-off errors to accumulate in long
computations. Maple ensures that the results of any single floating-point
arithmetic operation (+, −, ∗, or /) are fully accurate. Further, many
of the basic functions in Maple, such as the trigonometric functions and
their inverses, the exponential and logarithm functions, and some of the
other standards special functions for mathematics, are accurate to within
.6 units of last place (ulps), meaning that if the Digits + 1st digit
of the true result is a 4, Maple may round it up, or if it is a 6, Maple
may round it down. Most mathematical functions in Maple, including
numerical integration, achieve this accuracy on nearly all inputs.
Sometimes a definite integral has no closed form solution in terms of
standard mathematical functions. You can use evalf to obtain an answer
via numerical integration.
> r := Int(exp(x^3), x=0..1);
Z 1
3
r := e(x ) dx
0
> value(r);
Z 1
3
e(x ) dx
0
> evalf(r);
1.341904418
In other cases, Maple can find an exact solution, but the form of the
exact solution is almost incomprehensible. The function Beta below is one
of the special functions that appear in mathematical literature.
> q := Int( x^99 * (1-x)^199 / Beta(100, 200), x=0..1/5 );
1/5 99
− x)199
Z
x (1
q := dx
0 B(100, 200)
> value(q);
278522905457805211792552486504343059984\
03849800909690342170417622052715523897\
76190682816696442051841690247452471818\
79720294596176638677971757463413490644\
25727501861101435750157352018112989492\
.
972548449 217741280910371516468873\
84971552115934384961767251671031013243\
12241148610308262514475552524051323083\
13238717840332750249360603782630341376\
82537367383346083183346165228661133571\
76260162148352832620593365691185012466\
14718189600663973041983050027165652595\
68426426994847133755683898925781250000\
1
0
B(100, 200)
> evalf(q);
.3546007367 10−7
Note that the two examples above use the Int command rather than
int for the integration. If you use int, Maple first tries to integrate your
expression symbolically. Thus, when evaluating the commands below,
Maple spends time finding a symbolic answer and then converts it to
a floating-point approximation, rather than performing straight numeri-
cal integration.
8.2 Hardware Floating-Point Numbers • 399
> evalf( int(x^99 * (1-x)^199 / Beta(100, 200), x=0..1/5) );
.3546007367 10−7
When you want Maple to perform numerical calculations, you should

not use commands like int, limit, and sum that evaluate their arguments
symbolically.
In general, results from evalf(Int(...)) , evalf(Sum(...)) , and
evalf(Limit(...)), will be more accurate than results obtained from the
corresponding evalf(int(...)), evalf(sum(...)), and evalf(limit(...))
operations. More generally, symbolic evaluation can be suppressed by us-
ing unevaluation quotes. For example, evalf(sin(Pi/3)) = evalf(1/2
* 3^(1/2)) while evalf(’sin’(Pi/3)) computes a floating-point ap-
proximation to sin(evalf(Pi/3)).
8.2 Hardware Floating-Point Numbers

Maple offers an alternative to software floating-point numbers: your com-
puter’s hardware floating-point arithmetic. Hardware floating-point calcu-
lations are typically much faster than software floating-point calculations.
However, hardware floating-point arithmetic depends on your particular
type of computer, and you cannot increase the precision.
The evalhf command evaluates an expression using hardware floating-
point arithmetic.
> evalhf( 1/3 );
.333333333333333314
> evalhf( Pi );
3.14159265358979312
Your computer most likely does hardware floating-point arithmetic us-

ing a certain number of binary digits. The special construct, evalhf(Digits),
approximates the corresponding number of decimal digits.
> d := evalhf(Digits);
d := 15.
Therefore, evalhf and evalf return similar results if evalf uses a

setting of Digits that is close to evalhf(Digits). Maple usually shows
you two or three digits more than the value of evalhf(Digits) specifies.
When you perform hardware floating-point calculations, Maple must con-
vert all the base-10 software floating-point numbers to base-2 hardware
floating-point numbers, and then convert the result back to base 10. The
extra decimal digits allow Maple to reproduce the binary number precisely
if you use it again in a subsequent hardware floating-point calculation.
> expr := ln( 2 / Pi * ( exp(2)-1 ) );
e2 − 1
expr := ln(2 )
π
> evalhf( expr );
1.40300383684168617
> evalf[round(d)]( expr );
1.40300383684169
The results that evalhf returns, even including for evalhf(Digits),

are not affected by the value of Digits.
> Digits := 4658;
Digits := 4658
> evalhf( expr );
1.40300383684168617
> evalhf(Digits);
15.
> Digits := 10;

Digits := 10
You can use the evalhf(Digits) construct to tell whether hardware

floating-point arithmetic provides sufficient precision in a particular ap-
plication. If Digits is less than evalhf(Digits), then you may be able
to take advantage of the faster hardware floating-point calculations; oth-
erwise, you should use software floating-point arithmetic to perform the
calculation, with sufficient digits. The evaluate procedure below takes an
unevaluated parameter, expr. Without the uneval declaration, Maple
would evaluate expr symbolically before invoking evaluate.
> evaluate := proc(expr::uneval)
> if Digits < evalhf(Digits) then
> evalf(evalhf(expr));
> else
> evalf(expr);
> end if;
> end proc:
The evalhf command knows how to evaluate many of Maple’s func-

tions, but not all. For example, you cannot evaluate an integral using
hardware floating-point arithmetic.
> evaluate( Int(exp(x^3), x=0..1) );
Error, (in evaluate) unable to evaluate function ‘Int‘
in evalhf
You can improve the evaluate procedure so that it traps such

errors and tries to evaluate the expression using software floating-point
numbers instead.
> evaluate := proc(expr::uneval)
> local result;
> try
> evalf(evalhf(expr));
> catch:
> evalf(expr);
> end try;
> else
> evalf(expr);
> end if;
> end proc:
> evaluate( Int(exp(x^3), x=0..1) );

1.341904418
The evaluate procedure provides a model of how to write procedures

that take advantage of hardware floating-point arithmetic whenever pos-
sible.
Newton Iterations
You can use Newton’s method to find numerical solutions to equations.
As section 3.1 describes, if xn is an approximate solution to the equation
f (x) = 0, then xn+1 , given by the following formula, is typically a better
approximation.
f (xn )
xn+1 = xn − 0
f (xn )
This section illustrates how to take advantage of hardware floating-point
arithmetic to calculate Newton iterations.
The iterate procedure below takes a function, f, its derivative, df,
and an initial approximate solution, x0, as input to the equation f (x) = 0.
iteration calculates at most N successive Newton iterations until the
difference between the new approximation and the previous one is small.
The iterate procedure prints out the sequence of approximations so you
can follow the workings of the procedure.
> iterate := proc( f::procedure, df::procedure,
> x0::numeric, N::posint )
> local xold, xnew;
> xold := x0;
> xnew := evalf( xold - f(xold)/df(xold) );
> to N-1 while abs(xnew-xold) > 10^(1-Digits) do
> xold := xnew;
> print(xold);
> xnew := evalf( xold - f(xold)/df(xold) );
> end do;
> xnew;
> end proc:
The procedure below calculates the derivative of f and passes all the
necessary information to iterate.
> Newton := proc( f::procedure, x0::numeric, N::posint )
> local df;
> df := D(f);
> print(x0);
> iterate(f, df, x0, N);
> end proc:
Now you can use Newton to solve the equation x2 − 2 = 0.

> f := x -> x^2 - 2;
f := x → x2 − 2
> Newton(f, 1.5, 15);
1.5
1.416666667
1.414215686
1.414213562
1.414213562
The version of Newton below uses hardware floating-point arithmetic

if possible. Since iterate only tries to find a solution to an accuracy of
10^(1-Digits), Newton uses evalf to round the result of the hardware
floating-point computation to an appropriate number of digits.
> Newton := proc( f::procedure, x0::numeric, N::posint )
> local df, result;
> df := D(f);
> print(x0);
> try
> evalf(evalhf(iterate(f, df, x0, N)));
> catch:
> end try;
> else
> end if;
> end proc:
Below, Newton uses hardware floating-point arithmetic for the iter-

ations and rounds the result to software precision. You can tell which
numbers are hardware floating-point numbers because they have more
digits than the software floating-point numbers, given the present setting
of Digits.
> Newton(f, 1.5, 15);
1.5
1.41666666666666674
1.41421568627450988
1.41421356237468987
1.41421356237309514
1.414213562
You may find it surprising that Newton must use software floating-
point arithmetic to find a root of the Bessel function below.
> F := z -> BesselJ(1, z);
F := z → BesselJ(1, z)
> Newton(F, 4, 15);
3.826493523
3.831702467
3.831705970
3.831705970
The reason is that evalhf does not know about BesselJ and the
symbolic code for BesselJ uses the type command and remember tables,
which evalhf does not allow.
> evalhf( BesselJ(1, 4) );
Error, remember tables are not supported in evalhf
Using a try-catch block as in the Newton procedure above, allows

your procedure to work even when evalhf fails.
You may wonder why the Newton procedure above prints out so many
digits when it is trying to find a ten-digit approximation. The reason is
that the print command is located inside the iterate procedure which
is inside a call to evalhf, where all numbers are hardware floating-point
numbers, and print as such.
Computing with Arrays of Numbers

Use the evalhf command for calculations with numbers. The only struc-
tured Maple objects allowed in a call to evalhf are arrays of numbers,
either table-based arrays or rtable-based Arrays. If an array has undefined
entries, evalhf initializes them to zero. The procedure below calculates
the polynomial 2 + 5x + 4x2 .
> p := proc(x)
> local a, i;
> a := array(0..2);
> a[0] := 2;
> a[1] := 5;
> a[2] := 4;
> sum( a[i]*xî, i=0..2 );
> end proc:
> p(x);
2 + 5 x + 4 x2
If you intend to enclose p in a call to evalhf, you cannot define the

local array a using array(1..3, [2,5,4]), since lists are not allowed
inside evalhf. You can, however, enclose p in a call to evalhf if the
parameter x is a number.
> evalhf(p(5.6));
155.439999999999997
You can also pass an array of numbers as a parameter inside a call

to evalhf. The procedure below calculates the determinant of a 2 × 2
matrix. The (2,2) entry in the array a below is undefined.
> det := proc(a::array(2))
> a[1,1] * a[2,2] - a[1,2] * a[2,1];
> end proc:
> a := array( [[2/3, 3/4], [4/9]] );

2 3 
3 4 
a := 
4

a2, 2
9
> det(a);
2 1
a2, 2 −
3 3
If you call det from inside a call to evalhf, Maple uses the value 0
for the undefined entry, a[2,2].
> evalhf( det(a) );
−.333333333333333314
evalhf passes arrays by value, so the (2,2) entry of a is still unde-

fined.
> a[2,2];
a2, 2
If you want evalhf to modify an array that you pass as a parameter

to a procedure, you must enclose the name of the array in a var construct.
The var construct is special to evalhf and is necessary only if you want
evalhf to modify an array of numbers that is accessible at the session
level.
> evalhf( det( var(a) ) );
−.333333333333333314
Now a is an array of floating-point numbers.

> eval(a);
[.666666666666666629 , .750000000000000000]
[.444444444444444420 , 0.]
The evalhf command always returns a single floating-point number,
but the var construct allows you to calculate a whole array of numbers
with one call to evalhf. Section 9.7 illustrates the use of var to calculate
a grid of function values that you can use for plotting.
You can also create arrays of hardware floating-point values directly
with the Array command. Proper use of this command can save significant
8.3 Floating-Point Models in Maple • 407
amounts of time, especially in plotting routines, which rely heavily on

arrays of floating-point values. See the help pages for Array, Matrix, and
Vector for more details and examples.
8.3 Floating-Point Models in Maple

In addition to being able to represent symbolic constants, such as π and
γ, and exact integers and rational numbers, such as 37 and 3/4, Maple
has the ability to represent approximations to numeric values, using its
floating-point system. Numbers in this system are represented by pairs of
integers, (m,e). The first integer is called the significand or mantissa.
The second integer is called the exponent. The number represented is
m × 10e .
Examples of floating-point numbers in Maple are 3.1415, 1.0, −0.0007,

1.0e0, and 2e1234567. The last two are examples of floating-point num-
bers entered in scientific notation: the "e" separates the mantissa and
exponent parts of the number. Such numbers can also be used to repre-
sent complex numbers (as can exact integers and rationals): 1.0 + 2.7 ∗ I,
etc.
In some contexts, Maple distinguishes between software floats and
hardware floats. The evalhf evaluator (discussed in section 8.2), for ex-
ample, works with hardware floats, and Maple can construct certain kinds
of matrices and vectors with hardware float entries. Generally, however,
Maple works with software floats to carry out approximate (but usually
very accurate) numerical calculations.
Floating-point number systems are approximations to the mathe-
matical set of real (and complex) numbers, and hence necessarily have
some limitations. Most importantly, such systems have limited range
(there are largest and smallest representable numbers) and limited preci-
sion (the entire set of representable floating-point numbers is finite). One
very important feature of Maple’s software floating-point system is that
you have control over at least the latter of these: you can specify the pre-
cision at which Maple is to work when doing floating-point computation.
Some of the specific details of these computation systems are provided
in the next few sections.
Software Floats
Maple’s software floating-point computations are carried out in base 10.
The precision of a computation is determined by the setting of Digits.
The maximum exponent, minimum exponent, and maximum value for
Digits are machine wordsize dependent. You can obtain the values for
these limits from the Maple_floats command.
This software floating-point system is designed as a natural extension
of the industry standard for hardware floating-point computation, known
as IEEE 754. Thus, there are representations for infinity and undefined
(what IEEE 754 calls a "NaN", meaning "Not a Number"). Complex num-
bers are represented by using the standard x + I*y format.
One important feature of this system is that the floating-point rep-
resentation of zero, 0., retains its arithmetic sign in computations. That
is, Maple distinguishes between +0. and -0. when necessary. In most
situations, this difference is irrelevant, but when dealing with functions
such as ln(x), which have a discontinuity across the negative real axis,
preserving the sign of the imaginary part of a number on the negative
real axis is important.
For more intricate applications, Maple implements extensions of the
IEEE 754 notion of a numeric event, and provides facilities for mon-
itoring events and their associated status flags. The "Maple Numerics
Overview" help page is a good starting place to learn more about this
system. See ?numerics.
Roundoff Error
When you perform floating-point arithmetic, whether using software or
hardware floats, you are using approximate numbers rather than pre-
cise real numbers or expressions. Maple can work with exact (symbolic)
expressions. The difference between an exact real number and its floating-
point approximation is called the roundoff error . For example, suppose
you request a floating-point representation of π.
> pi := evalf(Pi);
π := 3.141592654
Maple rounds the precise value π to ten significant digits because

Digits is set to its default value of 10. You can approximate the roundoff
error above by temporarily increasing the value of Digits to 15.
> evalf[15](Pi - pi);
8.3 Floating-Point Models in Maple • 409
−.41021 10−9
Roundoff errors arise not only from the representation of input data,
but also as a result of performing arithmetic operations. Each time
you perform an arithmetic operation on two floating-point numbers, the
infinitely-precise result usually will not be representable in the floating-
point number system and therefore the computed result will also have an
associated roundoff error.
For example, suppose you multiply two ten-digit numbers with Digits
= 10. The result can easily have nineteen or twenty digits, but Maple will
only store the first ten digits.
> 1234567890 * 1937128552;
2391516709101395280
> evalf(1234567890) * evalf(1937128552);
.2391516709 1019
Whenever you apply one of the four basic arithmetic operations (ad-
dition, subtraction, multiplication, or division) to two floating-point num-
bers, the result is the correctly rounded representation of the infinitely
precise result, unless overflow or underflow occurs. Of course, Maple may
need to compute an extra digit or two behind the scenes to ensure that
the answer is correct.
Even so, sometimes a surprising amount of error can accumulate, par-
ticularly when subtracting two numbers which are of similar magnitude. In
the calculation below, the accurate sum of x, y, and z is y = 3.141592654.
> x := evalf(987654321);
x := .987654321 109
> y := evalf(Pi);
y := 3.141592654
> z := -x;
z := −.987654321 109
> x + y + z;
3.1
Catastrophic cancellation is the name of this phenomenon. During

the subtraction the eight leading digits cancel out, leaving only two sig-
nificant digits in the result.
One advantage of Maple’s software floats, in contrast to fixed-precision
floating-point numbers systems, is that the user can increase the precision
to alleviate some of the consequences of roundoff errors. For example,
increasing Digits to 20 dramatically improves the result.
> Digits := 20;
Digits := 20
> x + y + z;
3.141592654
You should employ standard numerical analysis techniques to avoid

large errors accumulating in your calculations. Often, reordering the oper-
ations leads to a more accurate final result. For example, when computing
a sum, add the numbers with the smallest magnitude first.
8.4 Extending the evalf Command

The evalf command knows how to evaluate many functions and con-
stants, such as sin and Pi. You can also define your own functions or
constants, and extend evalf by adding information about how to com-
pute such functions or constants.
Defining Your Own Constants

You may define a new constant and write procedures that manipulate this
constant symbolically. You could then write a procedure that can calculate
a floating-point approximation of your constant to any number of digits. If
you assign the procedure a name of the form ‘evalf/constant/name ‘,
then Maple invokes the procedure when you use evalf to evaluate an
expression containing your constant, name.
8.4 Extending the evalf Command • 411
Suppose you want the name MyConst to represent the following infinite
series:
∞
X (−1)i π i
M yConst = .
2i i!
i=1
You can calculate approximations to the above series in many ways; the
procedure below is one implementation. Note that if ai is the ith term in
the sum, then ai+1 = −ai (π/2)/i gives the next term. You can calculate
an approximation to the series by adding terms until Maple’s model for
software floating-point numbers cannot distinguish the new partial sum
from the previous one. Using numerical analysis, you can prove that this
algorithm calculates Digits accurate digits of MyConst if you use two ex-
tra digits inside the algorithm. Therefore, the procedure below increments
Digits by two and uses evalf to round the result to the proper number
of digits before returning. The procedure does not have to reset the value
of Digits because Digits is an environment variable.
> ‘evalf/constant/MyConst‘ := proc()
> local i, term, halfpi, s, old_s;
> halfpi := evalf(Pi/2);
> old_s := 1;
> term := 1.0;
> s := 0;
> for i from 1 while s <> old_s do
> term := -term * halfpi / i;
> old_s := s;
> s := s + term;
> end do;
> evalf[Digits-2](s);
> end proc:
When you invoke evalf on an expression containing MyConst, Maple

invokes ‘evalf/constants/MyConst‘ to calculate an approximate value.
> evalf(MyConst);
−.7921204237
> evalf[40](MyConst);
−.7921204236492380914530443801650212299661
You can express the particular constant, MyConst, in closed form and,
in this case, you can use the closed-form formula to calculate approxima-
tions to MyConst more efficiently.
> Sum( (-1)î * Piî / 2î / i!, i=1..infinity );

∞
X (−1)i π i
2i i!
i=1
> value(%);
e(−1/2 π) (1 − e(1/2 π) )
> expand(%);
1
√ −1
eπ
> evalf(%);
−.7921204237
Defining Your Own Functions

If you define your own functions, you may want to write your own pro-
cedure for calculating numerical approximations to the function values.
When you invoke evalf on an expression containing an unevaluated call
to a function F, then Maple calls the procedure ‘evalf/F ‘ if such a
procedure exists.
Suppose you want to study the function x 7→ (x − sin(x))/x3 .
> MyFcn := x -> (x - sin(x)) / x^3;
x − sin(x)
MyFcn := x →
x3
This function is not defined at x = 0, but you can extend it as a

continuous function by placing the limiting value in MyFcn’s remember
table.
> MyFcn(0) := limit( MyFcn(x), x=0 );
1
MyFcn(0) :=
6
8.4 Extending the evalf Command • 413
For small values of x, sin(x) is almost equal to x, so the subtraction

x − sin(x) in the definition of MyFcn can lead to inaccuracies due to catas-
trophic cancellation. When you evaluate v below to ten digits, only the
first two are correct.
> v := ’MyFcn’( 0.000195 );
v := MyFcn(.000195)
> evalf(v);
.1618368482
> evalf(v, 2*Digits);
.16666666634973617222
If you depend on accurate numerical approximations of MyFcn, you

must write your own procedure to provide them. You could write such a
procedure by exploiting the series expansion of MyFcn.
> series( MyFcn(x), x=0, 11 );
1 1 2 1 1
− x + x4 − x6 + O(x8 )
6 120 5040 362880
The general term in the series is
x2i
ai = (−1)i , i ≥ 0.
(2i + 3)!
Note that ai = −ai−1 x2 /((2i + 2)(2i + 3)). For small values of x, you
can then calculate an approximation to MyFcn(x) by adding terms un-
til Maple’s model for software floating-point numbers cannot distinguish
the new partial sum from the previous one. For larger values of x, catas-
trophic cancellation is not a problem, so you can use evalf to evaluate
the expression. Using numerical analysis, you can prove that this algo-
rithm calculates Digits accurate digits of the function value if you use
three extra digits inside the algorithm. Therefore, the procedure below
increments Digits by three and uses evalf to round the result to the
proper number of digits before returning.
> ‘evalf/MyFcn‘ := proc(xx::algebraic)

> local x, term, s, old_s, xsqr, i;
> x := evalf(xx);
> Digits := Digits+3;
> if type(x, numeric) and abs(x)<0.1 then
> xsqr := x^2;
> term := evalf(1/6);
> s := term;
> old_s := 0;
> for i from 1 while s <> old_s do
> term := -term * xsqr / ((2*i+2)*(2*i+3));
> old_s := s;
> s := s + term;
> end do;
> else
> s := evalf( (x-sin(x))/x^3 );
> end if;
> eval[Digits-3](s);
> end proc:
When you invoke evalf on an expression containing an unevaluated

call to MyFcn, Maple invokes ‘evalf/MyFcn‘.
> evalf( ’MyFcn’(0.000195) );
.1666666663498
You should now recode the symbolic version of MyFcn so that it takes
advantage of ‘evalf/MyFcn‘ if the argument is a floating-point number.
> MyFcn := proc(x::algebraic)
> if type(x, float) then
> evalf(’MyFcn’(x));
> else
> (x - sin(x)) / x^3;
> end if;
> end proc:
The evalf command automatically looks for ‘evalf/MyFcn‘ when

used in the evalf(‘MyFcn‘) syntax.
> MyFcn(0) := limit( MyFcn(x), x=0 );
1
MyFcn(0) :=
6
Now you can properly evaluate MyFcn with numeric as well as symbolic
arguments.
8.5 Using the Matlab Package • 415
> MyFcn(x);
x − sin(x)
x3
> MyFcn(0.099999999);
.1665833531735
> MyFcn(0.1);
.1665833531700
Section 3.4 describes how to extend many other Maple commands.
8.5 Using the Matlab Package

Another way to accomplish numerical computations in Maple is to use
the Matlab package, which provides a way to access several of the MAT-
LAB built-in functions (assuming you have a copy of MATLAB properly
installed on your computer). The mathematical functions provided are:
• chol: Cholesky factorization
• defined: test whether varaible exists
• det: determinant
• dimensions: compute dimensions of matrix
• eig: eigenvalues and eigenvectors
• evalM: evaluate expression
• fft: discrete Fourier transforms
• getvar: get numeric array or matrix
• inv: matrix inverse
• lu: LU decomposition
• ode45: solve ordinary differential equation

• qr: QR orthogonal-triangular decomposition

• size: compute size of matrix
• square: determine whether matrix is square
• transpose: matrix transposition
There are also a handful of support and utility commands provided.
Matlab converts all Maple structures to hfarrays — arrays of hard-
ware floating-point values — before it performs any computations. The
results you get will usually be in terms of hfarrays, not standard Maple
matrices. The convert( ,array) command has been extended to handle
any conversions between the two.
For more information on all these commands and the Matlab package
in general, please refer to the online help for Matlab. To learn how to
start the MATLAB application from your Maple session, see the online
help for Matlab[openlink].
8.6 Conclusion
The various techniques described in this chapter afford an important ex-
tension to Maple’s programming language and its ability to perform sym-
bolic manipulations. With numerical techniques at your disposal, you can
solve equations which are otherwise unsolvable, investigate the properties
of complicated solutions, and quickly obtain numerical estimates.
Symbolic calculations give precise representations, but in some cases
can be expensive to compute even with such a powerful tool as Maple.
At the other extreme, hardware floating-point arithmetic allows you fast
computation directly from Maple. This involves, however, limited accu-
racy. Software floating-point offers a balance. As well as sometimes being
much faster than symbolic calculations, you also have the option to con-
trol the precision of your calculations, thus exerting control over errors.
Software floating-point calculations and representations mimic the
IEEE 754 standard representation closely, except for the great advan-
tage of arbitrary precision. Because of the similarity with this popular
standard, you can readily apply the knowledge of accumulation of error
and numerical analysis principles that numerous texts and papers contain.
When you need to know that your calculations are precise, this wealth of
information at your disposal should provide you with confidence in your
results.
9 Programming with Maple
Graphics
Maple has a wide range of commands for generating both two- and
three-dimensional plots. For mathematical expressions, you can use li-
brary procedures, such as plot and plot3d, or one of the many spe-
cialized graphics routines found in the plots and plottools packages,
the DEtools package (for working with differential equations), and the
stats package (for statistical data). The input to these commands is typ-
ically one or more Maple formulæ, operators, or functions, along with
information about domains and possibly ranges. In all cases, the graphic
commands allow for the setting of options, specifying such attributes as
coloring, shading, or axes style.
The purpose of this chapter is to reveal the structure of the procedures
that Maple uses to generate graphical output, and allow you to generate
your own graphics procedures. This chapter includes basic information
about argument conventions, setting defaults, and processing of plotting
options. A major part of the material describes the data structures that
Maple uses for plotting, along with various techniques to build such data
structures in order to produce graphics in Maple. In addition, you will see
how some of the existing functions in the plots and plottools packages
produce specific plotting data structures.
9.1 Basic Plot Functions

This section illustrates some of the basic workings of the graphics pro-
cedures in Maple, as well as some of the properties that are common to
417
418 • Chapter 9: Programming with Maple Graphics
all Maple plotting commands. Also, it discusses plotting Maple opera-

tors or functions versus formula expressions and the setting of optional
information.
Several of Maple’s graphics procedures take mathematical expres-
sions as their input. Examples of such commands include plot, plot3d,
animate, animate3d, and complexplot. All these commands allow the
input to be in one of two forms: formulæ or functions. The former con-
sists of expressions such as x2 y −y 3 +1 or 3 sin(x) sin(y)+x, both formulæ
in the variables x and y. If p and q are functions with two arguments, then
p + q is an example of a function expression. The graphics procedures use
the way you specify the domain information to determine if the input
is a function expression or a formula in a specified set of variables. For
example, the command below generates a three-dimensional plot of the
surface which sin(x) sin(y) defines. This formula is in terms of x and y.
> plot3d( sin(x) * sin(y), x=0..4*Pi, y=-2*Pi..2*Pi );
If instead, you define two functions, each with two arguments,

> p := (x, y) -> sin(x): q := (x, y) -> sin(y):
then you can plot the surface that p ∗ q determines in the following
manner.
> plot3d( p * q, 0..4*Pi, -2*Pi..2*Pi );
Both cases produce the same three-dimensional plot. In the first ex-
ample, you supply the information that the input is an expression in x
and y by giving the second and third arguments in the form x = range
and y = range , while in the second example, there are no variable names.
Working with formula expressions is simple, but in many cases, func-
tions provide a better mechanism for constructing mathematical func-
tions. The following constructs a mathematical function which, for a given
9.1 Basic Plot Functions • 419
input, computes the required number of iterations (to a maximum of 10)

for the sequence zn+1 = zn2 + c to exit the disk of radius 2 for various
complex starting points c = x + iy.
> mandelbrotSet := proc(x, y)
> local z, m;
> z := evalf( x + y*I );
> m := 0;
> to 10 while abs(z) < 2 do
> z := z^2 + (x+y*I);
> m := m + 1;
> end do:
> m;
> end proc:
You now have a convenient method for computing a three-dimensional

Mandelbrot set on a 50 × 50 grid.
> plot3d( mandelbrotSet, -3/2..3/2, -3/2..3/2, grid=[50,50] );
Creating a Maple graphic at the command level displays it on the

plotting device (that is, your terminal). In many cases, you can then
interactively alter the graph using the tools available with these plotting
devices. Examples of such alterations include changing the drawing style,
the axes style, and the view point. You can include this information by
using optional arguments to plot3d.
> plot3d( sin(x)*sin(y), x=-2*Pi..2*Pi, y=-2*Pi..2*Pi,
> style=patchnogrid, axes=frame );
1
0.5
0
–0.5
–1
–6 –6
–4 –4
–2 –2
y 0 2 0 x
2
4 4
6 6
> plot3d( mandelbrotSet, -1.5..1.5, -1.5..1.5, grid=[50,50],

> style=wireframe, orientation=[143,31] );
Every plotting procedure allows for optional arguments. You give the
optional information in the form name =option . Some of these options
affect the amount of information concerning the function that you give
to the plotting procedures. The grid option that the Mandelbrot set ex-
ample uses is an instance of using an optional argument. You can use
other options for specifying visual information once you have determined
the graphical points. The type of axes, shading, surface style, line styles,
and coloring are but a few of the options available in this category. Ob-
tain information about all the allowable options for the two-dimensional
and three-dimensional cases using the help pages ?plot,options and
?plot3d,options.
Any graphics routine you create should allow users a similar set of op-
tions. When writing programs that call existing Maple graphics routines,
simply pass the potential optional arguments directly to these routines.
9.2 Programming with Plotting Library Functions • 421
9.2 Programming with Plotting Library Functions

This section gives examples of programming with the graphics procedures
in Maple.
Plotting a Loop
Consider the first problem of plotting a loop from a list of data.
> L1 := [ [5,29], [11,23], [11,36], [9,35] ];
L1 := [[5, 29], [11, 23], [11, 36], [9, 35]]
The plot command draws lines between the listed points.

> plot( L1 );
36
34
32
30
28
26
24
5 6 7 8 9 10 11
You may want to write a procedure that also draws a line from the
last to the first point. All you need to do is append the first point in L1
to the end of L1.
> L2 := [ op(L1), L1[1] ];
L2 := [[5, 29], [11, 23], [11, 36], [9, 35], [5, 29]]
> plot( L2 );
36
34
32
30
28
26
24
5 6 7 8 9 10 11
The procedure loopplot automates this technique.

> loopplot := proc( L )
> plot( [ op(L), L[1] ] );
> end proc;
loopplot := proc(L) plot([op(L), L1 ]) end proc
This procedure has a number of shortcomings. You should always

verify the input, L, to loopplot to be a list of points, where a point
is a list of two constants. That is, L should be of type list([constant,
constant]). The loopplot command should also allow a number of plot-
ting options. All loopplot has to do is pass on the options to plot. Inside
a procedure, args is the sequence of arguments in the call to the proce-
dure, and nargs is the number of arguments. Thus args[2..nargs] is the
sequence of options passed to loopplot. The loopplot procedure should
pass all but its first argument, L, directly to plots.
> loopplot := proc( L::list( [constant, constant] ) )
> plot( [ op(L), L[1] ], args[2..nargs] );
> end proc:
The above version of loopplot gives an informative error message if

you try to use it with improper arguments, and it also allows plotting
options.
> loopplot( [[1, 2], [a, b]] );
Error, invalid input: loopplot expects its 1st
argument, L, to be of type list([constant, constant]),
but received [[1, 2], [a, b]]
9.2 Programming with Plotting Library Functions • 423
> loopplot( L1, linestyle=3 );
36
34
32
30
28
26
24
5 6 7 8 9 10 11
Exercise
1. Improve loopplot so that it can handle the empty list as input.
A Ribbon Plot Procedure

This section ends with the creation of a ribbonplot procedure, a three-
dimensional plot of a list of two-dimensional formulæ or functions.
The ribbonplot procedure uses the the display procedure from the
plots package to display the plots. This procedure is called explicitly
using its full name so that ribbonplot will work even when the short
names for the functions in the plots package have not been loaded.
The hasoption command helps you handle options. In the ribbonplot
procedure, hasoption returns false if numpoints is not among the op-
tions listed in opts. If opts contains a numpoints option, then hasoption
assigns the value of the numpoints option to n, and returns the remaining
options in the fourth argument (in this case, modifying the value of the
list opts).
> ribbonplot := proc( Flist, r1::name=range )
> local i, m, p, y, n, opts;
> opts := [ args[3..nargs] ];
> if not hasoption( opts, ’numpoints’, ’n’, ’opts’ )
> then n := 25 # default numpoints
> end if;
>
> m := nops( Flist );
> # op(opts) is any additional options
> p := seq( plot3d( Flist[i], r1, y=(i-1)..i,
> grid=[n, 2], op(opts) ),
> i=1..m );
> plots[display]( p );
> end proc:
Now ribbonplot uses the number of grid points you ask it to.
> ribbonplot( [cos(x), cos(2*x), sin(x), sin(2*x)],

> x=-Pi..Pi, numpoints=16 );
The input to ribbonplot above must be a list of expressions. You

should extend ribbonplot so that it also accepts a list of functions. One
difficulty with this extension is that you need to create two-dimensional
functions from one-dimensional functions, something that was not a prob-
lem in the initial examples of ribbonplot. For this you can create an
auxiliary procedure, extend, which makes use of the unapply command.
> extend := proc(f)
> local x,y;
> unapply(f(x), x, y);
> end proc:
For example, the extend procedure converts the R → R function

x 7→ cos(2x) to a R2 → R function.
> p := x -> cos(2*x):
> q := extend(p);
q := (x, y) → cos(2 x)
The following gives the new ribbonplot code.

> ribbonplot := proc( Flist, r1::{range, name=range} )
> local i, m, p, n, opts, newFlist;
> if type(r1, range) then
> # Functional input.
> if not hasoption( opts, ’numpoints’, ’n’, ’opts’ )
> then n := 25 # default numpoints
> end if;
> m := nops( Flist );
> # change plot3d for functional input
> p := seq( plot3d( extend( Flist[i] ), r1, (i-1)..i,
9.3 Maple’s Plotting Data Structures • 425
> grid=[n,2], op(opts) ),

> i=1..m );
> plots[display]( p );
> else
> # Expressions. Convert each to a function of lhs(r1).
> newFlist := map( unapply, Flist, lhs(r1) );
> # Use lhs(r1) as the default x-axis label.
> opts := [ ’labels’=[lhs(r1), "", "" ],
> args[3..nargs] ];
> ribbonplot( newFlist, rhs(r1), op(opts) )
> end if
> end proc:
Here is a ribbon plot of three functions.

> ribbonplot( [cos, sin, cos + sin], -Pi..Pi );
9.3 Maple’s Plotting Data Structures

Maple generates plots by sending the user interface an unevaluated PLOT
or PLOT3D function call. The information included inside these functions
determines the objects they will graph. Every command in the plots
package creates such a function. View this flow of information in the fol-
lowing manner. A Maple command produces a PLOT structure and passes
it to the user interface. In the user interface, Maple constructs primitive
graphic objects based on the PLOT structure. It then passes these objects
to the chosen device driver for display. This process is shown schematically
in figure 9.1.
You can assign the plotting data structures to variables, transform
them into other structures, save them, or even print them out.
You can see examples of a plot structure in either two- or three-
dimensions by line printing such a structure.
Figure 9.1 How plots are displayed
> lprint( plot(2*x+3, x=0..5, numpoints=3, adaptive=false) );

PLOT(CURVES([[0., 3.], [2.61565849999999989, 8.2313170\
0000000066], [5., 13.]],COLOUR(RGB,1.0,0.,0.)),
AXESLABELS("x",""),VIEW(0. .. 5.,DEFAULT))
Here, plot generates a PLOT data structure that includes the

information for a single curve defined by three points, with the curve
colored with the red-green-blue (RGB) values (1.0, 0, 0), which corresponds
to red. The plot has a horizontal axis running from 0 to 5. Maple, by
default, determines the scale along the vertical axes using the information
that you provide in the vertical components of the curve. The numpoints
= 3 and adaptive = false settings ensure that the curve consists of only
three points.
The second example is the graph of z = xy over a 3 × 4 grid. The
PLOT3D structure contains a grid of z values over the rectangular region
[0, 1] × [0, 2].
> lprint( plot3d(x*y, x=0..1, y=0..2, grid=[3,4]) );
PLOT3D(GRID(0. .. 1.,0. .. 2.,Array(1 .. 3,1 .. 4,{(2,
2) = .333333333333333314, (2, 3) = .666666666666666629
, (2, 4) = 1., (3, 2) = .666666666666666629, (3, 3) =
1.33333333333333326, (3, 4) = 2.},datatype = float[8],
storage = rectangular,order = C_order)),AXESLABELS(x,y
,""))
The structure includes labels x and y for the plane but no label
for the z-axis.
The third example is again the graph of z = xy but this time in
cylindrical coordinates. The PLOT3D structure now contains a mesh of
points that make up the surface, along with the information that the
plotting device should display the surface in a point style.
> lprint( plot3d( x*y, x=0..1, y=0..2, grid=[3,2],

> coords=cylindrical, style=point ) );
PLOT3D(MESH(Array(1 .. 3,1 .. 2,1 .. 3,{(1, 2, 3) = 2.
, (2, 2, 1) = .877582561890372758, (2, 2, 2) = .479425\
538604203006, (2, 2, 3) = 2., (3, 2, 1) = 1.0806046117\
3627952, (3, 2, 2) = 1.68294196961579300, (3, 2, 3) =
2.},datatype = float[8],storage = rectangular,order =
C_order)),STYLE(POINT))
Since the plot is not in cartesian coordinates there are no de-

fault labels, it follows that the PLOT3D structure does not contain any
AXESLABELS.
The PLOT Data Structure

You can construct and manipulate a plotting data structure directly to
create two- and three-dimensional plots. All you need is a correct arrange-
ment of the geometric information inside a PLOT or PLOT3D function. The
information inside this function determines the objects that the plotting
device displays. Here Maple evaluates the expression
> PLOT( CURVES( [ [0,0], [2,1] ] ) );
0.8
0.6
0.4
0.2
0 0.5 1 1.5 2
and passes it to the Maple interface which determines that this is a plot
data structure. The Maple interface then dismantles the contents and
passes the information to a plot driver which then determines the graph-
ical information that it will render onto the plotting device. In the latest
example, the result is a single line from the origin to the point (2, 1). The
CURVES data structure consists of one or more lists of points each generat-
ing a curve, along with some optional arguments (for example, line style
or line thickness information). Thus, the expression
> n := 200:
> points := [ seq( [2*cos(i*Pi/n), sin(i*Pi/n) ], i=0..n) ]:
> PLOT( CURVES( evalf(points) ) );
0.8
0.6
0.4
0.2
–2 –1 0 1 2
generates the plot of a sequence of n + 1 points in the plane. The points

found inside the PLOT data structure must be numeric. If you omit the
evalf statement, then non-numeric objects within the PLOT structure,
such as sin(π/200), cause an error.
> PLOT( CURVES( points ) );
Plotting error, non-numeric vertex definition
> type( sin(Pi/n), numeric );
false
Hence, no plot is generated.

In general, the arguments inside a PLOT structure are all of the form
ObjectName ( ObjectInformation, LocalInformation )
where ObjectName is a function name; for example, one of CURVES,

POLYGONS, POINTS, or TEXT; ObjectInformation contains the basic geomet-
ric point information that describes the particular object; and the optional
LocalInformation contains information about options that apply only to
this particular object. ObjectInformation depends on the ObjectName.
In the case where the ObjectName is CURVES or POINTS, the ObjectInfor-
mation consists of one or more lists of two-dimensional points. Each list
supplies the set of points making up a single curve in the plane. Similarly,
when ObjectName is POLYGONS, then the object information consists of
one or more lists of points where each list describes the vertices of a
single polygon in the plane. When ObjectName is TEXT, the object infor-
mation consists of a point location along with a text string. The optional
information is also in the form of an unevaluated function call. In the

two-dimensional case, the options include AXESSTYLE, STYLE, LINESTYLE,
THICKNESS, SYMBOL, FONT, AXESTICKS, AXESLABELS, VIEW, and SCALING.
You can also place some of these as LocalInformation inside a POINTS,
CURVES, TEXT, or POLYGONS object; LocalInformation overrides the global
option for the rendering of that object. The COLOR option allows for a fur-
ther format when you place it on an object. In the case of an object having
multiple subobjects (for example multiple points, lines, or polygons), you
can supply one color value for each object.
Here is a simple way to generate a filled histogram of sixty-three values
of the function y = sin(x) from 0 to 6.3. Maple colors each trapezoid
individually by the HUE value corresponding to y = | cos(x)|.
> p := i -> [ [(i-1)/10, 0], [(i-1)/10, sin((i-1)/10)],
> [i/10, sin(i/10)], [i/10, 0] ]:
Now p(i) is the list of corners of the ith trapezoid. For example, p(2)
contains the corners of the second trapezoid.
> p(2);
1 1 1 1 1 1
[[ , 0], [ , sin( )], [ , sin( )], [ , 0]]
10 10 10 5 5 5
Define the function h to give the color of each trapezoid.

> h := i -> abs( cos(i/10) ):
> PLOT( seq( POLYGONS( evalf( p(i) ),
> COLOR(HUE, evalf( h(i) )) ),
> i = 1..63) );
0.5
0 1 2 3 4 5 6
–0.5
–1
A Sum Plot
You can create procedures that directly build PLOT data structures. For
example, given an unevaluated sum you can compute the partial sums,
and place the values in a CURVES structure.
> s := Sum( 1/k^2, k=1..10 );

10
X 1
s :=
k2
k=1
You can use the typematch command to pick the unevaluated sum
apart into its components.
> typematch( s, ’Sum’( term::algebraic,
> n::name=a::integer..b::integer ) );
true
The typematch command assigns the parts of the sum to the given
names.
> term, n, a, b;
1
, k, 1, 10
k2
You can now calculate the partial sums.

> sum( term, n=a..a+2 );
49
36
The following defines a procedure, psum, which calculates a floating-

point value of the mth partial sum.
> psum := evalf @ unapply( Sum(term, n=a..(a+m)), m );
1+m
À !
X 1
psum := evalf @ m →
k2
k=1
You can now create the necessary list of points.

> points := [ seq( [[i,psum(i)], [i+1,psum(i)]],
> i=1..(b-a+1) ) ];
points := [[[1, 1.250000000], [2, 1.250000000]],

[[2, 1.361111111], [3, 1.361111111]],
[[3, 1.423611111], [4, 1.423611111]],
[[4, 1.463611111], [5, 1.463611111]],
[[5, 1.491388889], [6, 1.491388889]],
[[6, 1.511797052], [7, 1.511797052]],
[[7, 1.527422052], [8, 1.527422052]],
[[8, 1.539767731], [9, 1.539767731]],
[[9, 1.549767731], [10, 1.549767731]],
[[10, 1.558032194], [11, 1.558032194]]]
> points := map( op, points );
points := [[1, 1.250000000], [2, 1.250000000],

[2, 1.361111111], [3, 1.361111111], [3, 1.423611111],
[4, 1.423611111], [4, 1.463611111], [5, 1.463611111],
[5, 1.491388889], [6, 1.491388889], [6, 1.511797052],
[7, 1.511797052], [7, 1.527422052], [8, 1.527422052],
[8, 1.539767731], [9, 1.539767731], [9, 1.549767731],
[10, 1.549767731], [10, 1.558032194], [11, 1.558032194]]
This list has the right form.
1.55
1.5
1.45
1.4
1.35
1.3
1.25
2 4 6 8 10
The sumplot procedure automates this technique.

> sumplot := proc( s )
> local term, n, a, b, psum, m, points, i;
> if typematch( s, ’Sum’( term::algebraic,
> n::name=a::integer..b::integer ) ) then
> psum := evalf @ unapply( Sum(term, n=a..(a+m)), m );
> points := [ seq( [[i,psum(i)], [i+1,psum(i)]],

> i=1..(b-a+1) ) ];
> points := map(op, points);
> else
> error "expecting a Sum structure as input"
> end if
> end proc:
Here is a sumplot of an alternating series.

> sumplot( Sum((-1)^k/k, k=1..25 ));
–0.5
–0.55
–0.6
–0.65
–0.7
–0.75
–0.8
5 10 15 20 25
The limit of this sum is − ln 2.

> Sum((-1)^k/k, k=1..infinity): % = value(%);
∞
X (−1)k
= −ln(2)
k
k=1
See ?plot,structure for more details on the PLOT data structure.
The PLOT3D Data Structure

The three-dimensional plotting data structure has a form similar to the
PLOT data structure. Thus, for example, the Maple expression below gen-
erates a three-dimensional plot of three lines and axes of type frame.
> PLOT3D( CURVES( [ [3, 3, 0], [0, 3, 1],

> [3, 0, 1], [3, 3, 0] ] ),
> AXESSTYLE(FRAME) );
1
0.8
0.6
0.4
0.2
0
0 0
0.5 0.5
1 1
1.5 1.5
2 2
2.5 2.5
3 3
The following procedure creates the sides of a box and colors them
yellow.
> yellowsides := proc(x, y, z, u)
> # (x,y,0) = coordinates of a corner.
> # z = height of box
> # u = side length of box
> POLYGONS(
> [ [x,y,0], [x+u,y,0], [x+u,y,z], [x,y,z] ],
> [ [x,y,0], [x,y+u,0], [x,y+u,z], [x,y,z] ],
> [ [x+u, y,0], [x+u,y+u,0], [x+u,y+u,z], [x+u,y,z] ],
> [ [x+u, y+u,0], [x,y+u,0], [x,y+u,z], [x+u,y+u,z] ],
> COLOR(RGB,1,1,0) );
> end proc:
The redtop procedure generates a red lid for the box.

> redtop := proc(x, y, z, u)
> # (x,y,z) = coordinates of a corner.
> # u = side length of square
> POLYGONS( [ [x,y,z], [x+u,y,z], [x+u,y+u,z], [x,y+u,z] ],
> COLOR(RGB, 1, 0, 0) );
> end proc:
You can now put the sides and the top inside a PLOT3D structure to
display them.
> PLOT3D( yellowsides(1, 2, 3, 0.5),
> redtop(1, 2, 3, 0.5),
> STYLE(PATCH) );
Using yellowsides and redtop you can create a three-dimensional

histogram plot. Here is the histogram corresponding to z = 1/(x + y + 4),
for 0 ≤ x ≤ 4 and 0 ≤ y ≤ 4.
> sides := seq( seq( yellowsides(i, j, 1/(i+j+4), 0.75),
> j=0..4), i=0..4):
> tops := seq( seq( redtop( i, j, 1/(i+j+4), 0.75),
> j=0..4 ), i=0..4 ):
Histograms look nice when you enclose them in a box of axes. Axes
are generated using AXESSTYLE.
> PLOT3D( sides, tops, STYLE(PATCH), AXESSTYLE(BOXED) );
0.25
0.2
0.15
0.1
0.05
0
0 0
1 1
2 2
3 3
4 4
You can modify the above construction to create a listbarchart3d

procedure which, for a given list of lists of heights, gives a three-
dimensional bar chart as above for its output.
The names of the objects that can appear inside a PLOT3D data struc-
ture include all those that you can use in the PLOT data structure. Thus
POINTS, CURVES, POLYGONS, and TEXT are also available for use inside
an unevaluated PLOT3D call. As in the two-dimensional case, when the
object name is CURVES or POINTS, the point information consists of one
or more lists of three-dimensional points, each list supplying the set of
points making up a single curve in three-dimensional space. In the case of

a POLYGONS structure, the point information consists of one or more lists
of points. In this case, each list describes the vertices of a single polygon
in three-dimensional space. There are two extra objects for PLOT3D struc-
tures. GRID is a structure that describes a functional grid. It consists of
two ranges defining a grid in the x–y plane and a list of lists of z values
over this grid. In the following example LL contains 4 lists each of length
3. Therefore the grid is 4 × 3, and x runs from 1 to 3 in increments of 2/3,
whereas y runs from 1 to 2 in increments of 1/2.
> LL := [ [0,1,0], [1,1,1], [2,1,2], [3,0,1] ]:
> PLOT3D( GRID( 1..3, 1..2, LL ), AXESLABELS(x,y,z),

> ORIENTATION(135, 45), AXES(BOXED) );
3
2.5
2
z
1.5
1
0.5
0
3 1
2.5 1.2
1.4
x2 1.6 y
1.5 1.8
1 2
The MESH structure contains a list of lists of three-dimensional points

describing a surface in three dimensions.1
> LL := [ [ [0,0,0], [1,0,0], [2,0,0], [3,0,0] ],
> [ [0,1,0], [1,1,0], [2.1, 0.9, 0],
> [3.2, 0.7, 0] ],
> [ [0,1,1], [1,1,1], [2.2, 0.6, 1],
> [3.5, 0.5, 1.1] ] ];
LL := [[[0, 0, 0], [1, 0, 0], [2, 0, 0], [3, 0, 0]],

[[0, 1, 0], [1, 1, 0], [2.1, .9, 0], [3.2, .7, 0]],
[[0, 1, 1], [1, 1, 1], [2.2, .6, 1], [3.5, .5, 1.1]]]
The MESH structure represents the quadrilaterals spanned by
LLi,j , LLi,j+1 , LLi+1,j , LLi+1,j+1
1
An n × m × 3 hfarray is also allowed as input to MESH.
for all meaningful values of i and j.

> PLOT3D( MESH( LL ), AXESLABELS(x,y,z), AXES(BOXED),
> ORIENTATION(-140, 45) );
1
0.8
0.6
z
0.4
0.2
0 3.5
1 3
0.8 2.5
0.6 2
y 0.4 1.5x
1
0.2 0.5
0 0
All the options available for PLOT are also available for PLOT3D. In ad-
dition, you can also use the GRIDSTYLE, LIGHTMODEL, and AMBIENTLIGHT
options. See ?plot3d,structure for details on the various options to the
PLOT3D structure.
9.4 Programming with Plot Data Structures

This section describes some of the tools that are available for programming
at the PLOT and PLOT3D data structure level. Plotting data structures
have the advantage of allowing direct access to all the functionality that
Maple’s plotting facilities provide. The examples in section 9.3 show the
extent of the facilities’ power. You could easily thicken the lines in the
sum plot by adding local information to the objects in that example. This
section provides a simple set of examples that describe how to program
at this lower level.
Writing Graphic Primitives

You can write procedures that allow you to work with plot objects at a
more conceptual level. For example, the line and disk commands in the
plottools package provide a model for programming primitives such as
points, lines, curves, circles, rectangles, and arbitrary polygons in both
two and three dimensions. In all cases, you can specify options, such as
line or patch style and color, in the same format as in other plotting
procedures in Maple.
9.4 Programming with Plot Data Structures • 437
> line := proc(x::list, y::list)

> # x and y represent points in either 2-D or 3-D
> local opts;
> opts := convert( opts, PLOToptions );
> CURVES( evalf( [x, y] ), op(opts) );
> end proc:
Inside a procedure, nargs is the number of arguments and args

is the actual argument sequence. Thus, in line, args[3..nargs] is
the sequence of arguments that follow x and y. The convert(...,
PLOToptions) command converts user-level options to the format that
PLOT requires.
> convert( [axes=boxed, color=red], PLOToptions );
[AXESSTYLE(BOX ), COLOUR(RGB , 1.00000000, 0., 0.)]
The disk procedure below is similar to line except that you can spec-
ify the number of points that disk should use to generate the disk. There-
fore disk must handle that option, numpoints, separately. The hasoption
command determines whether a certain option is present.
> disk := proc(x::list, r::algebraic)
> # draw a disk of radius r centered at x in 2-D.
> local i, n, opts, vertices;
> opts := [ args[3..nargs] ] ;
> if not hasoption( opts, numpoints, n, ’opts’ )
> then n := 50;
> end if;
> opts := convert(opts, PLOToptions);
> vertices := seq( evalf( [ x[1] + r*cos(2*Pi*i/n),
> x[2] + r*sin(2*Pi*i/n) ] ),
> i = 0..n );
> POLYGONS( [vertices], op(opts) );
> end proc:
You can now display two disks connected by a line as follows.

> with(plots):
Warning, the name changecoords has been redefined
> display( disk([-1, 0], 1/2, color=plum),

> line([-1, 1/2], [1, 1/2]),
> disk([1, 0], 1/2, thickness=3),
> scaling=constrained );
0.4
0.2
–1.5 –1 –0.5 0.5 1 1.5
–0.2
–0.4
Note how the options to the individual objects apply only to those
objects.
Plotting Gears
This example shows how you can manipulate plotting data structures to
embed two-dimensional plots into a three-dimensional setting. The pro-
cedure below creates a little piece of the boundary of a two-dimensional
graph of a gear-like structure.
> outside := proc(a, r, n)
> local p1, p2;
> p1 := evalf( [ cos(a*Pi/n), sin(a*Pi/n) ] );
> p2 := evalf( [ cos((a+1)*Pi/n), sin((a+1)*Pi/n) ] );
> if r = 1 then p1, p2;
> else p1, r*p1, r*p2, p2;
> end if
> end proc:
For example
> outside( Pi/4, 1.1, 16 );
[.9881327882, .1536020604], [1.086946067, .1689622664],

[1.033097800, .3777683623],
[.9391798182, .3434257839]
> PLOT( CURVES( [%] ), SCALING(CONSTRAINED) );

0.35
0.3
0.25
0.2
0.15
0.93918 1.01306 1.08695
When you put the pieces together, you get a gear. SCALING(CONSTRAINED),
which corresponds to the option scaling=constrained, is used to ensure
that the gear appears round.
> points := [ seq( outside(2*a, 1.1, 16), a=0..16 ) ]:
> PLOT( CURVES(points), AXESSTYLE(NONE), SCALING(CONSTRAINED) );
You can fill this object using the POLYGONS object. However, you must
be careful, as Maple assumes that the polygons are convex. Hence, you
should draw each wedge-shaped section of the gear as a triangular poly-
gon.
> a := seq( [ [0, 0], outside(2*j, 1.1, 16) ], j=0..15 ):
> b := seq( [ [0, 0], outside(2*j+1, 1, 16) ], j=0..15 ):
> PLOT( POLYGONS(a,b), AXESSTYLE(NONE), SCALING(CONSTRAINED) );
Adding STYLE(PATCHNOGRID) to the above structure and combining it

with the curve from the first picture gives you a filled gear-like structure.
To embed this in three dimensions, say at a thickness of t units, you can
use the utility procedures
> double := proc( L, t )
> local u;
> [ seq( [u[1], u[2], 0], u=L ) ],
> [ seq( [u[1], u[2], t], u=L ) ];
> end proc:
which take a list of vertices and create two copies in three-dimensional

space, one at height 0 and the second at height t, and
> border := proc( L1, L2 )
> local i, n;
> n := nops(L1);
> seq( [ L1[i], L2[i], L2[i+1], L1[i+1] ], i = 1..n-1 ),
> [ L1[n], L2[n], L2[1], L1[1] ];
> end proc:
which input two lists of vertices and join the corresponding vertices from
each list into vertices that make up quadrilaterals. You can create the top
and bottom vertices of the gear embedded into three-dimensional space
as follows.
> faces :=
> seq( double(p,1/2),
> p=[ seq( [ outside(2*a+1, 1.1, 16), [0,0] ],
> a=0..16 ),
> seq( [ outside(2*a, 1,16), [0,0] ], a=0..16 )
> ] ):
Now faces is a sequence of doubled outside values.
> PLOT3D( POLYGONS( faces ) );
As above, the following are the points on the outline of a gear.

> points := [ seq( outside(2*a, 1.1, 16), a=0..16 ) ]:
> PLOT( CURVES(points), AXESSTYLE(NONE), SCALING(CONSTRAINED) );
If you double these points, you get vertices of the polygons making
up the border of the three-dimensional gear.
> bord := border( double( [ seq( outside(2*a+1, 1.1, 16),
> a=0..15 ) ], 1/2) ):
> PLOT3D( seq( POLYGONS(b), b=bord ) );
To display the gear you need to put these together in a single PLOT3D
structure. Use STYLE(PATCHNOGRID) as a local option to the top and
bottom of the gear so that they do not appear as several triangles.
> PLOT3D( POLYGONS(faces, STYLE(PATCHNOGRID) ),
> seq( POLYGONS(b), b=bord ),
> STYLE(PATCH), SCALING(CONSTRAINED) );
Note that the global STYLE(PATCH) and SCALING(CONSTRAINED) op-

tions apply to the whole PLOT3D structure, except where the local
STYLE(PATCHNOGRID) option to the top and bottom of the gear over-
rides the global STYLE(PATCH) option.
Polygon Meshes
Section 9.3 describes the MESH data structure which you generate when
you use plot3d to draw a parametrized surface. This simple matter in-
volves converting a mesh of points into a set of vertices for correspond-
ing polygons. Using polygons rather than a MESH structure allows you to
modify the individual polygons. The procedure polygongrid creates the
vertices of a quadrangle at the (i, j)th grid value.
> polygongrid := proc(gridlist, i, j)
> gridlist[j][i], gridlist[j][i+1],
> gridlist[j+1][i+1], gridlist[j+1][i];
> end proc:
You can then use makePolygongrid to construct the appropriate poly-

gons.
> makePolygongrid := proc(gridlist)
> local m,n,i,j;
> n := nops(gridlist);
> m := nops(gridlist[1]);
> POLYGONS( seq( seq( [ polygongrid(gridlist, i, j) ],
> i=1..m-1), j=1..n-1) );
> end proc:
The following is a mesh of points in two-dimensional space.

> L := [ seq( [ seq( [i-1, j-1], i=1..3 ) ], j=1..4 ) ];
L := [[[0, 0], [1, 0], [2, 0]], [[0, 1], [1, 1], [2, 1]],
[[0, 2], [1, 2], [2, 2]], [[0, 3], [1, 3], [2, 3]]]
The makePolygongrid procedure creates the POLYGONS structure cor-
responding to L.
> grid1 := makePolygongrid( L );
grid1 := POLYGONS([[0, 0], [1, 0], [1, 1], [0, 1]],

[[1, 0], [2, 0], [2, 1], [1, 1]], [[0, 1], [1, 1], [1, 2], [0, 2]],
[[1, 1], [2, 1], [2, 2], [1, 2]], [[0, 2], [1, 2], [1, 3], [0, 3]],
[[1, 2], [2, 2], [2, 3], [1, 3]])
Put the polygons inside a PLOT structure to display them.
> PLOT( grid1 );
2.5
1.5
0.5
0 0.5 1 1.5 2
You can also use the convert(..., POLYGONS) command to con-

vert GRID or MESH structures to polygons; see ?convert,POLYGONS.
convert(..., POLYGONS) calls the procedure ‘convert/POLYGONS‘ which,
in the case of a MESH structure, works as the makePolygongrid procedure
above.
9.5 Programming with the plottools Package

While the plotting data structure has the advantage of allowing direct
access to all the functionality that Maple’s plotting facilities provide, it
does not allow you to specify colors (such as red or blue) in an intuitive
way, nor does√it allow you to use all the representations of numeric data,
such as π or 2, that you find in Maple.
This section shows you how to work with basic graphic objects at a
level higher than that of the plotting data structures. The plottools
package provides commands for creating lines, disks, and other two-
dimensional objects, along with commands to generate shapes such as
spheres, tori, and polyhedra. For example, one can draw a sphere of unit
radius and a torus at specified center using a patch style of rendering and
a frame style of axis
> with(plots): with(plottools):
Warning, the name arrow has been redefined
> display( sphere( [0, 0, 2] ), torus( [0, 0, 0] ),

> style=patch, axes=frame, scaling=constrained );
0
–3 –3
–2 –2
–1 –1
0 0
1 1
2 2
3 3
and rotate it at various angles via the functions in the plottools package.
> rotate( %, Pi/4, -Pi/4, Pi/4 );
9.5 Programming with the plottools Package • 445
2
1
0
–1
–2
–3
–2 –2
–1 –1
0 0
1 1
2 2
3
A Pie Chart
You can write a plotting procedure to build a pie chart of a list of inte-
ger data. The piechart procedure below uses the following partialsum
procedure which calculates the partial sums of a list of numbers up to a
given term.
> partialsum := proc(d, i)
> local j;
> evalf( Sum( d[j], j=1..i ) )
> end proc:
For example
> partialsum( [1, 2, 3, -6], 3 );
6.
The piechart procedure first computes the relative weights of the

data along with the centers of each pie slice. piechart uses a TEXT struc-
ture to place the data information at the center of each pie slice and
the pieslice command from the plottools package to generate the pie
slices. Finally, piechart also varies the colors of each slice by defining a
color function based on hue coloring.
> piechart := proc( data::list(integer) )
> local b, c, i, n, x, y, total;
>
> n := nops(data);
> total := partialsum(data, n);
> b := 0, seq( evalf( 2*Pi*partialsum(data, i)/total ),
> i =1..n );
> x := seq( ( cos(b[i])+cos(b[i+1]) ) / 3, i=1..n ):
> y := seq( ( sin(b[i])+sin(b[i+1]) ) / 3, i=1..n ):
> c := (i, n) -> COLOR(HUE, i/(n + 1)):
> PLOT( seq( plottools[pieslice]( [0, 0], 1,
> b[i]..b[i+1], color=c(i, n) ),

> i=1..n),
> seq( TEXT( [x[i], y[i]],
> convert(data[i], name) ),
> i = 1..n ),
> AXESSTYLE(NONE), SCALING(CONSTRAINED) );
> end proc:
Here is a piechart with six slices.

> piechart( [ 8, 10, 15, 10, 12, 16 ] );
10
15
8
10
16
12
The AXESSTYLE(NONE) option ensures that Maple does not draw any
axes with the pie chart.
A Dropshadow Procedure
You can use the existing procedures to create other types of plots that
are not part of the available Maple graphics library. For example, the
following procedure computes the three-dimensional plot of a surface,
z = f (x, y), that has a dropshadow projection onto a plane located below
the surface. The procedure makes use of the commands contourplot,
contourplot3d, display from the plots package, and transform from
the plottools package.
> dropshadowplot := proc(F::algebraic, r1::name=range,
> r2::name=range, r3::name=range)
> local minz, p2, p3, coption, opts, f, g, x, y;
>
> # set the number of contours (default 8)
> opts := [args[5..nargs]];
> if not hasoption( opts, ’contours’, coption, ’opts’ )
> then coption := 8;
> end if;
>
> # determine the base of the plot axes
> # from the third argument
> minz := lhs(‘if‘(r3::range, r3, rhs(r3)));
> minz := evalf(minz);
>
>
> # create 2d and 3d contour plots for F.
> p3 := plots[contourplot3d]( F, r1, r2,
> ’contours’=coption, op(opts) );
> p2 := plots[contourplot]( F, r1, r2,
> ’contours’=coption, op(opts) );
>
> # embed contour plot into R^3 via plottools[transform]
> g := unapply( [x,y,minz], x, y );
> f := plottools[transform]( g );
> plots[display]([ f(p2), p3 ]);
> end proc:
The filled=true option to contourplot and contourplot3d causes

these two commands to fill the regions between the level curves with a
color that indicates the level.
> expr := -5 * x / (x^2+y^2+1);
x
expr := −5
x2 + y2 + 1
> dropshadowplot( expr, x=-3..3, y=-3..3, z=-4..3,

> filled=true, contours=3, axes=frame );
2
1
0
–1
–2
–3
–4
–3 –3
–2 –2
–1 –1
y0 0x
1 1
2 2
3 3
The first section of the dropshadow procedure determines if you have

specified a contours option in the optional arguments (those after the
fourth argument), making use of the hasoption procedure. The next sec-
tion of dropshadowplot determines the z value of the base. Note that
you must take care since you specify ranges differently for formula that
for function input. The remaining sections create the correct plotting ob-
jects which represent the two types of contour plots. dropshadowplot
embeds the two-dimensional contour plot into three-dimensional space
using the transformation
(x, y) 7→ [x, y, minz]
going from R2 → R3 . Finally, it displays the two plots together in one

three-dimensional plotting object.
Note that you can either provide an alternate number of levels or even
specify the precise contour locations via the contours option. Thus,
> dropshadowplot( expr, x=-3..3, y=-3..3, z=-4..3,
> filled=true, contours=[-2,-1,0,1,2] );
produces a plot similar to that shown above, except now it produces

5 contours at levels −2, −1, 0, 1, and 2.
Creating a Tiling
The plottools package provides a convenient environment for program-
ming graphical procedures. For example, you can draw circular arcs in a
unit square.
> a := rectangle( [0,0], [1,1] ),

> arc( [0,0], 0.5, 0..Pi/2 ),
> arc( [1,1], 0.5, Pi..3*Pi/2 ):
> b := rectangle( [1.5,0], [2.5,1] ),
> arc( [1.5,1], 0.5, -Pi/2..0 ),
> arc( [2.5,0], 0.5, Pi/2..Pi ):
You must use display from plots to show the objects that rectangle
and arc create.
> display( a, b, axes=none, scaling=constrained );
You can tile the plane with a and b type rectangles. The following
procedure creates such a m × n tiling using a function, g, to determine
when to use an a-tile and when to use a b-tile. The function g should
return either 0, to use an a-tile, or 1, to use a b-tile.
> tiling := proc(g, m, n)
> local i, j, r, h, boundary, tiles;
>
> # define an a-tile
> r[0] := plottools[arc]( [0,0], 0.5, 0..Pi/2 ),
> plottools[arc]( [1,1], 0.5, Pi..3*Pi/2 );
> # define a b-tile
> r[1] := plottools[arc]( [0,1], 0.5, -Pi/2..0 ),
> plottools[arc]( [1,0], 0.5, Pi/2..Pi );
> boundary := plottools[curve]( [ [0,0], [0,n],
> [m,n], [m,0], [0,0]] );
> tiles := seq( seq( seq( plottools[translate](h, i, j),
> h=r[g(i, j)] ), i=0..m-1 ), j=0..n-1 );
> plots[display]( tiles, boundary, args[4..nargs] );
> end proc:
As an example, define the following procedure which randomly returns

either 0 or 1.
> oddeven := proc() rand() mod 2 end proc:
Create a 20 × 10 tiling (called a Truchet tiling) with no axes and

constrained scaling.
> tiling( oddeven, 20, 10, scaling=constrained, axes=none);
When you use the same procedure again, the random tiling is differ-
ent.
> tiling( oddeven, 20, 10, scaling=constrained, axes=none);
A Smith Chart
The commands in the plottools package allow for easy creation of such
useful graphs as a Smith Chart, used in microwave circuit analysis.
> smithChart := proc(r)
> local i, a, b, c ;
> a := PLOT( seq( plottools[arc]( [-i*r/4,0],
> i*r/4, 0..Pi ),
> i = 1..4 ),
> plottools[arc]( [0,r/2], r/2,
> Pi-arcsin(3/5)..3*Pi/2 ),
> plottools[arc]( [0,r], r, Pi..Pi+arcsin(15/17) ),
> plottools[arc]( [0,2*r], 2*r,
> Pi+arcsin(3/5)..Pi+arcsin(63/65) ),
> plottools[arc]( [0,4*r], 4*r,
> Pi+arcsin(15/17)..Pi+arcsin(63/65) )
> );
> b := plottools[transform]( (x, y) -> [x,-y] )(a);
> c := plottools[line]( [ 0, 0], [ -2*r, 0] ):
> plots[display]( a, b, c, axes = none,
> scaling = constrained,
> args[2..nargs] );
> end proc:
Here is a Smith Chart of radius 1.

> smithChart( 1 );
Exercise
1. Make a Smith Chart by building appropriate circular arcs above the
axes, creating a copy reflected on the axis (using the transform pro-
cedure), and then adding a final horizontal line. The parameter r
denotes the radius of the largest circle. Modifying the smithChart
procedure to add text to mark appropriate grid markers is a simple
operation.
Modifying Polygon Meshes

You can easily construct a new plot tool that works like those in the
plottools package. For example, you can cut out or modify polygon
structures by first working with individual faces and then mapping the
results onto entire polygons. Thus, you can have a procedure that cuts
out the inside of a single face of a polygon.
> cutoutPolygon := proc( vlist_in::{list, hfarray},
> scale::numeric )
> local vlist, i, center, outside, inside, n, edges, polys;
>
> vlist := ‘if‘(vlist_in::hfarray, op(3, eval(vlist_in)),
> vlist_in);
> n := nops(vlist);
> center := add( i, i=vlist ) / n;
> inside := seq( scale*(vlist[i]-center) + center,
> i=1..n);
> outside := seq( [ inside[i], vlist[i],
> vlist[i+1], inside[i+1] ],
> i=1..n-1 ):
> polys := POLYGONS( outside,
> [ inside[n], vlist[n],

> vlist[1], inside[1] ],
> STYLE(PATCHNOGRID) );
> edges := CURVES( [ op(vlist), vlist[1] ],
> [ inside, inside[1] ] );
> polys, edges;
> end proc:
Note that cutoutPolygon was written to handle input in either list

form or hfarray form.
The following are the corners of a triangle.
> triangle := [ [0,2], [2,2], [1,0] ];
triangle := [[0, 2], [2, 2], [1, 0]]
The cutoutPolygon procedure converts triangle to three polygons

(one for each side) and two curves.
> cutoutPolygon( triangle, 1/2 );
1 5 3 5
POLYGONS([[ , ], [0, 2], [2, 2], [ , ]],
2 3 2 3
3 5 2 2 1 5
[[ , ], [2, 2], [1, 0], [1, ]], [[1, ], [1, 0], [0, 2], [ , ]],
2 3 3 3 2 3
STYLE(PATCHNOGRID )), CURVES(
1 5 3 5 2 1 5
[[0, 2], [2, 2], [1, 0], [0, 2]], [[ , ], [ , ], [1, ], [ , ]])
2 3 2 3 3 2 3
Use the display command from the plots package to show the tri-
angle.
> plots[display]( %, color=red );
1.5
0.5
0 0.5 1 1.5 2
The cutout procedure below applies cutoutPolygon to every face of

a polyhedron.
> cutout := proc(polyhedron, scale)
> local v;
> seq( cutoutPolygon( v, evalf(scale) ), v=polyhedron);
> end proc:
You can now cut out 3/4 of each face of a dodecahedron.

> display( cutout( dodecahedron([1, 2, 3]), 3/4 ),
> scaling=constrained);
As a second example, you can take a polygon and raise or lower its
barycenter.
> stellateFace := proc( vlist::list, aspectRatio::numeric )
> local apex, i, n;
>
> n := nops(vlist);
> apex := add( i, i = vlist ) * aspectRatio / n;
> POLYGONS( seq( [ apex, vlist[i],
> vlist[modp(i, n) + 1] ],
> i=1..n) );
> end proc:
The following are the corners of a triangle in three-dimensional space.

> triangle := [ [1,0,0], [0,1,0], [0,0,1] ];
triangle := [[1, 0, 0], [0, 1, 0], [0, 0, 1]]
The stellateFace procedure creates three polygons, one for each side
of the triangle.
> stellateFace( triangle, 1 );
1 1 1
POLYGONS([[ , , ], [1, 0, 0], [0, 1, 0]],
3 3 3
1 1 1 1 1 1
[[ , , ], [0, 1, 0], [0, 0, 1]], [[ , , ], [0, 0, 1], [1, 0, 0]])
3 3 3 3 3 3
Since these POLYGONS belong in three-dimensional space, you must
put them inside a PLOT3D structure to display them.
> PLOT3D( % );
Again, you can extend stellateFace to work for arbitrary polyhedra

having more than one face.
> stellate := proc( polyhedron, aspectRatio)
> local v;
> seq( stellateFace( v, evalf(aspectRatio) ),
> v=polyhedron );
> end proc:
This allows for the construction of stellated polyhedra.

> stellated := display( stellate( dodecahedron(), 3),
> scaling= constrained ):
> display( array( [dodecahedron(), stellated] ) );
You can use convert(..., POLYGONS) to convert a GRID or MESH

structure to the equivalent set of POLYGONS. Here is a POLYGONS version
of the Klein bottle.
> kleinpoints := proc()
> local bottom, middle, handle, top, p, q;
>
> top := [ (2.5 + 1.5*cos(v)) * cos(u),
> (2.5 + 1.5*cos(v)) * sin(u), -2.5 * sin(v) ]:
> middle := [ (2.5 + 1.5*cos(v)) * cos(u),
> (2.5 + 1.5*cos(v)) * sin(u), 3*v - 6*Pi ]:
> handle := [ 2 - 2*cos(v) + sin(u), cos(u),
> 3*v - 6*Pi ]:
> bottom := [ 2 + (2+cos(u))*cos(v), sin(u),
> -3*Pi + (2+cos(u)) * sin(v) ]:
> p := plot3d( {bottom, middle, handle, top},
> u=0..2*Pi, v=Pi..2*Pi, grid=[9,9] ):
> p := select( x -> op(0,x)=MESH, [op(p)] );
> seq( convert(q , POLYGONS), q=p );
> end proc:
> display( kleinpoints(), style=patch,
> scaling=constrained, orientation=[-110,71] );
You can then use the commands for manipulation of polygons to alter
the view of the Klein bottle.
> display( seq( cutout(k, 3/4), k=kleinpoints() ),
9.6 Example: Vector Field Plots

This section describes the problem of plotting a vector field of two dimen-
sional vectors in the plane. The example herein serves to pinpoint some of
the tools available for plot objects on grids in two- and three-dimensional
space.
The command to plot a vector field should have the following syntax.
vectorfieldplot( F, r1, r2 , options )
The input, F, is a list of size two, giving the functions that make up the
horizontal and vertical components of the vector field. The arguments r1
and r2 describe the domain grid of the vectors. The three arguments F,
r1, and r2 are similar in form to the input you need to use for plot3d.
Similarly, the optional information includes any sensible specification that
plot or plot3d allows. Thus, options of the form grid = [m,n ], style
= patch, and color = colorfunction are valid options.
The first problem is to draw a vector. Let [x, y] represent a point, the
starting point of the arrow, and [a, b], the components of the vector. You
can determine the shape of an arrow by three independent parameters,
t1, t2, and t3. Here t1 denotes the thickness of the arrow, t2 the thickness
of the arrow head, and t3 the ratio of the length of the arrow head in
comparison to the length of the arrow itself.
The procedure arrow below from the plottools package constructs
seven vertices of an arrow. It then builds the arrow by constructing two
polygons: a triangle (spanned by v5 , v6 , and v7 ) for the head of the arrow
and a rectangle (spanned by v1 , v2 , v3 , and v4 ) for the tail; it then removes
9.6 Example: Vector Field Plots • 457
boundary lines by setting the style option inside the polygon structure.
It also constructs the boundary of the entire arrow via a closed curve
through the vertices.
> myarrow := proc( point::list, vect::list, t1, t2, t3)
> local a, b, i, x, y, L, Cos, Sin, v, locopts;
>
> a := vect[1]; b := vect[2];
> if has( vect, ’undefined’) or (a=0 and b=0) then
> RETURN( POLYGONS( [ ] ) );
> end if;
> x := point[1]; y := point[2];
> # L = length of arrow
> L := evalf( sqrt(a^2 + b^2) );
> Cos := evalf( a / L );
> Sin := evalf( b / L);
> v[1] := [x + t1*Sin/2, y - t1*Cos/2];
> v[2] := [x - t1*Sin/2, y + t1*Cos/2];
> v[3] := [x - t1*Sin/2 - t3*Cos*L + a,
> y + t1*Cos/2 - t3*Sin*L + b];
> v[4] := [x + t1*Sin/2 - t3*Cos*L + a,
> y - t1*Cos/2 - t3*Sin*L + b];
> v[5] := [x - t2*Sin/2 - t3*Cos*L + a,
> y + t2*Cos/2 - t3*Sin*L + b];
> v[6] := [x + a, y + b];
> v[7] := [x + t2*Sin/2 - t3*Cos*L + a,
> y - t2*Cos/2 - t3*Sin*L + b];
> v := seq( evalf(v[i]), i= 1..7 );
>
> # convert optional arguments to PLOT data structure form
> locopts := convert( [style=patchnogrid,
> args[ 6..nargs ] ],
> PLOToptions );
> POLYGONS( [v[1], v[2], v[3], v[4]],
> [v[5], v[6], v[7]], op(locopts) ),
> CURVES( [v[1], v[2], v[3], v[5], v[6],
> v[7], v[4], v[1]] );
> end proc:
Note that you must build the polygon structure for the arrow in two
parts, because each polygon must be convex. In the special case where the
vector has both components equal to zero or an undefined component,
such as a value resulting from a non-numeric value (for example, a complex
value or a singularity point), the myarrow procedure returns a trivial
polygon. Here are four arrows.
> arrow1 := PLOT(myarrow( [0,0], [1,1], 0.2, 0.4, 1/3,
> color=red) ):
> color=yellow) ):
> color=blue) ):

> color=green) ):
The display command from the plots package can show an array of
plots.
> with(plots):
> display( array( [[arrow1, arrow2], [arrow3, arrow4 ]] ),

1. 1.
0.0. 1. 0.0. 1.
1. 1.
0.0. 1. 0.0. 1.
The remainder of this section presents a number of solutions to the

programming problem of generating a vector field plot, each a bit more
powerful than its predecessors. The first and simplest solution requires
the input to be in functional (rather than expression) form. You first need
three utility procedures that process the domain information, generate a
grid of function values, and place the information in a PLOT3D structure.
The procedure domaininfo determines the endpoints and increments
for the grid. domaininfo takes as input the two ranges r1 and r2 and the
two grid sizes m and n, and returns the grid information as an expression
sequence of four elements.
> domaininfo := proc(r1, r2, m, n)
> lhs(r1), lhs(r2),
> evalf( (rhs(r1) - lhs(r1))/(m-1) ),
> evalf( (rhs(r2) - lhs(r2))/(n-1) );
> end proc:
Here is an example using multiple assignments to assign the four val-

ues returned to separate variables.
> a, b, dx, dy := domaininfo( 0..12, 20..100, 7, 9);
a, b, dx , dy := 0, 20, 2., 10.
Now a, b, dx, and dy have the following values.

> a, b, dx, dy;
0, 20, 2., 10.
For the conversion to a grid of numerical points, you can take advan-
tage of the extendibility of Maple’s convert command. The procedure
‘convert/grid‘ below takes a function f as input and evaluates it over
the grid which r1, r2, m, and n specify.
> ‘convert/grid‘ := proc(f, r1, r2, m, n)
> local a, b, i, j, dx, dy;
> # obtain information about domain
> a,b,dx,dy := domaininfo( r1, r2, m, n );
> # output grid of function values
> [ seq( [ seq( evalf( f( a + i*dx, b + j*dy ) ),
> i=0..(m-1) ) ], j=0..(n-1) ) ];
> end proc:
Now you can evaluate the undefined function, f , on a grid as follows.
> convert( f, grid, 1..2, 4..6, 3, 2 );
[[f(1., 4.), f(1.500000000, 4.), f(2.000000000, 4.)],

[f(1., 6.), f(1.500000000, 6.), f(2.000000000, 6.)]]
The final utility procedure determines the scalings which ensure that
the arrows do not overlap. Then generateplot calls upon the myarrow
procedure to draw the vectors. Note that generateplot moves the origin
of each arrow to center it over its grid-point.
> generateplot := proc(vect1, vect2, m, n, a, b, dx, dy)
> local i, j, L, xscale, yscale, mscale;
>
> # Determine scaling factors.
> L := max( seq( seq( vect1[j][i]^2 + vect2[j][i]^2,
> i=1..m ), j=1..n ) );
> xscale := evalf( dx/2/L^(1/2) );
> yscale := evalf( dy/2/L^(1/2) );
> mscale := max(xscale, yscale);
>
> # Generate plot data structure.
> # Each arrow is centered over its point.
> PLOT( seq( seq( myarrow(
> [ a + (i-1)*dx - vect1[j][i]*xscale/2,
> b + (j-1)*dy - vect2[j][i]*yscale/2 ],
> [ vect1[j][i]*xscale, vect2[j][i]*yscale ],

> mscale/4, mscale/2, 1/3 ), i=1..m), j=1..n) );
> # Thickness of tail = mscale/4
> # Thickness of head = mscale/2
> end proc:
With these utility functions in place, you are ready to make the first
vectorfieldplot command by putting them all together.
> vectorfieldplot := proc(F, r1, r2, m, n)
> local vect1, vect2, a, b, dx, dy;
>
> # Generate each component over the grid of points.
> vect1 := convert( F[1], grid, r1, r2 ,m, n );
> vect2 := convert( F[2], grid, r1, r2 ,m, n );
>
> # Obtain the domain grid information from r1 and r2.
> a,b,dx,dy := domaininfo(r1, r2, m, n);
>
> # Generate the final plot structure.
> generateplot(vect1, vect2, m, n, a, b, dx, dy)
> end proc:
Try the procedure on the vector field (cos(xy), sin(xy)).

> p := (x,y) -> cos(x*y): q := (x,y) -> sin(x*y):
> vectorfieldplot( [p, q], 0..Pi, 0..Pi, 15, 20 );
3
2.5
2
1.5
1
0.5
0 0.5 1 1.5 2 2.5 3
The vectorfieldplot code shows how to write a procedure that gen-

erates vector field plots based on alternative descriptions of the input. For
example, you could create a procedure listvectorfieldplot, with the
input consisting of a list of m lists, each of which consists of n pairs of
points. Each pair of points represents the components of a vector. The
domain grid would be 1, . . . , m in the horizontal direction and 1, . . . , n in
the vertical direction (as for listplot3d from the plots package).
> listvectorfieldplot := proc(F)
> local m, n, vect1, vect2;
>
> n := nops( F ); m := nops( F[1] );
> # Generate the 1st and 2nd components of F.
> vect1 := map( u -> map( v -> evalf(v[1]) , u) , F);
> vect2 := map( u -> map( v -> evalf(v[2]) , u) , F);
>
> # Generate the final plot structure.
> generateplot(vect1, vect2, m, n, 1, 1, m-1, n-1)
> end proc:
For example, the list

> l := [ [ [1,1], [2,2], [3,3] ],
> [ [1,6], [2,0], [5,1] ] ]:
plots
> listvectorfieldplot( l );
2.2
2
1.8
1.6
1.4
1.2
1
1 2 3 4 5
At this stage, the vectorfieldplot procedure still has problems. The

first is that the procedure only works with function input, rather than
with both function and formula input. You can solve this by converting
formula expressions to procedures, and then having vectorfieldplot call
itself recursively with the new output as in the ribbonplot procedure in
section 9.2.
A second problem is that vectorfieldplot only works with lists as
input, not hfarrays.
To overcome such problems, ensure that you first convert all input
functions to functions that only output a numeric real value or the value
undefined, the only type of data the Maple plotting data structure ac-
cepts. You may also want to use the more efficient hardware floating-
point calculations rather than software floating-point operations, when-
ever possible. Section 9.7 describes how to do this. Instead of writing your
own procedure for computing the grid, you can use the library function
convert(..., gridpoints) which, in the case of a single input, gener-
ates a structure of the following form.
[ a.. b, c.. d, [ [z11, ... , z1n ], ... ,

[ zm1 , ... , zmn ] ] ]
The third argument may also be an hfarray.

It uses either expressions or procedures as input. The output gives the
domain information a..b and c..d along with the z values of the input
that it evaluates over the grid.
> convert( sin(x*y), ’gridpoints’,
> x=0..Pi, y=0..Pi, grid=[2, 3] );
[0...3.14159265358979, 0...3.14159265358979, [
[0., 0., 0.],
[0., −.975367972083633571, −.430301217000074065]]]
When xy > 0 and ln(−xy) is complex, the grid contains the value
undefined.
> convert( (x,y) -> log(-x*y), ’gridpoints’,
> 1..2, -2..1, grid=[2,3] );
[1...2., −2...1., [[.693147180559945286,

−.693147180559945286, undefined ],
[1.386294361, 0., undefined ]]]
The version of vectorfieldplot below makes use of the convert(...,
gridpoints) procedure. The vectorfieldplot command should allow a
number of options. In particular, it should allow a grid = [m,n ] op-
tion. You can accomplish this by passing the options to convert(...,
gridpoints). The utility procedure makevectors handles the interface
to convert(..., gridpoints).
> makevectors := proc( F, r1, r2 )
> local v1, v2;
>
> # Generate the numerical grid
> # of components of the vectors.
> v1 := convert( F[1], ’gridpoints’, r1, r2,
> args[4 .. nargs] );
> v2 := convert( F[2], ’gridpoints’, r1, r2,
> args[4 .. nargs] );
>
> # The domain information is contained in first
> # two operands of v1. The function values in
> # the 3rd components of v1 and v2.
> [ v1[1], v1[2], v1[3], v2[3] ]
> end proc:
Here is the new version of vectorfieldplot.

> vectorfieldplot := proc(F, r1, r2)
> local R1, R2, m, n, a, b, v1, v2, dx, dy, v;
>
> v := makevectors( F, r1, r2, args[4..nargs] );
> R1 := v[1]; R2 := v[2]; v1 := v[3]; v2 := v[4];
>
> n := nops(v1); m := nops(v1[1]);
> a,b,dx,dy := domaininfo(R1, R2, m, n);
>
> generateplot(v1, v2, m, n, a, b, dx, dy);
> end proc:
Test this procedure.

> p := (x,y) -> cos(x*y):
> q := (x,y) -> sin(x*y):
> vectorfieldplot( [p, q], 0..Pi, 0..Pi,
> grid=[3, 4] );
3
2.5
2
1.5
1
0.5
0 0.5 1 1.5 2 2.5 3 3.5
All the versions of vectorfieldplot so far have scaled each arrow

so that each vector fits into a single grid box. No overlapping of arrows
occurs. However, the arrows still vary in length. Often this results in
graphs that have a large number of very small, almost invisible vectors.
For example, a plot of the gradient field of F = cos(xy) exhibits this
behavior.
> vectorfieldplot( [y*cos(x*y), x*sin(x*y)],
> x=0..Pi, y=0..Pi, grid=[15,20]);
3
2.5
2
1.5
1
0.5
0 0.5 1 1.5 2 2.5 3
The final version of vectorfieldplot differs in that all the arrows

have the same length—the color of each vector provides the information
about the magnitudes of the arrows. You must add a utility procedure
that generates a grid of colors from the function values.
> ‘convert/colorgrid‘ := proc( colorFunction )
> local colorinfo, i, j, m, n;
>
> colorinfo := op( 3, convert(colorFunction,
> ’gridpoints’, args[2..nargs] ) );
> map( x -> map( y -> COLOR(HUE, y), x) , colorinfo );
> end proc:
The above procedure uses the convert( ... , gridpoints) to gen-

erate a list of lists of function values that specify the colors (using hue
coloring).
> convert( sin(x*y), ’colorgrid’,
> x=0..1, y=0..1, grid=[2,3] );
[[COLOR(HUE , 0.), COLOR(HUE , 0.), COLOR(HUE , 0.)

], [COLOR(HUE , 0.),
COLOR(HUE , .479425538604203006),
COLOR(HUE , .841470984807896505)]]
Here is the final version of vectorfieldplot.
> vectorfieldplot := proc( F, r1, r2 )
> local v, m, n, a, b, dx, dy, opts, p, v1, v2,
> L, i, j, norms, colorinfo,
> xscale, yscale, mscale;
>
> v := makevectors( F, r1, r2, args[4..nargs] );
> v1 := v[3]; v2 := v[4];
> n := nops(v1); m := nops( v1[1] );
>
> a,b,dx,dy := domaininfo(v[1], v[2], m, n);
>
> # Determine the function used for coloring the arrows.
> opts := [ args[ 4..nargs] ];
> if not hasoption( opts, color, colorinfo, ’opts’ ) then
> # Default coloring will be via
> # the scaled magnitude of the vectors.
> L := max( seq( seq( v1[j][i]^2 + v2[j][i]^2,
> i=1..m ), j=1..n ) );
> colorinfo := ( F[1]^2 + F[2]^2 )/L;
> end if;
>
> # Generate the information needed to color the arrows.
> colorinfo := convert( colorinfo, ’colorgrid’,
> r1, r2, op(opts) );
>
> # Get all the norms of the vectors using zip.
> norms := zip( (x,y) -> zip( (u,v)->
> if u=0 and v=0 then 1 else sqrt(u^2 + v^2) end if,
> x, y), v1, v2);
> # Normalize v1 and v2 (again using zip ).
> v1 := zip( (x,y) -> zip( (u,v)-> u/v, x, y),
> v1, norms );
>
> v2 := zip( (x,y) -> zip( (u,v)-> u/v, x, y),
> v2, norms );
>
> # Generate scaling information and plot data structure.
> xscale := dx/2.0; yscale := dy/2.0;
> mscale := max(xscale, yscale);
>
> PLOT( seq( seq( myarrow(
> [ a + (i-1)*dx - v1[j][i]*xscale/2,
> b + (j-1)*dy - v2[j][i]*yscale/2 ],
> [ v1[j][i]*xscale, v2[j][i]*yscale ],
> mscale/4, mscale/2, 1/3,
> ’color’=colorinfo[j][i]
> ), i=1..m ), j=1..n ) );
> end proc:
With this new version you can obtain the following plots.
> x=0..Pi, y=0..Pi,grid=[15,20] );
3
2.5
2
1.5
1
0.5
0 0.5 1 1.5 2 2.5 3
You can color the vectors via a function, such as sin(xy).

> x=0..Pi, y=0..Pi, grid=[15,20], color=sin(x*y) );
3
2.5
2
1.5
1
0.5
0 0.5 1 1.5 2 2.5 3
Other vector field routines can be derived from the routines above.
For example, you can also write a complex vector field plot that takes
complex number locations and complex number descriptions of vectors
as input. You simply need to generate the grid of points in an alternate
manner.
In the plottools package, there is the arrow function, which gener-
ates arrows and vectors. This function is more versatile than the proce-
dures described in this section.
9.7 Generating Grids of Points

Section 9.6 points out that the simple operation of obtaining an array
of grid values for a given procedure, that is, the problem of computing
the values of a function you wish to plot over a grid of points, is not an
9.7 Generating Grids of Points • 467
obvious task. You must deal with efficiency, error conditions, and non-
numeric output. You can handle the case where the input is a formula
in two variables in the same way as in the ribbonplot procedure from
section 9.2. Thus, for simplicity of presentation, this section avoids this
particular case.
The goal is to compute an array of values for f at each point on a
m × n rectangular grid. That is, at the locations
xi = a + (i − 1)δx and yj = c + (j − 1)δy
where δx = (b − a)/(m − 1) and δy = (d − c)/(n − 1). Here i and j vary

from 1 to m and 1 to n, respectively.
Consider the function f : (x, y) 7→ 1/ sin(xy). You need to evaluate f
over the m × n grid with the ranges a, . . . , b and c, . . . , d.
> f := (x,y) -> 1 / sin(x*y);
1
f := (x, y) →
sin(x y)
The first step is to convert the function f to a numeric procedure.

Since Maple requires numeric values (rather than symbolic) for plots, ask
Maple to convert f to a form which returns numerical answers or the
special value undefined.
> fnum := convert( f , numericproc );
fnum := proc(_X , _Y )
local err ;
err := traperror(evalhf(f(_X , _Y ))) ;
if type([err ], [numeric]) then err
else
err := traperror(evalf(f(_X , _Y ))) ;
if type([err ], [numeric]) then err else undefined end if
end if
end proc
The above procedure, which is the result of this conversion, attempts
to calculate the numerical values as efficiently as possible. Hardware
floating-point arithmetic, although of limited precision, is more efficient
than software floating-point and is frequently sufficient for plotting. Thus,
fnum tries evalhf first. If evalhf is successful, it returns a numeric result;
otherwise, it generates an error message. If this happens, fnum attempts

the calculation again by using software floating-point arithmetic by call-
ing evalf. Even this calculation is not always possible. In the case of f ,
the function is undefined whenever x = 0 or y = 0. In such cases, the pro-
cedure fnum returns the name undefined. Maple’s plot display routines
recognize this special name.
At the point (1, 1), the function f has the value 1/ sin(1) and so fnum
returns a numerical estimate.
> fnum(1,1);
1.18839510577812123
However, if you instead try to evaluate this same function at (0, 0),
Maple informs you that the function is undefined at these coordinates.
> fnum(0,0);
undefined
Creating such a procedure is the first step in creating the grid of

values.
For reasons of efficiency, you should, whenever you can, compute not
only the function values but also the grid points by using hardware
floating-point arithmetic. In addition, you should do as much compu-
tation as possible in a single call to evalhf. Whenever you use hardware
floating-point arithmetic, Maple must first convert the expression to a
series of commands of hardware floating-point numbers, and then convert
the result of these back to Maple’s format for numbers.
Write a procedure that generates the coordinates of the grid in the
form of an array. Since the procedure is to plot surfaces, the array is
two-dimensional. The following procedure returns an array z of function
values.
> evalgrid := proc( F, z, a, b, c, d, m, n )
> local i, j, dx, dy;
>
> dx := (b-a)/m; dy := (d-c)/n;
> for i to m do
> for j to n do
> z[i, j] := F( a + (i-1)*dx, c + (j-1)*dy );
> end do;
> end do;
> end proc:
9.7 Generating Grids of Points • 469
This evalgrid procedure is purely symbolic and does not handle error
conditions.
> A := array(1..2, 1..2):
> evalgrid( f, ’A’, 1, 2, 1, 2, 2, 2 ):
> eval(A);
 1 1 
 sin(1) 3 
sin( ) 
2 

 1 1 

3 9
 
sin( ) sin( )
2 4
> evalgrid( f, ’A’, 0, Pi, 0, Pi, 15, 15 ):

Write a second procedure, gridpoints, which makes use of evalgrid.

The procedure should accept a function, two ranges, and the number of
grid points to generate in each dimension. Like the procedure fnum which
Maple generated from your function f above, this routine should attempt
to create the grid using hardware floating-point arithmetic. Only if this
fails, should gridpoints resort to software floating-point arithmetic.
> gridpoints := proc( f, r1, r2, m, n )
> local u, x, y, z, a, b, c, d;
>
> # Domain information:
> a := lhs(r1); b := rhs(r1);
> c := lhs(r2); d := rhs(r2);
>
> z := hfarray(1..m, 1..n);
> if Digits <= evalhf(Digits) then
> # Try to use hardware floats
> # - notice the need for var in this case.
> u := traperror( evalhf( evalgrid(f, var(z),
> a, b, c, d, m, n) ) );
> if lasterror = u then
> # Use software floats, first converting f to
> # a software float function.
> z := array( 1..m, 1..n );
> evalgrid( convert( f, numericproc ),
> z, a, b, c, d, m, n );
> end if;
> else
> # Use software floats, first converting f to
> # a software float function.
> z := array( 1..m, 1..n );
> evalgrid( convert(f, numericproc), z,
> a, b, c, d, m, n );
> end if;
> eval(z);
> end proc:
The second argument to evalgrid must be the array (or hfarray)

which receives the results; Maple must not convert it to a number before
it calls evalhf. Indicate this special status to Maple using the special
function var whenever you call evalgrid from within evalhf. Chapter 8
discusses numerical calculations in detail.
Test the procedures. Here gridpoints can use hardware floating-point
arithmetic to calculate two of the numbers, but it must resort to software
calculations in four cases where the function turns out to be undefined.
> gridpoints( (x,y) -> 1/sin(x*y) , 0..3, 0..3, 2, 3 );
[undefined , undefined , undefined ]

[undefined , 1.00251130424672485 ,
7.08616739573718667]
In the following example, gridpoints can use hardware floating-point
for all the calculations. Therefore, this calculation is faster, although the
difference will not be apparent unless you try a much larger example.
> gridpoints( (x,y) -> sin(x*y) , 0..3, 0..3, 2, 3 );
[0. , 0. , 0.]
[0. , .997494986604054445 , .141120008059867213]
If you ask for more digits than hardware floating-point arithmetic
can provide, then gridpoints must always use software floating-point
operations.
> Digits := 22:
> gridpoints( (x,y) -> sin(x*y) , 0..3, 0..3, 2, 3 );
[0. , 0. , 0.]
[0. , .9974949866040544309417 ,
.1411200080598672221007]
> Digits := 10:
9.8 Animation • 471
The gridpoints procedure is remarkably similar to the convert(...,

gridpoints) procedure which is part of the standard Maple library. The
library command includes more checking of the arguments and, therefore,
will likely suffice for many of your needs.
9.8 Animation
Maple has the ability to generate animations in either two or three dimen-
sions. As with all of Maple’s plotting facilities, such animations produce
user-accessible data structures. Data structures of the following type rep-
resent animations.
PLOT( ANIMATE( ... ) )
or
PLOT3D( ANIMATE( ... ) )
Inside the ANIMATE function is a sequence of frames; each frame is

a list of the same plotting objects that can appear in a single plotting
structure. Every procedure that creates an animation builds such a se-
quence of frames. You can see an example by printing the output of such
a procedure.
> lprint( plots[animate]( x*t, x=-1..1, t = 1..3,
> numpoints=3, frames = 3 ) );
PLOT(ANIMATE([CURVES([[-1., -1.], [0., 0.], [1.0000000\
00, 1.]],COLOUR(RGB,1.00000000,0.,0.))],[CURVES([[-1.,
-2.], [0., 0.], [1.000000000, 2.]],COLOUR(RGB,1.000000\
00,0.,0.))],[CURVES([[-1., -3.], [0., 0.], [1.00000000\
0, 3.]],COLOUR(RGB,1.00000000,0.,0.))]),AXESLABELS(x,
‘‘),VIEW(-1. .. 1.,DEFAULT))
The function points below is a parametrization of the curve (x, y) =

(1 + cos(tπ/180)2 , 1 + cos(tπ/180) sin(tπ/180)).
> points := t -> evalf(
> [ (1 + cos(t/180*Pi)) * cos(t/180*Pi ),
> (1 + cos(t/180*Pi)) * sin(t/180*Pi ) ] ):
For example,
> points(2);
[1.998172852, .06977773357]
You can plot a sequence of points.

> PLOT( POINTS( seq( points(t), t=0..90 ) ) );
1.2
1
0.8
0.6
0.4
0.2
0 0.5 1 1.5 2
You can now make an animation. Make each frame consist of the
polygon spanned by the origin, (0, 0), and the sequence of points on the
curve.
> frame := n -> [ POLYGONS([ [ 0, 0 ],
> seq( points(t), t = 0..60*n) ],
> COLOR(RGB, 1.0/n, 1.0/n, 1.0/n) ) ]:
The animation consists of six frames.
> PLOT( ANIMATE( seq( frame(n), n = 1..6 ) ) );
The display command from the plots package can show an anima-
tion in static form.
> with(plots):
> display( PLOT(ANIMATE(seq(frame(n), n = 1..6))) );

1.00 1.00 1.00

.50 .50 .50
0.0. 2. 0.0. 2. 0.0. 2.
1.00 1. 1.
.50 0.0. 0.0.
2. 2.
0.0. 2. –1. –1.
The varyAspect procedure below illustrates how a stellated surface

varies with the aspect ratio. The procedure takes a graphical object as
input and creates an animation in which each frame is a stellated version
of the object with a different aspect ratio.
> with(plottools):
> varyAspect := proc( p )

> local n, opts;
> opts := convert( [ args[2..nargs] ], PLOT3Doptions );
> PLOT3D( ANIMATE( seq( [ stellate( p, n/sqrt(2)) ],
> n=1..4 ) ),
> op( opts ));
> end proc:
Try the procedure on a dodecahedron.

> varyAspect( dodecahedron(), scaling=constrained );
Here is the static version.

> display( varyAspect( dodecahedron(),
> scaling=constrained ) );
The Maple library provides three methods for creating animations: the
animate and animate3d commands in the plots package, or the display
command with the insequence = true option set. For example, you can
show how a Fourier series approximates a function, f , on an interval
[a, b] by visualizing the function and successive approximations as the
number of terms increase with each frame. You can derive the nth partial
n
X 2π
sum of the Fourier series by using fn (x) = c0 /2 + ck cos( kx) +
b−a
k=1
2π
sk sin( kx), where
b−a
Z b ² ³
2 2π
ck = f (x) cos kx dx
b−a a b−a
and Z b ² ³
2 2π
sk = f (x) sin kx dx.
b−a a b−a
The fourierPicture procedure below first calculates and plots the kth
Fourier approximation for k up to n. Then fourierPicture generates an
animation of these plots, and finally it adds a plot of the function itself
as a backdrop.
> fourierPicture :=
> proc( func, xrange::name=range, n::posint)
> local x, a, b, l, k, j, p, q, partsum;
>
> a := lhs( rhs(xrange) );
> b := rhs( rhs(xrange) );
> l := b - a;
> x := 2 * Pi * lhs(xrange) / l;
>
> partsum := 1/l * evalf( Int( func, xrange) );
> for k from 1 to n do
> # Generate the terms of the Fourier series of func.
> partsum := partsum

> + 2/l * evalf( Int(func*sin(k*x), xrange) )
> * sin(k*x)
> + 2/l * evalf( Int(func*cos(k*x), xrange) )
> * cos(k*x);
> # Plot k-th Fourier approximation.
> q[k] := plot( partsum, xrange, color=blue,
> args[4..nargs] );
> end do;
> # Generate sequence of frames.
> q := plots[display]( [ seq( q[k], k=1..n ) ],
> insequence=true );
> # Add the function plot, p, to each frame.
> p := plot( func, xrange, color = red, args[4..nargs] );
> plots[display]( [ q, p ] );
> end proc:
You can now use fourierPicture to see, for example, the first six
Fourier approximations of ex .
> fourierPicture( exp(x), x=0..10, 6 ):
This is the static version.

> display( fourierPicture( exp(x), x=0..10, 6 ) );
.200e5 .200e5 .200e5

.150e5 .150e5 .150e5
.100e5 x.100e5 x.100e5 x
.50e4 .50e4 .50e4
0.0.2.4.6.8.10. 0.0.2.4.6.8.10. 0.0.2.4.6.8.10.
.200e5 .200e5 .200e5

.150e5 .150e5 .150e5
.100e5 x.100e5 x.100e5 x
.50e4 .50e4 .50e4
0.0.2.4.6.8.10. 0.0.2.4.6.8.10. 0.0.2.4.6.8.10.
Below are the first six Fourier approximations of x -> signum(x-1).

The signum function is discontinuous, so the discont=true option is
called for.
> fourierPicture( 2*signum(x-1), x=-2..3, 6,
> discont=true );
Again, these pages require a static version.

> display( fourierPicture( 2*signum(x-1), x=-2..3, 6,
> discont=true ) );
2. 2.
1. 2.
1. 1.
–2. 0.
–1.0.1.x2.3. 0.0.1.x2.3.
–1. –2. 0.0.1.x2.3. –2.
–1. –1.
–1. –1.
–2. –2.
–2.
2. 2. 2.
1. 1. 1.
0.0.1.x2.3. –2. 0.0.1.x2.3. –2. 0.0.1.x2.3.
–1.
–2.
–1. –1.
–1. –1. –1.
–2. –2. –2.
You can also create similar animations with other series approxima-
tions, such as Taylor, Padé, and Chebyshev–Padé, with the generalized
series structures that Maple uses.
Animation sequences exist in both two and three dimensions. The
procedure below ties a trefoil knot by using the tubeplot function in the
plots package.
> TieKnot := proc( n:: posint )
> local i, t, curve, picts;
> curve := [ -10*cos(t) - 2*cos(5*t) + 15*sin(2*t),
> -15*cos(2*t) + 10*sin(t) - 2*sin(5*t),
> 10*cos(3*t) ]:
> picts := [ seq( plots[tubeplot]( curve,
> t=0..2*Pi*i/n, radius=3),
> i=1..n ) ];
> plots[display]( picts, insequence=true, style=patch);
> end proc:
You can tie the knot in, say, six stages.

> TieKnot(6);
Here is the static version.

> display( TieKnot(6) );
You can combine the graphical objects from the plottools pack-
age with the display in-sequence option to animate physical objects in
motion. The springPlot procedure below creates a spring from a three-
dimensional plot of a helix. springPlot also creates a box and a copy of
this box and moves one of the boxes to various locations depending on a
value of u. For every u, you can locate these boxes above and below the
spring. Finally springPlot makes a sphere and translates it to locations
above the top of the top box with the height again varying with a param-
eter. Finally, it produces the entire animation by organizing a sequence
of positions and showing them in sequence by using display.
> springPlot := proc( n )
> local u, curve, springs, box, tops, bottoms,
> helix, ball, balls;
> curve := (u,v) -> spacecurve(
> [cos(t), sin(t), 8*sin(u/v*Pi)*t/200],
> t=0..20*Pi,
> color=black, numpoints=200, thickness=3 ):
> springs := display( [ seq(curve(u,n), u=1..n) ],
> insequence=true ):
> box := cuboid( [-1,-1,0], [1,1,1], color=red ):
> ball := sphere( [0,0,2], grid=[15, 15], color=blue ):
> tops := display( [ seq(
> translate( box, 0, 0, sin(u/n*Pi)*4*Pi/5 ),
> u=1..n ) ], insequence=true ):
> bottoms := display( [ seq( translate(box, 0, 0, -1),
> balls := display( [ seq( translate( ball, 0, 0,
> 4*sin( (u-1)/(n-1)*Pi ) + 8*sin(u/n*Pi)*Pi/10 ),
> display( springs, tops, bottoms, balls,
> style=patch, orientation=[45,76],
> end proc:
The code above uses the short names of the commands from the plots
and plottools packages in order to improve readability. You must either
use long names or remember to load these two packages before using
springPlot.
> springPlot(6);
> display( springPlot(6) );
Section 9.5 describes how the commands in the plottools package

can help you with graphics procedures.
9.9 Programming with Color

As well as coloring each type of object in the plot data structures, you
can also add colors to plotting routines. The color option allows you to
specify colors in the form of a solid color, by name, by RGB or HUE values,
or via a color function in the form of a Maple formula or function. Try
each of the following commands for yourself.
> plot3d( sin(x*y), x=-3..3, y=-3..3, color=red );

> plot3d( sin(x*y), x=-3..3, y=-3..3,
> color=COLOUR(RGB, 0.3, 0.42, 0.1) );
> p := (x,y) -> sin(x*y):

> q := (x,y) -> if x < y then 1 else x - y end if:
> plot3d( p, -3..3, -3..3, color=q );
Although usually less convenient, you may also specify the color at-
tributes at the lower level of graphics primitives. At the lowest level, you
can accomplish a coloring of a graphical object by including a COLOUR
function as one of the options inside the object.
> PLOT( POLYGONS( [ [0,0], [1,0], [1,1] ],
> [ [1,0], [1,1], [2,1], [2,0] ],
> COLOUR(RGB, 1/2, 1/3, 1/4 ) ) );
9.9 Programming with Color • 479
0.8
0.6
0.4
0.2
0 0.5 1 1.5 2
You can use different colors for each polygon via either
PLOT( POLYGONS( P1, ... , Pn ,

COLOUR(RGB, p1, ..., pn )) )
or
PLOT( POLYGONS( P1, COLOUR(RGB, p1 ) ), ... ,

POLYGONS( Pn, COLOUR(RGB, pn )) )
Thus, the following two PLOT structures represent the same picture of
a red and a green triangle.
> PLOT( POLYGONS( [ [0,0], [1,1], [2,0] ],
> COLOUR( RGB, 1, 0, 0 ) ),
> POLYGONS( [ [0,0], [1,1], [0,1] ],
> COLOUR( RGB, 0, 1, 0 ) ) );
> PLOT( POLYGONS( [ [0,0], [1,1], [2,0] ],

> [ [0,0], [1,1], [0,1] ],
> COLOUR( RGB, 1, 0, 0, 0, 1, 0 ) ) );
0.8
0.6
0.4
0.2
0 0.5 1 1.5 2
The three RGB values must be numbers between 0 and 1.

Generating Color Tables

The following procedure generates an m × n color table of RGB values.
Specifically, colormap returns a sequence of two elements: a POLYGONS
structure and a TITLE.
> colormap := proc(m, n, B)
> local i, j, points, colors, flatten;
> # points = sequence of corners for rectangles
> points := seq( seq( evalf(
> [ [i/m, j/n], [(i+1)/m, j/n],
> [(i+1)/m, (j+1)/n], [i/m, (j+1)/n] ]
> ), i=0..m-1 ), j=0..n-1 ):
> # colors = listlist of RGB color values
> colors := [seq( seq( [i/(m-1), j/(n-1), B],
> i=0..m-1 ), j=0..n-1 )] ;
> # flatten turns the colors listlist into a sequence
> flatten := a -> op( map(op, a) );
> POLYGONS( points,
> COLOUR(RGB, flatten(colors) ) ),
> TITLE( cat( "Blue=", convert(B, string) ) );
> end proc:
Here is a 10 × 10 table of colors; the blue component is 0.

> PLOT( colormap(10, 10, 0) );
Blue=0
1
0.8
0.6
0.4
0.2
0 0.2 0.4 0.6 0.8 1
You can use animation to vary the blue component as well. The
colormaps procedure below uses animation to generate an m × n × f
color table.
> colormaps := proc(m, n, f)
> local t;
> PLOT( ANIMATE( seq( [ colormap(m, n, t/(f-1)) ],
> t=0..f-1 ) ),
> AXESLABELS("Red", "Green") );
> end proc:
The following gives you a 10 × 10 × 10 color table.

> colormaps(10, 10, 10);
You can visualize the color scale for HUE coloring as follows.
> points := evalf( seq( [ [i/50, 0], [i/50, 1],
> [(i+1)/50, 1], [(i+1)/50, 0] ],
> i=0..49)):
> PLOT( POLYGONS(points, COLOUR(HUE, seq(i/50, i=0..49)) ),

> AXESTICKS(DEFAULT, 0), STYLE(PATCHNOGRID) );
0.2 0.4 0.6 0.8 1
The AXESTICKS(DEFAULT, 0) specification eliminates the axes label-

ing along the vertical axis but leaves the default labeling along the hori-
zontal axis.
You can easily see how to create a colormapHue procedure which
creates the color scale for any color function based on HUE coloring.
> colormapHue := proc(F, n)
> local i, points;
> points := seq( evalf( [ [i/n, 0], [i/n, 1],
> [(i+1)/n, 1], [(i+1)/n, 0] ]
> ), i=0..n-1 ):
> PLOT( POLYGONS( points,
> COLOUR(HUE, seq( evalf(F(i/n)), i=0.. n-1) )),
> AXESTICKS(DEFAULT, 0), STYLE(PATCHNOGRID) );
> end proc:
The basis of this color scale is y(x) = sin(πx)/3 for 0 ≤ x ≤ 40.

> colormapHue( x -> sin(Pi*x)/3, 40);
0.2 0.4 0.6 0.8 1
Visualizing the grayscale coloring is a simple matter of using an ar-

bitrary procedure, F , since gray levels are simply those levels that have
equal parts of red, green, and blue.
> colormapGraylevel := proc(F, n)
> local i, flatten, points, grays;
> points := seq( evalf([ [i/n, 0], [i/n, 1],
> [(i+1)/n, 1], [(i+1)/n, 0] ]),
> i=0..n-1):
> flatten := a -> op( map(op, a) );
> grays := COLOUR(RGB, flatten(
> [ seq( evalf([ F(i/n), F(i/n), F(i/n) ]),
> i=1.. n)]));
> PLOT( POLYGONS(points, grays),
> AXESTICKS(DEFAULT, 0) );
> end proc:
The identity function, x 7→ x, yields the basic gray scale.

> colormapGraylevel( x->x, 20);
0.2 0.4 0.6 0.8 1
Adding Color Information to Plots

You can add color information to an existing plot data structure. The
procedure addCurvecolor colors each curve in a CURVES function via the
scaled y coordinates.
> addCurvecolor := proc(curve)

> local i, j, N, n , M, m, curves, curveopts, p, q;
>
> # Get existing point information.
> curves := select( type, [ op(curve) ],
> list(list(numeric)) );
> # Get all options but color options.
> curveopts := remove( type, [ op(curve) ],
> { list(list(numeric)),
> specfunc(anything, COLOR),
> specfunc(anything, COLOUR) } );
>
> # Determine the scaling.
> # M and m are the max and min of the y-coords.
> n := nops( curves );
> N := map( nops, curves );
> M := [ seq( max( seq( curves[j][i][2],
> i=1..N[j] ) ), j=1..n ) ];
> m := [ seq( min( seq( curves[j][i][2],
> i=1..N[j] ) ), j=1..n ) ];
> # Build new curves adding HUE color.
> seq( CURVES( seq( [curves[j][i], curves[j][i+1]],
> i=1..N[j]-1 ),
> COLOUR(HUE, seq((curves[j][i][2]
> - m[j])/(M[j] - m[j]),
> i=1..N[j]-1)),
> op(curveopts) ), j=1..n );
> end proc:
For example
> c := CURVES( [ [0,0], [1,1], [2,2], [3,3] ],
> [ [2,0], [2,1], [3,1] ] );
c := CURVES([[0, 0], [1, 1], [2, 2], [3, 3]],

[[2, 0], [2, 1], [3, 1]])
> addCurvecolor( c );
CURVES([[0, 0], [1, 1]], [[1, 1], [2, 2]], [[2, 2], [3, 3]],
1 2
COLOUR(HUE , 0, , )), CURVES([[2, 0], [2, 1]],
3 3
[[2, 1], [3, 1]], COLOUR(HUE , 0, 1))
You can then map such a procedure over all CURVES structures of an
existing plot structure to provide the desired coloring for each curve.
> addcolor := proc( aplot )
> local recolor;
> recolor := x -> if op(0,x)=CURVES then
> addCurvecolor(x)
> else x end if;
> map( recolor, aplot );
> end proc:
Try addcolor on a plot of sin(x) + cos(x).

> p := plot( sin(x) + cos(x), x=0..2*Pi,
> linestyle=2, thickness=3 ):
> addcolor( p );
0.5
0 1 2 3 4 5 6
x
–0.5
–1
If you add color to two curves simultaneously, the two colorings are
independent.
> q := plot( cos(2*x) + sin(x), x=0..2*Pi ):
> addcolor( plots[display](p, q) );
1
0.5
1 2 3x 4 5 6
0
–0.5
–1
–1.5
–2
The addcolor procedure also works on three-dimensional space

curves.
> spc := plots[spacecurve]( [ cos(t), sin(t), t ], t=0..8*Pi,
> numpoints=100, thickness=2, color=black ):
> addcolor( spc );
You can easily alter the coloring of an existing plot by using coloring
functions. Such coloring functions should either be of the form CHue : R2 →
[0, 1] (for Hue coloring) or of the form CRGB : R2 → [0, 1] × [0, 1] × [0, 1].
The example above uses the color function CHue (x, y) = y/ max(yi ).
Creating A Chess Board Plot

The final example of programming with color shows how to make a chess
board type grid with red and white squares in a three-dimensional plot.
You do not simply assign a coloring function as an argument to plot3d.
A coloring function, in such a case, provides colors for vertices of a grid,
which does not yield color patches. You must first convert the grid or
mesh into polygonal form. The rest of the procedure assigns either a red
or white color to a polygon, depending on which grid area it represents.
> chessplot3d := proc(f, r1, r2)
> local m, n, i, j, plotgrid, p, opts, coloring, size;
>
> # obtain grid size
> # and generate the plotting data structure
> if hasoption( [ args[4..nargs] ], grid, size) then
> m := size[1];
> n := size[2];
> else # defaults
> m := 25;
> n := 25;
> end if;
>
> p := plot3d( f, r1, r2, args[4..nargs] );
>
> # convert grid data (first operand of p)
> # into polygon data
> plotgrid := op( convert( op(1, p), POLYGONS ) );
> # make coloring function - alternating red and white
> coloring := (i, j) -> if modp(i-j, 2)=0 then
> convert(red, colorRGB)
> else
> convert(white, colorRGB)

> end if;
> # op(2..-1, p) is all the operands of p but the first
> PLOT3D( seq( seq( POLYGONS( plotgrid[j + (i-1)*(n-1)],
> coloring(i, j) ),
> i=1..m-1 ), j=1..n-1 ),
> op(2..-1, p) );
> end proc:
Here is a chess board plot of sin(x) sin(y).

> chessplot3d( sin(x)*sin(y), x=-Pi..Pi, y=-Pi..Pi,
> style=patch, axes=frame );
1
0.5
0
–0.5
–1
–3 –3
–2 –2
–1 –1
y0 0x
1 1
2 2
3 3
Note that chessplot3d works when the plotting structure from

plot3d is either a GRID or MESH output type. The latter is the type of
output that comes from parametric surfaces or from surfaces that use
alternate coordinate systems.
> chessplot3d( (4/3)^x*sin(y), x=-1..2*Pi, y=0..Pi,
> coords=spherical, style=patch,
> lightmodel=light4 );
9.10 Conclusion
In this chapter, you have seen how you can make graphics procedures
based on the commands plot and plot3d, as well as the commands found
in the plots and plottools packages. However, for ultimate control, you
must create PLOT and PLOT3D data structures directly; these are the prim-
itive specifications of all Maple plots. Inside the PLOT and PLOT3D data
structures you can specify points, curves, and polygons, as well as grids
of values and meshes of points. You have also seen how to handle plot op-
tions, create numerical plotting procedures, work with grids and meshes,
manipulate plots and animations, and apply non-standard coloring to
your graphics.
10 Input and Output
Although Maple is primarily a system and language for performing

mathematical manipulations, many situations arise where such manipu-
lations require the use of data from outside of Maple, or the production of
data in a form suitable for use by other applications. You may also need
Maple programs to request input directly from the user and/or present
output directly to the user. To meet these needs, Maple provides a com-
prehensive collection of input and output (I/O) commands. The Maple
I/O library is the term which refers to these commands as a group.
10.1 A Tutorial Example

This section illustrates some of the ways you can use the Maple I/O
library in your work. Specifically, the examples show how to write a table
of numerical data to a file, and how to read such a table from a file. The
examples refer to the following data set, given in the form of a list of lists
and assumed to represent a list of (x, y) pairs, where each x is an integer
and each y is a real number.
> A := [[0, 0],
> [1, .8427007929],
> [2, .9953222650],
> [3, .9999779095],
> [4, .9999999846],
> [5, 1.000000000]]:
In a real application, this list would have been generated by a Maple com-
mand you executed or by a Maple procedure you wrote. In this example,
the list was simply typed in as you see it above.
If you want to use some other program (like a presentation graphics
program, or perhaps a custom C program) to process data that Maple
489
490 • Chapter 10: Input and Output
has generated, then you often need to save the data to a file in a format
that the other program recognizes. Using the I/O library, you will find it
easy to write such data to a file.
> for xy in A do fprintf("myfile", "%d %e\n", xy[1], xy[2]) end do:
> fclose("myfile");
If you print the file myfile, or view it with a text editor, it looks like this:
0 0.000000e-01
1 8.427008e-01
2 9.953223e-01
3 9.999779e-01
4 1.000000e+00
5 1.000000e+00
The fprintf command wrote each pair of numbers to the file. This com-
mand takes two or more arguments, the first of which specifies the file
that Maple is to write, and the second of which specifies the format for
the data items. The remaining arguments are the actual data items that
Maple is to write.
In the example above, the file name is myfile. The first time a given
file name appears as an argument to fprintf (or any of the other output
commands described later), the command creates the file if it does not
already exist, and prepares (opens) it for writing. If the file exists, the
new version overwrites the old one. You can override this behavior (for
example, if you want to append to an already existing file) by using the
fopen command, described later.
The format string, "%d %e\n", specifies that Maple should write the
first data item as a decimal integer (%d), and the second data item in
Fortran-like scientific notation (%e). A single space should separate the
first and second data items, and a line break (\n) should follow the second
data item (to write each pair of numbers on a new line). By default, as
in our example, Maple rounds floating-point numbers to six significant
digits for output. You can specify more or fewer digits by using options
to the %e format. The section on fprintf describes these options in more
detail.
When you are finished writing to a file, you must close it. Until you
close a file, the data may or may not actually be in the file, because output
is buffered under most operating systems. The fclose command closes a
file. If you forget to close a file, Maple automatically closes it when you
exit.
For a simple case like the one presented here, writing the data to a
file by using the writedata command is easier.
> writedata("myfile2", A, [integer,float]):
The writedata command performs all the operations of opening the file,
writing the data in the specified format, an integer and a floating-point
number, and closing the file. However, writedata does not provide the
precise formatting control that you may need in some cases. For this, use
fprintf directly.
In some applications, you may want to read data from a file. For
example, some data acquisition software may supply data that you may
want to analyze. Reading data from a file is almost as easy as writing to
it.
> A := [];
A := []
> do
> xy := fscanf("myfile2", "%d %e");
> if xy = 0 then break end if;
> A := [op(A),xy];
> end do;
xy := [0, 0.]
A := [[0, 0.]]
xy := [1, .8427007929]
A := [[0, 0.], [1, .8427007929]]
xy := [2, .995322265]
A := [[0, 0.], [1, .8427007929], [2, .995322265]]
xy := [3, .9999779095]
A := [[0, 0.], [1, .8427007929], [2, .995322265],

[3, .9999779095]]
xy := [4, .9999999846]
A := [[0, 0.], [1, .8427007929], [2, .995322265],

[3, .9999779095], [4, .9999999846]]
xy := [5, 1.000000000]
A := [[0, 0.], [1, .8427007929], [2, .995322265],

[3, .9999779095], [4, .9999999846], [5, 1.000000000]]
xy := []
A := [[0, 0.], [1, .8427007929], [2, .995322265],

[3, .9999779095], [4, .9999999846], [5, 1.000000000],
[]]
xy := 0
> fclose("myfile2");
This example starts by initializing A to be the empty list. Upon entering

the loop, Maple reads a pair of numbers at a time from the file.
The fscanf command reads characters from a specified file, and parses
them according to the specified format (in this case, "%d %e", indicating a
decimal integer and a real number). It either returns a list of the resulting
values or the integer 0 to indicate that it has reached the end of the file.
10.2 File Types and Modes • 493
The first time you call fscanf with a given file name, Maple prepares
(opens) the file for reading. If it does not exist, Maple generates an error.
The second line of the loop checks if fscanf returned 0 to indicate
the end of the file, and breaks out of the loop if it has. Otherwise, Maple
appends the pair of numbers to the list of pairs in A. (The syntax A :=
[op(A),xy] tells Maple to assign to A a list consisting of the existing
elements of A, and the new element xy.)
As when you wrote to a file, you can read from a file more easily by
using the readdata command.
> A := readdata("myfile2", [integer,float]);
A := [[0, 0.], [1, .8427007929], [2, .995322265],

[3, .9999779095], [4, .9999999846], [5, 1.000000000]]
The readdata command performs all the operations of opening the file,
reading the data and parsing the specified format (an integer and a
floating-point number), and closing the file. However, readdata does not
provide the precise parsing control that you may need in some cases. For
this, use fscanf directly.
These examples illustrate some of the basic concepts of Maple’s I/O li-
brary, and you can do a great deal by using only the information presented
in this section. However, to make more effective and efficient use of the
I/O library, an understanding of a few more concepts and commands is
useful. The remainder of this chapter describes the concepts of file types,
modes, descriptors, and names, and presents a variety of commands for
performing both formatted and unformatted file I/O.
10.2 File Types and Modes

Most of the Maple I/O library commands operate on files. This chapter
uses the term file to denote not just files on a disk, but also Maple’s user
interface. In most cases, you cannot distinguish between the two from the
point of view of the I/O commands. Almost any operation that you can
perform on a real file you can perform on the user interface, if appropriate.
Buffered Files versus Unbuffered Files

The Maple I/O library can deal with two different kinds of files: buffered
(STREAM) and unbuffered (RAW). No difference exists in how Maple uses
them, but buffered files are usually faster. In buffered files, Maple collects
characters in a buffer and writes them to a file all at once when the buffer
is full or the file is closed. Raw files are useful when you wish to explicitly
take advantage of knowledge about the underlying operating system, such
as the block size on the disk. For general use, you should use buffered files,
and they are used by default by most of the I/O library commands.
Commands that provide information about I/O status use the identi-
fiers STREAM and RAW to indicate buffered and unbuffered files, respectively.
Text Files versus Binary Files

Many operating systems, including DOS/Windows and the Macintosh
operating system, distinguish between files containing sequences of char-
acters (text files ) and files containing sequences of bytes (binary files ).
The distinction lies primarily in the treatment of the newline character.
Other distinctions may exist on some platforms, but they are not visible
when using the Maple I/O library.
Within Maple, the newline character, which represents the concept of
ending one line and beginning a new one, is a single character (although
you can type it as the two characters “\n” within Maple strings). The
internal representation of this character is the byte whose value is 10, the
ASCII linefeed character. Many operating systems, however, represent the
concept of newline within a file using a different character, or a sequence
of two characters. For example, DOS/Windows represents a newline with
two consecutive bytes whose values are 13 and 10 (carriage return and
line feed). The Macintosh represents a newline with the single byte with
value 13 (carriage return).
The Maple I/O library can deal with files as either text files or bi-
nary files. When Maple writes to a text file, any newline characters that
it writes to the file are translated into the appropriate character or char-
acter sequence that the underlying operating system uses. When Maple
reads this character or character sequence from a file, it translates back
into the single newline character. When Maple writes to a binary file, no
translation takes place; it reads newline characters and writes them as
the single byte with value 10.
When running Maple under the UNIX operating system or one of its
many variants, Maple makes no distinction between text and binary files.
It treats both in the same way, and no translation takes place.
Commands which can specify or query whether a file is a text file or
a binary file use the identifiers TEXT and BINARY, respectively.
10.3 File Descriptors versus File Names • 495
Read Mode versus Write Mode

At any given time, a file may be open either for reading or for writing.
You cannot write to a file that is open only for reading, but you can write
to and read from a file that is open for writing. If you attempt, using
the Maple I/O library, to write to a file which is open only for reading,
Maple closes and reopens the file for writing. If the user does not have
the necessary permissions to write to the file (if the file is read-only, or
resides on a read-only file system), errors occur at that point.
Commands where you can specify or query whether a file is open for
reading or writing use the identifiers READ and WRITE, respectively.
The default and terminal Files

The Maple I/O library treats the Maple user interface as a file. The iden-
tifiers default and terminal refer to this file. The default identifier
refers to the current input stream, the one from which Maple reads and
processes commands. The terminal identifier refers to the top-level input
stream, the one which was the current input stream when you first started
Maple.
When Maple is run interactively, default and terminal are equiv-
alent. Only when reading commands from a source file using the read
statement does a distinction arise. In that case, default refers to the file
being read; whereas, terminal refers to the session. Under UNIX, if input
is redirected from a file or pipe, terminal refers to that file or pipe.
Note that only the symbols default and terminal are special;
the strings "default" and "terminal" simply refer to files with those
names.
10.3 File Descriptors versus File Names

The commands of the Maple I/O library refer to files in one of two ways:
by name or by descriptor.
Referring to a file by name is the simpler of the two methods. The
first time Maple performs an operation on the file, it opens the file, either
in READ mode or in WRITE mode and as a TEXT file or a BINARY file, as
appropriate to the operation that it is performing. The primary advantage
of referring to files by name is simplicity. However, you will experience a
slight performance penalty for using this method, especially if performing
many small operations on a file (such as writing individual characters).
Referring to a file by descriptor is only slightly more complex and

is a familiar concept to those who have programmed in more traditional
environments. A descriptor simply identifies a file after you have opened
it. Use the name of the file once to open it and create a descriptor. When
you subsequently manipulate the file, use the descriptor instead of the file
name. An example in section 10.4 illustrates the use of a file descriptor.
The advantages of the descriptor method include more flexibility when
opening the file (you can specify whether the file is TEXT or BINARY, and
if Maple should open the file in READ mode or in WRITE mode), improved
performance when performing many operations on a file, and the ability
to work with unbuffered files. The disadvantage is a slight increase in the
amount of programming that you must do.
Which approach is best depends on the task at hand. You can perform
simple file I/O tasks most easily by using names; whereas, more complex
tasks can benefit from the use of descriptors.
In subsequent sections, the term fileIdentifier refers to either a file
name or a file descriptor.
10.4 File Manipulation Commands

Opening and Closing Files
Before you can read from or write to a file, you must open it. When refer-
ring to files by name, this happens automatically with the first attempt at
any operation on the file. When you use descriptors, however, you must
explicitly open the file first in order to create the descriptor.
The two commands for opening files are fopen and open. The fopen
command opens buffered (STREAM) files; whereas, the open command
opens unbuffered (RAW) files.
Use the fopen command as follows.
fopen( fileName, accessMode, fileType )
The fileName specifies the name of the file to open. This name is specified
as a string, and follows the conventions that the underlying operating
system uses. The accessMode must be one of READ, WRITE, or APPEND,
indicating whether you should initially open the file for reading, writing,
or appending. The optional fileType is either TEXT or BINARY.
If you try to open the file for reading and it does not exist, fopen
generates an error.
10.4 File Manipulation Commands • 497
If you try to open the file for writing and it does not exist, Maple
first creates it. If it does exist and you specify WRITE, Maple truncates the
file to zero length; if you specified APPEND, subsequent calls to commands
that write to the file append to it.
Call the open command as follows.
open( fileName, accessMode )
The arguments to open are the same as those to fopen, except that you
cannot specify a fileType (TEXT or BINARY). Maple opens an unbuffered
file with type BINARY.
Both fopen and open return a file descriptor. Use this descriptor to
refer to the file for subsequent operations. You can still use the file’s name,
if you desire.
When you have finished with a file, you should tell Maple to close
it. This ensures that Maple actually writes all information to the disk. It
also frees up resources of the underlying operating system, which often
imposes a limit on the number of files that you can open simultaneously.
Close files by using the fclose or close commands. These two com-
mands are equivalent, and you can call them as follows.
fclose( fileIdentifier )
close( fileIdentifier )
The fileIdentifier is the name or descriptor of the file you wish to close.
Once you close a file, any descriptors referring to the file are no longer
valid.
> f := fopen("testFile.txt",WRITE):
> writeline(f,"This is a test"):
> fclose(f);
> writeline(f,"This is another test"):
Error, (in fprintf) file descriptor not in use
When you exit Maple or issue a restart command, Maple automatically

closes any open files, whether you opened them explicitly by using fopen
or open, or implicitly through a file I/O command.
Position Determination and Adjustment

Associated with each open file is the concept of its current position. This
is the location within the file to which a subsequent write takes place,
or from which a subsequent read takes place. Any reading or writing
operation advances the position by the number of bytes read or written.
You can determine the current position within a file by using the
filepos command. Use this command in the following manner.
filepos( fileIdentifier, position )
The fileIdentifier is the name or descriptor of the file whose position

you wish to determine or adjust. If you give a file name, and that file is
not yet open, Maple opens it in READ mode with type BINARY.
The position is optional. If you do not specify the position, Maple
returns the current position. If you supply the position, Maple sets the
current position to your specifications and returns the resulting position.
In that case, the returned position is the same as the specified position
unless the file is shorter than the specified position, in which case the
returned position is that of the end of the file (that is, its length). You
can specify the position either as an integer, or as the name infinity,
which specifies the end of the file.
The following command returns the length of the file myfile.txt.
> filepos("myfile.txt", infinity);
36
Detecting the End of a File

The feof command determines whether you have reached the end of a file.
Only use the feof command on files that you have opened as STREAMs,
either implicitly or explicitly via fopen. Call feof in the following manner.
feof( fileIdentifier )
The fileIdentifier is the name or descriptor of the file that you wish to
query. If you give a file name, and that file is not yet open, Maple opens
it in READ mode with type BINARY.
The feof command returns true if and only if you have reached the
end of the file during the most recent readline, readbytes, or fscanf
operation. Otherwise, feof returns false. This means that if 20 bytes
remain in a file and you use readbytes to read these 20 bytes, then feof
10.4 File Manipulation Commands • 499
still returns false. You only encounter the end-of-file when you attempt
another read.
Determining File Status

The iostatus command returns detailed information about all the files
currently in use. Call the iostatus command with the following syntax.
iostatus()
The iostatus command returns a list. The list contains the following
elements:
iostatus()[1] The number of files that the Maple I/O library is cur-
rently using.
iostatus()[2] The number of active nested read commands (when read
reads a file which itself contains a read statement).
iostatus()[3] The upper bound on iostatus()[1] + iostatus()[2]
that the underlying operating system imposes.
iostatus()[n ] for n > 3. A list giving information about a file currently
in use by the Maple I/O library.
When n > 3, the lists that iostatus()[n ] return each contain the
following elements:
iostatus()[n ][1] The file descriptor which fopen or open returned.

iostatus()[n ][2] The name of the file.
iostatus()[n ][3] The kind of file (STREAM, RAW, or DIRECT).
iostatus()[n ][4] The file pointer or file descriptor that the underlying
operating system uses. The pointer is in the form FP=integer or
FD=integer .
iostatus()[n ][5] The file mode (READ or WRITE).
iostatus()[n ][6] The file type (TEXT or BINARY).
Removing Files
Many files are solely for temporary use. Often, you no longer need such
files when you complete your Maple session and thus, you should remove
them. Use the fremove command to do this.
fremove( fileIdentifier )
The fileIdentifier is the name or descriptor of the file you wish to remove.
If the file is currently open, Maple closes it before removing it. If the file
does not exist, Maple generates an error.
To remove a file without knowing whether it exists or not, use a
try/catch statement to trap the error that fremove might create.
> try fremove("myfile.txt") catch: end try:
10.5 Input Commands

Reading Text Lines from a File
The readline command reads a single line of text from a file. Characters
are read up to and including a new line. The readline command then
discards the new line character, and returns the line of characters as a
Maple string. If readline cannot read a whole line from the file, then it
returns 0 instead of a string.
Call the readline command by using the following syntax.
readline( fileIdentifier )
The fileIdentifier is the name or descriptor of the file that you wish to
read. For compatibility with earlier versions of Maple, you can omit the
fileIdentifier, in which case Maple uses default. Thus readline() and
readline(default) are equivalent.
If you use -1 as the fileIdentifier, Maple also takes input from the
default stream, except that Maple’s command line preprocessor runs
on all input lines. This means that lines beginning with “!” pass to the
operating system instead of returning through readline, and that lines
beginning with “?” translate to calls to the help command.
If you call readline with a file name, and that file is not yet open,
Maple opens it in READ mode as a TEXT file. If readline returns 0 (indi-
cating the end of the file) when called with a file name, it automatically
closes the file.
The following example defines a Maple procedure which reads a text
file and displays it on the default output stream.
10.5 Input Commands • 501
> ShowFile := proc( fileName::string )

> local line;
> do
> line := readline(fileName);
> if line = 0 then break end if;
> printf("%s\n",line);
> end do;
> end proc:
Reading Arbitrary Bytes from a File

The readbytes command reads one or more individual characters or bytes
from a file, returning either a string or a list of integers. If there are
no more characters remaining in the file when you call readbytes, the
command returns 0, indicating that you have reached the end of the file.
Use the following syntax to call the readbytes command.
readbytes( fileIdentifier, length, TEXT )
The fileIdentifier is the name or descriptor of the file that Maple is to

read. The length, which you may omit, specifies how many bytes Maple
needs to read. If you omit length, Maple reads one byte. The optional
parameter TEXT indicates that the result is to be returned as a string
rather than a list of integers.
You can specify the length as infinity, in which case Maple reads
the remainder of the file.
If you specify TEXT when a byte with value 0 resides among the bytes
being read, the resulting string contains only those characters preceding
the 0 byte.
If you call readbytes with a file name, and that file is not yet open,
Maple opens it in READ mode. If you specify TEXT, Maple opens it as
a TEXT file; otherwise, Maple opens it as a BINARY file. If readbytes
returns 0 (indicating the end of the file) when you call it with a file name,
it automatically closes the file.
The following example defines a Maple procedure which reads an en-
tire file, by using readbytes, and copies it to a new file.
> CopyFile := proc( sourceFile::string, destFile::string )
> writebytes(destFile, readbytes(sourceFile, infinity))
> end proc:
Formatted Input
The fscanf and scanf commands read from a file, parsing numbers and
substrings according to a specified format. The commands return a list of
these parsed objects. If no more characters remain in the file when you
call fscanf or scanf, they return 0 instead of a list, indicating that it
has reached the end of the file.
Call the fscanf and scanf commands as follows.
fscanf( fileIdentifier, format )

scanf( format )
The fileIdentifier is the name or descriptor of the file you wish to read.
A call to scanf is equivalent to a call to fscanf with default as the
fileIdentifier.
If you call fscanf with a file name, and that file is not yet open, Maple
opens it in READ mode as a TEXT file. If fscanf returns 0 (indicating the
end of the file) when you call it with a file name, Maple automatically
closes the file.
The format specifies how Maple is to parse the input. The format is
a Maple string made up of a sequence of conversion specifications, that
may be separated by other characters. Each conversion specification has
the following format, where the brackets indicate optional components.
%[*][width ][modifiers ]code
The “%” symbol begins the conversion specification. The optional “*”
indicates that Maple is to scan the object, but not return it as part of the
result. It is discarded.
The optional width indicates the maximum number of characters to
scan for this object. You can use this to scan one larger object as two
smaller objects.
The optional modifiers are used to indicate the type of the value to
be returned:
l or L The letters l and L are supported for compatibility with the C

scanf function, and indicate that a “long int” or “long long” is to be
returned. In Maple, these flags have no effect.
z c or Z One of these flags can precede any of the numeric formats,

namely d, o, x, e, f, or g, indicating that a complex value is to
be scanned. Each of the real and imaginary parts of the complex
value are scanned by using the specified format with the z or Z elided.
The z format scans the real part, followed by a the character specified
by c, followed by the imaginary part. The Z format scans the real
part, followed by a “+” or “-” sign, followed by the imaginary part,
followed by a string of character corresponding to the current setting
of interface(imaginaryunit).
The z and Z options can result in one of the few conditions in which
scanf will raise an exception. If scanf is part way through scanning a
complex value (for example, the real part has already been successfully
scanned), and is unable to finish scanning the remainder (for exam-
ple, there is no imaginary part after the real part), scanf will raise
an exception of the form " ‘%1‘ expected in input for complex
format ", where %1 will be replaced by the expected character (for
example, a comma).
The code indicates the type of object you wish to scan. It determines
the type of object that Maple returns in the resulting list. The code can
be one of the following:
d The next non-blank characters in the input must make up a signed or

unsigned decimal integer. A Maple integer is returned.
o The next non-blank characters in the input must make up an unsigned
octal (base 8) integer. The integer is converted to decimal and returned
as a Maple integer.
x The next non-blank characters in the input must make up an unsigned
hexadecimal (base 16) integer. The letters A through F (either capital
or lower case) represent the digits corresponding to the decimal num-
bers 10 through 15. The integer is converted to decimal and returned
as a Maple integer.
y The next non-blank characters in the input must make up an IEEE hex-
dump format floating-point value. This value must consist of sixteen
hexadecimal characters. The value is converted to and returned as a
Maple float.
e, f, or g The next non-blank characters in the input must make up
a signed or unsigned decimal number, possibly including a decimal
point, and possibly followed by E or e, an optional sign, and a decimal
integer indicating a power of ten. The number is returned as a Maple
floating-point value.
In addition to numeric values, the e, f, and g formats also recognize
the special values “inf” and “NaN”. If an i or N is encountered when
scanf is looking for the first digit of a number, it assumes that one
of these special values has been found, and proceeds to look for the
subsequent nf or aN. If the rest of the special value is not found, an
exception is raised.
he, hf, or hg These are special formats for reading one or two-dimensional
numeric arrays. In general, such arrays should be read by using the
more sophisticated functionality provided by the {} format, but the
he, hf, and hg formats are provided for backward compatibility with
hfarrays, and provide some intelligence in automatically dealing with
a variety of textual layouts of such arrays.
The following input must make up a one or two-dimensional array
of floating-point (or integer) values. Characters encountered during
scanning are categorized into three classes: numeric, separator, and
terminator. All the characters that can appear within a number (the
digits, decimal point, signs, E, e, D, and d) are numeric. Any white
space, commas, or square brackets are separators. A square bracket
not immediately followed by a comma, and any other character, are
terminators. If a backslash is encountered, it and the following char-
acter are ignored completely.
The dimensions of the array are determined by the number of lines
read, and the number of values in the first line. If either of these is 1,
or if the number of rows multiplied by the number of columns does
not equal the total number of values read, a one-dimensional array is
produced.
The definition of “the first line” is “everything read up to the first
line break that does not immediately follow a comma or a backslash,
or up to the first closing square bracket that is immediately followed
by a comma”.
The kinds of things that can be read this way include anything that
was written by the corresponding printf, “typical” tables of numbers,
and lprinted or saved (in text form) Maple lists and lists of lists.
The result is returned as an hfarray of one or two dimensions.
hx The following input must make up a one or two dimensional array of

floating-point numbers in IEEE hex-dump format (16 characters per
number). The dimensions of the array are determined as described for
the "%he", "%hf", and "%hg" formats above.
s The next non-blank characters, up to but not including the following

blank characters (or the end of the string), are returned as a Maple
string.
a Maple collects and parses the next non-blank characters, up to but not
including the following blank characters (or the end of the string). An
unevaluated Maple expression is returned.
m The next characters must be a Maple expression encoded in Maple’s .m

file format. Maple reads enough characters to parse a single complete
expression; it ignores the width specification. The Maple expression
is returned.
c This code returns the next character (blank or otherwise) as a Maple

string. If a width is specified, that many characters (blank or other-
wise) are returned as a single Maple string.
[. . .] The characters between “[” and “]” become a list of characters that
are acceptable as a character string. Maple scans characters from the
input until it encounters one that is not in the list. The scanned
characters are then returned as a Maple string.
If the list begins with a “^” character, the list represents all those
characters not in the list.
If a “]” is to appear in the list, it must immediately follow the opening
“[” or the “^” if one exists.
You can use a “-” in the list to represent a range of characters. For
example, “A-Z” represents any capital letter. If a “-” is to appear as
a character instead of representing a range, it must appear either at
the beginning or the end of the list.
{. . .}wft The characters between the left brace, "{", and the right brace,
"}", are options for scanning Arrays, Matrices, or Vectors (i.e., the
various classes of rtable). The optional w is an integer specifying the
width to scan for each element (any width specified before the opening
"{" would apply to the entire rtable being scanned, but is ignored).
The character f specifies the format code, and can be any format code
supported by scanf except [...] or {...}. The character t, which
must be one of a, m, c, or r, specifies the type of object to be created
(Array, Matrix, Vector[column], or Vector[row] respectively).
Details on rtable formatting options are described in the help page
?rtable_scanf.
M The next sequence of characters must correspond to a well formed XML

element. The result is a Maple function call whose name is constructed
from the XML element, whose arguments are either function calls for
the child elements or the CDATA as strings, and whose attributes are
equations defining the XML attributes of the object.
n The total number of characters scanned up to the “%n” is returned as

a Maple integer.
Maple skips non-blank characters in the format but not within a con-
version specification (where they must match the corresponding charac-
ters in the input). It ignores white space in the format, except that a space
immediately preceding a “%c” specification causes the “%c” specification
to skip any blanks in the input.
If it does not successfully scan any objects, Maple returns an empty
list.
The fscanf and scanf commands use the underlying implementation
that the hardware vendor provides for the “%o” and “%x” formats. As a
result, input of octal and hexadecimal integers is subject to the restrictions
of the machine architecture.
The following example defines a Maple procedure that reads a file
containing a table of numbers, in which each row can have a different
width. The first number in each row is an integer specifying how many
real numbers follow it in that row, and commas separate all the numbers
in each row.
> ReadRows := proc( fileName::string )
> local A, count, row, num;
> A := [];
> do
> # Determine how many numbers are in this row.
> count := fscanf(fileName,"%d");
> if count = 0 then break end if;
> if count = [] then
> error "integer expected in file"
> end if;
> count := count[1];
>
> # Read the numbers in the row.
> row := [];
> while count > 0 do
> num := fscanf(fileName,",%e");
> if num = 0 then
> error "unexpected end of file"
> end if;
> if num = [] then
> error "number expected in file"
> end if;

> row := [op(row),num[1]];
> count := count - 1
> end do;
>
> # Append the row to the accumulated result.
> A := [op(A),row]
> end do;
> A
> end proc:
Reading Maple Statements

The readstat command reads a single Maple statement from the
terminal input stream. Maple parses and evaluates the statement, and
returns the result. Call the readstat command as follows.
readstat( prompt, ditto3, ditto2, ditto1 )
The prompt argument specifies the prompt that readstat is to use. If you
omit the prompt argument, Maple uses a blank prompt. You can either
supply or omit all of the three arguments ditto3, ditto2, and ditto1. If you
supply them, they specify the values which Maple uses for %%%, %%, and
% in the statement that readstat reads. Specify each of these arguments
as a Maple list containing the actual value for substitution. This allows
for values that are expression sequences. For example, if % is to have the
value 2*n+3 and %% is to have the value a,b, then use [2*n+3] for ditto1
and [a,b] for ditto2.
The response to readstat must be a single Maple expression. The
expression may span more than one input line, but readstat does not
permit multiple expressions on one line. If the input contains a syntax
error, readstat returns an error describing the nature of the error, and
its position in the input.
The following example shows a trivial use of readstat within a pro-
cedure.
> InteractiveDiff := proc( )
> local a, b;
> a := readstat("Please enter an expression: ");
> b := readstat("Differentiate with respect to: ");
> printf("The derivative of %a with respect to %a is %a\n",
> a,b,diff(a,b))
> end proc:
Reading Tabular Data

The readdata command reads TEXT files containing tables of data. For
simple tables, you will find this more convenient than writing your own
procedure by using a loop and the fscanf command.
Use the following syntax to call the readdata command.
readdata( fileIdentifier, dataType, numColumns )
The fileIdentifier is the name or descriptor of the file from which readdata
reads the data. The dataType must be one of integer or float, or you
can omit it, in which case readdata assumes float. If readdata needs
to read more than one column, you can specify the type of each column
by using a list of data types.
The numColumns argument indicates how many columns of data are
to be read from the file. If you omit numColumns, readdata reads the
number of columns specified by the number of data types that you spec-
ified (one column if you did not specify any dataType).
If Maple reads only one column, readdata returns a list of the values
read. If Maple reads more than one column, readdata returns a list of
lists, each sublist of which contains the data read from one line of the file.
If you call readdata with a file name, and that file is not yet open,
Maple opens it in READ mode as a TEXT file. Furthermore, if you call
readdata with a file name, it automatically closes the file when readdata
returns.
The following two examples are equivalent uses of readdata to read
a table of (x, y, z)-triples of real numbers from a file.
> A1 := readdata("my_xyz_file.text",3);
A1 := [[1.5, 2.2, 3.4], [2.7, 3.4, 5.6], [1.8, 3.1, 6.7]]
> A2 := readdata("my_xyz_file.text",[float,float,float]);
A2 := [[1.5, 2.2, 3.4], [2.7, 3.4, 5.6], [1.8, 3.1, 6.7]]

10.6 Output Commands • 509
10.6 Output Commands

Configuring Output Parameters by using the interface
Command
The interface command is not an output command. It is a mechanism
to provide communication between Maple and the user interface (called
Iris). You can use it to configure several parameters affecting the output
produced by various commands within Maple.
To set a parameter, call the interface command as follows.
interface( variable = expression )
The variable argument specifies which parameter you wish to change, and
the expression argument specifies the value that the parameter is to have.
See the following sections or ?interface for which parameters you can
set. You may set multiple parameters by giving several arguments of the
form variable = expression , with commas separating them.
To query the setting of a parameter, use the following syntax.
interface( variable )
The variable argument specifies the parameter to query. The interface

command returns the current setting of the parameter. You can query
only one parameter at a time.
One-Dimensional Expression Output

The lprint command prints Maple expressions in a one-dimensional no-
tation very similar to the format Maple uses for input. In most cases,
you could return this output to Maple as input, and the same expression
would result. The single exception is if the expression contains Maple
names containing non-alphanumeric characters.
The lprint command is called as follows.
lprint( expressionSequence )
The expressionSequence consists of one or more Maple expressions. Each

of the expressions is printed in turn, with three spaces separating each of
them. Maple prints a new line character after the last expression.
Maple always sends the output that lprint produces to the default
output stream. You can use the writeto and appendto commands, de-
scribed later, to temporarily redirect the default output stream to a
file.
The interface parameter screenwidth affects the output of lprint.

If possible, Maple wraps the output between tokens. If a single token is
too long to display (for example, a very long name or number), Maple
breaks it across lines, and prints a backslash, “\”, before each such break.
The following example illustrates lprint output, and how screenwidth
affects it.
> lprint(expand((x+y)^5));
x^5+5*x^4*y+10*x^3*y^2+10*x^2*y^3+5*x*y^4+y^5
> interface(screenwidth=30);
> lprint(expand((x+y)^5));
x^5+5*x^4*y+10*x^3*y^2+10*x^2
*y^3+5*x*y^4+y^5
Two-Dimensional Expression Output

The print command prints Maple expressions in a two-dimensional no-
tation. Depending on the version of Maple you are running, and the user
interface you are using, this notation is either the standard math notation
that appears in text books and other typeset mathematical documents, or
an approximation of standard math notation using only text characters.
Use the following method to call the print command.
print( expressionSequence )
The expressionSequence consists of one or more Maple expressions. Maple

prints each expression, in turn, with commas separating them.
The output produced by print is always sent to the default output
stream. You can use the writeto and appendto commands, described
later, to temporarily redirect the default output stream to a file.
Several interface parameters affect the output of print. They are
set using the syntax
interface( parameter = value )
They include:
prettyprint This selects the type of output that print is to produce.

If you set prettyprint to 0, print produces the same output as
lprint. If you set prettyprint to 1, print produces a simulated
math notation using only text characters. If you set prettyprint to
2, and the version of Maple you are running is capable of it, print
produces output using standard math notation. The default setting
of prettyprint is 2.
indentamount This specifies the number of spaces that Maple uses to in-
dent the continuation of expressions that are too large to fit on a single
line. This parameter takes effect only when you set prettyprint (see
above) to 1, and/or when Maple is printing procedures. The default
setting of indentamount is 4.
labelling or labeling You can set this to true or false, indicating

whether or not Maple should use labels to represent common subex-
pressions in large expressions. The use of labels can make large expres-
sions easier to read and comprehend. The default setting of labelling
is true.
labelwidth This indicates the size that a subexpression must have in

order for Maple to consider it for labeling (if labelling is true). The
size is the approximate width, in characters, of the expression when
printed with print and prettyprint = 1.
screenwidth This indicates the width of the screen in characters. When

prettyprint is 0 or 1, Maple uses this width to decide when to wrap
long expressions. When prettyprint is 2, the user interface must deal
with pixels instead of characters, and determines the width automat-
ically.
verboseproc Use this parameter when printing Maple procedures. If you

set verboseproc to 1, Maple only prints user defined procedures;
Maple shows system defined procedures in a simplified form giving
only the arguments, and possibly a brief description of the procedure.
If you set verboseproc to 2, Maple prints all procedures in full. Set-
ting verboseproc to 3 prints all procedures in full, and prints the
contents of a procedure’s remember table in the form of Maple com-
ments after the procedure.
When you use Maple interactively, it automatically displays each com-

puted result. The format of this display is the same as if you used the
print command. Therefore, all the interface parameters that affect the
print command also affect the display of results.
The following example illustrates print output, and how prettyprint,
indentamount, and screenwidth affect it.
> print(expand((x+y)^6));
x6 + 6 x5 y + 15 x4 y 2 + 20 x3 y 3 + 15 x2 y 4 + 6 x y 5 + y 6
> interface(prettyprint=1);
6 5 4 2 3 3 2 4 5
x + 6 x y + 15 x y + 20 x y + 15 x y + 6 x y
6
+ y
> interface(screenwidth=35);
6 5 4 2 3 3
x + 6 x y + 15 x y + 20 x y
2 4 5 6
+ 15 x y + 6 x y + y
> interface(indentamount=1);
6 5 4 2 3 3
x + 6 x y + 15 x y + 20 x y
2 4 5 6
+ 15 x y + 6 x y + y
> interface(prettyprint=0);
x^6+6*x^5*y+15*x^4*y^2+20*x^3*y^3+
15*x^2*y^4+6*x*y^5+y^6
Writing Maple Strings to a File

The writeline command writes one or more Maple strings to a file. Each
string appears on a separate line. Call the writeline command as follows.
writeline( fileIdentifier, stringSequence )
The fileIdentifier is the name or description of the file to which you want
to write, and stringSequence is the sequence of strings that writeline
should write. If you omit the stringSequence, then writeline writes a
blank line to the file.
Writing Arbitrary Bytes to a File

The writebytes command writes one or more individual characters or
bytes to a file. You may specify the bytes either as a string or a list of
integers.
The following syntax calls the writebytes command.
writebytes( fileIdentifier, bytes )
The fileIdentifier is the name or descriptor of the file to which writebytes

is writing. The bytes argument specifies the bytes that writebytes writes.
This can be either a string or a list of integers. If you call writebytes
with a file name, and that file is not yet open, Maple opens it in WRITE
mode. If you specify the bytes as a string, Maple opens the file as a TEXT
file; if you specify the bytes as a list of integers, Maple opens the file as a
BINARY file.
The following example defines a Maple procedure which reads an en-
tire file and copies it to a new file using writebytes.
> CopyFile := proc( sourceFile::string, destFile::string )
> writebytes(destFile, readbytes(sourceFile, infinity));
> end proc:
Formatted Output
The fprintf and printf commands write objects to a file, using a spec-
ified format.
Call the fprintf and printf commands as follows.
fprintf( fileIdentifier, format, expressionSequence )

printf( format, expressionSequence )
The fileIdentifier is the name or descriptor of the file to which Maple is to

write. A call to printf is equivalent to a call to fprintf with default
as the fileIdentifier. If you call fprintf with a file name, and that file is
not yet open, Maples opens it in WRITE mode as a TEXT file.
The format specifies how Maple is to write the elements of the expres-
sionSequence. This Maple string is made up of a sequence of formatting
specifications, possibly separated by other characters. Each format spec-
ification has the following syntax, where the brackets indicate optional
components.
%[flags ][width ][.precision ][modifiers ]code
The “%” symbol begins the format specification. One or more of the fol-
lowing flags can optionally follow the “%” symbol:
+ A signed numeric value is output with a leading “+” or “-” sign, as

appropriate.
- The output is left justified instead of right justified.
blank A signed numeric value is output with either a leading “-” or a

leading blank, depending on whether the value is negative or non-
negative.
0 The output is padded on the left (between the sign and the first digit)
with zeroes. If you also specify a “-”, the “0” is ignored.
{} The braces enclose a set of detailed formatting options for printing

an rtable. These are described in more detail in the help page
?rtable_printf.
The optional width indicates the minimum number of characters to

output for this field. If the formatted value has fewer characters, Maple
pads it with blanks on the left (or on the right, if you specify “-”).
The optional precision specifies the number of digits that appear after
the decimal point for floating-point formats, or the maximum field width
for string formats.
You may specify both width and/or precision as “*”, in which case
Maple takes the width and/or precision from the argument list. The width
and/or precision arguments must appear, in that order, before the argu-
ment that is being output. A negative width argument is equivalent to
the appearance of the “-” flag.
The optional modifiers are used to indicate the type of the value to
be printed:
l or L The letters l and L are supported for compatibility with the C

printf function, and indicate that a "long int" or "long long" is to
be formatted. In Maple, these flags have no effect.
z c or Z One of these flags can precede any of the numeric formats,
namely d, o, x, e, f, or g, indicating that a complex value is to be
formatted. Each of the real and imaginary parts of the complex value
are formatted using the specified format, with the z or Z elided. The
z format prints the real part, followed by a the character specified by
c, followed by the imaginary part. The Z format prints the value in
the form x+yi, where x is the real part, y is the imaginary part, and i
is the current setting of interface(imaginaryunit). If y is negative,
a "-" is output instead of a "+". If a supplied value is not complex,
it is treated as a complex value with a zero imaginary part.
The code indicates the type of object that Maple is to write. The code
can be one of the following.
d Formats the object as a signed decimal integer.

o Formats the object as an unsigned octal (base 8) integer.
x or X Formats the object as an unsigned hexadecimal (base 16) integer.
Maple represents the digits corresponding to the decimal numbers 10
through 15 by the letters “A” through “F” if you use “X”, or “a”
through “f” if you use “x”.
e or E Formats the object as a floating-point number in scientific nota-
tion. One digit will appear before the decimal point, and precision
digits will appear after the decimal point (six digits if you do not
specify a precision). This is followed by the letter “e” or “E”, and a
signed integer specifying a power of 10. The power of 10 will have a
sign and at least three digits, with leading zeroes added if necessary.
If the value being formatted is infinity, -infinity, or undefined, the

output is "Inf", "-Inf", or "NaN" respectively.
f Formats the object as a fixed-point number. The number of digits spec-

ified by the precision will appear after the decimal point.
g or G Formats the object using “d”, “e” (or “E” if you specified “G”), or
“f” format, depending on its value. If the formatted value does not
contain a decimal point, Maple uses “d” format. If the value is less
than 10−4 or greater than 10precision , Maple uses “e” (or “E”) format.
Otherwise, Maple uses “f” format.
y or Y The floating-point object is formatted in byte-order-independent

IEEE hex dump format (16 characters wide). At least 16 characters
will always be output, regardless of the specified width. The preci-
sion is ignored. The digits corresponding to the decimal numbers 10
through 15 are represented by the letters "A" through "F" if uppercase
Y was specified, or "a" through "f" if lowercase y was specified.
c Outputs the object, which must be a Maple string containing exactly

one character, as a single character.
s Outputs the object, which must be a Maple string of at least width

characters (if specified) and at most precision characters (if specified).
a or A Outputs the object, which can be any Maple object, in correct

Maple syntax. Maple outputs at least width characters (if specified)
and at most precision characters (if specified). Note: truncating a
Maple expression by specifying a precision can result in an incomplete
or syntactically incorrect Maple expression in the output.
The "%a" and "%A" formats are identical, except that "%A" will omit
any quotes that would normally appear around Maple symbols that
require them.
q or Q These are similar to "%a" or "%A", except that "%q" or "%Q" will
consume all remaining arguments and print them as an expression
sequence, with each element formatted in "%a" or "%A" format re-
spectively. No additional format specifiers can appear after "%q" or
"%Q", since there will be no arguments left to format.
m The object, which can be any Maple object, is output in Maple’s “.m”
file format. Maple outputs at least width characters (if specified),
and at most precision characters (if specified). Note: truncating a
Maple “.m” format expression by specifying a precision can result in
an incomplete or incorrect Maple expression in the output.
% A percent symbol is output verbatim.
Maple outputs characters that are in format but not within a format
specification verbatim.
All of the formats apply to Arrays (type Array), Matrices (type
Matrix), Vectors (type Vector), and hfarrays (type hfarray), all of which
are objects of type rtable.
If no rtable-specific formatting options are specified (via the {...} op-
tion, see ?rtable_printf), the %a, %A, %m, and %M format codes will print
a representation of the rtable structure itself. For example, %a would print
a Matrix, Vector, or Array call.
If no additional rtable-specific formatting options are specified for a
format code other than %a, %A, %m, and %M, or if an empty rtable option
sequence (i.e., just {}) is specified for any format code, the following
default formatting is applied:
One-dimensional objects are formatted as one long line, with the ele-
ments separated by at least one space.
Objects of N dimensions, where N > 1, are formatted as a sequence of
(N −1)-dimensional objects separated by N −2 blank lines. Therefore, two-
dimensional objects are formatted in the obvious way, three-dimensional
objects are formatted as a series of two-dimensional objects separated by
blank lines, and so on.
Any of the floating-point formats can accept integer, rational, or
floating-point objects; Maple converts the objects to floating-point val-
ues and outputs them appropriately.
The fprintf and printf commands do not automatically start a
new line at the end of the output. If you require a new line, the format
string must contain a new line character, “\n”. Output from fprintf
and printf is not subject to line wrapping at interface(screenwidth)
characters.
The “%o”, “%x”, and “%X” formats use the underlying implementation
that the hardware vendor provides. As a result, output of octal and hex-
adecimal values is subject to the restrictions of the machine architecture.
Writing Tabular Data

The writedata command writes tabular data to TEXT files. In many cases,
this is more convenient than writing your own output procedure by using
a loop and the fprintf command.
Call the writedata command in the following manner.
writedata( fileIdentifier, data, dataType, defaultProc )
The fileIdentifier is the name or descriptor of the file to which writedata

writes the data.
If you call writedata with a filename, and that file is not yet open,
Maple opens it in WRITE mode as a TEXT file. Furthermore, if you call
writedata with a file name, the file automatically closes when writedata
returns.
The data must be a vector, matrix, list, or list of lists.1 If the data
is a vector or list of values, writedata writes each value to the file on a
separate line. If the data is a matrix or a list of lists of values, writedata
writes each row or sublist to the file on a separate line, with tab characters
separating the individual values.
The dataType is optional, and specifies whether writedata is to write
the values as integers, floating-point values (the default), or strings. If you
specify integer, the values must be numeric, and writedata writes them
as integers (Maple truncates rational and floating-point values to inte-
gers). If you specify float, the values must be numeric, and writedata
writes them as floating-point values (Maple converts integer and rational
values to floating-point). If you specify string, the values must be strings.
When writing matrices or lists of lists, you can specify the dataType as
a list of data types, one corresponding to each column in the output.
The optional defaultProc argument specifies a procedure that writedata
calls if a data value does not conform to the dataType you specified
(for example, if writedata encounters a non-numeric value when the
dataType is float). Maple passes the file descriptor corresponding to
the fileIdentifier, along with the non-conforming value, as an argument to
the defaultProc. The default defaultProc simply generates the error, Bad
data found. A more useful defaultProc might be the following.
> UsefulDefaultProc := proc(f,x) fprintf(f,"%a",x) end proc:
1
For information about how to read and write rtable-based Matrices and Vectors,
see the help pages ?ImportMatrix and ?ImportVector.
This procedure is a sort of “catch-all”; it is capable of writing any kind

of value to the file.
The following example computes a 5 by 5 Hilbert matrix, and writes
its floating-point representation to a file.
> writedata("hilbertFile.txt",linalg[hilbert](5)):
Examining the file shows:

1 .5 .3333333333 .25 .2
.5 .3333333333 .25 .2 .1666666667
.3333333333 .25 .2 .1666666667 .1428571429
.25 .2 .1666666667 .1428571429 .125
.2 .1666666667 .1428571429 .125 .1111111111
Flushing a Buffered File

I/O buffering may result in a delay between when you request a write
operation and when Maple physically writes the data to the file. This is
to capitalize on the greater efficiency of performing one large write instead
of several smaller ones.
Normally, the I/O library chooses when to write to a file automatically.
In some situations, however, you may desire to ensure that the data you
write has actually made it into the file. For example, under UNIX, a
common procedure is to run a command, such as “tail -f fileName”, in
another window in order to monitor the information as Maple is writing
it. For cases such as these, the Maple I/O library provides the fflush
command.
Call the fflush command using the following syntax.
fflush( fileIdentifier )
The fileIdentifier is the name or descriptor of the file whose buffer Maple
is to flush. When you call fflush, Maple writes all information that is in
the buffer but not yet in the physical file to the file. Typically, a program
would call fflush whenever something significant is written (for example,
a complete intermediate result or a few lines of output).
Note that you do not need to use fflush; anything you write to a
file will physically be written no later than when you close the file. The
fflush command simply forces Maple to write data on demand, so that
you can monitor the progress of a file.
Redirecting the default Output Stream

The writeto and appendto commands redirect the default output
stream to a file. This means that any operations that write to the default
stream write to the file you specify instead.
You can call the writeto and appendto commands as follows.
writeto( fileName )
appendto( fileName )
The fileName argument specifies the name of the file to which Maple is
to redirect the output. If you call writeto, Maple truncates the file if it
already exists, and writes subsequent output to the file. The appendto
command appends to the end of the file if the file already exists. If the
file you specify is already open (for example, it is in use by other file I/O
operations), Maple generates an error.
The special fileName terminal (specified as a name, not a string)
causes Maple to send subsequent default output to the original default
output stream (the one that was in effect when you started Maple). The
calls writeto(terminal) and appendto(terminal) are equivalent.
Issuing a writeto or appendto call directly from the Maple prompt
is not the best choice of action. When writeto or appendto are in effect,
Maple also writes any error messages that may result from subsequent
operations to the file. Therefore, you cannot see what is happening. You
should generally use the writeto and appendto commands within proce-
dures or files of Maple commands that the read command is reading.
10.7 Conversion Commands

C or Fortran Generation
Maple provides commands to translate Maple expressions into two other
programming languages, C and Fortran. Conversion to other program-
ming languages is useful if you have used Maple’s symbolic techniques to
develop a numeric algorithm, which then may run faster as a C or Fortran
program than as a Maple procedure.
Perform a conversion to Fortran or C by using the fortran or C com-
mands in the codegen package, respectively. There are also several sup-
port commands for code generation, which can be found in the codegen
package.
Call the fortran and C commands using the following syntax.
10.7 Conversion Commands • 521
codegen[fortran]( expression, options )

codegen[C]( expression, options )
The expression can take one of the following forms:
1. A single algebraic expression: Maple generates a sequence of C or

Fortran statements to compute the value of this expression.
2. A list of expressions of the form name = expression : Maple generates

a sequence of statements to compute each expression and assigns it
to the corresponding name.
3. A named array of expressions: Maple generates a sequence of C or

Fortran statements to compute each expression and assigns it to the
corresponding element of the array.
4. A Maple procedure: Maple generates a C function or Fortran subrou-

tine.
The fortran command uses the ‘fortran/function_name‘ com-

mand when translating function names to their Fortran equivalents. This
command takes three arguments: the Maple function name, the number
of arguments, and the precision, and returns a single Fortran function
name. You can override the default translations by assigning values to
the remember table of ‘fortran/function_name‘.
> ‘fortran/function_name‘(arctan,1,double) := datan;
fortran/function_name(arctan, 1, double) := datan
> ‘fortran/function_name‘(arctan,2,single) := atan2;
fortran/function_name(arctan, 2, single) := atan2
When translating arrays, the C command re-indexes all array indices

to begin with 0, since the base of C arrays is 0. The fortran command
re-indexes arrays to begin with 1, but only when Maple is translating a
procedure.
Here Maple symbolically calculates the anti-derivative.
> f := unapply( int( 1/(1+x^4), x), x );
√
1√ x2 + x 2 + 1 1√ √
f := x → 2 ln( √ )+ 2 arctan(x 2 + 1)
8 x2 − x 2 + 1 4
1 √ √
+ 2 arctan(x 2 − 1)
4
The fortran command generates a Fortran routine.
> codegen[fortran](f, optimized);
c The options were : operatorarrow
doubleprecision function f(x)
doubleprecision x
doubleprecision t1
doubleprecision t12
doubleprecision t16
doubleprecision t2
doubleprecision t3
doubleprecision t8
t1 = sqrt(2.D0)
t2 = x**2
t3 = x*t1
t8 = log((t2+t3+1)/(t2-t3+1))
t12 = atan(t3+1)
t16 = atan(t3-1)
f = t1*t8/8+t1*t12/4+t1*t16/4
return
end
Now translate the same expression to C.

> codegen[C](f, optimized);
/* The options were : operatorarrow */
double f(x)
double x;
{
double t1;
double t12;
double t16;
double t2;
double t3;
double t8;
{
t1 = sqrt(2.0);
t2 = x*x;
t3 = x*t1;
t8 = log((t2+t3+1.0)/(t2-t3+1.0));
t12 = atan(t3+1.0);
t16 = atan(t3-1.0);
return(t1*t8/8.0+t1*t12/4.0+t1*t16/4.0);
}
}
LATEX Generation
Maple supports conversion of Maple expressions to the LATEX typesetting
language. Conversion to typesetting languages is useful when you need to
insert a result in a scientific paper.
You can perform conversion to LATEX by using the latex command.
Call the latex command as follows.
latex( expression, fileName )
The expression can be any mathematical expression. Maple-specific ex-

pressions, such as procedures, are not translatable. The fileName is op-
tional, and specifies that Maple writes the translated output to the file
you specified. If you do not specify a fileName, Maple writes the output
to the default output stream (your session).
The latex command knows how to translate most types of mathemat-
ical expressions, including integrals, limits, sums, products, and matrices.
You can expand the capabilities of latex by defining procedures with
names of the form ‘latex/functionName ‘. Such a procedure is respon-
sible for formatting calls to the function called functionName. You should
produce the output of such formatting functions with printf. latex uses
writeto to redirect the output when you specify a fileName.
The latex command does not generate the commands that LATEX
requires to put the typesetting system into mathematics mode ($...$, for
example).
The following example shows the generation of LATEX for an equation
for an integral and its value. Notice the use of Int, the inert form of int,
to prevent evaluation of the left hand side of the equation that Maple is
formatting.
> Int(1/(x^4+1),x) = int(1/(x^4+1),x);
√
1√ x2 + x 2 + 1
Z
1
dx = 2 ln( √ )
x4 + 1 8 x2 − x 2 + 1
1√ √ 1√ √
+ 2 arctan(x 2 + 1) + 2 arctan(x 2 − 1)
4 4
> latex(%);
\int \! \left( {x}^{4}+1 \right) ^{-1}{dx}=1/8
\,\sqrt {2}\ln \left( {\frac {{x}^{2}+x\sqrt
{2}+1}{{x}^{2}-x\sqrt {2}+1}} \right) +1/4\,
\sqrt {2}\arctan \left( x\sqrt {2}+1 \right) +
1/4\,\sqrt {2}\arctan \left( x\sqrt {2}-1
\right)
You can also export a whole Maple worksheet in LATEX format by

choosing Export As from the File menu, then LaTeX. For more infor-
mation, see Section 7.3 of the Learning Guide.
Conversion between Strings and Lists of Integers

The readbytes and writebytes commands described in sections 10.5
and 10.6 can work with either Maple strings or lists of integers. You
can use the convert command to convert between these two formats as
follows.
convert( string, bytes )

convert( integerList, bytes )
If you pass convert(...,bytes) a string, it returns a list of integers; if

you pass it a list of integers, it returns a string.
Due to the way strings are implemented in Maple, the character cor-
responding to the byte-value 0 cannot appear in a string. Therefore, if
integerList contains a zero, convert returns a string of only those char-
acters corresponding to the integers preceding the occurrence of 0 in the
list.
Conversion between strings and lists of integers is useful when Maple
must interpret parts of a stream of bytes as a character string, while it
must interpret other parts as individual bytes.
In the following example, Maple converts a string to a list of integers.
Then, it converts the same list, but with one entry changed to 0, back to
a string. Notice that the string is truncated at the location of the 0.
> convert("Test String",bytes);
[84, 101, 115, 116, 32, 83, 116, 114, 105, 110, 103]
> convert([84,101,115,116,0,83,116,114,105,110,103],bytes);
“Test”
Parsing Maple Expressions and Statements

The parse command converts a string of valid Maple input into the cor-
responding Maple expression. The expression is simplified, but not eval-
uated.
Use the parse command as follows.
parse( string, options )
The string argument is the string that needs parsing. It must describe a
Maple expression (or statement, see below) by using the Maple language
syntax.
You may supply one or more options to the parse command:
statement This indicates that parse is to accept statements in addition

to expressions. However, since Maple does not allow the existence of
unevaluated statements, parse does evaluate the string if you specify
statement.
nosemicolon Normally, parse supplies a terminating semicolon, “;” if

the string does not end in a semicolon or a colon, “:”. If you
specify nosemicolon, this does not happen, and Maple generates
an unexpected end of input error if the string is incomplete. The
readstat command, which uses readline and parse, makes use of
this facility to allow multi-line inputs.
If the string passed to parse contains a syntax error, parse generates

an error (which you can trap with traperror) of the following form.
incorrect syntax in parse:

errorDescription (errorLocation )
The errorDescription describes the nature of the error (for example, ‘+‘
unexpected, or unexpected end of input). The errorLocation gives the
approximate character position within the string at which Maple detected
the error.
When you call parse from the Maple prompt, Maple displays the
parsed result depending on whether the call to parse ends in a semicolon
or a colon. Whether the string passed to parse ends in a semicolon or a
colon does not matter.
> parse("a+2+b+3");
a+5+b
> parse("sin(3.0)"):
> %;
.1411200081
Formatted Conversion to and from Strings

The sprintf and sscanf commands are similar to fprintf/printf and
fscanf/scanf, except that they read from or write to Maple strings in-
stead of files.
Call the sprintf command using the following syntax.
sprintf( format, expressionSequence )
The format specifies how Maple is to format the elements of the expres-
sionSequence. This Maple string is made up of a sequence of formatting
specifications, possibly separated by other characters. See 10.6.
The sprintf command returns a string containing the formatted re-
sult.
Call the sscanf command as follows.
sscanf( sourceString, format )
The sourceString provides the input for scanning. The format specifies
how Maple is to parse the input. A sequence of conversion specifications
(and possibly other anticipated characters) make up this Maple string.
See 10.5. The sscanf command returns a list of the scanned objects, just
as fscanf and scanf do.
The following example illustrates sprintf and sscanf by converting a
floating-point number and two algebraic expressions into a floating-point
format, Maple syntax, and Maple .m format, respectively. This string is
then parsed back into the corresponding objects using sscanf.
> s := sprintf("%4.2f %a %m",evalf(Pi),sin(3),cos(3));
s := “3.14 sin(3) -%$cosG6#\“\ “$”
> sscanf(s,"%f %a %m");

10.8 A Detailed Example • 527
[3.14, sin(3), cos(3)]
10.8 A Detailed Example

This section provides an example that uses several of the I/O facilities
described in this chapter to generate a Fortran subroutine in a text file. In
this example, you can find all of the required Maple commands typed on
the command line. In general, for such a task you would write a procedure
or, at the very least, a file of Maple commands.
Suppose you wish to compute values of the function 1 − erf(x) +
exp(−x) for many points on the interval [0,2], accurate to five decimal
places. By using the numapprox package from the Maple library, you can
obtain a rational approximation for this function as follows.
> f := 1 - erf(x) + exp(-x):
> approx := numapprox[minimax](f, x=0..2, [5,5]);
approx := (1.872569003 + (−2.480756984+

(1.455338215 + (−.4103981070 + .04512734455 x) x) x
)x)/(.9362855506 + (−.2440826049+
(.2351099626 + (.00115111416 − .01091329716 x) x) x
)x)
You can now create the file and write the subroutine header to the
file.
> file := "approx.f77":
> fprintf(file, "real function f(x)\nreal x\n"):
Before you can write the actual Fortran output to the file, you must close
the file. Otherwise, the fortran command attempts to open the file in
APPEND mode, which results in an error if the file is already open.
> fclose(file):
Now you can write the actual Fortran statements to the file.
> codegen[fortran]([’f’=approx], filename=file):
Finally, you add the remainder of the Fortran subroutine syntax.

> fopen(file, APPEND):
> fprintf(file, "return\nend\n"):
> fclose(file):
If you now examine the file, it looks like this:

real function f(x)
real x
f = (0.187258E1+(-0.2480777E1+(0.1455351E1+
#(-0.4104024E0+0.4512788E-1*x)*x)*x)*x)/(0.9
#362913E0+(-0.2440864E0+(0.235111E0+(0.11504
#53E-2-0.1091373E-1*x)*x)*x)*x)
return
end
This subroutine is now ready for you to compile and link into a Fortran
program.
10.9 Notes to C Programmers

If you have experience programming in the C or C++ programming lan-
guages, many of the I/O commands described in this chapter will seem
familiar. This is not coincidental, as the Maple I/O library design pur-
posely emulates the C standard I/O library.
In general, the Maple I/O commands work in a similar manner to
their C counterparts. The differences that arise are the result of differences
between the Maple and C languages, and how you use them. For example,
in the C library, you must pass the sprintf function a buffer into which
it writes the result. In Maple, strings are objects that you can pass around
as easily as numbers, so the sprintf command simply returns a string
that is sufficiently long to hold the result. This method is both easier to
work with and less error prone, as it removes the danger of writing past
the end of a fixed length buffer.
Similarly, the fscanf, scanf, and sscanf commands return a list of
the parsed results instead of requiring you to pass references to variables.
This method is also less error prone, as it removes any danger of passing
the wrong type of variable or one of insufficient size.
Other differences include the use of a single command, filepos, to

perform the work of two C functions, ftell and fseek. You can do this
in Maple, since functions can take a variable number of arguments.
In general, if you have C or C++ programming experience, you should
have very little trouble using the Maple I/O library.
10.10 Conclusion
This chapter has revealed the details of importing and exporting data and
code into and out of Maple. Most commands discussed in this chapter are
more primitive than those commands which you are likely to use, such
as save and writeto. The aforementioned Maple commands ensure that
you are properly equipped to write specialized exporting and importing
procedures. Their basis is similar to the commands found in the popular
C programming language, although they have been extended to allow easy
printing of algebraic expressions.
Overall, this book provides an essential framework for understanding
Maple’s programming language. Each chapter is designed to teach you to
use a particular area of Maple effectively. However, a complete discussion
of Maple can not fit into a single book. The Maple help system is an
excellent resource and complements this volume. While this book teaches
fundamental concepts and provides a pedagogical introduction to topics,
the help system provides the details on each command and feature. It
explains such things as the options and syntax of Maple commands and
serves as a resource for use of the Maple interface.
Also, numerous authors have published many books about Maple.
These include not only books, such as this one, on the general use of
Maple, but also books directed toward the use of Maple in a particular
field or application. Should you wish to consult books that parallel your
own area of interest, this book will still serve as a handy reference and
guide to Maple programming.
11 Using Compiled Code in
Maple
It is possible to call routines written in C or Fortran by using Maple’s

external calling facility. Maple makes extensive use of this facility to call
NAG and other numerical libraries. You can augment the rich suite of
functions in Maple by writing your own libraries. There are also third
party external libraries available to extend the power of Maple. External
functions can be used to improve performance because compiled code is
usually much faster than interpreted code.
Maple’s external calling facility is very extensive, allowing a user to
translate most data from a Maple format to a hardware format for use
in an external program. To balance power and flexibility with ease of
use, the external calling application program interface (API) is divided
into three methods. All use the same basic interface, but each has very
different back-end support routines.
Using any of the methods, a user can call a C routine. Using Methods
1 and 3, a user can call a Fortran routine. Method 2 is not applicable
to Fortran routines (as discussed below). The structure of the procedure
that defines the external call depends on the routine, and thus whether
it was written in C or Fortran. However, the choice to use C or Fortran
routines is dependent on only the availability of routines that perform the
desired function or, in the case of custom routines, the preference of the
developer writing the external library.
Method 1: Calling External Functions In most cases, compiled func-

tions use only standard hardware types like integers, floating-point num-
bers, strings, pointers (to strings, integers, and floating-point numbers),
matrices, and vectors. In these cases, Maple can automatically translate
between its internal representation and the hardware representation. Be-
cause this method does not require the use of a compiler, it is efficient
531
532 • Chapter 11: Using Compiled Code in Maple
and easy to use. This method of directly calling the external code allows
the use of an external library without modification.
Method 2: Wrapper Generation Method 1 is limited by its use of only

standard data types. When dealing with more complicated compound
types or passing functions or records as parameters, a compiled wrap-
per is needed. Because Fortran does not use these data structures, this
method applies only to C routines. The wrapper performs the conversion
between Maple’s internal representation and the hardware representation.
Maple automatically generates and compiles wrappers, which are based
on your specifications, to interface with libraries of compiled code. This
allows you to use a greater diversity of external libraries than you can by
simply directly calling the external function. External calls that use these
wrappers require that a C compiler is installed.
Method 3: Customizing Wrappers For flexibility beyond that available

by either of the other methods, an external API is provided for users who
want to augment existing wrappers or write custom wrappers . The user
can write the wrapper in C or Fortran as desired. This powerful method
also allows direct access to Maple data structures from the wrapper.
This chapter provides a technical overview of Maple’s external calling

mechanism, and provides a context in which to explain the external calling
facilities.
11.1 Method 1: Calling External Functions

To illustrate how Maple’s external calling facility is used, start by looking
at a trivial example of a function written in C. Consider the following code
that adds two numbers and returns the result. Obviously, such a function
would never be used since Maple’s + operator exists, but working through
this example will demonstrate the steps that are required to use compiled
code in Maple.
int add( int num1, int num2 )

{
return num1+num2;
}
There are 3 basic steps required to call an external function.

11.1 Method 1: Calling External Functions • 533
Step 1: DLL Creation First, this function needs to be compiled into a

Dynamic Link Library (Windows XXX.DLL), or Shared Library (Unix
libXXX.so or Macintosh XXX.ShLib). For the rest of this chapter, the
compiled library will be referred to as a DLL. If the sources are down-
loaded from the internet or purchased, a DLL may already have been
built. Otherwise, consult the compiler’s documentation for help on how
to build a DLL. When building the DLL ensure that you export the func-
tion that Maple is intended to be able to call. In this case, the function
name is add.
This is the only step that requires the user to have knowledge of
a specific compiler. For the remaining steps, it does not matter if the
function was written in C or Fortran.
Maple expects that the external library functions are compiled by
using the _stdcall calling convention, which is the default under UNIX
but must be specified explicitly on Windows.
Step 2: Function Specification To make the appropriate argument con-

versions, Maple requires some details about the function that it is going
to call. At a minimum, Maple needs to know the following:
• name of the function,
• type of parameters the function passes and returns, and
• name of the DLL containing the function.
The specification of the parameter types are independent of the com-

piler. The same specification can be used regardless of the language that
was used to compile the DLL. The example uses the C type int. In Maple,
this is specified as integer[4]. The 4 in the square brackets denotes the
number of bytes used to represent the integer. Most C compilers use 4-
byte ints, but some older compilers may use 2-byte ints. Maple’s type
specification is flexible enough to support both types of compiler integer
sizes. See Table 11.1 for a map of the most common type relations.
Since num1 and num2 are both ints, they can be specified as the fol-
lowing in Maple.
num1::integer[4]
num2::integer[4]
The return type does not have a name so the keyword RETURN is used.
RETURN::integer[4]
Using all of this information, the complete function can be defined by

calling the Maple function define_external.
> myAdd := define_external(
> ’add’,
> ’num1’::integer[4],
> ’RETURN’::integer[4],
> ’LIB’="mylib.dll"
> );
It is important to specify the function exactly, and to ensure that the

arguments are in the correct order. Failure to do this may result in strange
behavior or program crashes when executing step 3.
Step 3: Function Invocation Executing the define_external call for

myAdd returns a Maple procedure that translates Maple types to hardware
types that can work with an external function. This procedure can be used
the same way as any other procedure in Maple.
> myAdd(1,2);
> a := 33:
> b := 22:
> myAdd(a,b);
55
> r:= myAdd(a,11);
r := 44
Procedures generated in this manner contain run-time information

and thus cannot be saved. The define_external command must be reis-
sued after exiting or restarting Maple.
The following subsections provide additional information for Step 2,

the function specification.
External Definition
The define_external function constructs and returns another function
which can be used to make the actual call. The define_external function
is called as follows.
define_external( functionName, arg1::type1, ...,

argN::typeN, options, ‘LIB‘=dllName )
define_external( functionName, ‘MAPLE‘,
options, ‘LIB‘=dllName )
The functionName parameter specifies the name of the actual external

function to be called. This name can be specified as a Maple string or
name.
The parameters arg1 through argN describe the arguments of the
function to be called. These should be specified in the order they ap-
pear in the documentation or source code for the external function,
without regard to issues such as actual passing order (left to right ver-
sus right to left). The intent is that the Maple procedure returned by
define_external will have the same calling sequence as the actual ex-
ternal function when used in the language for which it was written. The
only exception is that one argument may be given the name RETURN. This
specifies the type returned by the function rather than a parameter passed
to the function. For more information about how each argi is specified,
see the section on type specification.
The options are used to specify argument passing conventions, li-
braries, calling methods, etc. See the appropriate sections of this chapter
for details.
If instead of the arg parameters, the single word MAPLE is specified,
the external function is assumed to accept the raw Maple data struc-
tures passed without conversion. This assumes that the wrapper has been
manually generated and compiled into a shared library. Various support
functions for writing such external functions are described in Section 11.3.
Using MAPLE instead of specifying arguments is the basis of method 3.
The location of the DLL containing the function to be called is spec-
ified by using the LIB option to define_external. The dllName is a
string that specifies the filename of the library in which the function is to
be found. The format of this name is highly system dependent. Likewise,
whether a full pathname is required depends on the system. In general,
the name should be in the same format as would be specified to a compiler
on the same system.
Type Specification
Step two of the introductory example indicated how to specify types using
Maple notation. Maple uses its own notation to provide a generic well-
defined interface for calling compiled code in any language.
The format of each arg parameter is as follows.
argumentIdentifier :: dataDescriptor
The return value description is also described by using a data de-

scriptor, with the name RETURN as the argumentIdentifier. If the function
returns no value, no RETURN parameter is specified. Also, if no parameters
are passed, no argument identifiers are required.
Scalar Data Formats

External libraries generally deal with scalar data supported directly by the
underlying machine. All array, string, and structured formats are built up
from these. The data descriptors used to represent scalar formats usually
contain a type name and size. The size represents the number of bytes
needed to represent the given hardware type. Table 11.1 lists the basic
type translations for standard C and Fortran compilers.
Structured Data Formats

In addition to the basic types listed in Table 11.1, Maple also recognizes
some compound types that can be derived from the basic types, such as
arrays and pointers. These compound types are listed in Table 11.2.
Character String Data Formats Strings are similar to both scalar and
array data. A string in C is an array of characters, but it is often manip-
ulated as if it were an object. A string in Maple is an atomic object, but
it can be manipulated as if it were an array of characters.
Parameter n in string[n] indicates that the called function is expecting
a fixed size string. Otherwise, a pointer to a character buffer (char*) will
be used.
Strings are implicitly passed by reference (only a pointer to the string
is passed), but any changes made to the string are not copied back to
Maple unless the string is declared with a size.
Table 11.1 Basic Type Translations

Maple Data C Type Fortran Type
Descriptor
integer[1] char BYTE
integer[2] short INTEGER*2
integer[4] int INTEGER
long1 INTEGER*4
integer[8] long1 INTEGER*8
long long INTEGER*8
float[4] float REAL
REAL*4
float[8] double DOUBLE PRECISION
REAL*8
char[1] char CHARACTER
boolean[1] char LOGICAL*1
boolean[2] short LOGICAL*2
boolean[4] int LOGICAL
long LOGICAL*4
boolean[8] long LOGICAL*8
long long LOGICAL*8
1 Type long is typically (but not necessarily) 4-bytes on
32-bit machines and 8-bytes on 64-bit machines. Use the

sizeof operator or consult your compiler manual to
verify sizeof(long).
Table 11.2 Compound Types

Maple Data Descriptor C Type Fortran Type
ARRAY(datatype=typename, char BYTE
order=..., etc. )
string[n] char x[n] CHARACTER*2
complex[4] struct COMPLEX
{ float r, i; } COMPLEX*8
complex[8] struct DOUBLE
{ double r, i; } COMPLEX
COMPLEX*16
REF(typename) TYPENAME* NA
Array Data Formats An array of data is a homogeneous, n-rectangular

structure matching the Maple rtable formats. Any datatype that is ac-
cepted by Maple’s Array, Matrix, or Vector constructor will be accepted.
The options are used to specify array conventions. They are the same
optional arguments that can be passed to the Array constructor in Maple.
The only significant difference is that indexing functions must be specified
with indfn= (and are not allowed when using wrapper-generated external
calling). These options override any defaults normally assumed by the
Array constructor.
datatype=... Only hardware datatypes are allowed. This field is re-

quired, but the equation form of entry is not necessary. For example,
simply specifying integer[4] is sufficient.
order=... This may be left unspecified for vectors since Fortran and C
representation is the same. Otherwise, this will default to Fortran_order
when calling a Fortran library and C_order when calling a C library.
storage=... If this is not specified, the default is full rectangular storage
subtype=... This is optional and restricts the subtype to Array, Matrix,

Vector[row], or Vector[column].
indfn=(..., ...) This specifies the indexing functions that the Array,
Matrix, or Vector must have.
Other Compound Types There are other types, including records

(structs), and procedures that are supported when using wrapper gen-
erated external linking. These data descriptors are described in section
11.2.
Specifying Argument Passing Conventions

Different programming languages have different conventions for parameter
passing. C always uses pass-by-value; pass-by-reference must be done
explicitly by passing an address. Fortran uses pass-by-reference. Pascal
uses either, depending on how the parameter was declared.
Maple’s external calling mechanism currently supports both C and
Fortran calling conventions. Automatic wrapper generation is only sup-
ported for C, but an external API exists for both C and Fortran. The
default convention used is C. To use Fortran calling, specify the name
Fortran as a parameter to define_external.
11.2 Method 2: Wrapper Generation • 539
> f := define_external(‘my_func‘,‘FORTRAN‘, ...);
Some other compiler implementations (such as Pascal, C++, and

Java) may be made to work with C external calling by using the cor-
rect definitions and order of passed parameters.
11.2 Method 2: Wrapper Generation

There are some types in Maple that do not lend well to automatic “on
the fly” conversions. Two of these types are procedures (callbacks), and
records (structs). Maple provides an alternate mechanism for handling
this kind of data.
For a description of the steps required to use compiled code in Maple,
refer to Section 11.1. The same three basic steps (DLL creation, function
specification, and function invocation as described on pages 533-534) are
used in this method. The information in this section extends the basic
information by describing the use of wrappers.
Specifying the keyword WRAPPER in the call to define_external will
cause Maple to generate code to do the necessary data translations.
Maple will compile this code into a shared library and dynamically
link to the new library. Subsequently invoking the procedure returned
by define_external will call the newly generated conversion routine be-
fore calling the external function in the library you provided.
The C code generated by Maple “wraps” the Maple data structures
by translating them to hardware equivalent types. Hence, the code file is
called the wrapper, and the library generated from this code is called the
wrapper library.
Additional Types and Options

Generating a wrapper file allows Maple to translate more complicated
types that are difficult to handle without compilation technology. It also
allows greater flexibility when dealing with pointers and passed data that
do not exactly fit the required type.
Table 11.3 (located at the end of the chapter) lists additional types
are supported when the keyword WRAPPER is specified.
Structured Data Formats

A structure is a non-homogeneous collection of members, corresponding
to a struct in C, or a record in Pascal. A union is similar, except that
all the members start at the same memory address.

Each member :: descriptor pair describes one member of the structure
or union. The descriptor is any of the types described in this chapter.
The options are used to specify what kind of datatype the wrap-
per should expect for conversion purposes. The following two options are
supported.
TABLE Tables will be used as the corresponding Maple type. Using
tables is the default behavior, and they are friendlier to use than lists.
When tables are used, the member names will correspond to table
indices.
LIST Lists will be used as the corresponding Maple type. Lists are pri-
marily used in a read-only basis. Lists cannot be modified in-place, so
making updates to a list structure in external code requires a copy to
be made. When structures must be passed back to Maple, or if they
contain pointer types, it is better to use tables.
Lists and tables cannot be used interchangeably. Once the wrapper
has been generated, it will accept only the declared type, not both.
Enumerated Types
Maple’s external calling mechanism does not directly support enumerated
types (such as enum in C). Instead, use the integer[n] type with n of an
appropriate size to match the size of the enumerated type of the compiler
with which the external function was compiled (usually this is the same
size as the int type).
Procedure Call Formats

Some languages, like C, support passing functions as arguments. A Maple
procedure can be passed to an external function in the same way. The
wrapper will set up a C style procedure to call Maple to execute the
passed procedure with the given arguments. This C callback is given to
the external call to be used like any other C function.
Each member :: descriptor pair describes one parameter of the proce-
dure. The descriptor is any of the types described in this chapter.
It is not permitted to declare a procedure that itself takes a procedure
parameter. In other words, a callback cannot itself call back to the external
code.
Call by Reference
Unless modified as described below, each argument is passed by value.
The REF modifier can be used to override this.
argumentIdentifer :: REF( dataDescriptor, options )
The REF modifier can take the following options.

ANYTHING This option must be first in the list of options. Use this
option to declare the equivalent of a C void* parameter. The wrapper
code will attempt to convert passed arguments to simple types, (4-
byte integer, 8-byte float, complex, or string), when encountered. If
no conversion to one of these types is possible, then NULL is passed to
the external function.
CALL_ONLY This option specifies that although the object is to be
passed by reference, any changes made by the external function will
not be written back to the Maple symbol that was passed. This can
be used both to protect the objects being passed (see section on array
options), and to reduce overhead (as no translation back to Maple
data structures need be made).
RETURN_ONLY This option specifies that no data is actually passed
to the external function. Instead, only a reference to the allocated
space is passed, and the external function is expected to fill the space
with data. The result is converted back into an appropriate Maple
object.
Array Options
If an ARRAY argument is declared as CALL_ONLY and an Array, Matrix,
or Vector with proper settings is passed to the external function (so
that no copying is required), then CALL_ONLY will have no effect and thus
will not prevent the called function from overwriting the original array.
To prevent this from occurring, the option COPY can be included in the
ARRAY descriptor.
The ARRAY descriptor accepts extra options when used with wrapper
generation. These options can be specified as follows.
ARRAY( dim1, ..., dimN, datatype=typename,

order=..., ..., options )
The dim1 through dimN parameters are each integer ranges, specify-
ing the range of each dimension of the array. Any of the upper or lower
bounds may be the name of another argument, in which case the value of
that argument will specifiy the corresponding array bound at run-time.
The options are used to specify how an array should be passed. The
following are valid options.
COPY Do not operate in-place on the given array. That is, make a copy
first, and use the copy for passing to and from the external function.
NO_COPY This ensures that a copy of the data is never made. Usually,
when using a wrapper generated external call, if the Array, Matrix,
or Vector is of the wrong type, (say the order is wrong), then a copy
is made with the correct properties before passing it to the external
function. Also, the “returned” array will have the properties of the
copy. If NO_COPY is specified, and an Array, Matrix, or Vector with
incorrect options is passed, an exception is raised. Arrays are always
passed by reference. If no options are given (via a REF descriptor),
they are passed by using the CALL_ONLY behavior of REF with the
noted exception described at the beginning of this section.
Non-Passed Arguments
Sometimes it will be necessary to pass additional arguments to the Maple
wrapper that should not be passed on to the external function. For ex-
ample, consider the following hypothetical C function:
int sum( int *v1, int *v2 )
This function takes two integer vectors, v1 and v2, and adds the ele-
ments of v2 to v1, stopping when it finds an entry that is zero. It might
be nice for the generated wrapper to make sure the vectors are the same
size. The Maple definition for this function is as follows.
> Sum := define_external( ‘sum‘,
> v1 :: ARRAY(1..size,integer[4]),
> v2 :: ARRAY(1..size,integer[4]),
> size :: NO_PASS(integer[4]),
> RETURN :: integer[4],
> LIB="libsum.dll");
The NO_PASS modifier indicates that the size argument should not be
passed to the external function. The Sum function could then be called by
the following statement,
> Sum(v1,v2,op(1,v1));
where v1 and v2 are vectors. Maple will pass the vector data, or a copy
of the vector data, and pass it to the external sum function. It will not
pass the size element to the external function, but size will be used for
argument checking (since the NO_CHECK option was not specified).
Note that this option can only be used for top-level arguments. That
is, it is invalid to declare a callback procedure’s arguments as NO_PASS.
Argument Checking and Efficiency Considerations

It is intended that the time and space costs of calling an external function
not be any higher than the costs for calling an equivalent built-in function
with the same degree of argument type checking. Clearly, the amount of
type checking done by a generated Maple-language wrapper exceeds that
done by most internal functions, so there will be some additional overhead.
The define_external function has an option NO_CHECK which, if
used, will disable the type checking done by the Maple-language wrapper.
For frequently called external functions that perform simple operations
this can significantly improve performance. However, there is a risk asso-
ciated with using the NO_CHECK option. If you pass an object of the wrong
type, the generated C-language wrapper might misinterpret what it has
received, resulting in erroneous translations to external types, and hence
unpredictable behavior of the external function.
Conversions
When the procedure returned by define_external is actually called,
the Maple arguments that are passed are converted to the corresponding
arguments of the external function. Likewise, the value returned from the
external function is converted back to the corresponding Maple type.
The following table describes each of the external types and the Maple
types that can be converted into that type. The first listed Maple type
is the one that a result of the corresponding external type would be con-
verted into.
External Type Allowed Maple Type(s)
boolean[n] boolean
integer[n] integer
float[n] float, rational, integer, numeric
complex[n] complex, numeric, float, rational, integer
char[n] one-character string
string[n] string, symbol, 0
ARRAY() Array, Vector, Matrix, name, 0
STRUCT() list, table
UNION() table
PROC() procedure
For STRUCTs, either lists or tables are valid for a particular declaration.
Once declared, only one of the types–a list or a table–will be acceptable.
They cannot be used interchangeably unless the wrapper is regenerated.
For UNIONs, only tables are permitted, and the table must contain ex-
actly one entry when passed (corresponding to one of the members of the
union).
If an argument of an incompatible type is passed, an error occurs,
and the external function will not be called. Likewise, if a value is passed
that would be out of range for the specified type (e.g., integer too large),
an error occurs. When passing floating-point values, precision in excess of
that supported by the external type is discarded, provided the magnitude
of the value is within the range of the external type.
Arguments that were declared as REFerences may be passed either
a name, a zero, or the declared kind of Maple expression. If a name is
passed, it is evaluated, and the value is passed by reference to the ex-
ternal function. After the external function returns, the revised value is
converted back to the type specified for the argument and assigned back
to the name. If the name passed has no value, then either NULL is passed,
or a pointer to newly allocated space for the structure is passed. This
behavior is determined by the presence or absence of ALLOC in the REF
declaration. If a zero is passed, NULL is passed to the external function.
If any other Maple expression is passed, its value is passed by reference,
and the revised value is discarded.
Compiler Options
To compile the wrapper library, Maple requires the use of a C compiler
installed on the same machine that is running Maple. Maple will generate
a system command to call the compiler. The compiler needs to be well
known to the system. It should be in the system PATH and all associated
environment variables need to be set.
The compile and link commands are completely customizable pro-
vided that your compiler has a command-line interface available. Default
configurations are provided, which should make most cases work “out of
the box.” Maple is preprogrammed to use the vendor-supplied C compiler
to compile wrappers on most platforms.1
All default compile and link options are stored in a module that can be
obtained by using the command define_external(‘COMPILE_OPTIONS‘).
When the module returned by this command is modified, the modifica-
tion will affect all wrapper generation commands via define_external
for the remainder of the session. All of the names exported by the compile
options module can also be specified as a parameter to define_external.
When specified as a parameter, the effect lasts only for the duration of
that call.
The compile and link commands are assembled by calling the COMPILE_COMMAND
1
Under Microsoft Windows, Maple uses the Microsoft C Compiler.
and LINK_COMMAND procedures defined in the compile options module.

These procedures make use of the definitions in the compile options mod-
ule to formulate a command string that will be executed using ssystem.2
To customize the compile and link commands to suit your situation,
any of the following options can be modified. All option values must be
strings or NULL, except for COMPILE_COMMAND and LINK_COMMAND, which
must be procedures or NULL.
COMPILER This specifies the name of the compiler executable.
CFLAGS This specifies miscellaneous flags passed to the compiler.
COMPILE_ONLY_FLAG This is the flag to indicate that the file

is only to be compiled. On most platforms it is “-c”, which will tell
the compiler to generate an object file, but not link it to form any
executable or library. Linking will happen in a separate command.
COBJ_FLAG This is the flag used by the compiler to specify the ob-
ject filename. The compiler command uses COBJ_FLAG || FILE ||
OBJ_EXT to name the object file. On most platforms it is “-o”.
LOBJ_FLAG This is the flag used by the linker to specify the tar-
get library name. The link command uses LOBJ_FLAG || FILE ||
DLL_EXT to name the shared library.
FILE This is the base name of the file to be compiled. The file extension
should not be included in this name. For example, if you want to
compile “foo.c”, set FILE="foo" and FILE_EXT=".c". When FILE is
set to NULL the system generates a file name based on the function
name.
FILE_EXT This is the program file extension. If you want to compile

“foo.c”, set FILE_EXT=".c", and FILE="foo".
OBJ_EXT This is the object file extension. Common extensions are

“.o” and “.obj”.
DLL_EXT This is the dynamic library extension. Common extensions

are “.dll” and “.so”.
INC_FLAG This precedes directories in the INC_PATH. On most plat-

forms it is “-I”.
2
If using the Microsoft C compiler, the LINK_COMMAND is set to NULL since the
COMPILE_COMMAND does both the compiling and linking.
INC_PATH This specifies the directories to search for header files.

Use an expression sequence to specify more than one directory.
For example, INC_PATH=("/usr/local/maple/extern/include",
"/users/jdoe/include").
COMPILE_COMMAND This is set to the procedure that generates

the compiler command. The procedure must return a string. Un-
less you are using an unusual compiler, it should not be necessary
to change the default.
LINKER This specifies the name of the linker executable.
LINK_FLAGS This specifies miscellaneous flags passed to the linker,

including those that tell the linker to build a dynamic (shared) library.
LIB_FLAG This precedes directories in the LIB_PATH. On most plat-

forms it is “-L”.
LIB_PATH This specifies the directories to search for libraries. Use an

expression sequence to specify more than one directory. For example,
LIB_PATH=("/usr/local/maple/extern/lib","/users/jdoe/lib").
LIB This names the library which contains the external function you
want to call. This option must be specified in every call to define_external.
LIBS This specifies other libraries that need to be linked with the wrap-
per library to resolve all external symbols. Use an expression sequence
to specify more than one library. For example, LIBS=("/usr/local/
maple/extern/lib/libtest.so","/users/jdoe/libdoe.so").
SYS_LIBS This specifies system libraries that need to be linked with

the wrapper library to resolve all external symbols. Use an ex-
pression sequence to specify more than one library. For example,
LIBS=("-lc","-lm").
EXPORT_FLAG This flag is used in combination with FUNCTION to

name the function to be exported from the shared library. This is
unassigned or set to NULL on platforms that export all symbols by
default.
FUNCTION This is the name of the external function defined in the

wrapper library. The system will generate a FUNCTION name if this is
left unassigned or set to NULL.
LINK_COMMAND This is set to the procedure that generates the

linker command. The procedure must return a string. Set this to NULL
if the compile command also does the linking.
A common use of these options as parameters to define_external

with a standard compiler would be to specify the filename. For example,
the following will generate a wrapper file named “foo.c”.
> f := define_external(‘myfunc‘,‘WRAPPER‘,‘FILE‘="foo",‘LIB‘=
> "mylib.dll"):
To use a non-standard compiler or to alter compile flags, it may be

easier to assign directly to the compile options module. The following
example shows how to setup the GNU compiler on a machine running
Solaris.
> p := define_external(‘COMPILE_OPTIONS‘):
> p:-COMPILER := "gcc";
> p:-COBJ_FLAG := "-o ":
> define_external(‘mat_mult‘,‘WRAPPER‘,‘LIB‘="libcexttest.so"):
The tricky part in the above example is that gcc likes to have a space
between -o and the object name. Modifying the COBJ_FLAG allows this to
be easily done. All other option default values were acceptable.
The best way see what commands are actually being executed is to
set the infolevel for define_external to 3 or higher. Repeating the
above example you might see the following.
> infolevel[define_external] := 3:
> define_external(‘mat_mult‘,‘WRAPPER‘,‘LIB‘="libcexttest.so"):
"COMPILE_COMMAND"
"gcc -g -c -I/user/local/maple/extern/include -o \
mwrap_mat_mult.o mwrap_mat_mult.c"
"LINK_COMMAND"
"ld -znodefs -G -dy -Bdynamic
-L/user/local/maple/bin/bin.SUN_SPARC_SOLARIS \
-omwrap_mat_mult.so mwrap_mat_mult.o -lc -lmaplec"
An alternate way to see the compile and link commands is to call the
command-builder procedures directly. Make sure to set or unassign the
variables that will be filled in, otherwise they will be left blank.
> p:-COMPILE_COMMAND();
"gcc -g -c -I/u/maple/extern/include -o .o .c"
> unassign(’p:-FILE’);
> p:-COMPILE_COMMAND();
"gcc -g -c -I/u/maple/extern/include -o FILE.o FILE.c"

The following example shows two calls to define_external separated
by the restart command. The first call does not use the WRAPLIB option
and thus generates quad.c and compiles the wrapper library quad.dll.
The second call uses the WRAPLIB option in order to reuse the existing
quad.dll. No compilation or wrapper generation is done in the second
call.
> quadruple_it := define_external(’quadruple_it’,
> WRAPPER,FILE="quad",
> x::float[4],
> RETURN::float[4],
> LIB="test.dll"):
> quadruple_it(2.2);
8.80000019073486328
> restart;
> quadruple_it := define_external(’quadruple_it’,
> WRAPPER,FILE="quad",
> x::float[4],
> RETURN::float[4],
> WRAPLIB="quad.dll",
> LIB="test.dll"):
> quadruple_it(2.2);
8.80000019073486328
When DLLs are created and compiled at runtime it is important not
to duplicate the name of a previously generated DLL without restarting
Maple (either by closing Maple down or issuing the restart command).
Maple will maintain an open connection with the first DLL opened
with any given name. Attempting to create a new DLL of the same
name without restarting may lead to unexpected results. The Maple
command dlclose can be used to avoid restarting, but subsequently
calling any external function in that closed DLL without reissuing the
define_external command will likely crash Maple.
Evaluation Rules
External functions follow normal Maple evaluation rules in that the argu-
ments are evaluated during a function call. It therefore may be necessary
to quote assigned names when passing by-reference. For example, consider
the following function that multiplies a number by two in-place.
void double_it( int *i )

{
if( i == NULL ) return;
*i *= 2;
}
In Maple, the wrapperless definition of this function might appear as

follows.
> double_it := define_external(’double_it’, i::REF(integer[4]),
> LIB="libtest.dll");
When an integer is passed to this function, a pointer to the hardware

conversion is sent to the external routine. In this case, the result will be
lost since there is no way to refer back to it from Maple.
> double_it(3);
The solution is to name the value you want to pass out. The name
needs to be quoted in order to prevent evaluation and thus have only the
value passed out.
> double_it(n); # n is evaluated so 3 gets passed
> n;
> double_it(’n’); # used unevaluation quotes to pass the ’n’

> n;
For numeric data, the string "NULL" can be passed as a parameter to

represent the address 0 (the C NULL). For strings, since "NULL" is a valid
string, the integer 0 represents address 0.
> double_it("NULL");
>
> concat := define_external(’concat’,
> RETURN::string, a::string, b::string,
> LIB="libtest.dll"):
> concat("NULL","x");
"NULLx"
> concat(0,0);
In the concat example above, the C code might look like the following.
Note that this function does not clean up memory as it should.
char * concat( char* a, char *b )

{
char *r;
if( !a || !b ) return( NULL );
r = (char*)malloc((strlen(a)+strlen(b)+1)*sizeof(char));
strcpy(r,a);
strcat(r,b);
return( r );
}
11.3 Method 3: Customizing Wrappers

For complete control over data conversions, Maple allows modification of
existing wrappers and creation of custom wrappers. There is an extensive
API of functions available to external programs called from Maple using
define_external.
To use this method, you must be familiar with the steps required to
use compiled code in Maple, described in Section 11.1. For this method,
you do not perform a function specification because Maple passes one
data structure containing all the passed information. Therefore, there are
11.3 Method 3: Customizing Wrappers • 551
only two basic steps (DLL creation and function invocation as described
on pages 533-534) in addition to wrapper generation. Wrappers were in-
troduced in Section 11.2.
External Function Entry Point

Maple will look up the symbol name given as the first argument to
define_external in the DLL specified in the LIB= argument. Maple also
looks up the MWRAP_symbolName in the wrapper library. This MWRAP_symbolName
function prototype has the following format.
ALGEB MWRAP_quadruple_it(
MKernelVector kv,
FLOAT32 (*fn) ( FLOAT32 a1 ),
ALGEB fn_args
);
The prototype above was taken from the wrapper quad.c described in
the previous section. The first argument kv is a handle to the Maple kernel
function vector. The second argument fn is a function pointer assigned
the symbol looked up in the external DLL. In this case, fn will be assigned
the quadruple_it external function. The last argument is a Maple ex-
pression sequence data structure containing all the arguments passed to
the function during any given call to the Maple procedure generated by
the define_external command.
The above entry point is the format used when wrappers are auto-
matically generated, and when WRAPLIB is specified. An alternate external
entry point that excludes the function pointer is available when the pa-
rameter MAPLE is specified instead of WRAPPER or WRAPLIB.
ALGEB MWRAP_quadruple_it(
MKernelVector kv,
ALGEB fn_args
);
The API function prototypes for manipulating Maple data structures

can be found in $MAPLE/extern/include where $MAPLE is the path of your
Maple installation. The header file maplec.h should be included when
writing custom C wrappers. One of the header files, maplefortran.hf
or maplefortran64bit.hf, should be included when writing custom For-
tran wrappers. Other header files, mplshlib.h, and mpltable.h contain
macros, types, and data structures that are needed for direct manipula-
tion of Maple data structures.
Maple uses directed acyclic graphs (dags) to represent all objects such
as integers, floating point numbers, sums, modules, procedures, etc. (See
Appendix A for more details about Maple’s internal representation of
objects.) These dags have the type ALGEB in C wrappers, and INTEGER
or INTEGER*8 in Fortran wrappers. Fortran 77 has no user type defi-
nition semantics so ALGEB pointers must be “faked” by using machine
word-sized integers. If the machine word size is 64-bit (e.g., as on a
DEC Alpha), then the header maplefortran64bit.hf must be used and
INTEGER*8 must be used as the dag datatype. Execute the Maple com-
mand kernelopts(wordsize) to see if you should be using 32-bit or
64-bit integer-dag types in Fortran. When working with C, the datatype
is ALGEB regardless of the machine word size.
For the most part, treat these dags as black boxes. In other words,
you do not have to know the internal details of dags to manipulate and
work with them. The only exception is the argument sequence passed to
the wrapper entry point. This is an expression seqence (EXPSEQ) dag, and
can be treated as an array of dags starting at index 1 (not 0). Thus,
fn_args[1] is the first parameter passed to the external function. Use
MapleNumArgs to determine the number of arguments passed. Note that
the Fortran API uses a slightly different naming convention. The equiva-
lent Fortran call is maple_num_args. The C API names will be used for
the remainder of this chapter. Refer to the API listing to find equivalent
Fortran names.
The easiest way to start writing custom wrappers is to inspect auto-
matically generated wrappers. Consider the add function that was intro-
duced at the beginning of this chapter. Use the WRAPPER option to tell
define_external to generate a wrapper. Also use the NO_COMPILE op-
tion to tell define_external not to compile the generated wrapper. The
name of the generated file will be returned.
> myAdd := define_external(
> ’add’,
> ’WRAPPER’,
> ’NO_COMPILE’,
> ’RETURN’::integer[4]
> );
myAdd := "mwrap_add.c"
The file mwrap_add.c will look something like the following.
/* MWRAP_add Wrapper
Generated automatically by Maple

Do not edit this file. */
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <mplshlib.h>
#include <maplec.h>
MKernelVector mapleKernelVec;
typedef void *MaplePointer;
ALGEB *args;
/* main - MWRAP_add */
ALGEB MWRAP_add( MKernelVector kv,
INTEGER32 (*fn) ( INTEGER32 a1, INTEGER32 a2 ),
ALGEB fn_args )
{
INTEGER32 a1;
INTEGER32 a2;
INTEGER32 r;
ALGEB mr;
int i;
mapleKernelVec = kv;
args = (ALGEB*) fn_args;
if( MapleNumArgs(mapleKernelVec,(ALGEB)args) != 2 )
MapleRaiseError(mapleKernelVec,"Incorrect number
of arguments");
/* integer[4] */
a1 = MapleToInteger32(mapleKernelVec,args[1]);
/* integer[4] */
a2 = MapleToInteger32(mapleKernelVec,args[2]);
r = (*fn)(a1, a2);
mr = ToMapleInteger(mapleKernelVec,(long) r);
return( mr );
}
The generated wrapper is human readable, and thus a good starting

point for creating your own wrapper. There may be some extra variables
and declarations used since the wrapper generation is generic and may
be heavy handed at times. For example, the use of args rather than
fn_args avoids the need for a cast with args[1], but it also is a static
global which is useful when working with callbacks which may need access
to the argument sequence outside the main entry point.
Remember that the add function simply added the arguments a1 and
a2 and returned the result. This can be done directly in the wrapper. By
removing the second argument fn so the MAPLE option can be used, plus
inlining the a1+a2 functionality and cleaning up the code, the wrapper
may look like the following.
/* Program to add two numbers from Maple */
#include <stdio.h>
#include <stdlib.h>
#include <maplec.h>
/* main entry point - MWRAP_add */

ALGEB myAdd( MKernelVector kv, ALGEB fn_args )
{
INTEGER32 a1; /* INTEGER32 => int (defined in mpltable.h) */
INTEGER32 a2;
INTEGER32 r;
if( MapleNumArgs(kv,fn_args) != 2 )
MapleRaiseError(kv,"Incorrect number of arguments");
/* convert from Maple integer to C int */

a1 = MapleToInteger32(kv,((ALGEB*)fn_args)[1]);
/* convert from Maple integer to C int */

a2 = MapleToInteger32(kv,((ALGEB*)fn_args)[2]);
r = a1 + a2;
return( ToMapleInteger(kv,(long) r) );
}
This program first checks to make sure there were exactly two ar-
guments passed in the Maple function call. It then converts the two
arguments to hardware integers and adds them together. The result is

converted back to a Maple integer and returned.
This program can be compiled into a DLL using your favorite C com-
piler. Ensure that you link with the Maple API shared library. The DLL
can be placed into the Maple bin.$SYSTEM directory, or somewhere else
in the PATH. When using DLLs outside of bin.$SYSTEM directory, you
may have to specify the full path to the DLL in the LIB argument to
define_external. Unix developers may need to set their load-library-
path.
Table 11.4 (located at the end of the chapter) lists the Maple API
Libraries for C and Fortran.
After compiling the DLL, the function can be used from within Maple.
No type desciptors are needed in the define_external call since Maple
does no conversion on arguments passed to the custom wrapper.
> myAdd := define_external(’myAdd’,’MAPLE’,’LIB’=
> "myAdd.dll"):
> myAdd(2,3);
> myAdd(2.2,1);
Error, (in myAdd) integer expected for integer[4] parameter
> myAdd(2^80,2^70);
Error, (in myAdd) integer too large in context
The equivalent Fortran wrapper would look like the following.
Program to add two numbers from Maple
INTEGER FUNCTION myAdd(kv, args)
INCLUDE "maplefortran.hf"
INTEGER kv
INTEGER args
INTEGER arg
INTEGER a1, a2, r

CHARACTER ERRMSG*20
INTEGER ERRMSGLEN
ERRMSGLEN = 20
IF ( maple_num_args(kv, args) .NE. 2 ) THEN

ERRMSG = ’Incorrect number of arguments’
CALL maple_raise_error( kv, ERRMSG, ERRMSGLEN )
myAdd = to_maple_null( kv )
RETURN
ENDIF
arg = maple_extract_arg( kv, args, 1 )

a1 = maple_to_integer32(kv, arg)
arg = maple_extract_arg( kv, args, 2 )

a2 = maple_to_integer32(kv, arg)
r = a1 + a2
myAdd = to_maple_integer( kv, r )
END
Once compiled into a DLL, the same syntax can be used in Maple
to access the function. The only difference is the additional keyword
’FORTRAN’ in the define_external call.
> myAdd := define_external(’myAdd’,’MAPLE’,’FORTRAN’,’LIB’=
> "myAdd.dll"):
> myAdd(2,3);
External API
An external API is provided for users who want to augment existing
wrappers or write their own custom wrappers. This section describes the
functions available when linking with the Maple API library (see Ta-
ble 11.4) and including either maplec.h or maplefortran.hf.
Argument Checking The following C function can be used to query

the number of arguments contained in the argument expression sequence
passed as the last argument to the external function entry point. The ex-
pression sequence passed to this entry point can be queried directly (e.g.,
((ALGEB*)expr)[1]). If n = MapleNumArgs(kv,expr), then the last ar-
gument is ((ALGEB*)expr[n].
M_INT MapleNumArgs( MKernelVector kv, ALGEB expr );
The arguments passed to the Fortran entry point cannot be queried

directly. The maple_extract_arg function must be used to get at the
argument data (e.g., arg1 = maple_extract_arg(kv,args,1)). If n =
maple_num_args(kv,s), then the last argument is maple_extract_arg(kv,args,n).
INTEGER maple_num_args( kv, s )

INTEGER maple_extract_arg( kv, s, i )
The following functions indicate the type of the given Maple object.
M_BOOL IsMapleAssignedName( MKernelVector kv, ALGEB s );

M_BOOL IsMapleComplexNumeric( MKernelVector kv, ALGEB s );
M_BOOL IsMapleNumeric( MKernelVector kv, ALGEB s );
M_BOOL IsMapleInteger( MKernelVector kv, ALGEB s );
M_BOOL IsMapleInteger8( MKernelVector kv, ALGEB s );
M_BOOL IsMapleName( MKernelVector kv, ALGEB s );
M_BOOL IsMapleNULL( MKernelVector kv, ALGEB s );
M_BOOL IsMaplePointer( MKernelVector kv, ALGEB s );
M_BOOL IsMaplePointerNULL( MKernelVector kv, ALGEB s );
M_BOOL IsMapleProcedure( MKernelVector kv, ALGEB s );
M_BOOL IsMapleRTable( MKernelVector kv, ALGEB s );
M_BOOL IsMapleString( MKernelVector kv, ALGEB s );
M_BOOL IsMapleTable( MKernelVector kv, ALGEB s );
M_BOOL IsMapleUnassignedName( MKernelVector kv, ALGEB s );
M_BOOL IsMapleUnnamedZero( MKernelVector kv, ALGEB s );
Equivalent Fortran functions are as follows. The C functions, IsMaplePointer,

IsMaplePointerNULL, IsMapleUnassignedName, and IsMapleUnnamedZero
are not available in the Fortran API.
INTEGER is_maple_assigned_name( kv, s )

INTEGER is_maple_complex_numeric( kv, s )

INTEGER is_maple_numeric( kv, s )
INTEGER is_maple_integer( kv, s )
INTEGER is_maple_integer8( kv, s )
INTEGER is_maple_name( kv, s )
INTEGER is_maple_null( kv, s )
INTEGER is_maple_procedure( kv, s )
INTEGER is_maple_rtable( kv, s )
INTEGER is_maple_string( kv, s )
INTEGER is_maple_table( kv, s )
INTEGER is_maple_unassigned_name( kv, s )
These functions all return TRUE (1) when the Maple dag s fits the
description given by the function name. If s is not of the correct type,
FALSE (0) is returned. Maple’s NULL is not the same as a C Pointer-
NULL. The former case is the empty expression sequence in the Maple
language. The latter case is a pointer variable set to the address zero. Since
there is no concept of real pointers in the Maple Language, the idea of
Pointer-NULL in this context means the Maple integer zero, or an unas-
signed Maple name. The IsMaple...Numeric routines use the Maple type
numeric definition. All other checks use the dag type definition. For ex-
ample, type(t[1],name) returns true in Maple, but IsMapleName checks
for a NAME dag and will return FALSE since t[1] is internally represented
as a TABLEREF dag. Integer query routines with the bit size specified in the
name will check to make sure the given Maple object s is a Maple integer
and also that it could fit into the specified number of bits if converted to
a hardware integer.
Conversions From Maple Objects The following functions return the

specified type when given a dag s that can be converted to that type.
COMPLEXF32 MapleToComplexFloat32( MKernelVector kv, ALGEB s );

COMPLEXF64 MapleToComplexFloat64( MKernelVector kv, ALGEB s );
CXDAG MapleToComplexFloatDAG( MKernelVector kv, ALGEB s );
FLOAT32 MapleToFloat32( MKernelVector kv, ALGEB s );
FLOAT64 MapleToFloat64( MKernelVector kv, ALGEB s );
INTEGER8 MapleToInteger8( MKernelVector kv, ALGEB s );

M_BOOL MapleToM_BOOL( MKernelVector kv, ALGEB s );
M_INT MapleToM_INT( MKernelVector kv, ALGEB s );
void* MapleToPointer( MKernelVector kv, ALGEB s );
char* MapleToString( MKernelVector kv, ALGEB s );
The following are the equivalent Fortran routines. Note that com-
plex and string conversion are done by reference. That is, the third
argument passed to the function will be set to the converted value
rather than the function returning the value. Equivalent functions for
MapleToComplexFloatDAG and MapleToPointer are not available.
SUBROUTINE maple_to_complex_float32( kv, s, c )

SUBROUTINE maple_to_complex_float64( kv, s, c )
REAL maple_to_float32( kv, s )
DOUBLEPRECISION maple_to_float64( kv, s )
INTEGER maple_to_integer8( kv, s )
INTEGER*8 maple_to_integer64( kv, s )
INTEGER maple_to_m_bool( kv, s )
INTEGER maple_to_m_int( kv, s )
INTEGER maple_to_string( kv, s, string )
Floating Point numbers may lose precision during the conversion to

hardware size data.
Conversion from a STRING dag to an integer will return the ASCII
value of the first character in that string. Conversion from a Maple
Boolean to an integer will return 1 for true or 0 for false.
Conversions from a STRING dag to a string should not be modified
in-place. A copy should be made if any modifications are necessary.
The MapleToPointer conversion returns the pointer value stored in a
Maple BINARY dag.
Conversions To Maple Objects The following functions return a dag

of the specified dag type when given a dag the corresponding hardware
data.
ALGEB ToMapleBoolean( MKernelVector kv, long b );

ALGEB ToMapleChar( MKernelVector kv, long c );
ALGEB ToMapleComplex( MKernelVector kv, double re,
double im );
ALGEB ToMapleComplexFloat( MKernelVector kv, ALGEB re,
ALGEB im );
ALGEB ToMapleExpressionSequence( MKernelVector kv, int
nargs, /* ALGEB arg1, ALGEB arg2, */ ... );
ALGEB ToMapleInteger( MKernelVector kv, long i );
ALGEB ToMapleInteger64( MKernelVector kv, INTEGER64 i );
ALGEB ToMapleFloat( MKernelVector kv, double f );
ALGEB ToMapleName( MKernelVector kv, char *n, M_BOOL
is_global );
ALGEB ToMapleNULL( MKernelVector kv );
ALGEB ToMapleNULLPointer( MKernelVector kv );
ALGEB ToMaplePointer( MKernelVector kv, void *v );
ALGEB ToMapleRelation( MKernelVector kv, const char *rel,
ALGEB lhs, ALGEB rhs );
ALGEB ToMapleString( MKernelVector kv, char *s );
ALGEB ToMapleUneval( MKernelVector kv, ALGEB s );
The equivalent Fortran routines are as follows. The Fortran API

does not support ToMapleExpressionSequence, ToMapleNULLPointer,
ToMaplePointer, ToMapleRelation, and ToMapleUneval.
to_maple_boolean( kv, b )
to_maple_char( kv, c )
to_maple_complex( kv, re, im )
to_maple_complex_float( kv, re, im )
to_maple_integer( kv, i )
to_maple_integer64( kv, i )
to_maple_float( kv, f )
to_maple_name( kv, s, s_len )
to_maple_null( kv )
to_maple_string( kv, s, s_len )
ToMapleBoolean is three valued. When b is zero, it return Maple’s

false dag. If n is -1, Maple’s FAIL dag is returned. If n is non-zero (and
not -1), Maple’s true dag is returned.
ToMapleChar returns a single character Maple string dag.
ToMapleComplex converts the pair of doubles, re and im, to the Maple
expression re + I*im, and returns this dag.
ToMapleComplexFloat converts a pair of FLOAT dags to the same
structure.
ToMapleExpressionSequence create and returns a Maple expression
sequence and fills it with the N algebraics, arg1, arg2, ..., argN.
ToMapleName returns a Maple NAME dag with the name n. If is_global

is set to TRUE, the name will be global in Maple’s name space. Otherwise,
if is_global is FALSE, the name will be a unique exported local.
ToMapleNULL returns Maple’s NULL dag (an empty EXPSEQ).
ToMapleNULLPointer returns Maple’s zero dag. This is the wrapper
representation of a NULL pointer passed to a procedure. This is not to be
confused with the value returned by ToMapleNULL.
ToMapleString copies the character string s to a Maple STRING dag
and returns it. When using the Fortran API, the length of the given string
must also be passed.
Rectangular Table (Vector, Matrix, Array) Manipulation Rtables are

the container class of Vector, Matrix, and Array data structures in
Maple. The basic access functions are as follows.
ALGEB RTableCreate( MKernelVector kv, RTableSettings *s,

void *pdata, M_INT *bounds );
void* RTableDataBlock( MKernelVector kv, ALGEB rt );
M_INT RTableNumElements( MKernelVector kv, ALGEB rt );
M_INT RTableNumDimensions( MKernelVector kv, ALGEB rt );
M_INT RTableLowerBound( MKernelVector kv, ALGEB rt,
M_INT dim );
M_INT RTableUpperBound( MKernelVector kv, ALGEB rt,
M_INT dim );
M_BOOL RTableIsReal( MKernelVector kv, ALGEB rt );
The Fortran API contains the following functions.
SUBROUTINE copy_to_array( kv, rt, a, num_rdims,

rbounds, num_fdims, fbounds, data_type )
SUBROUTINE copy_to_rtable( kv, a, rt, num_fdims,
fbounds, num_rdims, rbounds, data_type )
INTEGER convert_to_rtable( kv, a, num_rdims,
rbounds, num_fdims, fbounds, data_type )
INTEGER rtable_num_elements( kv, s )
INTEGER rtable_num_dimensions( kv, s )
INTEGER rtable_lower_bound( kv, s, dim )
INTEGER rtable_upper_bound( kv, s, dim )
INTEGER rtable_is_real( kv, s )
RtableDataBlock returns a pointer to the data block of a given

rtable. The returned value should be casted to the known data type
of the rtable. The data block can be manipulated directly instead of
using RtableAssign or RtableSelect. Users who directly manipulate

the data block must be aware of the storage type, order, data type, and
presence of indexing functions to do this properly.
In Fortran, there is no way to return an ARRAY pointer. To work
with an array created in Maple, the data-block must be copied to a pre-
allocated Fortran data block using the copy_to_array function. It copies
the contents of the rtable rt to the ARRAY, a. See maplefortran.hf for a
complete explanation of the parameters that are passed. To copy an array
back to Maple, the copy_to_rtable function can be used.
RtableCreate returns a newly created RTABLE as specified by:
1. The definitions given in the RtableSettings structure s.
2. A pointer to an existing block of data. If pdata is NULL, then a data

block is allocated and initialized to s->fill. When providing an al-
ready created block of data, it is important that s->foreign is set to
TRUE. Size, storage, data type, order, and indexing functions should all
be considered when managing your own data block. Usually, let Maple
create the data-block, then use RtableDataBlock to gain access to it.
3. The bounds array, bounds. An m x n matrix would have bounds =

1,m,1,n (i.e.,. both the upper and lower bounds must be specified).
The Fortran equivalent function is convert_to_rtable. It creates an

rtable from an existing Fortran array. The data is not copied in; instead,
the rtable maintains a pointer to the external data.
RtableNumElements returns the number of elements in a given
rtable. This may be different in sparse versus dense rtables.
1. For dense rtables, return the number of elements of storage allocated

for this rtable.
2. If rt is in NAG-sparse format, then this returns the number of ele-

ments in the data vector specified for the rtable, (which is the same
as the length of each index vector). Note that the number returned
here represents the number of data elements that are actually filled
in, not the number of elements allocated. Some of the elements may
have the value zero.
3. For Maple-sparse rtables, this always returns zero.
RtableNumDimensions returns the number of dimensions in a given

rtable.
RtableUpperBound and RtableLowerBound give the upper and lower

bound of the dimth dimension of the RTABLE, rt. For a 2 x 3 matrix,
RtableLowerBound(rt,1) will return 1 since the first dimension bounds
are 1..2, and the lower bound is 1.
RtableIsReal checks the elements of the RTABLE rt to see if they are
all real or not. If there are imaginary numbers present this returns FALSE.
If there are only real numbers, this returns TRUE. Note that this routine
will return immediately when given a non-complex hardware type.
In addition to the above functions, there is an extensive C API for
working rtable data types.
void RTableAppendAttribute( MKernelVector kv, RTableSettings

*s, char *name );
void RTableAppendIndFn( MKernelVector kv, RTableSettings
*s, ALGEB indfn );
void RTableGetDefaults( MKernelVector kv, RTableSettings
*s );
void RTableGetSettings( MKernelVector kv, RTableSettings
*s, ALGEB rt );
M_INT RTableIndFn( MKernelVector kv, ALGEB rt, M_INT num );
ALGEB RTableIndFnArgs( MKernelVector kv, ALGEB rt, M_INT num );
void RTableSetAttribute( MKernelVector kv, RTableSettings
*s, char *name );
void RTableSetIndFn( MKernelVector kv, RTableSettings *s,
ALGEB indfn );
void RTableSetType( MKernelVector kv, RTableSettings *s,
M_INT id, char *name );
RTableData RTableSelect( MKernelVector kv, ALGEB rt, M_INT
*index );
RTableData RTableAssign( MKernelVector kv, ALGEB rt, M_INT
*index, RTableData val );
void RTableSparseCompact( MKernelVector kv, ALGEB rt );
NAG_INT* RTableSparseIndexRow( MKernelVector kv, ALGEB rt,
M_INT dim );
ALGEB RTableSparseIndexSort( MKernelVector kv, ALGEB rt,
M_INT by_dim );
void RTableSparseSetNumElems( MKernelVector kv, ALGEB rt,
M_INT num );
M_INT RTableSparseSize( MKernelVector kv, ALGEB rt );
ALGEB RTableCopy( MKernelVector kv, RTableSettings *s,

ALGEB rt );
ALGEB RTableCopyImPart( MKernelVector kv, RTableSettings
*s, ALGEB rt );
ALGEB RTableCopyRealPart( MKernelVector kv, RTableSettings
*s, ALGEB rt );
ALGEB RTableZipReIm( MKernelVector kv, RTableSettings *s,
ALGEB rt_re, ALGEB rt_im );
Most Rtable access functions use the RtableSettings structure de-

fined in mpltable.h. This struct corresponds directly to the options avail-
able to the rtable constructor in Maple.3
RtableAppendAttribute appends the name attribute to the list of
attributes in the RtableSettings structure.
RtableAppendIndFn appends the indexing function, infn to the list
of indexing functions in the RtableSettings structure. Note that infn
must be a valid Maple name or table-reference. For example,
RTableAppendIndFn(kv,&settings,ToMapleName(kv,"symmetric",
TRUE));
RTableAppendIndFn(kv,&settings,EvalMapleStatement(kv,
"triangular[upper]"));
RtableGetDefaults fills the RtableSettings structure s with stan-

dard default values. These defaults are as follows:
data_type = RTABLE_DAG
maple_type = ’anything’ (Maple name ’anything’)
subtype = RTABLE_ARRAY
storage = RTABLE_RECT
p1 = -1, p2 = -1
order = RTABLE_FORTRAN
read_only = FALSE
foreign = FALSE
num_dimensions = -1
index_functions = ’NULL’ (Maple NULL)
attributes = ’NULL’ (Maple NULL)
transpose = FALSE
fill = 0
RtableGetSettings fills the RtableSettings structure s with the

settings held by the RTABLE, rt.
3
For more information, see ?rtable.
RtableIndFn returns the i th indexing function code. The indexing

codes are defined in mpltable.h in the form RTABLE_INDEX_XXXX. If there
are no indexing functions, this will give an error for any value of i. If there
is one indexing function, then rtableIndFun(rt,1) will return the code
for the only indexing function. Use MapleNumArgs to find out how many
indexing functions there are.
RtableIndFnArgs returns the argument expression sequence for in-
dexing function ’num’ in rtable ’rt’. If no arguments exist, Maple ’NULL’
is returned. The result can be further converted to a hardware type using
the MapleToXXX function(s). The number of arguments returned can be
determined using MapleNumArgs. Note that some knowledge about the
indexing functions is required in order to convert the return value to the
appropriate hardware type. For example, RTableIndFnArgs(kv,rt,1) of
a band[b1,b2] rtable will return the b1 part of the expression sequence
(b1,b2). The user must know that b1 and b2 are always integers. Con-
versely, c in constant[c] will always be the same type as the rtable’s
datatype. Thus for float[8] rtables, MapleToFloat64 should be used to
do the conversion to a hardware type.
RtableSetAttribute sets all the attributes of the RtableSettings
structure s to the single NAME attribute, name.
RtableSetIndFn sets all the indexing functions of the RtableSettings
structure s and resets it to the single indexing function infn.
RtableSetType sets the data_type field in the given RtableSettings
structure s to id, and when id=RTABLE_DAG, sets the maple_type to name.
For example, to set the data type to float[8], RTableSetType(kv,&s,
RTABLE_FLOAT,NULL) would be called. To set the type to numeric,
RTableSetType(kv,&s,RTABLE_DAG,"numeric") would be called. Ba-
sic type ids are defined in mpltable.h. To set compound types, the
RtableSettings data structure can be manipulated directly as follows:
settings.data_type = RTABLE_DAG;
settings.maple_type = EvalMapleStatement(kv,
"complex(numeric)");
RtableSelect returns the value rt[index], where rt is an RTABLE,
and index is an integer array.
RtableAssign assigns the value val to rt[index]. This function
should be used instead of assigning directly to the rtable data-block
whenever the given rtable has an indexing function or unusual storage
format (e.g., sparse). The index is an integer array. For example, the
following code assigns the value 3.14 to the [2,1] element of the given
datatype=float[8] rtable.
RTableData val;
M_INT *index;
index[0] = 2;
index[1] = 1;
val.float64 = 3.14;
RTableAssign(kv,rt,index,val);
RtableSparseCompact removes any zeros in the sparse rtable data

block. This should be called after an external routine that modifies the
sparse data block directly.
RtableSparseIndexRow returns the vector of indices for the i th di-
mension of rt. The rt must be a NAG sparse rtable.
RtableSparseIndexSort sorts the N th index vector for the NAG
sparse rtable rt. This is done in-place, and the other index vectors are
adjusted accordingly so that the index/value mapping is preserved.
RtableSparseSetNumElems sets the number of non-zero entries in the
NAG sparse rtable rt to N. This should only be done if the number of
elements has actually changed.
RtableSparseSize returns the number of entries allocated to store
data in the NAG sparse rtable rt. This is not necessarily the same as
RtableNumElems.
RtableCopy returns a copy of the rtable rt with new settings as given
by the RtableSettings structure s.
RtableCopyImPart returns a copy of the imaginary part of the rtable
rt with new settings as given by the RtableSettings structure s. The
copy returned is purely real, but contains only the imaginary parts of the
given rtable.
RtableCopyRealPart returns a copy of the real part of the rtable rt
with new settings as given by the RtableSettings structure s.
RtableZipReIm combines two real RTABLEs, rt_re and rt_im, into
a complex rtable of the form rt_re + I*rt_im. The settings of the new
rtable that is returned are determined by the RtableSettings structure
s.
List Manipulation To work with Maple lists, the following API functions
can be used. These functions are only available using the C API.
ALGEB MapleListAlloc( MKernelVector kv, M_INT num_members );

void MapleListAssign( MKernelVector kv, ALGEB list,
M_INT i, ALGEB val );
ALGEB MapleListSelect( MKernelVector kv, ALGEB list,

M_INT i );
MapleListAlloc creates a LIST dag with space for num_members ele-
ments. This list must be filled up before it can be passed back to Maple.
MapleListAssign sets the i th element of the given list to the value
val. That is, list[i] := val.
MapleListSelect returns the i th element of the given list.
Table Manipulation To work with Maple tables, the following API func-
tions can be used. These functions are only available using the C API.
ALGEB MapleTableAlloc( MKernelVector kv );
void MapleTableAssign( MKernelVector kv, ALGEB table,
ALGEB ind, ALGEB val );
ALGEB MapleTableSelect( MKernelVector kv, ALGEB table,
ALGEB ind );
void MapleTableDelete( MKernelVector kv, ALGEB table,
ALGEB ind );
M_BOOL MapleTableHasEntry( MKernelVector kv, ALGEB table,
ALGEB ind );
MapleTableAlloc creates a TABLE dag. The table is initially empty.
MapleTableAssign sets the ind element of the given table to the
value val. That is, table[ind] := val, where ind can be a NAME or an
expression sequence of numbers, or any other valid index into a Maple
table.
MapleTableSelect returns the ind element of the given table.
MapleTableDelete removes the ind element from the table.
MapleTableHasEntry queries the table to see if it contains an el-
ement at index ind. If it does, TRUE is returned; otherwise, FALSE is
returned.
Data Selection The following functions are available when using the
C API only and deal with selecting from various kinds of Maple data
structures.
ALGEB MapleSelectImaginaryPart( MKernelVector kv, ALGEB s );
ALGEB MapleSelectRealPart( MKernelVector kv, ALGEB s );
ALGEB MapleSelectIndexed( MKernelVector kv, ALGEB s, M_INT
dim, M_INT *ind );
MapleSelectImaginaryPart and MapleSelectRealPart return the
imaginary and real parts of a complex number dag, respectively.
MapleSelectIndexed returns a value from any indexable object in

Maple (list, array, set, etc.). The index is specified by filling in the ind
array with the desired index. The second parameter dim is the number of
dimensions in the array s (also the number of elements in ind).
For example, to lookup a[1,2,3], the following code could be used
(assuming arg1 points to the array a).
ALGEB val;
M_INT ind[3];
ind[0] = 1;
ind[1] = 2;
ind[2] = 3;
val = k->selectIndexed(arg1, 3, ind);
Unique Data The following function is available only in the C API.
ALGEB MapleUnique( MKernelVector kv, ALGEB s );
This function processes the given Maple expression s, and returns

the unique copy of that expression from Maple’s simpl table. For ex-
ample, if you create the number num = one-billion, then you com-
pute the number val = 2*500-million. An address comparison of num
and val will not indicate equality. After calling simplify as in num =
MapleUnique(kv,num);, both num and val will point to the same mem-
ory.
Error Handling The following functions will raise a Maple-style error

message.
void MapleRaiseError( MKernelVector kv, char *msg );

void MapleRaiseError1( MKernelVector kv, char *msg,
ALGEB arg1 );
void MapleRaiseError2( MKernelVector kv, char *msg,
ALGEB arg1, ALGEB arg2 );
The Fortran equivalent is:
SUBROUTINE maple_raise_error( kv, msg, len )
These functions display the message msg, stop execution, and return
to Maple’s input loop. A call to MapleRaiseError does not return.
The character string msg may contain wildcards of the form %N, where
N is a non-zero integer. These wildcards will be replaced by the extra ar-
gument, arg1 or arg2, before displaying the message. If %-N is specified,
then the optional argument will be displayed with st, nd, rd, or th ap-
pended to it. For example:
MapleRaiseError2(kv, "the %-1 argument, ’%2’, is not valid",

ToMapleInteger(i), args[i]);
This, if invoked, will raise the error, "the 4th argument, ’foo’, is not
valid", assuming i=4, and args[i] is set to the Maple name foo.4
The only option not allowed is %0 since the function cannot know how
many optional arguments are left to parse.
The C API also provides a mechanism for trapping errors raised by
Maple.
void* MapleTrapError( MKernelVector kv, void *(*proc)

P(( void *data )), void *data, M_BOOL *errorflag );
MapleTrapError executes the C function proc, passing it the data,

data. If an error occurs, errorflag is set to TRUE and traperror returns
immediately. If no error occurs, the result of proc(data) is returned and
errorflag is FALSE.
For example, the following code attempts to execute a Maple proce-
dure. If an error occurs, a separate branch of code is taken.
typedef struct {
MKernelVector k;
ALGEB fn, arg1, arg2;
} CallbackArgs;
void *tryCallback( void *data )

{
/* calls the maple procedure ’fn’ with arguments ’arg1’ */
/* and ’arg2’ */
return (void*)
EvalMapleProc( ((CallbackArgs*)data)->k,
((CallbackArgs*)data)->fn, 2,
((CallbackArgs*)data)->arg1,
((CallbackArgs*)data)->arg2);
4
For more information, see ?error.
void MainProc( MKernelVector k, ALGEB fn )

{
M_BOOL errorflag;
ALGEB result;
CallbackArgs a;
a.k = k;
a.fn = fn;
a.arg1 = ToMapleFloat(k,3.14);
a.arg2 = ToMapleInteger(k,44);
result = (ALGEB)MapleTrapError(k,tryCallback,&a,&errorflag);
if( errorflag ) {
/* do something */
}
else {
/* do something else */
}
}
Hardware Float Evaluation The following procedures evaluate a Maple
Procedure or statement using hardware floats.
double MapleEvalhf( MKernelVector kv, ALGEB s );
double EvalhfMapleProc( MKernelVector kv, ALGEB fn,
int nargs, double *args );
The equivalent Fortran functions are as follows.
DOUBLEPRECISION maple_evalhf( kv, s)
DOUBLEPRECISION evalhf_maple_proc( kv, fn, nargs, args )
MapleEvalhf applies evalhf to the given dag s. Then evalhf will
either evaluate an expression using hardware floats to produce a hardware
float result, or it will return the handle to an evalhfable rtable that can
be used as a parameter to EvalhfMapleProc.
EvalhfMapleProc calls the evalhf computation engine directly to
evaluate the given procedure fn without converting the hardware float
parameters to software floats. The procedure fn is a valid Maple PROC
dag, nargs is the number of parameters to pass to fn, and args is the list
of parameters. Note that args starts at 1; args[1] is the first parameter,
args[nargs] is the last, and args[0] is not used.
Setting up a callback may require the use of static local variables in

the wrapper module so that the callback will have access to the kernel
vector (unless it is passed via a data parameter that the callback receives).
The following is an example of a wrapper that uses EvalhfMapleProc to
evaluate a function that takes an hfarray and some numeric values.
#include "maplec.h"
static MKernelVector kv; /* kernel vector */

static ALGEB fn; /* function handle */
static double hfparams[HF_MAX_PARAMS+1]; /* parameters */
void callback( int N, double X, double Y[] )

{
hfparams[1] = (double)N;
hfparams[2] = X;
/* hfparams[3] is already set */
EvalhfMapleProc(kv,fn,3,hfparams);
}
/* main wrapper function called from Maple */

ALGEB test( MKernelVector k, ALGEB args )
{
/* skip arg checking for the sake of brevity */
kv = k; /* save kernel vector */

/* get the hfarray handle */
hfparams[3] = MapleEvalhf(DAG(args[1]));
fn = DAG(args[2]); /* save the function handle */
do_stuff(callback); /* start the routine that */

/* calls callback() */
return( k->toMapleNULL() );
}
In Maple, the external routine would be accessed just like any other,
except an error will be raised if the given procedure is not able to use
evalhf.
> f := proc(n,x,y) y[1] := n*sin(x); end:
> y := Vector([1,2],datatype=float[8]):
> p := define_external(’test’,MAPLE,LIB="libtest.so"):
> p(y,f):
General Evaluation The following procedures evaluate a Maple proce-

dures or statements. These routines are not available in the Fortran API.
ALGEB MapleEval( MKernelVector kv, ALGEB s );
ALGEB EvalMapleProc( MKernelVector kv, ALGEB fn, int nargs,
/* ALGEB arg1, ALGEB arg2, */ ... );
ALGEB EvalMapleStatement( MKernelVector kv, char *statement );
EvalMapleProc is a callback to Maple. The first argument fn is a
Maple PROC or FUNCTION dag, which will be evaluated with the arguments,
arg1 .. argN. For example, consider the following Maple function.
> f := proc(x) x^2; end:
If this function is passed to the external function as args[1], then

the following code would execute the given function at x := 3.14.
ALGEB a1, MapleResult;
double CResult;
a1 = ToMapleFloat(kv,3.14);
MapleResult = EvalMapleProc(kv,args[1],1,a1);
CResult = MapleToFloat64(kv,MapleResult);
EvalMapleStatement enables you to enter a single parsable Maple
statement and have it evaluated. For example, the following call will eval-
uate the integral x3 in the range x = 0..1.
ALGEB MapleResult;
double CResult;
MapleResult = EvalMapleStatement(kv,"int(x^3,x=0..1)");
CResult = mapleToFloat64(kv,MapleResult);
MapleEval evaluates a Maple expression. It is especially useful for
getting at the value of an assigned name.
Assignment to Maple Variables The following assignment functions are

available only when using the C API.
ALGEB MapleAssign( MKernelVector kv, ALGEB lhs, ALGEB rhs );
ALGEB MapleAssignIndexed( MKernelVector kv, ALGEB lhs,
M_INT dim, M_INT *ind, ALGEB rhs );
MapleAssign sets the value dag rhs to the name dag lhs. This is
equivalent to the Maple statement
> lhs := rhs;
MapleAssignIndexed sets the value rhs to the indexed variable lhs.

The second parameter dim tells the number of dimensions in the array
(or 1 if lhs is a table). The third parameter ind is a hardware array of
indices.
For example, to make the assignment a[1][2][3] = 3.14, the fol-
lowing code could be used (assuming arg1 points to the array a).
ALGEB rhs;
M_INT ind[3];
ind[0] = 1;
ind[1] = 2;
ind[3] = 3;
rhs = ToMapleFloat(kv,3.14);
MapleAssignIndexed(kv,arg1,3,ind,rhs);
User Information The MapleUserInfo command displays "msg" when

infolevel[’name’] is set to level. This command is only available in
the C API.
void MapleUserInfo( MKernelVector kv, int level, char

*name, char *msg );
Memory Management The following functions are available only when

using the C API.
void* MapleAlloc( MKernelVector kv, M_INT nbytes );

void MapleDispose( MKernelVector kv, ALGEB s );
void MapleGcAllow( MKernelVector kv, ALGEB a );
void MapleGcProtect( MKernelVector kv, ALGEB a );
MapleAlloc allocates nbytes bytes of memory and returns a pointer

to it. Garbage collection of this memory is handled by Maple. Note that
to allocate this memory, a new BINARY dag structure is created, and a
pointer to the data part of the dag is returned.
The following code snapshot might be seen in a wrapper that converts
a integer reference (a name) in Maple to C.
ALGEB arg1;
INTEGER32 *i;
i = MapleAlloc(kv,sizeof(INTEGER32));
*i = MapleToInteger32(kv,arg1);
MapleDispose frees the memory allocated to the structure s. This

should only be used on data structures created using MapleAlloc, or
those that were created externally and are guaranteed not to be pointed
to by any other Maple structure. Maple’s garbage collector will reclaim
any memory not pointed to by any other data structure, so in typical
cases it is not necessary to use MapleDispose.
MapleGcProtect prevents the algebraic a from being collected by
Maple’s garbage collector. The memory pointed to by a will not be freed
until Maple exits, or a call to MapleGcAllow is issued. Any dags that must
persist between external function invocations should be protected. This
includes any external global or static ALGEB variables that will be referred
to in a later external call. Failure to protect such a persistent variable
will lead to unexpected results if Maple’s garbage collector cleans it up
between function calls.
MapleGcAllow allows the algebraic structure a to be collected by
Maple’s garbage collector. Any algebraic structure that is not referenced
by another algebraic structure will automatically be destroyed and its
memory reclaimed. Algebraics are protected from garbage collection if
they are used somewhere (i.e., the value of a global name, part of an
array’s data, etc). The normal state of an algebraic is to have garbage
collection enabled on it.
11.4 System Integrity

The Maple kernel has no control over the quality or reliability of external
functions. If an external function performs an illegal operation, such as
accessing memory outside of its address space, that operation will most
likely result in a segmentation fault. The external routine will crash, tak-
ing Maple along with it.
If an external routine accesses memory outside of its address space but
inside Maple’s address space, the external routine will likely not crash,
but Maple will have become corrupted, resulting in inexplicable behavior
or a crash later in the Maple session. Similarly, an external routine that
deals directly with Maple data structures can corrupt Maple by misusing
the data structure manipulation facilities.
Therefore, external calling is a feature to use at your own risk.
Whether an external routine is one that you have written, or is one
supplied by a third party to which you have declared an interface (via
define_external), Maple must rely on the integrity of the external rou-
tine when it is called.
11.5 Conclusion
This chapter outlined the three methods for using compiled C or Fortran
routines in Maple. You can extend the power of Maple by using your own
or third party libraries.

0 # ,-
6 # ' 4 '
# #
* 4 4
01H! @ $$ @4 3334
.@ @) 3334 >2
> $$ >4 " .> >)
1>,>! @ $$ @4 3334
.@ @) 3334 >2
> $$ >4 " .> >)
H! @ $$ @4 3334 . !/" !.@ @4 3334 >2
> $$ >4 .>4 > ")

1> $$ "

¯

4 ' /" 1. 8 &

! " # " #

! + +44 ' !
= > = ! >
' ' ( ! (
. 0'8 '8
9
18'8E 0'*< !
<02.4 <0 8
.
4 <
1 0'*< !
8
4 5

0 .
4 9
18'8E 0'*< !
'
4
1 8
.
4
10'*< !
:#G*,3 9 '0<' .
4 9
18'8E 0'*< !

,-
1
*
.4 9
18'8E 0'*< !
5
' # ,- .///
'
¯

A Internal Representation
and Manipulation
The following is a list of the structures currently implemented in

Maple.
AN D ASSIGN BIN ARY BREAK CAT EN AT E

COM P LEX CON T ROL DCOLON DEBU G EQU AT ION
ERROR EXP SEQ F LOAT F OR F OREIGN
F U N CT ION GARBAGE HASH HASHT AB HF LOAT
IF IN EQU AT IN T N EG IN T P OS LESSEQ
LESST HAN LEXICAL LIST LOCAL M EM BER
M ODDEF M ODU LE N AM E N EXT N OT
OR P ARAM P OW ER P ROC P ROD
RAN GE RAT ION AL READ RET U RN RT ABLE
SAV E SERIES SET ST AT SEQ ST OP
ST RIN G SU M T ABLE T ABLEREF T RY
U N EV AL U SE ZP P OLY
Each of these structures, along with the constraints on its length and
contents, is described in the following sections.
A.1 Internal Organization

Maple appears to the user as an interactive calculator. The user interface
reads input, parses it, and then calls the math engine for each complete
statement encountered. Maple will read and evaluate an infinite number
of statements until a quit statement is evaluated, or the user interface is
shut down.
579
580 • Appendix A: Internal Representation and Manipulation
Maple consists of three main components: a kernel, a library, and a

user interface. The kernel and library together are known as the math
engine .
The kernel is written in the C language and is responsible for low-level
operations such as arbitrary precision arithmetic, file I/O, execution of the
Maple language, and the performance of simple mathematical operations
such as differentiation of polynomials.
Most of Maple’s mathematical knowledge is represented in the Maple
library, which is written in the Maple language. The library is stored in
an archive, and pieces of it are loaded and interpreted by the kernel on
demand.
The user interface is the part of Maple that the user sees, and is
conceptually separate from the math engine. The same math engine can
be used with different user interfaces. Usually, Maple is provided with
a graphical user interface (GUI) and a text based interface. The GUI
is more useful for interactive use, especially when working with plots or
large matrices. The textual interface is practical for batch processing, or
solving large problems where you wish to devote all the resources of your
machine to computation.
The internal functions in Maple are divided into four distinct groups:
1. Evaluators The evaluators are the main functions responsible for

evaluation. There are six types of evaluations: statements, algebraic
expressions, boolean expressions, name forming, arbitrary precision
floating-point arithmetic, and hardware floating-point arithmetic. The
user interface calls only the statement evaluator, but thereafter, there
are many interactions between evaluators. For example, the statement,
if a > 0 then b||i := 3.14/a end if
is first analyzed by the statement evaluator, which calls the Boolean

evaluator to resolve the if condition. Once completed (for example,
with a true result), the statement evaluator is invoked again to do the
assignment, for which the name-forming evaluator is invoked with the
left-hand side of the assignment, and the expression evaluator with
the right-hand side. Since the right hand side involves floating-point
values, the expression evaluator calls the arbitrary precision floating-
point evaluator.
Normally, the user will not specifically invoke any of the evaluators,
but in some circumstances, when a non-default type of evaluation
is needed, the user can directly call evalb (the Boolean evaluator),
A.1 Internal Organization • 581
evaln (the name-forming evaluator), evalf (the arbitrary precision

floating-point evaluator), or evalhf (the hardware floating-point eval-
uator).
2. Algebraic Functions These are commonly called basic functions.

Some examples are: taking derivatives (diff), dividing polynomials
(divide), finding coefficients of polynomials (coeff), series compu-
tation (series), mapping a function (map), expansion of expressions
(expand), and finding indeterminates (indets).
3. Algebraic Service Functions These functions are algebraic in

nature, but serve as subordinates of the functions in the above group.
In most cases, these functions cannot be explicitly called by the user.
Examples of such functions are the internal arithmetic packages, the
basic simplifier, and retrieval of library functions.
4. Data Structure Manipulation Functions These are like the

algebraic functions, but instead of working on mathematical objects
(polynomials, sets, etc.), they work on data structures (expression
sequences, sums, products, lists, etc.). Examples of such functions
are operand selection (op), operand substitution (subsop), searching
(has), and length determination (length),
5. General Service Functions Functions in this group are at the

lowest hierarchical level. That is, they may be called by any other
function in the system. They are general purpose, and not necessar-
ily specific to symbolic or numeric computation. Some examples are:
storage allocation and garbage collection, table manipulation, internal
I/O, and exception handling.
The flow of control need not remain internal to the Maple kernel. In
many cases, where appropriate, a decision is made to call functions written
in Maple and residing in the library. For example, many uses of the expand
function will be handled in the kernel. However, if an expansion of a sum
to a large power is required, then the internal expand will call the external
Maple library function ‘expand/bigpow‘ to resolve it. Functions such as
diff, evalf, series, and type make extensive use of this feature.
Thus, for example, the basic function diff does not know how to dif-
ferentiate any function. All of that knowledge resides in the Maple library
in procedures named ‘diff/functionName ‘. This is a fundamental fea-
ture of Maple since it permits flexibility (changing the library), personal
tailoring (by defining your own handling functions), readability (much of
Maple’s knowledge is visible at the user level), and it allows the kernel to
remain small by unloading non-essential functions to the library.
A.2 Internal Representations of Data Types

The parser and some internal functions are responsible for building all
of the data structures used internally by Maple. All of the internal data
structures have the same general format:
Header Data1 ... Datan
The header field, stored in one or more machine words, encodes the
length of the structure and its type. Additional bits are used to record
simplification status, garbage collection information, persistent store sta-
tus, and various information about specific data structures (e.g., whether
or not a for loop contains a break or next).
The length is encoded in 26 bits on 32-bit architectures, resulting in
a maximum single object size of 67, 108, 863 words (268, 435, 452 bytes,
or 256 megabytes). On 64-bit architectures, the length is stored in 32
bits, for a maximum object size of 4, 294, 967, 295 words (34, 359, 738, 360
bytes, or 32 gigabytes).
Every structure is created with its own length, and that length will not
change during the existence of the structure. Furthermore, the contents
of most data structures are never changed during execution, because it is
unpredictable how many other data structures may be referring to it, and
relying on it not to change. The normal procedure to modify a structure
is to copy it, and then to modify the copy. Structures that are no longer
used will eventually be reclaimed by the garbage collector.
The following figures describe each of the 58 structures currently im-
plemented in Maple, along with the constraints on their length and con-
tents. The 6-bit numeric value identifying the type of structure is of little
interest, so symbolic names will be used.
Logical AND
AND ∧ expr ∧ expr
Maple syntax: expr and expr

Length: 3
A.2 Internal Representations of Data Types • 583
Assignment Statement
ASSIGN ∧ name ∧ expr
Maple syntax: name := expr

Length: 3
The name entry should evaluate to an assignable object: a NAME,

FUNCTION, or TABLEREF structure.
Binary Object
BINARY data ...
Maple syntax: none

Length: arbitrary
The BINARY structure can hold any arbitrary data. It is not used
directly as a Maple object, but is used as storage for large blocks of data
inside other Maple objects (currently only RTABLEs). It is also sometimes
used as temporary storage space during various kernel operations.
Break Statement
BREAK
Maple syntax: break

Length: 1
Name Concatenation
CATENATE ∧ name ∧ expr
Maple syntax: name || expr

Length: 3
If the name entry is one of NAME, CATENATE, LOCAL, or PARAM, and if

the expr entry evaluates to an integer, NAME, or STRING, then the result
is a NAME. If the name entry is a STRING or CATENATE that will resolve to
a STRING, and if the expr entry evaluates to an integer, NAME, or STRING,
then the result is a STRING. If expr is a RANGE, then the result is to
generate an EXPSEQ of NAMEs or STRINGs.
Complex Value
COMPLEX ∧ re ∧ im
COMPLEX ∧ im
Maple syntax: Complex(re,im) or re + im * I

Length: 2 or 3
The re and im fields must point to INTPOS, INTNEG, RATIONAL, or

FLOAT structures, one of the NAMEs infinity or undefined, or a SUM
structure representing -infinity. In the length 3 case, if either re or im
is a FLOAT, the other must be a FLOAT as well.
Communications Control Structure

CONTROL ∧ integer
Maple syntax: none

Length: 2
This is an internal structure used in kernel to user-interface commu-

nication. Such a structure will never reach the user level, or even the
mathematical parts of the kernel.
Type Specification or Test
DCOLON ∧ expr ∧ type − expr
Maple syntax: expr :: typeExpr

Length: 3
This structure has three interpretations depending on the context in

which it is used. When it appears in the header of a procedure definition, it
is a typed parameter declaration. When it appears in the local section of a
procedure or on the left hand side of an assignment, it is a type assertion.
When it appears elsewhere (specifically in a conditional expression), it is
a type test.
Debug
DEBUG ∧ expr ∧ expr ...
Maple syntax: none

Length: 2 or more
This is another internal-only structure. It is used by the kernel when

printing error traceback information to transmit that information up the
call stack.
Equation or Test for Equality
EQUATION ∧ expr ∧ expr
Maple syntax: expr = expr

Length: 3
This structure has two interpretations depending on the context in

which it is used. It can be either a test for equality, or a statement of
equality (not to be confused with an assignment).
Error Statement
ERROR ∧ expr
Maple syntax: error "msg", arg, . . . arg

Length: 2
This represents the Maple error statement. The expr is either a single
expression (if only a message was specified in the error statement), or an
expression sequence (if arguments were also specified). The actual internal
tag used for the ERROR structure is MERROR, to prevent collision with a
macro defined by some C compilers.
Expression Sequence
EXPSEQ ∧ expr ∧ expr ...
Maple syntax: expr, expr, . . .

Length: 1 or more
Expression sequences are available to the user as a data structure, and

are also used to pass arguments to procedures. Effectively, procedures take
a single argument that is an expression sequence. An expression sequence
may be of length 1 (i.e., an empty sequence), which is represented by the
Maple symbol NULL, or in some contexts (such as parameters to a function
call) as nothing at all.
Floating-Point Number
FLOAT ∧ integer1 ∧ integer2 ∧ attrib − expr
Maple syntax: 1.2, 1.2e3, Float(12,34), Float(infinity)
Length: 2 (or 3 with attributes)
A floating-point number is interpreted as integer1 ∗ 10integer2 . A

floating-point number may optionally have attributes, in which case the
length of the structure will be 3, and the third word will point to a Maple
expression. This suggests that several floating-point numbers with the
same value but different attributes can exist simultaneously.
The integer2 field can optionally be one of the names undefined
or infinity, in which case the FLOAT structure represents an undefined
floating-point value (not-a-number, or NaN, in IEEE terminology), or a
floating-point infinity. When integer2 is undefined, integer1 can take
on different small integer values, allowing the existence of different NaNs.
When integer2 is infinity, integer1 must be 1 or −1.
For/While Loop Statement

FOR ∧ name ∧ f rom− ∧ by− ∧ to− ∧ cond− ∧ stat−
expr expr expr expr seq
FOR ∧ name ∧ in − expr ∧ cond − expr ∧ stat − seq

Maple syntax:
for name from fromExpr by byExpr to toExpr
while condExpr do
statSeq
od
Maple syntax:
for name in inExpr
while condExpr do
statSeq
od
Length: 7 or 5
The name follows the same rules as in ASSIGN, except that it can
also be the empty expression sequence (NULL), indicating that there is no
controlling variable for the loop.
The from-expr, by-expr, to-expr, and cond-expr entries are general

expressions. All are optional in the syntax of for loops and can thus be
filled in with default values (1, 1, NULL, and true respectively) by the
parser.
The stat-seq entry can be a single Maple statement or expression, a
STATSEQ structure, or NULL indicating an empty loop body. An additional
bit in the FOR structure’s header is used to indicate whether the stat-seq
contains any break or next statements.
Foreign Data
FOREIGN ...
Maple syntax: none

Length: 1 or more
This is similar to the BINARY structure, except that it is for use by

components of Maple outside the kernel, such as the user interface. A
FOREIGN structure is exempt from garbage collection, and it is up to the
external component to free it when it is done with it.
Function Call
FUNCTION ∧ name ∧ expr − seq ∧ attrib − expr
Maple syntax: name( exprSeq )

This structure represents a function invocation (as distinct from a

procedure definition that is represented by the PROC structure). The name
entry follows the same rules as in ASSIGN, or it may be a PROC structure.
The expr-seq entry gives the list of actual parameters, and is always an
expression sequence (possibly of length 1, indicating no parameters).
Garbage
GARBAGE ...
Maple syntax: none

Length: 1 or more
This structure is used internally by Maple’s garbage collector as a

temporary object type for free space.
Hardware Float
HFLOAT floatword
HFLOAT floatword floatword

Maple syntax: none
Length: 2 on 64-bit architectures, 3 on 32-bit architectures
This structure is used to hold a hardware floating-point value. The

one or two words (always 8 bytes) after the header hold the actual double-
precision floating-point value. HFLOAT objects are currently not available
directly to the user, but they are used internally to more efficiently trans-
fer hardware floating-point values between RTABLEs of such values, and
Maple’s I/O facilities (for example, the printf and scanf families of
functions).
If Statement
IF ∧ cond− ∧ stat− ∧ cond− ∧ stat− ... ... ∧ stat−
expr seq expr seq ... ... seq
Maple syntax:
if condExpr then
statSeq
elif condExpr then
statSeq
...
else statSeq
end if
Length: 3 or more
This structure represents the if . . . then . . . elif . . . else . . . end if

statement in Maple. If the length is even, the last entry is the body of an
else clause. The remaining entries are interpreted in pairs, where each
pair is a condition of the if or elif clause, followed by the associated
body.
Not Equal or Test for Inequality
INEQUAT ∧ expr ∧ expr
Maple syntax: expr < expr

Length: 3
This structure has two interpretations, depending on the context in

which it is used. It can be either a test for inequality or a statement of
inequality (not to be confused with an assignment).
Negative Integer
INTNEG integer integer ...
Maple syntax: −123

Length: 2 or more
This data structure represents a negative integer of arbitrary preci-

sion. See the comments below about the representation of integers.
Positive Integer
INTPOS integer integer ...
Maple syntax: 123

Length: 2 or more
This data structure represents a positive integer of arbitrary preci-

sion. Integers are represented internally in a base dependent on the host
machine. On 32-bit architectures, this base is 10, 000. On 64-bit architec-
tures, the base is 1, 000, 000, 000. The base is chosen such that the square
of the base is still representable in a machine integer. Each integer field
represents either 4 or 9 digits. The least significant digits are represented
first. For example, on a 32-bit platform, the integer 123, 456, 789, 638, 747
is represented as:
INTPOS 8747 8963 4567 123
Small integers are not represented by data structures at all. Instead of a

pointer to an INTPOS or INTNEG structure, a small integer is represented
by the bits of what would normally be a pointer. The least significant
bit is 1, which makes the value an invalid pointer (since pointers must be
word-aligned). Such an integer is called an immediate integer .
The range of integers representable in this way is −1, 073, 741, 823
to 1, 073, 741, 823 (i.e., about +/ − 109 ) on 32-bit architectures, and
−4, 611, 686, 018, 427, 387, 903 to 4, 611, 686, 018, 427, 387, 903 (i.e., about
+/ − 4 ∗ 1018 ) on 64-bit architectures. (These numbers may not seem
"small", but consider that Maple’s maximum integer magnitude is about

10268,435,448 on 32-bit architectures and 1038,654,705,646 on 64-bit architec-
tures.)
Less Than or Equal
LESSEQ ∧ expr1 ∧ expr2
Maple syntax: expr1 <= expr2, expr2 = expr1

Length: 3
This structure has two interpretations, depending on the context. It

can be interpreted as a relation (i.e., an inequation), or as a comparison
(e.g., in the condition of an if statement, or the argument to a call to
evalb). Maple does not have a greater-than-or-equal structure. Any input
of that form is turned around into a LESSEQ structure.
Less Than
LESSTHAN ∧ expr ∧ expr
Maple syntax: expr1 < expr2, expr2 expr1

Length: 3
Like the LESSEQ structure above, this structure has two interpreta-
tions, depending on the context. It can be interpreted as a relation (i.e.,
an inequation), or as a comparison (e.g., in the condition of an if state-
ment, or the argument to a call to evalb).
Maple does not have a greater-than structure. Any input of that form
is turned around into a LESS structure.
Lexically Scoped Variable within an Expression
LEXICAL integer
Maple syntax: name

Length: 2
This represents an identifier within an expression in a procedure that

is not local to that procedure, but is instead declared in a surrounding
procedure scope. The integer field identifies which lexically scoped vari-
able of the current procedure is being referred to. The integer, multiplied
by 2, is an index into the lexical-seq structure referred to by the PROC
DAG of the procedure. Specifically, |integer| * 2 - 1 is the index to

the NAME of the identifier, and |integer| * 2 is the index to a descrip-
tion (LOCAL, PARAM, or LEXICAL) relative to the surrounding scope. The
value of integer can be positive or negative. If positive, the original iden-
tifier was a local variable of a surrounding procedure; if negative, it was
a parameter of a surrounding procedure.
List
LIST ∧ expr − seq ∧ attrib − expr
Maple syntax: [ expr, expr, ... ]

The elements of the expr-seq are the elements of the list. The list can
optionally have attributes.
Local Variable within an Expression
LOCAL integer
Maple syntax: name

Length: 2
This indicates a local variable when it appears within an expression

in a procedure or module. The integer is an index into the procedure’s
local-seq. At procedure execution time, it is also an index into the internal
data structure holding the active locals on the procedure activation stack,
and holds private copies of the NAMEs of the local variables (private copies
in the sense that these NAMEs are not the same as the global NAMEs of the
same name).
Member
MEMBER ∧ module ∧ name
Maple syntax: module :- name

Length: 3
This structure represents a module member access in an expression.

MEMBER objects typically do not persist when a statement is simplified.
Instead, they are replaced by the actual member that they refer to (an
instance of a NAME).
Module Definition
MODDEF param- local- option- export- stat- desc-
...
seq seq seq seq seq seq
global- lexical- mod-

seq seq name
Maple syntax:
module modName ( )
description descSeq;
local localSeq;
export exportSeq;
global globalSeq;
option optionSeq;
statSeq
end module
Length: 10
The param-seq points to an expression sequence describing the formal

parameters of the module. Currently, Maple doesn’t support parameter-
ized modules, so this field always points to the sequence containing just
an instance of the name thismodule.
The local-seq points to an expression sequence listing the explicitly
and implicitly declared local variables. Each entry is a NAME. The explicitly
declared variables appear first. Within the module, locals are referred to
by LOCAL structures, the local variable number being the index into the
local-seq.
The export-seq points to an expression sequence listing the exported
module members. Each entry is a NAME. Within the module, exports are
referred to by LOCAL structures, the local variable number being the num-
ber of elements in the local-seq, plus the index into the export-seq.
The option-seq points to an expression sequence of options to the
module (for modules, options are the same thing as attributes). Each
entry is a NAME or EQUATION specifying an option. Typical options are
load= . . . and unload= . . .
The stat-seq field points to a single statement or a statement sequence
(STATSEQ). If the module has an empty body, this is a pointer to NULL
instead.
The desc-seq field points to an expression sequence of NAMEs or

STRINGs. These are meant to provide a brief description of what the mod-
ule does, and are displayed even when interface(verboseproc) is less
than 2.
The global-seq field points to a list of the explicitly declared global
variables in the module (those that appeared in the global statement).
This information is never used at run-time, but it is used when simplify-
ing nested modules and procedures to determine the binding of lexically
scoped identifiers (for example, an identifier on the left-hand side of an as-
signment in a nested procedure can be global if it appears in the global
statement of a surrounding context). This information is also used at
printing time, so that the global statement will contain exactly the global
identifiers that were declared in the first place.
The lexical-seq field points to an expression sequence of links to iden-
tifiers in the surrounding scope, if any. The sequence consists of pairs of
pointers. The first pointer of each pair is to the globally unique NAME of
the identifier; this is needed at simplification and printing time. The sec-
ond pointer is a pointer to a LOCAL, PARAM, or LEXICAL structure which is
understood to be relative to the surrounding scope. When a module def-
inition is evaluated, the lexical-seq is updated by replacing each of the
second pointers with a pointer to the actual object represented. The name
pointers are not touched, so that the actual identifier names are still avail-
able. The lexical-seq for a module contains entries for any surrounding-
scope identifiers used by that module or by any procedures or modules
contained within it.
The mod-name field points to the optional name of the module. If
a module name was specified when the module was declared, the name
will appear there. If no module name was specified, this field will contain
NULL.
Module Instance
MODULE ∧ export − seq ∧ mod − def ∧ local − seq
Maple syntax: none

Length: 4
Executing a module definition (MODDEF) results in a module instance.

Each local or exported member of the module is instantiated and be-
longs to that instance of the module. The export-seq field points to an
expression sequence of names of the instantiated exports (as opposed to
the global names, as stored in the module definition). The mod-def field
points back to the original module definition. The local-seq field points
to an expression sequence of names of the instantiated local variables of
the module.
Identifier
NAME ∧ assigned− ∧ attrib− characters characters ...
expr expr
Maple syntax: name

Length: 4 or more
The assigned-expr field points to the assigned value of the name. If

the name has no assigned value, this field is a null pointer (not a pointer
to NULL). The next field points to an expression sequence of attributes
of the name. If there are no attributes, this field points to the empty
expression sequence (NULL). The remaining fields contain the characters
making up the name, stored 4 or 8 per machine word (for 32-bit and 64-bit
architectures respectively). The last character is followed by a zero-byte.
Any unused bytes in the last machine word are also zero. The maximum
length of a name is 268,435,447 characters on 32-bit architectures and
34,359,738,351 characters on 64-bit architectures.
Next Statement
NEXT
Maple syntax: next

Length: 1
Logical NOT
NOT ∧ expr
Maple syntax: not expr

Length: 2
Logical OR
OR ∧ expr ∧ expr
Maple syntax: expr or expr

Length: 3
Procedure Parameter within an Expression
PARAM integer
Maple syntax: name

Length: 2
This indicates a parameter when it appears within a procedure. The

integer is an index into the procedure’s param-seq. Several special PARAM
structures exist:
PARAM 0
This represents the Maple symbol nargs, the number of arguments passed
when the procedure was called.
PARAM −1
This represents the Maple symbol args, the entire sequence of arguments
passed when the procedure was called.
PARAM −2
This represents the Maple symbol procname, referring to the currently

active procedure.
At procedure execution time, the integer (if positive) is used as an
index into the internal data structure Actvparams which is part of the
Maple procedure activation stack, and holds pointers to the values (which
are also Maple structures, of course) of the actual parameters passed to
the procedure.
Power
POWER ∧ expr ∧ expr
Maple syntax: expr∧ expr

Length: 3
This structure is used to represent a power when the exponent is
not an integer, rational, or floating-point value. When the exponent is
numeric, the POWER structure is converted to a length 3 PROD structure.
Procedure Definition
PROC ∧ param− ∧ local− ∧ option− ∧ rem− ∧ stat− ∧ desc−
...
seq seq seq table seq seq
∧ global− ∧ lexical−
seq seq
Maple syntax:
proc ( paramSeq )
description descSeq;
local localSeq;
export exportSeq;
global globalSeq;
option optionSeq;
statSeq
end proc
Length: 9
The param-seq points to an expression sequence describing the formal

parameters of the procedure. Each entry is either a NAME or a DCOLON
(which in turn contains a NAME and an expression specifying a type).
Within the procedure, parameters are referred to by PARAM structures,
the parameter number being the index into the param-seq.
The local-seq points to an expression sequence listing the explicitly
and implicitly declared local variables. Each entry is a NAME. The explicitly
declared variables appear first. Within the procedure, locals are referred
to by LOCAL structures, the local variable number being the index into
the local-seq.
The option-seq field points to an expression sequence of options to
the procedure (for procedures, options are the same thing as attributes).
Each entry is a NAME or EQUATION specifying an option. Typical options
are remember, operator, and ‘Copyright ...‘.
The rem-table field points to a hash table containing remembered
values of the procedure. Entries in the table are indexed by the procedure
arguments, and contain the resulting value. If there is no remember table,
this field contains a pointer to NULL, the empty expression sequence.
The stat-seq field points to a single statement or a statement sequence
(STATSEQ). If the procedure has an empty body, this is a pointer to NULL
instead. For each procedure that is built into the kernel, there is a wrapper
PROC that has the option builtin in its option-seq, and a single Maple
integer pointed to by its stat-seq. The integer gives the built-in function
number.
The desc-seq field points to an expression sequence of NAMEs or
STRINGs. These are meant to provide a brief description of what the pro-
cedure does, and are displayed even when interface(verboseproc) is
less than 2.
The global-seq field points to a list of the explicitly declared global
variables in the procedure (those that appeared in the global statement).
This information is never used at run-time, but it is used when simplifying
nested procedures to determine the binding of lexically scoped identifiers.
For example, an identifier on the left-hand side of an assignment in a
nested procedure can be global if it appears in the global statement of a
surrounding procedure. This information is also used at procedure print-
ing time, so that the global statement will contain exactly the same
global identifiers that were declared in the first place.
The lexical-seq field points to an expression sequence of links to iden-
tifiers in the surrounding scope, if any. The sequence consists of pairs of
pointers. The first pointer of each pair is to the globally unique NAME of
the identifier; this is needed at simplification and printing time. The sec-
ond pointer is a pointer to a LOCAL, PARAM, or LEXICAL structure which is
understood to be relative to the surrounding scope. When a procedure is
evaluated (not necessarily called), the lexical-seq is updated by replacing
each of the second pointers with a pointer to the actual object repre-
sented. The name pointers are not touched, so that the actual identifier
names are still available. The lexical-seq for a procedure contains entries
for any surrounding-scope identifiers used by that procedure or by any
procedures contained within it.
Product, Quotient, Power
PROD ∧ expr ∧ expon ∧ expr ∧ expon ... ...
Maple syntax: expr ^ expon * expr ^ expon ...

Length: 2n + 1
This structure is interpreted as pairs of factors and their numeric ex-

ponents. Rational or integer expressions to an integer power are expanded.
If there is a rational constant in the product, this constant will be moved
to the first entry by the simplifier. A simple power, such as a^2, is repre-
sented as a PROD structure. More complex powers involving non-numeric
exponents are represented as POWER structures.
Range
RANGE ∧ expr ∧ expr
Maple syntax: expr .. expr

Length: 3
Rational
RATIONAL ∧ integer ∧ pos − integer
Maple syntax: 1/2

Length: 3
This structure is one of the basic numeric objects in Maple. Note that
this is not a division operation, but only a representation for rational
numbers. Both fields must be integers (INTPOS, INTNEG or an immediate
integer) and the second must be positive.
Read Statement
READ ∧ expr
Maple syntax: read expr

Length: 2
The Maple read statement. The expression must evaluate to either a

string or symbol (STRING or NAME structure), and specifies the name of
the file to read.
Return Statement
RETURN ∧ expr − seq
Maple syntax: return "msg", arg, ... arg

Length: 2
The Maple return statement. The expression sequence is evaluated,

giving the value(s) to return.
Rectangular Table
RTABLE ∧ data ∧ maple− ∧ ind− ∧ attrib flags num-
...
type fn elems
L1 U1 ... ... LN UN P1 P2
Maple syntax: rtable(...)

Length: 2n + p, 2n + p, or 2n + p, where n is the number of di-
mensions (0 to 63), and p is 0, 1, or 2, depending on the number of Pi
parameters.
The data field points to either a block of memory (for dense and NAG-
sparse RTABLEs), or to a HASHTAB structure (for Maple-sparse RTABLEs).
The data block is either an object of type BINARY, or memory allocated
directly from the operating system’s storage manager when the block
would be too large to be allocated as a Maple data structure. If the data
block is a BINARY object, the data pointer points to the first data word,
not to the object header.
The maple-type field points to a Maple structure specifying the data
type of the elements of an RTABLE of Maple objects. If the RTABLE con-
tains hardware objects, the maple-type field points to the Maple NAME
anything.
The ind-fn pointer points to either an empty expression sequence
(NULL), or an expression sequence containing at least one indexing func-
tion and a pointer to a copy of the RTABLE structure. The copy of the
RTABLE is identical to the original, except that its ind-fn field refers to
one less indexing function (either NULL, or another expression sequence
containing at least one indexing function and a pointer to another copy
of the RTABLE with one less indexing function again).
The attrib pointer points to an expression sequence of zero or more
arbitrary attributes, which can be set by the setattribute function, and
queried by attributes.
The flags field is a bit field containing the following sub-fields:
• data type - 4 bits - indicates one of several hardware datatypes or

that a Maple data type (as specified by maple-type) is being used.
• sub type - 2 bits - indicates if the RTABLE is an Array, Matrix, or

Vector.
• storage - 4 bits - describes the storage layout (e.g. sparse, upper tri-
angular, etc.)
• order - 1 bit - indicates C or Fortran ordering of RTABLE elements.
• read only - 1 bit - indicates the RTABLE is to be read-only once created.

• foreign - 1 bit - indicates that the space pointed to by the data field
does not belong to Maple, so Maple should not garbage collect it.
• number of dimensions - 6 bits - the number of dimensions of the
RTABLE, from 0 to 63.
The num-elems field indicates the total number of elements of stor-
age allocated for the data. For a Maple-sparse RTABLE, num-elems is
not used. For a NAG-sparse RTABLE, num-elems specifies the number of
elements currently allocated, some of which might not be in use.
The remaining fields specify the upper and lower bounds of each di-
mension, and are stored directly as signed machine integers. The limits
on bounds are −2, 147, 483, 648 to 2, 147, 483, 647 for 32-bit architectures
and −9, 223, 372, 036, 854, 775, 808 to 9, 223, 372, 036, 854, 775, 807 for 64-
bit architectures. The total number of elements cannot exceed the upper
limit numbers either.
Save Statement
SAVE ∧ expr − seq
Maple syntax: save expr, expr, ...
Length: 2
The Maple save statement. The expression sequence gives a list of

names of objects to save, and either a file name or repository name in
which to save them. The file or repository name can be specified as a
NAME or STRING.
Series
SERIES ∧ expr ∧ expr integer ∧ expr integer ... ...
Maple syntax: none
Length: 2n + 2
This is the internal representation of a series in Maple. There is no

input syntax for a series; one can only arise from a computation. The first
expression has the general form x-a, where x denotes the variable of the
series used to do that expansion, and a denotes the point of expansion.
The remaining entries are interpreted as pairs of coefficients and expo-
nents. The exponents are integers, not pointers to integers or immediate
integers. The exponents appear in increasing order. A coefficient O(1) (a
function call to the function O, with parameter 1) is interpreted specially
by Maple as an order term.
Set
SET ∧ expr − seq ∧ attrib − expr
Maple syntax: {expr, expr, ...}

The entries in the set’s expression sequence are sorted in order of

increasing memory address. This is an arbitrary but consistent order,
necessary for efficiently working with sets.
Statement Sequence
STATSEQ ∧ stat ∧ stat ...
Maple syntax: stat; stat; ...

Length: 3 or more
This structure represents a sequence of two or more statements, and

can be used wherever a single statement (e.g., ASSIGN, IF, FOR) can ap-
pear. A statement sequence, containing only a single statement, is re-
placed by that statement. A statement sequence containing no statements
is replaced by the empty expression sequence (NULL). Nested STATSEQ
structures are flattened. All of the above transformations are made by
the simplifier.
Stop Maple
STOP
Maple syntax: quit, done, or stop

Length: 1
String
STRING reserved ∧ attrib − expr characters characters ...
Maple syntax: "This is a string"

Length: 4 or more
A Maple string is structurally similar to a NAME, except that it has

no assigned-value field. The attrib-expr field points to an expression
sequence of attributes of the string. If there are no attributes, this field
points to the empty expression sequence (NULL). The remaining fields

contain the characters making up the string, stored 4 or 8 per machine
word (for 32-bit and 64-bit architectures respectively). The last character
is followed by a zero-byte. Any unused bytes in the last machine word are
also zero.
The maximum length of a string is 268, 435, 447 characters on 32-bit
architectures and 34, 359, 738, 351 characters on 64-bit architectures.
Sum, Difference
SUM ∧ expr ∧ f actor ∧ expr ∧ f actor ... ...
Maple syntax: expr * factor + expr * factor ...

Length: 2n + 1
This structure is interpreted as pairs of expressions and their nu-

meric factors. Rational or integer expressions with an integer factor are
expanded and the factor replaced with 1. If there is a rational constant
in the sum, this constant will be moved to the first entry by the simpli-
fier. Simple products, such as a*2, are represented as SUMs. More complex
products involving non-numeric factors are represented as PROD struc-
tures.
Table
TABLE ∧ index − f unc ∧ array − bounds ∧ hash − tab
Maple syntax: N/A

Length: 4
This is a general table type, as created by the table and array func-
tions in Maple. The index-func will point to either a NAME or a PROC.
For general tables, the array-bounds field points to the empty expression
sequence (NULL). For arrays (not to be confused with Arrays, which are
implemented as RTABLEs), the array-bounds field refers to an expression
sequence of RANGEs of integers. The hash-tab field points to a HASHTAB
structure containing the elements.
Table Reference
TABLEREF ∧ name ∧ expr − seq ∧ attrib − expr
Maple syntax: name [ expr ]

This data structure represents a table reference, or indexed name.

The name entry follows the same rules as for ASSIGN, or it may be a
TABLE structure. (The parser will not generate a TABLEREF with a TABLE
structure for the name entry, but this can arise internally.) The expression
sequence contains the indices.
Try Statement
TRY ∧ try− ∧ catch− ∧ catch− ... ... ∧ f inal−
stat− −str stat− stat−

seq seq seq
Maple syntax:
try tryStat
catch "catchStr": catchStat
...
finally finalStat;
end try
Length: 3 or more
This structure represents a try statement, and can have an arbitrary

length, depending on how many catch blocks there are within it, and
whether or not it has a finally block. The catch-strs point to the catch
string of the corresponding catch block. If no catch string was specified,
the catch-str points to NULL. Empty catch-stat-seqs are also represented
by pointers to NULL, as is an empty (but present) finally block.
The actual internal tag used for the TRY structure is MTRY, to prevent
collision with a macro defined by some C exception handling libraries.
Unevaluated Expression
UNEVAL ∧ expr
Maple syntax: ’ expr ’

Length: 2
Use Statement
USE ∧ bindings ∧ statseq
Maple Syntax:
use bindings in
statseq
end use
Length: 3
The bindings component points to an expression sequence of equa-

tions whose left hand sides are symbols, and the statseq component points
to a sequence of statements that form the body of the use statement. The
right hand sides of the binding equations can be arbitary expressions.
The use statement introduces a new binding contour and binds the
names that appear on the left hand side of the equations in bindings. For
convenience, on input, a module ‘m’ can appear among the bindings,
and is treated as if it were the sequence e1 = m:-e1, e2 = m:-e2, ...,
where the ei are the exports of ‘m’. Within the sequence statseq of
statements, the symbols appearing on the left hand side of the equations
in bindings are bound to the corresponding right hand sides. The previous
bindings of those symbols are restored upon exit from the use statement.
Bindings are resolved during automatic simplification.
Polynomials with Integer Coefficients modulo n
ZPPOLY ∧ indet mod coef 0 coef 1 ...
ZPPOLY ∧ indet_seq mod ∧ zppoly0 ∧ zppoly1 ...
Maple Syntax: modp1( ConvertIn( expr, indet ), n );

Maple Syntax: modp2( ConvertIn( expr, indet1, indet2 ), n );
Length: degree(zppoly) +2 (for the zero polynomial)

Length: degree(zppoly) +3 (otherwise)
This is the internal representation of univariate and bivariate polyno-

mials modulo some integer. The modp1() and modp2() front ends provide
a suite of functions to work on this data structure operating in the domain
of polynomials in one or two variables with integer coefficients modulo n,
written Zn[x] or Zn[x, y], respectively. indet_seq is an expression se-
quence of the indeterminates of the polynomial (x), or (x,y). mod is the
integer modulus of the integer domain. In a univariate polynomial the
coefficients are stored in the following order.
(coef0*indet^0 + coef1*indet^1 + ... + coefi*indetî) mod n
A.3 The Use of Hashing in Maple • 605
A bivariate polynomial contains pointers to univariate ZPPOLY struc-

tures representing the coefficients of the first indeterminate.
(coef0(indet2)*indet1^0 + coef1(indet2)*indet1^1 + ...) mod n
where each coefi is a univariate polynomial in indet1 mod n.
All coefficients are stored including zero coefficients. The leading co-
efficient is always non-zero.
A.3 The Use of Hashing in Maple

An important factor in achieving Maple’s overall performance is the use
of hash table based algorithms for critical functions. Tables are used in
both simplification and evaluation, as well as for less critical functions.
For simplification, Maple keeps a single copy of each expression, or sub-
expression, during a session. This is done by keeping all objects in a
table. In procedures, the remember option specifies that the result of each
computation of the procedure is to be stored in a "remember table" asso-
ciated with the procedure. Finally, tables are available to the user as one
of Maple’s data types.
All of the table searching is done by hashing. The algorithm used is
direct chaining, except that the chains are dynamic vectors instead of
the typical linked lists. The two data structures used to implement hash
tables are HASHTAB and HASH.
Hash Table
HASHTAB ∧ hash − chain ∧ hash − chain ...
Maple syntax: none

Length: 2n + 1
This is an internal data structure with no Maple syntax equivalent. It

is used in the representation of tables within Maple. Each entry points to
a hash chain (a HASH structure), or is a null pointer if no entry has been
created in that bucket yet. The size of a HASHTAB structure depends on
the type of table and the platform, but is always a power of 2 plus one.
Hash Chain
HASH key ∧ expr key ∧ expr ... ...

Maple syntax: none

Length: 2n + 1
Each table element is stored as a pair of consecutive entries in a hash

bucket vector. The first entry of this pair is the hash key, and the second
is a pointer to a stored value. In some cases (e.g., procedure remember
tables, user defined tables), the key is also a pointer. In other cases, the
key is a hashed value (e.g., the simplification table, the symbol table).
The key cannot have the value zero (or the null pointer) since this is used
to indicate the bottom of the bucket.
The Simplification Table

By far, the most important table maintained by the Maple kernel is the
simplification table. All simplified expressions and subexpressions are
stored in the simplification table. The main purpose of this table is to en-
sure that simplified expressions have a unique instance in memory. Every
expression, which is entered into Maple or generated internally, is checked
against the simplification table and, if found, the new expression is dis-
carded and the old one is used. This task is done by the simplifier which
recursively simplifies (applies all the basic simplification rules) and checks
against the table. Garbage collection deletes the entries in the simplifi-
cation table that cannot be reached from a global name or from a "live"
local variable.
The task of checking for equivalent expressions within thousands of
subexpressions would not be feasible if it were not done with the aid of
hashing. Every expression is entered in the simplification table using its
signature as a key. The signature of an expression is a hashing function
itself, with one very important attribute: signatures of trivially equivalent
expressions are equal. For example, the signatures of the expressions a +
b + c and c + a + b are identical; the signatures of a ∗ b and b ∗ a are
also identical. If two expressions’ signatures disagree then the expressions
cannot be equal at the basic level of simplification.
Searching for an expression in the simplification table is done by:
• simplifying recursively all of its components,
• applying the basic simplification rules, and
• computing its signature and searching for this signature in the table.
If the signature is found then a full comparison is performed (taking

into account that additions and multiplications are commutative, etc.) to
A.3 The Use of Hashing in Maple • 607
verify that it is the same expression. If the expression is found, the one
in the table is used and the searched one is discarded. A full comparison
of expressions has to be performed only when there is a "collision" of
signatures.
Since simplified expressions are guaranteed to have a unique occur-
rence, it is possible to test for equality of simplified expressions using a
single pointer comparison. Unique representation of identical expressions
is a crucial ingredient to the efficiency of tables, hence also the remember
option. Also, since the relative order of objects is preserved during garbage
collection, this means that sequences of objects can be ordered by machine
address. For example, sets in Maple are represented this way. The set op-
erations union, intersection, etc. can be done in linear time by merging
sorted sequences. Sorting by machine address is also available to the user
with the sort command.
The Name Table

The simplest use of hashing in the Maple kernel is the name table. This
is a symbol table for all global names. Each key is computed from the
name’s character string and the entry is a pointer to the data structure
for the name. The name table is used to locate global names formed by
the lexical scanner or by name concatenation. It is also used by functions
that perform operations on all global names. These operations include:
1. marking for garbage collection,
2. the saving of a Maple session environment in a file, and
3. the Maple functions anames and unames which return all assigned and
unassigned global names, respectively.
Remember Tables
A remember table is a hash table in which the argument(s) to a procedure
call are stored as the table index, and the result of the procedure call is
stored as the table value. Because a simplified expression in Maple has a
unique instance in memory, the address of the arguments can be used as
the hash function. Hence, searching a remember table is very fast.
There are eight kernel functions which use remember tables: evalf,
series, divide, normal, expand, diff, readlib, and frontend. The
internal handling of the latter five is straightforward. There are some
exceptions with the first three, namely:
• evalf and series need to store some additional environment infor-
mation (’Digits’ for evalf and ’Order’ for series). Consequently, the
entries for these are extended with the precision information. If a re-
sult is requested with the same or less precision than what is stored in
the table, it is retrieved anyway and "rounded". If a result is produced
with more precision than what is stored, it is replaced in the table.
• evalf only remembers function calls (this includes named constants);

it does not remember the results of arithmetic operations.
• If the division succeeded and the divisor was a non-trivial polyno-

mial, the divide function stores the quotient in its remember table.
Otherwise nothing is stored in the remember table.
If option remember is specified together with option system, at

garbage collection time the remember table entries which refer to ex-
pressions no longer in use elsewhere in the system are removed. This
provides a relatively efficient use of remembering that will not waste stor-
age for expressions that have disappeared from the expression space.
Maple Language Arrays and Tables

Arrays and tables are provided as data types in the Maple language via
the array and table functions. An array is a table for which the compo-
nent indices must be integers lying within specified bounds. Arrays and
tables are implemented using Maple’s internal hash tables. Because of
this, sparse arrays are equally as efficient as dense arrays. A table object
consists of
1. index bounds (for arrays only),
2. a hash table of components, and
3. an indexing function.
The components of a table T are accessed using a subscript syntax

(e.g., T[a,b*cos(x)]). Since a simplified expression is guaranteed to have
a unique instance in memory, the address of the simplified index is used
as the hash key for a component. If no component exists for a given index,
then the indexed expression is returned.
The semantics of indexing into a table are described by its indexing
function. Aside from the default, general indexing, some indexing func-
tions are provided by the Maple kernel. Other indexing functions are
loaded from the library or are supplied by the user.
A.4 Portability • 609
Maple Language Rectangular Tables

Rectangular tables (as implemented by the RTABLE structure), can use a
variety of storage formats. One format, Maple-sparse, is identical to that
used in tables and arrays, namely a hash table. There is another sparse
format, NAG-sparse, which uses one vector for each dimension to record
indices, and a third vector to record the values of the entries. The majority
of RTABLE storage formats are dense, the simplest being the rectangular.
Other dense formats include upper-triangular and band, where storage is
allocated only for the upper triangle or a band of elements respectively.
To the user, rectangular tables manifest themselves as objects of type
Array, Matrix, Vector[row], and Vector[column]. Note that an Array
is not the same thing as an array.
A.4 Portability
The Maple kernel and the textual user interface are not tied to any one
operating system or hardware architecture. The Maple kernel was de-
signed to be portable to any system which supports a C compiler, a flat
address space, and a 32-bit or 64-bit word size. Some platforms on which
Maple is supported are (refer to the installation instructions for currently
supported OS versions):
Hardware Operating System

Intel P entium Based P C M icrosof t W indows
Linux
Apple P ower M acintosh M ac OS
Sun SP ARC Sun OS/Solaris
Silicon Graphics Iris IRIX
Hewlett P ackard P A − RISC HP − U X
IBM RS/6000 AIX
DEC Alpha Digital U N IX/Compaq T rue 64
The majority of the source code comprising the kernel is the same
across all platforms. Extensive use of macros and conditional compila-
tion take care of platform dependencies, such as word size, byte ordering,
storage alignment requirements, differences in hardware floating point
support, and sometimes, C compiler bugs.
The Maple library is interpreted by the Maple kernel. Therefore, other
than issues such as maximum object size, it is completely independent of
the underlying architecture.

Maple’s graphical user interface is implemented in C and C + +, and
makes use of a platform independent user interface class library to achieve
portability. There are only very few pieces of platform-specific source code.
Index
!, 116, 122, 170 ||, 115, 125

$, 115, 116, 156 {}, 121
%, 52, 116, 153, 170, 507 ^, 115, 165
evaluating, 52 ‘, 121
%%, 53
%%%, 53 abs, 34, 35
&, 100, 116 accuracy, 395, 397, 401, 467
&*, 115, 168 actual parameters, 205, 207
&*(), 168 add, 14, 195
’, 16, 120, 128, 184 algebraic functions, 581
(), 121 anames, 607
*, 115, 165 and, 115, 582
**, 115 animations, 471
+, 115, 116, 165 data structures of, 471
,, 115, 121 static form of, 472
-, 115, 116, 165 with backdrops, 474
->, 115, 123, 202 with display, 474, 477
., 115, 116, 165 appendto, 520
.., 115 approximations, 9
/, 115, 165 args, 207, 240
:, 120 arguments, 8, 205, 207, 422
::, 19, 115, 206 not passed, 542
:=, 3, 11, 115, 123 sequence of, 85, 207
;, 2, 120 arithmetic, 165
<, 115 Array, 406
<=, 115 array, 179, 608
<>, 115, 121 Arrays, 405
=, 11, 115 arrays, 90, 121, 179, 191, 405, 517
>, 115 and hardware floats, 405
>=, 115 and indexed names, 124
?, 122 evaluating, 179
@, 115, 164, 169 hardware floating-point, 406
@@, 115, 169 initializing, 90
[], 121, 159 selecting elements from, 159
#, 122 arrow, 214
\, 119, 122 arrow notation, 202, 203
|, 115 ASSERT, 385
611
612 • Index
assertions call by reference, 540

in procedures, 385 ANYTHING, 540
warnings, 387 CALL_ONLY, 540
assign, 129 RETURN_ONLY, 540
assigned, 129 Cartesian product, 89
assignment cases, 130
multiple, 458 cat, 126
assignment statements, 3, 11, 123, catch, 224
583 characters, 113
vs. equation definitions, 11 special, 114
assume, 40, 88, 104 Chebyshev polynomials, 36, 232
assumptions, 40, 88 checking results, 41
checking, 40 close, 497
on variables, 40 code, indenting, 16
atomic, 251 coeff, 34, 581
automatic simplification, 87 coeffs, 34, 35
collect, 39
backdrops, 474 COLOR, 478
BesselJ, 404 HUE, 429
Beta, 398 POLYGONS, 479
binary, 583 RGB, 433
binary search, 29 color, 426, 478
binding list, 294 adding, 482, 485
blanks, 118 color tables, 480
Boolean expressions, 177 gray scale, 482
break, 137, 239, 583 HUE, 481
breakpoints columns
explicit, 363 printing, 491, 518
removing, 363 reading, 508
removing explicit, 365 commands, 2, 3
setting, 362 long names of, 423
buffered files, 493 split lines, 2
flushing, 519 comments, 119, 215
builtin, 214 Complex, 584
by, 133 complex numbers, 9, 151, 408
default value, 134 evaluating, 152
imaginary part, 10, 151
C, 520, 531
imaginary unit, 101
and arrays, 521
norm, 10
generating, 520
real part, 10, 151
notes, 528
types of, 151
Index • 613
composition, 164, 169 saving to file, 490

repeated, 169 data structures, 25, 33, 106
computing and program design, 75
areas, 283 basic, 25
circumferences, 283 choosing, 70
computing with formulæ, 33 for animations, 471
concatenation, 125, 146, 261, 583 for plotting, 425–427, 432
conditional statements, 15, 130 in graph theory, 70
nested, 16 length, 582
constants manipulation, 581
defining numeric, 410 matrices as, 73
numeric, 144 sequences, 26
symbolic, 186 sets as, 72
constructors, 297 tables as, 73
cont, 371 data types
continuation character, 119 AND, 582
continuation of lines, 119 ASSIGN, 583
control, flow of, 581 BINARY, 583
conversions, 543 BREAK, 583
converting CATENATE, 583
expressions to C, 520 COMPLEX, 584
expressions to Fortran, 520 CONTROL, 584
expressions to LATEX, 523 DCOLON, 584
expressions to strings, 445, DEBUG, 584
480 EQUATION, 585
grids to polygons, 485 ERROR, 585
integers to string, 524 EXPSEQ, 585
meshes to polygons, 443, 455, FLOAT, 586
485 FOR, 586
strings to bytes, 524 FOREIGN, 587
strings to expressions, 97, 525 FUNCTION, 587
to formatted strings, 526 GARBAGE, 587
to PLOToptions, 437, 457 HASH, 605
CopyFile, 501, 513 HASHTAB, 605
Copyright, 214, 215, 249 HFLOAT, 588
coverage, 275 IF, 588
INEQUAT, 588
data INTNEG, 589
from other applications, 489 INTPOS, 589
reading from files, 491 LESSEQ, 590
representing in Maple, 70 LESSTHAN, 590
614 • Index
LEXICAL, 590 watchpoints, 366, 367

LIST, 591 debugging
LOCAL, 591 procedures, 349, 361
MEMBER, 591 debugopts, 275
MODDEF, 592 decimal numbers, 148
MODULE, 593 declared formal parameters, 206
NAME, 594 declaring types, 19, 233
NEXT, 594 declaring variables, 209
NOT, 594 define_external, 535
OR, 594 degree, 34
PARAM, 595 description, 242
POWER, 595 description field, 214
PROC, 596 detecting
PROD, 597 errors, 380
RANGE, 598 diff, 11, 261, 581
RATIONAL, 598 extending, 106
READ, 598 differential operator, 203
RETURN, 598 Digits, 396, 408, 409
RTABLE, 599, 609 evalhf, 399, 400
SAVE, 600 digits, number of, 396
SERIES, 600 display
SET, 601 insequence, 474
STATSEQ, 601 ditto operators, 52, 170
STOP, 601 divide, 581
STRING, 601 do, 133, 135, 137
SUM, 602 done, 601
TABLE, 602, 608 double quotes, 116
TABLEREF, 602
TRY, 603 efficiency, 1, 14, 22, 23, 30, 33,
UNEVAL, 603 46, 76, 90, 197, 211, 213,
DEBUG, 363 396, 401, 407, 411, 467,
debugger, 349, 361 468
breakpoints, 362, 363 elif, 16
clearing watchpoints, 367 embedding 2d graphics in 3d, 438,
controlling execution, 371 440, 447
explicit breakpoints, 363 encapsulation, 235, 257
invoking, 362 end, 239
numbering statements, 361 end do, 133, 135, 137
removing explicit breakpoints, end module, 239
365 end proc, 4
tutorial, 349 enumerated types, 540
Index • 615
environment variables, 53 tables, 51, 179, 228

evaluating, 53 unevaluated expressions, 185
listing, 54 evaluation, 46
naming, 54 and substitution, 198
equality, 585 Boolean, 175
error, 220, 388, 585 delayed, 16, 120, 184, 189
error checking, 18 exceptions, 52, 62
error messages, 18, 206 full, 6, 46, 84, 210, 211, 227
error returns, 219 global variables, 210
errors in interactive sessions, 46
catastrophic cancellation, 410, in procedures, 46, 47
413 last name, 6, 51, 227, 228
detecting, 380 levels of, 46, 129, 211
roundoff, 408 local variables, 210
trapping, 368, 390 numerical, 150, 396
eval, 6, 84, 179, 211, 228, 233 of special types, 64, 65
evalb, 38, 87, 175, 217, 580 of variables, 233
evalc, 152 one-level, 47, 51, 210, 211
evalf, 3, 396, 399, 581, 607 to a name, 63, 64, 128, 216
extending, 410 using hardware floats, 399
new constants, 410 using software floats, 396
new functions, 412 evaluators, 580
evalhf, 399, 581 event, numeric, 408
arrays, 405, 406 exception handling, 223
Digits, 399, 400 exceptions, 388
structured objects, 405 trapping of, 221
var, 406 untrappable, 222
evalm, 168 expand, 581
evaln, 128, 216, 581 explicit returns, 24, 218
evaluating exponent, 149
%, 52 export, 236, 244
arrays, 51, 179, 228 exported local variables, 237, 244
complex numbers, 152 exported variables
environment variables, 53 vs. local variables, 307
expressions, 143 exporting
global variables, 52 LATEX, 524
local variables, 47, 50, 51 exports, 244
matrices, 168 expression trees, 140
parameters, 49, 50, 216 expressions, 139
parsed strings, 525 Boolean, 177
procedures, 51, 228 converting from strings, 97
616 • Index
evaluating, 143 filepos, 498

internal structure of, 140 files, 493
operands of, 142 appending to, 497
reading from file, 138 binary, 494
removing parts of, 192 buffered, 493, 519
saving to file, 138 closing, 490, 497
selecting parts of, 192 creating, 490, 497
simplifying, 143 current position in, 498
unevaluated, 184 default, 495, 500
extending deleting, 499
commands, 106 descriptors of, 495, 497
convert, 459 detecting end of, 498
diff, 106 flushing, 519
evalf, 410 length of, 498
simplify, 107 opening, 490, 496, 497
type, 99 printing bytes to, 513
extensibility, 257 printing columns to, 491, 518
extension mechanism, 261 printing formatted, 490, 514
external calling, 531 printing strings to, 513
argument passing conventions, RAW, 493
538 READ, 495
array data formats, 536 reading, 138, 139
custom wrapper, 532, 550 reading bytes from, 501
direct calling, 531, 532 reading columns from, 493,
Maple-generated wrapper, 532, 508
539 reading formatted, 492
methods, 531 reading lines from, 500
other data formats, 538 reading remainder of, 501
scalar data formats, 536 redirecting default, 520
string data formats, 536 removing, 499
structured data formats, 536 saving, 138
types, 536 saving selected values, 138
scanning, 492, 502
Factor, 172 status of, 499
factorial, 170 STREAM, 493
fail returns, 225 terminal, 495
fclose, 224, 497 text, 494
feof, 498 truncating, 497
fflush, 519 unbuffered, 493
Fibonacci numbers, 22, 77 WRITE, 495
file descriptors, 495, 497
Index • 617
floating-point numbers, 148, 395, full evaluation, 6, 46, 227

586 function call, 587
n-digit machine, 396 function table, 256
accuracy of, 397 functions, 160, 202
and new constants, 410 algebraic, 581
and new functions, 412 composition of, 164, 169
bases of, 397 defining numeric, 412
digits of, 396, 397 evaluating, 161
evaluating, 150 numeric and symbolic, 414
hardware, 399, 408, 467 zeroth operand of, 162
hardware or software, 401
limitations, 407 Galois fields, 172
models of, 407 garbage collection, 213, 582, 587,
parts, 149 606, 608
precision, 407, 408 generic programming, 306, 312,
representation of zero, 408 314, 318
roundoff errors, 408 generic programs, 236
software, 396, 408, 467 global, 8, 243
values, 9 global options, 429, 442
zero, 408 global variables, 7, 87, 88, 208,
flow of control, 581 231
fopen, 496 assigning values to, 210
for, 133, 135, 137 evaluating, 52
default value, 134 in procedures, 51
from, 133 referencing, 243
in, 137 vs. local, 209
for loop, 586 graph theory, 70
foreign data, 587 graphics, programming with, 417
formal parameters, 201, 205, 206, GRID, 426
230 converting to polygons, 443
format strings, 490, 492, 502, 514 gridpoints, 461, 466
Fortran, 520, 531 group, 266
and arrays, 521
Hamiltonians, 101, 102, 106
generating, 520, 527
associativity of, 106
fprintf, 490, 514
inverse of, 104
fractions, 147
hardware float, 588
freeing resources, 224
hardware floating-point numbers,
fremove, 499
399, 407, 408, 467
from, 133
and arrays, 405
default value, 134
and structured objects, 405
fscanf, 492, 502
618 • Index
base of, 399 prompting for, 94

digits of, 399, 404 input data, 4, 5
has, 581 inputs, 8
hash tables, 76, 605 Int, 398, 399
hashing, 607 int, 227, 398, 399
hfarray, 517 integers, 118, 589
structured objects, 405 immediate, 589
histograms, 429, 434 natural, 118
negative, 589
I, 9, 101 positive, 589
identity matrix, 168 prime, 144
IEEE standard, 396, 408 signed, 118
if, 202, 588 types of, 144
if/then/else/endif, 130 integration
Im, 10, 31 by parts, 37
imaginary part, 10, 151 numerical, 397, 399
sign preservation, 408 interactive
imaginary unit, 9 input, 94
changing representation of, session, 87
152 interface, 6, 102, 152, 214, 387,
immediate integer, 589 509
immutable state, 299 indentamount, 511
implementations labelling, 511
vs. interfaces, 306 labelwidth, 153, 511
implicit scoping rules, 249 prettyprint, 511
in, 137 screenwidth, 510, 511
indets, 581 verboseproc, 511
index variables, 209 interfaces, 307
indexed names, 124 manipulation, 308
inequality, 588 vs. implementations, 306
infinite recursion, 102 interfaces and platforms, 2
infinity, 408 internal functions, 580
infolevel internal organization, 579
all, 107 internal representations of data
simplify, 107 types, 582
input, 2, 8 intersect, 115
formatted, 502 into, 371
from a file, 95 invoking the debugger, 362
from the terminal, 95 iostatus, 499
interactive, 94 iquo, 19
non-numeric, 15 irem, 19
Index • 619
is, 40, 175 merging, 194

iterations, number of, 21 of lists, 31
selecting elements from, 159
kernel, 580 load, 249
commands in, 214 local, 8, 243
supported platforms, 609 local options, 428, 429, 442, 457
kernelopts, 385 local variables, 5, 7, 50, 94, 208,
keywords, 115 218, 230, 233, 591
Klein bottle, 455 assertion failures, 222
evaluating, 210
labeling common subexpressions,
exported, 237, 244
153
outside their procedure, 86
labels, 153
referencing, 243
width of, 153
returning, 89
last name evaluation, 6, 51, 227,
vs. exported variables, 307
228
vs. global, 209
last_name_eval, 251
logical AND, 582
lasterror, 220
logical operators
lastexception, 220, 389
truth tables of, 178
LATEX, 524
loop invariants, 385
LATEX, generating code, 523
looping constructs, 13
Laurent series, 182
add, 195
length, 581
breaking out of, 137
lexical scoping, 55
for, 13, 133, 137
rules, 56, 249
map, 191
lexically scoped variable, 590
mul, 195
lexorder, 30
remove, 192, 193
library, 580
select, 192, 193
Limit, 399
selectremove, 192
limit, 399
seq, 195
limits
skipping iterations of, 137
numerical, 399
while, 19, 135
line continuation, 119
zip, 194
line-feeds, 118
looping statements, 133
LinearAlgebra, 266
lprint, 509
LinkedList, 268
list, 374 MakeIteration, 83
lists, 25, 121, 158, 591 MakeZn, 250
appending elements to, 421 Mandelbrot set, 419
empty, 158 mantissa, see significand
mapping over, 191 map, 17, 35, 191, 203, 581
620 • Index
and sequences, 65 modules, 20, 235

in procedures, 54 and types, 251
Maple Debugger, 349, 361 declarations, 242
tutorial, 349 definition, 236, 239, 243
Maple language, 111 definition syntax, 238
semantics, 112 description, 242
syntax, 111 error reporting, 242
Maple_floats, 408 exports, 298
maplemint, 393 implicit scoping rules, 249
mapping, 191, 202 lexical scoping rules, 249
over lists, 191 manipulation, 275
over sequences, 34, 65 members of, 244
over terms, 35 membership tests, 247
matching, 223 named, 240
math engine, 580 nested, 249
Matlab, 415 options, 249
Matrices, 517 parameterized, 250
reading, 518 referencing global variables,
writing, 518 243
matrices referencing local variables, 243
as data structures, 73 referring to, 240
identity, 168 types of variables in, 307
multiplying, 168 use for new packages, 266
max, 35 modulo, 171
member, 247, 288 positive representation, 171
MEMBER objects, 591 symmetric representation, 171
merging mul, 195
lists, 194 multiple assignments, 458
vectors, 194 multiplication
MESH, 426 non-commutative, 115, 168
converting to polygons, 443
messages, 298 NAG, 531
microwave circuit analysis, 450 name, 595
mint, 393 name table, 607
minus, 115 named modules, 240
minus signs, adjacent, 111 names, 116, 123
missing parameters, 216 currently assigned, 129
mod, 115, 171 indexed, 124
modeling objects, 296 of procedure, 4
module, 236, 239, 251, 592 protected, 127
moduledefinition, 251 protecting, 127
Index • 621
unassigning, 128 number of, 141

unprotecting, 127 of arithmetic operators, 165
value of, 6 of arrays, 180
with a tilde, 88 of concatenations, 146
nargs, 208, 240 of fractions, 148
negative integer, 589 of functions, 162
nested conditional statements, 16 of indexed names, 145
nested modules, 249 of integers, 144
nested procedures, 54, 82, 85, 209 of lists, 158
neutral operators, 172 of logical operators, 176
defining, 101 of procedures, 228
infix, 101 of ranges, 183
naming conventions for, 172 of relations, 174
newline character, 494 of series, 181
Newton’s method, 82, 402 of sets, 158
next, 137, 239, 594 of strings, 145
next, 371 of tables, 179
nops, 25, 154 of unevaluated expressions,
not, 116, 594 185
number of iterations, 21 zeroth, 142, 145, 162, 181
numbered parameters, 220 operator, 204, 206, 214
numbering statements, 361 operator rebinding, 294
numeric, 207 operators, 100
numeric estimate, 82 *, 102
numeric event, 408 arithmetic, 165
Numerical Algorithms Group, 531 arrow, 202
numerical constants, 144 binary, 115
numerical integration, 397, 399 concatenation, 125
numerical limits, 399 differential, 203
numerical programming, 395 ditto, 52, 170
numerical sums, 399 factorial, 170
numerics, 408 logical, 176
modulus, 171
O(xn ), 181 neutral, 100, 172
objects, 235, 297 nullary, 114
modeling, 296 relational, 173
one-level evaluation, 51, 210, 211 unary, 116
op, 142, 162, 231, 581 your own, 100, 172
open, 496 options
operands, 142 converting to PLOToptions,
negative, 160 437, 457
622 • Index
displaying of, 230 numbered, 220

global, 429, 442 passing, 205
local, 428, 429, 442, 457 sequence of, 85
passing, 422 symbolic, 39
processing, 423, 437 type checking, 67
specifying, 212 viewing the assignments, 218
options within procedures, 595
arrow, 214 parentheses, 202
builtin, 214 parse, 97, 525
Copyright, 214 partition, 94
operator, 214 pipes, 495
remember, 76, 212 placeholders, 220
system, 78, 212, 213 platforms
or, 115, 594 and interfaces, 2
order of evaluation, 48 plot, 11, 31, 225
organization plot drivers, 427
internal, 579 plotting, 417, 421
outfrom, 371 animations, 471
output AXESSTYLE, 432
controlling, 509 COLOR, 429
rounding, 490 colors, 445
suppressing, 6 CURVES, 427, 428, 432, 434
data structures, 425–427, 429,
package, 249 432
packages, 231, 235, 266 formulæ, 418, 424
exports, 266 functions, 418, 424, 458
in the standard library, 266 GRID, 435
table-based, 266 lists of lists, 31
use modules for, 266 MESH, 435
using interactively, 267 non-numeric values, 428, 457,
parameters, 8, 47 461, 467
actual, 202, 205, 207 numerical approximations, 395
assigning values to, 215 options, 419, 420
checking types of, 18 POINTS, 428, 434
declared, 206 POLYGONS, 428, 433, 434, 439
evaluating, 49, 216 roots of polynomials, 31
extra, 205 SCALING, 439
formal, 201, 202, 205, 206, STYLE, 433
230 TEXT, 428, 434, 445
missing, 205, 216 undefined values, 457, 461,
number of, 205, 208 467
Index • 623
with plottools, 444 adding descriptions to, 214

polygon meshes, 442, 451 and %, 52
cutting faces out of, 451 and environment variables, 53
stellating, 453 and floating-point input, 9
POLYGONS anonymous, 203
COLOR, 479 arguments of, 207
converting from grids or meshes, assertions, 385
443 bodies of, 5, 201
convex, 439, 457 body of, 231
polynomials, 33 breakpoints, 362, 363
Chebyshev, 232 call formats, 540
coefficients of, 33 calling, 5
definition of, 33 classes of, 214
degree of, 34 clearing watchpoints, 367
expanding, 37 controlling execution, 371
explicit, 39 debugging, 349, 361
height of, 34 defining, 4, 5, 201, 203, 596
plotting roots of, 31 dispatching, 287
symbolic, 39 displaying, 6
portablility, 609 displaying description, 230
positive integer, 589 displaying of, 230
powers, 595, 597 ditto operators in, 52
precision, 410 documenting, 215
floating-point numbers, 407, ending, 4
408 evaluating, 6, 228
print, 404, 510 evaluating in, 46, 47
printf, 514 exceptions, 388
printing, 509, 510 executing, 5
bytes, 513 execution details, 107
columns, 518 explicit breakpoints, 363
formatted, 490, 514 in the kernel, 214
library procedures, 6 invoking, 5, 202, 204
of procedures, 214, 215 loading, 231
strange expressions, 87 modular, 20
strings, 513 name of, 4, 230
to files, 490 naming, 202
printlevel, 380 nested, 54, 58, 82, 85, 209
priority queue, 300 numbering statements, 361
proc, 4, 201, 236 operands of, 228, 229
procedures, 3 options of, 212
actual parameters of, 202 parameters of, 201
624 • Index
parameters within, 595 numerical, 395

passing information, 89 with color, 478
passing input to, 94 with graphics, 417
printing of, 214, 215 with plot structures, 436
recursive, 22, 213 prompt, 2
reliability of, 45 protect, 127
remembering values, 76 protected names, 127
removing explicit breakpoints, Puisseux series, 182
365 punctuation marks, 120
returning unevaluated, 15
saving, 231 Quaternions, 101, 102, 106
side effects of, 8 associativity of, 106
simple, 3 inverse of, 104
simplifying, 204, 214 quaternions, 264
structure of, 229 quick-sort algorithm, 56
syntax, 393 quit, 579, 601
that return unevaluated, 225 quit, 371
tracing, 380, 382, 384 quotient field, 318
type of, 228 quotients, 597
unnamed, 203
raising exceptions, 388
values returned by, 5, 14, 24,
rand, 82
202, 215
random distribution, 60
variables in, 208
random numbers
watchpoints, 366, 367
generating, 59
with error returns, 219
randpoly, 32
with explicit returns, 218
range, 598
writing of, 231
ranges, 182
processing options, 423
rational, 598
procname, 240
Re, 10, 31
product, 197
read, 139, 231, 598
products, 195, 597
readability, 119
efficiency, 197
of code, 55
explicit, 195
readbytes, 501
symbolic, 197
readdata, 493, 508
program optimization, 204
reading
programming
bytes from files, 501
constructs, 11, 15
columns, 508
efficiency, 213
data, 491
efficiently, 22, 23, 30, 33
from default, 500
generic, 236, 306, 312, 314,
lines from files, 500
318
Index • 625
remainder of file, 501 rounding, 397

statements, 507 roundoff errors, 408, 409
readline, 95, 498, 500 catastrophic cancellation, 410,
ReadRows, 506 413
readstat, 95, 525 IEEE standard, 409
real part, 10, 151 increasing precision, 410
records, 263 similar magnitudes, 409
instantiating, 263 rtable, 599, 609
representing quaternions, 264 rtables, 121, 191, 405, 514, 517,
types, 264 518
rectangular tables, 609 and Matrices, 121, 168
recursive programming, 22, 38, and Vectors, 121
213
REF, 544 samples directory, 238
reference save, 231, 600
call by, 540 saving procedures, 231
remember, 212 protect, 502
remember tables, 23, 103, 212, scanning
229, 412, 607 files, 492, 502
adding entries to, 77 strings, 526
efficiency of, 76, 78 scoping rules, 55
options, 77 searching, 606
removing, 78 searching lists, 29
removing entries from, 78, 213 select, 192
viewing entries of, 76 selection operation, 159
remove, 192 selection statements, 130
repetition statements, 133 selectremove, 192
default values, 134 semantics, 201
omitting clauses from, 133 semicolon, 3, 5
reserved words, 115, 117 seq, 28, 155, 195
return, 239, 371 sequences, 26, 154, 195
return statement, 24, 218, 598 creating, 155, 156, 195
returns, 5, 118 efficiency, 197
error, 219 empty, 154
explicit, 24, 218 flattening of, 27, 154
fail, 225 mapping over, 65
unevaluated, 15 number of elements in, 155
RGB, 426 of arguments, 205, 207, 422
root finding, 82 of statements, 601
rootplot, 32 symbolic, 156, 197
rotating plots, 444 type of, 65
626 • Index
series, 180 sorting, 56

creating, 180 sprintf, 526
internal structure of, 181 sscanf, 526
Laurent, 182 step, 371
Puisseux, 182 stop, 601
zeroth operand of, 181 stopat, 362, 371
series, 581, 607 stoperror, 367
sets, 27, 87, 158 stopwhen, 366
as data structures, 72 strings, 145, 601
empty, 158 and double quotes, 116
order of elements in, 159 converting to expressions, 97
selecting elements from, 159 iterations over, 118
shadows, 446 length of, 116
Shapes, 249, 282 parsing, 97
object-oriented approach, 304 with backslash, 117
shift, 85 with double quote, 116
showstat, 361, 362 structured types, 66, 187
showstop, 377 submodules, 249
side effects, 8 subs, 197
sign subscripts, 124, 159
of zero, 408 subsop, 581
significand, 149 substitution, 197
simplification algebraic, 199
of expressions, 143 evaluation rules, 198
simplification table, 606 subtasks, 20
simplify, 41 Sum, 399
extending, 107 sum, 197, 399
sin, 227 sums, 195, 602
single quotes, 16, 184 efficiency, 197
Smith charts, 450 explicit, 195
software floating-point numbers, numerical, 399
396, 408, 467 symbolic, 197
accuracy of, 397 suppressing
base of, 397 of output, 6
digits of, 396, 397 symbolic evaluation, 399
software floats, 407 symbolic computations, 33
solutions symbolic polynomials, 39
analytical, 395 symbolic programming, 33
numerical, 395 symbolic sequences, 156
solve, 155 symbolic transformations, 17
sort, 607 syntax, 201
Index • 627
checking procedures, 393 tokens, 114

declaring variables, 208 trace, 249, 380, 382
error statement, 219 tracelast, 380, 384
for variables, 11 tracing
mapping, 202 procedures, 380, 382, 384
procedure description, 214 transforming plots, 451
return statement, 218 traperror, 389
specifying options, 212 trapping
try statement, 221 errors, 390
type declaration, 206 Truchet tilings, 449
while loop, 19 truth tables, 178
syntax errors, 111 try, 223, 368, 388, 390, 404, 603
in files, 113 try statement, 221, 223
system type
integrity, 574 record, 264
system, 213 type, 581
extending, 99
table, 178, 608 type checking, automatic, 67
table references, 602 typematch, 107
tables, 121, 178, 191, 602 types, 19, 140, 141
as data structures, 73 and modules, 251
creating, 178 checking, 18, 66, 67, 94, 99,
elements of, 179 206, 221, 422
evaluating, 179 declaration of, 233
indexing function, 179 declaring, 19
selecting elements from, 159 defining new, 99, 102
tabs, 118 enumerated, 540
taylor, 180 matching, 68, 430
terminators, 2, 96, 525 of arithmetic operators, 165
excluding, 5 of arrays, 179
text files of complex numbers, 151
reading, 139 of concatenations, 146
thismodule, 240 of constants, 187
tilde, 41, 88 of expression sequences, 65,
tilings, 448 154
Truchet, 449 of factorials, 170
time, 22 of floating-point numbers, 148
timelimit, 222 of fractions, 147
to, 133 of function calls, 160
default value, 134 of indexed names, 145
token separators, 118 of integers, 144
628 • Index
of lists, 158 environment, 53

of logical operators, 176 evaluation of, 233
of procedures, 228 exported vs. local, 307
of ranges, 182 global, 7, 87, 88, 208, 231
of relations, 173 identifying, 87
of sequences, 65, 154 interface, 6
of series, 180 lexically scoped, 590
of sets, 158 local, 5, 7, 86, 94, 208, 218,
of strings, 145 230, 233, 591
of tables, 179 scope of, 7, 55
of unevaluates expressions, 184 unassigning, 89, 128
set of, 67 undeclared, 56
structured, 66, 99, 187 vector fields, 456
with special evaluation, 64, Vectors, 517, 518
65 read, 518
typesetting, 523 write, 518
vectors
unapply, 83 merging, 194
unassigning, 128 verboseproc, 6, 214
unbuffered files, 493
undefined, 408 WARNING, 387
underscores, 116 warnlevel, 387
unevaluated expressions, 184, 401, watchpoints
603 clearing, 367
unevaluated returns, 11, 15 setting, 366, 367
uniform distribution, 59 where, 374
union, 115 while, 135
unprotect, 127 default value, 134
unsigned floats, 148 with, 231
unstopat, 363, 365 wrapper, 532
unstopwhen, 367 custom, 532, 550
untrace, 384 Maple-generated, 532, 539
use, 238, 291, 603 writebytes, 513
user input, 94, 97 writedata, 491, 518
user interface, 580 writeline, 513
userinfo, 107 writeto, 520
value, of name, 6 zero

variables floating-point representation,
classes of, 8 408
declaring, 209 sign of, 408
Index • 629
zip, 194
ZPPOLY, 604
630 • Index

Maple7prguide PDF

Uploaded by

Copyright:

Available Formats

Maple7prguide PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Maple7prguide PDF

Uploaded by

Copyright:

Available Formats

Maple 7

M. B. Monagan K. O. Geddes K. M. Heal

Waterloo Maple Inc.

Maple and Maple V are registered trademarks of Waterloo Maple Inc.

The use of general descriptive names, trade names, trademarks, etc.,

4 The Maple Language 111

4.6 Substitution . . . . . . . . . . . . . . . . . . . . . . . . . . 197

6 Programming with Modules 235

Lexical Scoping Rules . . . . . . . . . . . . . . . . . . . . 249

7 Debugging Maple Programs 349

8 Numerical Programming in Maple 395

9 Programming with Maple Graphics 417

Creating A Chess Board Plot . . . . . . . . . . . . . . . . 485

10 Input and Output 489

10.9 Notes to C Programmers . . . . . . . . . . . . . . . . . . . 528

11 Using Compiled Code in Maple 531

A Internal Representation and Manipulation 579

Error Statement . . . . . . . . . . . . . . . . . . . . . . . 585

Table Reference . . . . . . . . . . . . . . . . . . . . . . . . 602

important differences between Maple and traditional computer languages,

1.1 Getting Started

Ordinarily, you complete the command with a semicolon, then press

A Maple program is essentially just a prearranged group of commands

half := proc(x) evalf(1/2 ∗ x) end proc

> half(1) + half(2);

Merely enclosing the command evalf(x/2); between a proc(. . . )

The procedure needs no input.

• The name of this program (procedure) is f.

• The procedure definition starts with proc(). The empty parenthesis

• Semicolons separate the individual commands that make up the pro-

• The procedure definition that echoes as the value of the name f is

• The local a; statement declares a as a local variable. This means

The execution of a procedure is also referred to as an invocation or

Sometimes you may find it necessary to examine the body of a pro-

Locals and Globals

Next, define two nearly identical procedures: g, explicitly using b as

Therefore, the value of the global variable b is still 2. The procedure

The effect of using the procedure h (which uses global variables) is

h changes the global variable b, so it is no longer 2. When you invoke

If you do not indicate whether a variable used inside a procedure is

Inputs, Parameters, Arguments

Maple considers floating-point values to be approximations, rather

In addition to support for exact and floating-point approximate num-

> k(2 + 3*I, %);

> k(1.362, 5*I);

√ Suppose you want to write a procedure which calculates the norm,

abnorm := proc(a, b) sqrt(a2 + b2 ) end proc

Now abnorm can calculate the norm of 2 + 3i.

> abnorm(2, 3);

znorm := proc(z) sqrt(<(z)2 + =(z)2 ) end proc

If you do not specify enough information for Maple to calculate the

as complex numbers. If they were real numbers, then <(x + i y) would

Many Maple commands return unevaluated in such cases. Thus, you

1.2 Basic Programming Constructs

The Assignment Statement

Find the derivative of y with respect to x.

Plot y and yp together.

–1 –0.8–0.6–0.4–0.2 0.2 0.4 0.6 0.8 1

The following procedure combines this sequence of steps.