Scribd Logo
Upload

Welcome to Scribd!

  • 17 views

    Homework #5

    Uploaded by

    M shayan Javed
    This document describes Homework #5 problems for a numerical computing course. Problem 1 involves solving linear systems using Gaussian elimination, back substitution, and LU factorization. Problem 2 analyzes the arithmetic operations needed to solve multiple linear systems with the same matrix. Problem 3 compares directly inverting a matrix to Gaussian elimination. Problem 4 develops an LU factorization method for staircase matrices and implements it in MATLAB, counting operations.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Homework #5

    Uploaded by

    M shayan Javed
    17 views2 pages
    This document describes Homework #5 problems for a numerical computing course. Problem 1 involves solving linear systems using Gaussian elimination, back substitution, and LU factorization. Problem 2 analyzes the arithmetic operations needed to solve multiple linear systems with the same matrix. Problem 3 compares directly inverting a matrix to Gaussian elimination. Problem 4 develops an LU factorization method for staircase matrices and implements it in MATLAB, counting operations.

    Original Title

    Attachment_1627735769 (1)

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document describes Homework #5 problems for a numerical computing course. Problem 1 involves solving linear systems using Gaussian elimination, back substitution, and LU factorization. Problem 2 analyzes the arithmetic operations needed to solve multiple linear systems with the same matrix. Problem 3 compares directly inverting a matrix to Gaussian elimination. Problem 4 develops an LU factorization method for staircase matrices and implements it in MATLAB, counting operations.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    17 views2 pages

    Homework #5

    Uploaded by

    M shayan Javed
    This document describes Homework #5 problems for a numerical computing course. Problem 1 involves solving linear systems using Gaussian elimination, back substitution, and LU factorization. Problem 2 analyzes the arithmetic operations needed to solve multiple linear systems with the same matrix. Problem 3 compares directly inverting a matrix to Gaussian elimination. Problem 4 develops an LU factorization method for staircase matrices and implements it in MATLAB, counting operations.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    C&EE 103 Applied Numerical Computing and Modeling Summer 2021

    Homework #5
    Due at 11:59 pm on Wednesday, July 28, 2021

    Problem 1. Find the solution x to the linear system Ax D b with


    2 3 0 1 0 1
    1 2 4 3 x1 1
    6 8 57 B C B 3C
    AD6 7 ; x D Bx2 C ; and b D B C:
    6 0
    40 1 2 0 5 @ x3 A @ 7A
    2 0 6 2 x4 2

    a) Use Gaussian elimination to find the upper triangular matrix U , and use back substitution to find x.

    b) Use Gaussian elimination to find the inverse matrix A 1, and find x from x D A 1 b.

    c) Use the LU -factorization of A to find x from Lg D b and Ux D g.

    For each case, write down each elimination step (and/or the (forward and) back substitution) to find the
    solution x by hand calculations.
    Problem 2. In engineering problems, it is often the case that m linear systems Axi D bi , with the same
    n ⇥ n system matrix A, need to be solved for all i 2 f1; 2; 3; :::; mg to find, e.g., different displacements
    that are represented by xi of the same structure that is represented by A subjected to different load cases
    that are represented by bi . Give an operations count on how many additions and subtractions (AS) and
    multiplications and divisions (MD) are required to solve these m linear systems when Gaussian elimination
    is employed. How many arithmetic operations are required in total for n D 100000 and m D 10?
    Problem 3. Consider solving the linear system Ax D b, and assume that the n ⇥ n inverse system matrix
    A 1 is known. How many arithmetic operations are required to find x from x D A 1 b? Moreover,
    how many operations are required for n D 100, and how many operations are required when Gaussian
    elimination is employed for the same n? Compare the results, and give a brief statement.
    Problem 4. Consider the following 5 ⇥ 5 variant of a staircase matrix:
    2 3
    b1 c1
    6a2 b2 7
    6 7
    AD6 6 7:
    a3 b3 c3 7
    4 a4 b4 5
    a5 b5

    To factorize A as A D LU , consider the general forms of L and U , which are in this case given as
    2 3 2 3
    1 ˇ1 c1
    6˛2 1 7 6 ˇ2 7
    6 7 6 7
    6 7 6 7:
    LD6 ˛3 1 7 and U D 6 ˇ3 c3 7
    4 ˛4 1 5 4 ˇ4 5
    ˛5 1 ˇ5

    1
    C&EE 103 Applied Numerical Computing and Modeling Summer 2021

    a) Derive general formulas to determine the coefficients of both L and U for the case that A is a general
    n ⇥ n staircase matrix of the form given above.

    b) Use the result found in Part a) to write a M ATLAB script based on the LU -factorization of A that is
    able to solve the two linear systems Ax1 D f1 and Ax2 D f2 with user-defined staircase matrix A
    and user-defined right-hand side vectors f1 and f2 . Apply this M ATLAB script to the case in which
    A, f1 , and f2 are defined as follows:
    2 3 0 1 0 1
    3 5 5 3
    64 6 7 B 1C B9C
    6 7 B C B C
    6 9 1 7 7 B7C B4C
    AD6 6 7 ; f1 D B C ; and f2 D B
    B C C
    7 B 6C :
    6 2 5 7 B C3 B C
    4 8 7 15 @ 8A @5A
    6 3 4 2

    c) Give an operations count for all additions and subtractions (AS) as well as multiplications and divi-
    sions (MD) required to solve the two nth-order linear systems Ax1 D f1 and Ax2 D f2 based on
    the LU -factorization of A given above. How many operations are required to solve both linear sys-
    tems with n D 3000000? Compare the number of total operations with the number of total operations
    required to solve the same two linear systems using Gaussian elimination, and give a brief statement.

    d) Extend your M ATLAB script obtained in Part b) to include an operations counter into your script,
    and output the result of total operations and the result split into two parts according to additions and
    subtractions (AS) as well as multiplications and divisions (MD). Furthermore, output the respective
    results obtained from Gaussian elimination, and compute the factor by which Gaussian elimination
    is computationally more expensive (in terms of the counted operations) than the LU -factorization
    derived above.

    Hint: a M ATLAB template will be provided on CCLE.

    Turn in your M ATLAB script for this problem. Remember to copy all your files into a folder labeled Last-
    nameFirstnameHW5, zip that folder, and upload it to CCLE.

    You might also like