1.

Chapter 1

Trigonometry

Mathematics is the language of physics, so we must all

have a certain fluency. The first order of business is to
remind ourselves of some basic relations from trigonometry.

opposite
sin θ = ,
hypotenuse
adjacent
cos θ = ,
hypotenuse
sin ! opposite
tan θ = = .
cos ! adjacent

For right now just think of a vector as something with

both a magnitude and a direction:

!
Vector notation: A = (A 1, A 2 ) . Vector addition:

! ! ! !
( A + B )1 = A1 + B1 ; ( A + B )2 = A2 + B2 .
1.2

Unit Vectors:

^
e3

^
e2 2
^1
e

1
These are handy guys which point along the 1,2 or 3 directions
with unit magnitude. We always choose a right-handed
coordinate system in this class.

Matrices

A matrix is a collection of entries which can be arranged

in row/column form:
columns

! A 11 A 12 A13 $
rows # A 21 A 22 A 23 &
" A 31 A 32 A 33 %

A single generalized matrix element is denoted:

Aij

row column

Addition of matrices (number of rows and columns the same

for both A and B matrices):
 1.3

! A 11 A 12 # ! B11 B12 # ! A 11 + B11 A 12 + B12 #

" A 21 A 22 $ + " B21 B22 $ = " A 21 + B21 A 22 + B22 $ . (1.1)

In more abstract language ("index notation") this is just

(A + B)ij = Aij + Bij, (1.2)

where i and j are taking on all possible values independently.

In the above equation i and j are said to be "free" indices.
The free indices on one side of an equality must always be the
same on the other side.
Multiplication of matrices. (Here we only require that
the number of columns of A equal the number of rows of B):

! A 11 A 12 # ! B1 # !# A 11 B1 + A12 B2 $
" A 21 A 22 $ " B2 $ = . (1.3)
" A 21 B1 + A 22B2 %

Another example:

! A 11 A 12 # ! B11 B12 # ! A 11 B11 + A 12B21 A11 B12 + A 12B22 #

" A 21 A 22 $ " B21 B22 $ = " A 21 B11 + A 22B21 A 21 B12 + A 22B22 $ . (1.4)

Notice that the result has the same number of rows as A and
the same number of columns as B. In index language, these two
examples can be written much more compactly as:

2
(AB)i = ! Aik Bk (1.5)
k=1

"dummy" index, k
2
(AB)ij = ! Aik Bkj (1.6)
k=1

"dummy" index, k

Note that dummy indices are ones which are summed over all of
their values. Unlike free indices, which must be the same on
1.4

both sides of an equation, dummy indices can appear on either

side. Also notice that dummy indices always appear twice on a
given side of an equation. These rules trip up many beginning
students of mechanics.
For reference, here is a summary of the understood "index
jockey" rules for index manipulations:

1. "Dummy" indices are those which are summed. Each such index
always appears exactly twice. One can interpret this sum as
matrix multiplication only if the indices can be placed
directly next to each other. Separate summation symbols must
be used for independent summations.

2. In general, one can not change the order of indices on an

object, such as Aij. (Occasionally one knows effect of
interchanging indices; see later comments on symmetric and
antisymmetric matrices.)

3. Free indices are those that are unsummed. In general, each

free index appears once on both sides of a given equation.

Identity matrix (3 x 3 context):

! 1 0 0$
# 0 1 0&
" 0 0 1%

The identity matrix is often simply written as the number "1",

or is absent altogether in contexts where its presence is
unambiguous. (Physicists must learn to read behind the lines
for the meaning!)
We will need three additional matrix operations.

1. Inverse: AA-1 = 1. (1.7)

The "1" on the right hand side here means the identity matrix.
A Theorem from linear algebra establishes that AA-1 = 1
 1.5

implies A-1A = 1. (Can you prove this ?) Finding A-1 in general

is fairly complicated. For most of the matrices we will
encounter, finding A-1 will be easy. (I'm thinking of rotation
matrices, which will follow shortly.) Notice:

(AB)-1(AB) = 1,
(AB)-1A = B-1,
=> (AB)-1 = B-1A-1. (1.8)

2. Transpose: T = A .
Aij (1.9)
ji

Examples of the transpose operation:

! B1 $
B = # , BT = (B1 B2)
" B2 %
"column matrix" "row matrix"

!A
A = " A 11
A 12 # T ! A 11 A 21 #
21 A 22 $ , A = " A 12 A 22 $ .

Also

(AB) Tij = (AB)ji ,

= ! Ajk Bki ,
k
= ! T AT ,
Bik kj
k
T T
= (B A )ij ,
T T T (1.10)
=> (AB) = B A .

3. Determinant:

! A11 A 12 $
2x2 case: det # = A11A22 - A12A21. (1.11)
" A 21 A 22 %

A11 A 12
also written as
A21 A 22
1.6

! A 11 A 12 A13 $
3x3 case: det # A 21 A 22 A 23 & = A11 (A22A33 - A23A32)
" A 31 A 32 A 33 %
- A12(A21A33 - A23A31) + A13(A21A32 - A22A31). (1.12)

Note that det(AB) = (det A)(det B) and that det AT = det A and
det( A !1 )=1/det(A). We'll find a more elegant definition of the
determinant later. An important point to realize is that the
inverse of a matrix, A, exists only if det(A) is not zero.
An important point about linear algebra will also be
called upon in later chapters. A system of linear (only x1,2,3
appear, never (x1,2,3)2 or higher powers) homogeneous (the
right hand side of the following equations are zero)
equations,

!# A 11x1 + A 12x2 + A 13 x3 = 0,
" A21 x1 + A 22x 2 + A 23 x3 = 0, (1.13)
#$ A x + A x + A x = 0,
31 1 32 2 33 3

has a nontrivial (x1,2,3 are not all zero!) solution if and

only if

! A 11 A 12 A13 $
det # A 21 A 22 A 23 & = 0. (1.14)
" A 31 A 32 A 33 %

One often encounters matrices which are said to be

symmetric or antisymmetric. A symmetric matrix is one for
which

AT = A or Aji = Aij .

An antisymmetric matrix has

AT = -A or Aji = -Aij.
 1.7

Matrices are not generally symmetric or antisymmetric, but

such combinations can always be constructed. For example,

C = A + AT,

is a matrix combination which is symmetric (Cij = Cji) even

though A is not itself. Likewise

D = A - AT,

is antisymmetric (Dij = -Dji).

Orthogonal Transformations

Let us study a little about the above mentioned rotation

matrices. Special case:
2

P P at (x 1 ,0)
(old coordinates)
x1 1

Let's do a "passive rotation," where we rotate the axes and

not the point P. (Define θ >0 for counterclockwise rotations.)
1.8

2
1'
2'
-
x1

! P
1 (! >0 here)
x1

x2

P at ( x 1, x 2) = (x1cosθ, -x1sinθ)
!
= (x1cosθ, x1cos(θ + )) (new coordinates)
2

Another special case:

2

x2 P' 1'
2'

= =
x2 x1
!
1

P' at(0,x2) (old coordinates)

!
P' at ( x 1, x 2) = (x2sinθ, x2cosθ) =(x2cos( - θ), x2cosθ) (new)
2
 1.9

Let's now put these two special cases together into the
general case of rotation in a plane:
2
x2 P"
P'
P" at (x1 ,x 2 )
x
1
1 P P'
P
= (x 1,0) + (0,x 2 )
(old coordinates)

Now characterize the point P" with respect to rotated axes:

2
2' P" 1'

x'1
x'
2 !
1

P P'

P" at ( x'1 , x'2 ) = ( x 1, x 2) + ( x 1, x 2)

! !
= (x1cosθ, x1cos(θ + ))+(x2cos( - θ), x2cosθ) (new coordinates)
2 2
Therefore, comparing the old and new descriptions, we have

"
x'1 = x1 cos ! + x 2 cos( # !),$&
2
" % (1.15)
x'2 = x1 cos( + !) + x 2 cos !.&
2 '

To simplify the notation, introduce the concept of direction

cosines:
1.10

cos(x 'i ,xj) = cosine of the angle between the x'

i and xj axes.

start measurement of angle

The angle may be measured in a clockwise or counterclockwise

sense but must be consistent. We will adopt counterclockwise
as in the above examples. For the above rotation,

$cos(x'1,x 1) = cos !,
&&cos(x'1,x 2) " "
= cos(2" # + !) = cos( # !),
2 2
% "
'
&cos(x2,x 1) = cos( + !),
2
&'cos(x' ,x ) = cos !.
2 2

So, our general rotation may be written

!x' = x cos(x' ,x ) + x cos(x' ,x ),

" 1' 1 1 1 2 1 2 (1.16)
# 2
x = x 2 cos(x '
,x
2 1 ) + x 2 cos(x '
2,x 2).

In three dimensions, we have

!# x'1 = x1 cos(x'1, x1) + x2 cos(x'1, x2) + x3 cos(x'1, x3),

"x'2 = x1 cos(x'2, x1) + x2 cos(x'2, x2) + x3 cos(x'2, x3), (1.17)
#$x' = x cos(x' , x ) + x cos(x' ,x ) + x cos(x', x ).
3 1 3 1 2 3 2 3 3 3

I wrote this out explicitly so that you would notice a

pattern in the indices which indicates the above can be
written in matrix language. Let us define

λij ≡ cos(x 'i ,xj). (1.18)

If we arbitrarily choose to represent the position (x1,x2,x3)

as a column matrix, the above relationship can be written,
 1.11

! x'1 $ " ! 11 !12 ! 13 % ! x1 $

# x2 & = $ ! 21
' ! 22 ! 23 ' # x2 & . (1.19)
" x'3 % # ! 31 ! 32 ! 33 & " x3 %

or more abstractly

x' = λx, (1.20)

where it is understood x' and x are column matrices and λ is a

3x3 matrix. The easiest way to see this is just to do the
matrix multiplication on the right-hand side to see that it
reproduces the rotation.
Written out in explicit index notation, the above
relation may also be written as
3
x 'i = ! λij xj , (1.21)
j =1

where it is understood that the free index "i" takes on values

1,2,3.
Question: What if we knew the x 'i instead of the xj ? In
other words, what is the inverse relationship between these
quantities? Using matrix notation we have

x' = λ x
=> λ -1
x' = λ-1 λ x = x,

so*
x = λ-1 x'
or
3
xi = ! λ !1 '
ij x j . (1.22)
j=1

* We are assuming that λ-1 exists, which we know requires det(λ) ! 0; we will
establish this momentarily.
1.12

As stated above, there is a simple way of getting λ-1

(given λ) for rotations. Consider:

! x'i 2 = x '1 2 + x '2 2 + x '3 2.

i

By definition

# &# &
! x '2
i = ! % ! "ijxj(%! "ikx k (
% (% (
i i $j '$ k '

= ! λijλikxjxk . (1.23)
i,j,k

Require: ! x '2
i = ! xi2 (length preserved)
i i

=> ! λijλikxjxk = ! xi2. (1.24)

i,j,k i

The coefficient of each xjxk term (j,k independent) must be

the same on both sides:

$
&! " ij " ik = 0,j # k,
=> % i (1.25)
&! " ij " ik = 1,j = k.
' i
More simply

! λijλik = δjk , (1.26)

i

"Kronecker delta"
where

"0, i ! j ,
δij ≡ # (1.27)
$1, i = j .
 1.13

Notice that the above is not a matrix statement (why?), but

that the right-hand side, δjk, is just index notation for the
identity matrix. However, it may be cast into matrix language
with the help of the transpose:

λ ji
T
= λij ,

=> ! λ ji
T
λik = δjk or !T! = 1. (1.28)
i

This last statement establishes that

!T = λ-1, (1.29)

for rotation matrices. In fact, the above equation is

sometimes taken to define such transformations. Actually, the
types of coordinate transformations allowed by this equation
are a bit more general than simply rotations, as we will see
shortly. The name orthogonal transformations are usually used
for real matrices which satisfy !T = λ-1 and therefore preserve
the lengths of vectors. Using the above, the relationship
between the x 'i and the xi may now be cast as

x' = λx <=> x = !T x',

in matrix language, or

x 'i = ! λij xj <=> xi = ! λjix 'j ,

j j

in component language. It is important to note that in place

of

! λijλik = δjk , (1.30)

i
1.14

we may also write

! λjiλki = δjk . (1.31)

i

Deriving one from the other will be a homework problem.

How many independent λij elements are there? Notice that

the equation ! λijλik = δjk is symmetric under j ! k, so that

i

this system of equations actually represents 6, not 9

equations. (The number of independent elements of a real 3x3
matrix are 6.) This means the number of independent λij is 9 - 6
= 3. This makes sense from the physical point of view of
rotations in three dimensions, which require three independent
angles, in general.
Let me make three additional points about the λij:

1. x" = (λ2λ1)x = λ3x, (1.32)

λ3 is an orthogonal transformation if λ1 and λ2 also are. (We

can view this as a transformation λ1 followed by a second
transformation λ2.) That is, the product of orthogonal
matrices is also an orthogonal matrix. The proof of this
statement is left as a problem.

"x'= !1x1,
2. # (1.33)
$x"= ! 2x'= ! 2!1x .

Do the rotations in the opposite order:

"x= ! 2x,
# (1.34)
$x= ! 1 x= !1 !2x.

In general λ1λ2 ≠ λ2λ1, (“noncommutative”) and the order in

which rotations are performed is important.
 1.15

3. As stated above, we started out describing rotation, but

the λij can also represent operations which are not rotations.
For example,

" !1 0 0 %
λinv = $$ 0 !1 0 ' .
'
# 0 0 !1&

This satisfies !†inv ! inv = 1, but describes a spatial inversion.

This is not an operation which can be physically carried out
on objects in general. We will consider such orthogonal
transformations only sparingly.
All orthogonal transformations satisfy

"+1, rotations
det λ = # (1.35)
$!1, inversions

λij which have det λ = 1 are called proper orthogonal

transformations, and those that have det λ = -1 are called
improper orthogonal transformations. In general, improper
orthogonal transformations can be thought of as an inversion
plus an arbitrary rotation since

det (λinversion λrotation) = det (λinversion)(det λrotation),

which means the combination is still improper. Proving that

det( λ) = +! 1 for orthogonal transformations is left as a
homework problem at the end of the Chapter.

Scalar and Vector Fields

You probably have an intuitive feeling as to the meaning

of a "scalar". It is something which is unchanged under a
1.16

change in coordinates. Under rotations, a scalar field behaves

as

Scalar: φ'(x') = φ(x), (1.36)

where x 'i = ! λijxj. It is understood that the x and x'

j

represent the same point described in two coordinate systems.

See the below:

2'
2
1'
!'(x'
1
,x'
2
) = ! (x1,x 2)
x

lines of
constant density
(! = constant )

There are also pseudoscalars, which behave like scalars under

rotations, but acquire an extra minus sign under inversions.

Pseudoscalar: φ'(x') = (detλ) φ(x). (1.37)

Although we've informally defined a vector as a quantity

with both magnitude and direction, we need a quantitative
definition. Here it is:

Vector: A 'i(x')= ! λij Aj(x). (1.38)

j

The λij are elements of an orthogonal transformation.

Obviously, the transformation law for the coordinates x 'i , xi
is the template for this definition.
 1.17

There are also pseudovectors, which transform as

Pseudovector: A 'i(x')= (detλ) ! λij Aj(x). (1.39)

j

These quantities do not change direction under inversions, as

vectors do. We will see an example of a pseudovector shortly.

Vector Algebra and Scalar Differentiation

! !
Given two vectors, A and B , we may form either a scalar or a
vector. Let's study these two possibilities.

! !
A. B= ! AiBi. (1.40)
i

denoted defined

It is also called the "dot product." Here is the proof it is a

scalar:

$
& Ai! =
&
" #ijAj,
% j
&
&Bi! = " #ikBk.
' k
! ! # &# &
=> A ' . B ' = ! A 'i B 'i = ! % ! " ij Aj( % ! " ik Bk (
i i $ j '$ k '
# & ! !
= ! % ! " ij" ik ( AjBk = ! AjBj = A . B .
j,k $ i ' j

!jk
1.18

Notice I changed the order of the sums in some intermediate

expressions, which is always allowed for a finite number of
terms. Another, more geometrical, definition of the scalar
product follows from the diagrams below:
2

_
A ._
B = A 1 B1 + A 2 B 2
1
_
B

_
A

Same situation with a new coordinate system:

2
2'

B'
2

1
1
! B
B'
1
1
A
1'
Therefore

! ! ! !
A '. B ' = A '1 B '1 , | A | = ! A2i ,| B | = ! Bi2 .
i i

! ! ! ! ! ! B' $ ! !
A '1 = | A | here, so A '. B ' = | A || B | # !1 & . Then given A . B =
" B%
"#$
= cos '
! !
A ' . B ',
! ! ! !
=> A . B = | A | | B |cosθ. (1.41)
 1.19

! !
The vector product (or cross product) of A and B is defined
as:

! !
( A x B )i = ! εijkAjBk. (1.42)
j,k

denoted defined

Define the "permutation symbols", εijk:

#!123 = !231 = !312 = 1,

%
$!321 = !213 = !132 = "1, (1.43)
%all other ! © s = 0.
& ijk

Important property: εijk = εjki = εkij, εijk = -εikj = -εjik. This

is known as the cyclic property of the permutation symbols.
Using the εijk, a more formal definition of the determinant of
a 3x3 matrix, A, can be written

det A ≡ ! εnℓm A1n A2ℓ A3m. (1.44)

n,!,m

! !
We will use this definition shortly. In order to see how A x B
transforms, we will need the following identity for orthogonal
transformations,

! εijk λjℓλkm = (detλ) ! εnℓm λin, (1.45)

j,k n

This identity will be proven later (end of the Chapter.) Let’s

begin:
1.20

! !
( A 'x B ')i = ! εijk A 'j B 'k = ! εijk λjl A l λkm Bm
j,k j,k,!,m

= ! εijk λjℓ λkm AℓBm. (1.46)

j,k,!,m

Now use (1.45) on the right-hand side of (1.46):

! !
=> ( A ' x B ')i = (detλ) ! εnℓmλin AℓBm = (detλ) ! λin (εnjkAjBk)
n,!,m n,j,k
(j)(k) (j)(k)

notice I am
indicating a change
in the dummy indices!
! !
= (detλ) ! λin ( A x B )n.
n

Therefore we have that

! ! ! !
( A ' x B ')i = (det λ) ! λin ( A x B )n. (1.47)
n

! !
The extra factor of (det λ) indicates that A x B actually
! !
transforms as a pseudovector (assuming A and B are themselves
vectors.)
As an aside, there are many useful identities we can form
with the εijk by multiplying and summing on indices. For

example, let's evaluate: ! εijk εℓmk. Following the index jockey

k
rules, the only objects we can build out of δij and the εijk
which have 4 free indices are:

δijδℓm, δimδℓj, δiℓδjm.

 1.21

Therefore, we must have

! εijk εℓmk = C1δijδℓm + C2δimδℓj + C3δiℓδjm,

k

where the C1,2,3 are unknown constants. One immediately can see
that C1 = 0. (Why?) Now multiply both sides of the above by δij
and sum over i and j:

=> 0 = C2δℓm + C3δℓm => C2 = -C3.

Now multiply by δjm and sum over j and m:

=> ! εijk εℓjk = - 2C2δiℓ .

j,k

Then, taking a special case, say i = ℓ = 1, we then see that

! ε1jk ε1jk = 2,
j,k

so therefore C2=-1 and

! εijk εℓjk = 2δiℓ, (1.48)

j,k

=> ! εijk εℓmk = δiℓδjm - δimδℓj. (1.49)

k
From this last expression, we easily see that

! εijk εijk = # (!ii!jj " !ij!ij) = 3 x 3 - 3 = 6. (1.50)

i,j,k i,j

There is an alternate definition of the vector product as

well. It says:
1.22

! ! ! !
A x B = | A || B | |sinθ| ˆ
n, (1.51)

" A! = magnitude of A!
$! !
$B = " " B
where # ! !
$! = angle between A and B
$%nˆ = unit vector using the right hand rule

It doesn't matter in which direction θ is measured. This angle

! !
is measured in the plane defined by A and B if they aren't
colinear. This definition is illustrated below.

Let's now reconnect to the useful concept of unit vectors

in this context. The scalar products are summarized as

ei . ˆ
ˆ e j = δij. (1.52)

Now, using the second definition of the cross product, we

find

e1 x ˆ
ˆ e 2 = |sin90o| ˆ
e3 = ˆ
e 3.

Similarly,

e3 x ˆ
ˆ e1 = ˆ e2 x ˆ
e2 , ˆ e3 = ˆ
e 1.
 1.23

Notice that order matters here since

e2 x ˆ
ˆ e1 = - ˆ
e 3,

and similarly for the other non zero products. Also

e1 x ˆ
ˆ e2 x ˆ
e1 = ˆ e3 x ˆ
e2 = ˆ e 3 = 0.

We can summarize all properties of unit vectors under the

cross product as

ei x ˆ
ˆ ej = ! εijk ˆ
e k, (1.53)
k

and with a cyclic order wheel for permutation symbol indices:

!
=> 123 = 1,

3 => ! 213 =-1, etc.

2
I can now show the first and second definitions of the
!
cross product are equivalent by resolving the components of A
!
and B into unit vectors. From the second definition,

! !
( A x B ) = (A1 ˆ
e 1 + A2 ˆ e 3) x (B1 ˆ
e 2 + A3 ˆ e 1 + B2 ˆ
e 2 + B3 ˆ
e 3)

= (A2B3 - A3B2) ˆ
e 1 + (A3B1 - A1B3) ˆ
e 2 + (A1B2 - A2B1) ˆ
e 3.

On the other hand, from the first definition,

! !
( A x B )1 = ! ε1jkAjBk = ε123A2B3 + ε132A3B2 = (A2B3 - A3B2),
j,k
! !
( A x B )2 = ε231A3B1 + ε213A1B3, = (A3B1 - A1B3),
1.24

! !
( A x B )3 = ε312A1B2 + ε321A2B1, = (A1B2 - A2B1).

Comparing, we see the two definitions agree. By forming

the determinant and comparing with the above definitions for
the cross product, one may also show that

! ! ˆ
e1 ˆ
e2 ˆe3
A x B = A1 A 2 A 3 , (1.54)
B1 B2 B3

where the right hand side is understood as a symbolic

determinant.
From either definition of the cross product we have that

! ! ! !
A ! B = "B ! A , (1.55)

as well as
! ! ! ! ! ! ! ! !
A ! (B ! C) = B(A " C ) # C(A " B ) . (1.56)

The last statement follows from Eq.(1.49) above. In addition,

there is the cyclic rule,
! ! ! ! ! ! ! ! !
A ! (B " C) = B ! (C " A ) = C ! ( A " B) , (1.57)

which is a direct result of the cylclic nature of the

! ! !
permutation symbols. Given that A , B , and C are vectors, the
! ! !
quantity A .( B x C ) is an example of a pseudovector. (This is
part of a problem.)
So, we know how to add and multiply vectors together.
Question: does it make sense to define division? Consider

! ! !
Ax B= C.
 1.25

! !
Given that A and C are known quantities, this equation
! !
clearly does not uniquely define B since any component of B
!
along A will not contribute to the cross product. So the
! ! !
answer is in general, no. However, if we know both A . B and A x
! !
B , we can solve for B , say. (We will have a homework problem
along these lines.)
Differentiation of vectors with respect to scalars leads
to new vectors.

Given:

A 'i = ! λij Aj , s a scalar.

j

Then

dA'i dA j dA j
ds'
= ! λij
ds '
= ! λij
ds
,
j j

dA'i ! dAj $
=>
ds'
= ! λij #
" ds %
. (1.58)
j

The important implication is that velocity and acceleration

are vectors.

Alternate Coordinate Systems

!
The radius vector, r , is easy to characterize in the
"
three
r most common coordinate systems. We have ( r! !
dr
; t is considered a scalar ):
dt
1.26

!
Rectangular: r = x1 ˆ
e 1 + x2 ˆ
e 2 + x3 ˆ
e3 (all e!ˆ i = 0)

!
Cylindrical: r = ρˆ e z (only e!ˆ z = 0)
eρ + z ˆ
Additional unit vector: ˆ eφ

!
Spherical: r = rˆ er
Additional unit vectors: ˆe φ, ˆ
eθ
! !
v = r!˙ , a = v!˙ = r
˙!˙ .

! !
As an exercize, let's work out v and a in cylindrical
coordinates. Follow a particle’s trajectory at two closely
spaced moments in time:

2
ˆ
e1!

ˆ
e2! ˆ
e2"

ˆ
e1"

d!
!
3 (up) 1

Displace the unit vectors to the origin:

 1.27

d! > 0 here
2
dˆ
e!

ˆ
e1! ˆ
e2" dˆ
e"
d!
d!
ˆ
e2! ˆ
e1"
1

eρ = ˆ
Can now see that (d ˆ e 2! - ˆ
e 1! )

|d ˆ
e ρ| = |dφ|,
e ρ = direction x length,
dˆ
= ˆ
e φ dφ.

Likewise (d ˆ
eφ = ˆ
e 2! - ˆ
e 1! )
|d ˆ
e φ| = |dφ|,
dˆ
eφ = - ˆe ρ dφ.

Therefore

dˆ
e! d! ˆ
= e! or e!ˆ ρ = !! ê φ , (1.59)
dt dt
and
dˆ
e! d! ˆ
= - e ρ or e!ˆ φ = - !!ê ρ. (1.60)
d! dt

Thus in cylindrical coordinates,

!
v = !˙ eˆ ρ + ρ e!ˆ ρ + z˙ eˆ z,
!
=> v = !˙ eˆ ρ + ρ !˙ eˆ φ + z˙ eˆ z. (1.61)
and
1.28

!
a = !˙˙eˆ ρ + !˙ e˙ˆ ρ + !˙ "˙ eˆ φ + ρ !!!ê φ + ρ !˙ e˙ˆ φ + z˙˙eˆ z,
! !
=> a = ( !˙˙ - ρ !˙ 2) e ρ + (2 !˙ "˙ + ! ˙"˙ ) ˆ e φ + z˙˙eˆ z. (1.62)

Since the unit vectors in rectangular coordinates do not

change, it is very useful to know the decomposition of the
unit vectors in other coordinate systems in terms of
rectangular coordinates.

!
$e ! = (cos ", sin ", 0),
&!
Cylindrical: %e" = (# sin ", cos ", 0), (1.63)
&e
!
' z = (0, 0,1).

$&ˆer = (sin ! cos ",sin ! sin ",cos !),

Spherical: %ˆe! = (cos ! cos ", cos ! sin ",# sin !), (1.64)
&'ˆ
e" = (# sin ",cos ",0).

Also

e! x ˆ
=> ˆ e" = ˆ
ez ,

z "

cyclically in cylindrical coordinates and

e! x ˆ
=> ˆ e" = ˆ
er ,

r "

cyclically in spherical coordinates.

 1.29

Angular Velocity

Another important concept for particle motion is angular

velocity. Here we will rely mainly an intuition to understand
the concept. First, identify an instantaneous circular path:

Flip this picture on it's side. Choose a cylindrical

coordinate system with the origin along the axis of rotation.

!
Let ! be directed co-linear to the axis of rotation, which is
the third axis above. Angular velocity and velocity are
related by the equation

! ! !
v = ! x r. (1.65)
1.30

Notice, using a cylindrical coordinate system,

! ! !
! =!2eˆ z, v ="#˙ eˆ # , r ="eˆ " +zeˆ z,

!
! !
v = ! x r ! "#˙ eˆ # =$zeˆ z x (!eˆ ! +zeˆ z ), = "z!eˆ # .
This means

!z = "˙ , (1.66)

so that !z has the meaning of instantaneous time change in the

!
angular position. Notice the direction of ! is given by a
right hand rule. (Curl the fingers of you right hand in the
! !
direction of v and your thumb points in the direction of ! ).
A common case is uniform circular motion:

! " !
v ! " is a constant

# vector pointed along

1 3-axis cylindrical
!
" up coordinates.

!
v = !˙
#ˆ
e# ,

!
a = $!˙
#2ˆ
e! ,
"
" = ˙
#ˆ
ez .

Differential Operators and Leibnitz Rule

A frequently occurring mathematical operation is the

gradient:

! $
! " # ˆei $xi
. (1.67)
i
 1.31

The x,y,z components of this are just the usual partial

derivative operators. When operating on a scalar, it gives a
vector:
! #"' ! #"'
!'"'= $ ˆ
e'i
#x'i
or (!' "') i
=
#x'i
. (1.68)
i

Proof: The chain rule says,

!"' !"' !xj !" !xj

!x'i
= # !xj !x'i
= # !xj !x'i
.
j j

Now remember

xj = " ! kj x'k ,
k
!xj !x"k
=> = $ # kj = #ij . (1.69)
!xi" !xi"
k !
% ki
Therefore

!#" !# ! "
!xi"
= % $ij !x
j
or ("!#!)i = % $ij("#)j , (1.70)
j j

!
which states that ! " transforms as a vector. A physical
interpretation of the gradient operator operating on a scalar is
given by the following sketch. Think of lines of constant
density in 2 dimensions again. Construct the 1’ axis tangent to
!
the constant φ line, at x ’:
1.32

2 2
!
x'

1’

lines of local
constant ! coordinate
system
1

Clearly,

!" ! $# ˆ'
= 0 ! "# = e only.
!x'1 $x'2 2

We conclude that the gradient at each point is a vector

pointing perpendicular to lines of constant φ at that point
and points in the direction that φ is increasing.
In three dimensions we can orient a new coordinate system
so that the 1’2’ plane is now locally tangent to the φ =
!
constant surface at x':

3 ! = constant surface

3’
2’
!
x'
2
1’
1

Again

! #" ˆ'
!'" = e only.
#x'3 3

So the gradient is perpendicular to lines (2 dimensions) or

surfaces (3 dimensions) of constant φ, and points in the
direction φ is increasing. When dotted into a vector field,
the divergence gives a scalar. (You should be able to prove
 1.33

! ! !
this statement.) Thus, ! " A transforms as a scalar if A , is a
vector.
A further mathematical operation, the curl of a vector,
!
Curl: (! " A! ) # j,k
% $ijk !jA k , (1.71)

denoted defined

can also be defined using the gradient ! i, but will not be

v v v
used extensively in this course. If A is a vector, ! " A can
be shown to transform as a pseudovector.
Another important operator is the Laplacian:

! ! #2
!"! = $ #x 2 % !2 . (1.72)
i i

One can show it produces a scalar or vector when acting on a

scalar or vector, respectively. Here is the proof it
transforms as a scalar when acting on a scalar:

#2 "' # # '
!'2"' = $ #x'2
= $ #x' #x'i
".
i i i i

Since φ = φ, we have
’

$ #xj ' # $ #x ' #

!'2"' = * &% #x' )( #x &% #x'k )( #x " .
i,j,k i j i k

As shown above

!xj
= " ij,
!x'i
& ) , ,
! "'2 #' = % '% $ ij $ ik * #,
j,k ( i + ,xj ,xk
1.34

!2 "
= # !x 2 = $ 2" .
i i
& ) , ,
! "'2 #' = % '% $ij $ik * ,x ,xk
#,
j,k ( i + j

The proof that ∇2 operating on a vector yields another vector

is similar.
Finally, there is a useful rule for taking the derivative
with respect to a variable which is contained in the upper and
lower limits of an integral. This result, called the Leibnitz
rule for differentiation, states that

d b(t) b(t)
#f(x,t)
! dx f(x,t) = f(b(t), t)˙
b(t) " f(a(t), t)˙
a(t) + ! dx . (1.73)
dt a(t) a(t) #t

When the differentiating variable appears in the upper and

lower limits of an integral as well as in the integrand, one
obtains three terms. The first term is the value of the
integrand evaluated at the upper limit, f(b(t),t), times the
derivative of the upper limit, ˙
b(t). The second term is similar
but represents minus the integrand at the lower limit, -
f(b(t),t), times the derivative of the lower limit, ˙
a(t). The
third term represents the contribution from the t-dependence
in the integral itself. Eq.(1.73) is the simplest case of the
dependence on a single variable, t; we will use a more general
version of the Leibnitz rule in a three dimensional context in
the next Chapter.

Complex Variables

We will have occasion to use complex numbers and

variables in our study of mechanics. The imaginary number i is
given by

i= !1 ,
 1.35

and when raised to various powers gives,

i2 = !1, i3 = !i, i4 = 1.

These numbers can be arranged on a unit circle in a plane:

y
z=x+iy

-1 ! 1
x

-i

This is called the complex number plane. As the figure

suggests, any complex number can be written as a combination
of a real and an imaginary number,

z = x + iy, (1.74)

where x and y are both real numbers. The complex conjugate of

z is given by x - iy and is denoted as z*. These two numbers
specify a location in the above plane on the real (x) and
imaginary (y) axes. The real and imaginary parts of z are
separated off with

Re z ! x, Im z ! y,
1.36

and the distance from the origin to the point z is given as

|z| = z z! = x2 + y2 . (1.75)

From plane trigonometry, any complex number on the above unit

circle may be written as

z = cosφ + i sinφ, (1.76)

where the angle φ is measured from the real axis, x. It was

Euler who showed this may be written as

z = ei! , (1.77)

for 0 ! " < 2# . Given this result we can now represent any
number in the complex plane as

z = |z| ei! . (1.78)

We often encounter ratios of complex numbers,

z1
z = .
z2

We may always factorize such a ratio into real and imaginary

parts by multiplying top and bottom by the complex conjugate
of z2 :

z1 z!2
z = ,
| z2 |2

which may also be written as

z1 i(!1 " !2 )
z = e .
z2
 1.37

Complex numbers are extremely useful in solving linear

differential equations, which is where we will see them next
in this course.
1.38

Chapter 1 Problems

1. (a) Let

"0 1 3% "6 2 !2%

A = $ ' , B = $ 1 0 4 '.
$!1 2 0' $ '
$ ' $ '
#!2 4 0& #4 1 !1&

Find the matrix product AB.

(b) Find det (AB), and show that det (AB) = (det A)(det B).

(c) Form the matrix product BTAT and show that BT AT = (AB)T,
as it should.

2. Try proving that

A-1A=1,

implies that

AA-1=1.

3. Prove that the inverse of the transpose is the transpose of

the inverse of a square matrix (assuming it’s inverse
exists!).

4. Given # ! ij! ik = "jk , show that

i

# !ji! ki = "jk ,
i

follows immmediately. [Hint: Use matrix notation and remember

that λ−1 = λT for these matrices.]
 1.39

5. Starting with (λ is a matrix; T denotes transpose)

λ-1 = λT,

prove that

det λ = +
- 1.

[Hint: det λ = det λT.]

6. Given that the 3 x 3 matrices λ1 and λ2 represent orthogonal

transformations on vectors x and x',

!T
-1 T -1
1 = λ 1 , !2 = λ 2 ,
(x'= λ1x, x" = λ2x')

show that the matrix product, λ2λ1, is also an orthogonal

transformation.

7. Using the definition given in the text,

det A = " !ijk A1iA 2jA3k ,

i,j,k

show that the determinant of a 3 x 3 matrix may also be

written in the forms:

(a) det A = " !ijk Ai1Aj2A k3 ,

i,j,k

1
(b) det A = " !ijk!!mn A !iA mjA nk .
3 ! i,j,k,!,m,n

8. Argue that the determinant of an antisymmetric 3x3 matrix,

AT = ! A ,

is zero. Is the determinant of an antisymmetric 4x4 matrix

necessarily zero?
1.40

9. We need to complete the proof of the relation we found

! !
necessary when proving that A x B actually transformed as a
(pseudo) vector. This proof was based upon the supposed
identity,

# !ijk "j! " km = C# !n!m "in .

j,k n

By multiplying both sides by λip (and summing i) and choosing a

special case, show that this identity may be reduced to
Eq.(1.44) of the text (almost; see prob. 7), implying

det ! = C.

10. Given that

! ! ! ! ! !
A x ( B x C) = ( A x B)x C,
! ! !
for vectors A , B and C, show that this implies that

! !
A x C = 0.

11. Show that

! ! ! ! " ! 2
A ! [ B " ( A " B)] = ( A " B) .

! ! ! ! ! !
12. Given that A , B and C are vectors, argue that A ! (B " C)
transforms as a pseudoscalar.

!
13. Does the angular velocity of a particle, ! , transform as
a vector or pseudovector? Explain.

14. Evaluate the sum:

 1.41

" !ijk!!mk!njm .
k,j,m

[Hint: Use results already proven.]

15. (Adapted from Marion and Thorton.) Let's investigate a

point raised in the text. Consider

! ! !
A ! B=C ,

! !
where A and C are considered known quantities. Clearly, this
! !
equation does not uniquely define B since any component of B
!
along A will not contribute to the cross product. However,
consider

! !
A . B = S,

!
where S is a known scalar. Show that we can solve for B and
!
that ( A considered nonzero)

! ! ! !
B = ! 2 (SA + C ! A ) .
1
A

16. Express the spherical unit vector, ^

e θ, in terms of the
! ! !
cylindrical unit vectors eρ, eφ and ez. The x,y,z components of
these vectors are

!
eθ = (cosθcosφ, cosθsinφ, -sinθ) ,

and
!
eρ = (cosφ,sinφ,0),
!
eφ = (-sinφ,cosφ,0),
!
ez = (0,0,1).
1.42

17. Show that the velocity vector in spherical coordinates is

!
given by: v = ˙ er + r ˙
rˆ e! + r sinθ !! ˆ
!ˆ e φ. [Hint: Use the
decomposition of ê r found in the script.]

18. In spherical coordinates, show that

a) ˙
eθ = - ˙
ˆ e! + ˙
!ˆ ! cosθ ˆ
e r,

e! φ = - !!! cosθ ˆ
b) ˆ eθ - ˙
! sinθ ˆ
e r.

[Hint: See decomposition of ^

e θ, ^
e φ in the script.]

19. A particle follows the trajectory y = Kx2 (K is a constant)

with constant speed, s, in the xy plane as shown.

x
O
! !
a) Show that ˙x˙ . x˙ = 0 anywhere on the trajectory.

!
b) Find the acceleration, ˙x˙ , when the particle is at the
origin, 0.
!
c) Find the instantaneous angular velocity, ! , at 0.

!
20. Given the vector function v = 20xy ˆ
i + 25yz ˆ
j + 15xy ˆ
k , find:
! ! ! !
(a) ! " v , (b) ! " v .

21. Given that

 1.43

! i' = # !ij"j ,
j

and

A'j = ∑ λjk Ak ,
k

prove that:
! !
(a) ! '. A ' transforms as a scalar,
! !
(b) ! " A # transforms as a pseudovector.

22. Show that (∇2 ≡ ! !j2 )

j
! !
∇2(f g) = g∇2f + f∇2g + 2 ! f. ! g.

23. Show that

z = cosφ + i sinφ,

may be written as

z = ei! .

[Hint: Consider the power series expansion of the

exponential.]

24. Show that the angle φ in z = ei! is given by ( 0 ! " < 2# )

# Im(z)&
! = tan "1 % .
$ Re(z)'