Lecture Notes

Vector Analysis
ZCT 211

Lecturer: Yoon Tiem Leong

Triple cross product
● LHS of (a) ● RHS of (a)
● Expand fully the
expressions

The compare component by components on both sides, and the results follow.
The trick is: There is no trick involved. Just expand both sides and compare.
 The proof of (b)
know

Show this

(b) can be trivially proofed based on the (a)

●Simply swap the order of (A x B) with C in (a)

(hence introducing a minus sign) and then

rename B -> A, C -> B, A->C in the resultant
formula: A×B×C=
−A×B×C=−BA⋅CCA⋅B
●
B -> A, C -> B, A->C

−C×A×B=−AC⋅BBC⋅A
A×B×C=BA⋅C−AB⋅C
Vector Differentiation

If u = t, then dr/dt= v
Can you derive these equations using
differentiation by parts?
 Vector as a function of scalar or a
set of scalar

A= Ai i+ A j j + A k k
A i= Ai (u) A= A(u)
A= A(x , y , z)
Ai = A i ( x , y , z )
A=A (x , y , z , t)
A i= Ai ( x , y , z ,t )
 Topics

1. Calculations involving algebra or the taking the

derivations of vectors
2. Application on kinematics (involving r, v, a)
3. Application on dynamics / rotational dynamics,
F = ma etc.
4. Differential geometry
A= Ai i+ A j j+ A i j
A= A(u)
A= A( x , y , z )
A=A ( x , y , z , t)
Partial derivative of a function of
multiple variable
∂ f (x , y , z )=lim f (x+ Δ x , y , z)
Δ x →0
∂x (Δ x)
∂ f ( x , y , z)=lim f (x , y + Δ y , z )
Δx →0
∂y (Δ y )

∂ f (x , y , z)=lim f (x , y , z + Δ z)
Δz → 0
∂z (Δ z )
Example of partial derivative
2 3 4
f ( x , y , z )=x y z

∂ f ( x , y , z )=2 x ( y 3 z 4 )
∂x
∂ f ( x , y , z)=3 y2 (x 2 z 4 )
∂y
∂ f (x , y , z )=4 z 3 (x 2 y 3 )
∂z
We shall assume all vectors encountered are differentiable to any order n needed

An example would be A(x,y,z) as vector potential at fixed time.

 Differential of a function of single
and multi-variables
Recall your ZCA 110:

df ( x) For single variable

df ( x )= dx
dx function

∂ f (x , y, z) ∂ f ( x , y , z)
df ( x , y , z )= dx + dy For multiple
∂x ∂y
∂ f ( x , y , z)
variables function
+ dz
∂z
Comment: solve by expanding A and B into their component form

Ai and Bi are functions of variable u, Ai= Ai(u), Bi= Bi(u)

You can solve these questions by either (1) carry out the vector
cross or dot product first, then take the derivative next, or (2)
take the derivative of the bi-vectors first, then reduce the
resultant vector derivatives.

Method 1

Method 2
Visualise the curve of C =
using Mathematica
Uniform circular motion
 This is the vector of angular momentum/m,
fixed in direction and magnitude.
In other words, the angular speed is constant
and the circular object is not making any
angular acceleration.
Visualise uniform circular motion using Mathematica
Differential geometry

r
 Unit tangent vector at a point P on a
curve C

lim Δs →0 ∣d r∣=ds
curve C
arc length s

ds
Δs dr
A fixed point
The rate at which T changes with respect to the
arc length s measures the curvature of C

Te direction of is , where
is normal to the curve at
that point. is also
perpendicular to
 Principle normal
We call the unit vector N (which is defined as a
unit vector in the direction as that of )

The magnitude of at the point on the

curve is denoted as =

Hence, we write
Binormal to the curve B
 The coordinate system {T, N, B} is
called trihedral / triad at the point

As s changes, the triad

{T, N, B} changes along the curve C
– moving trihedral
 To characterise the geometry of a
curve in 3D space
●Two quantities are required to describe the
geometry of a 3D curve
●

It tells you how B varies with s It tells you how T varies with s
To show this, need to prove the following:
dB
⊥B
ds
dB
⊥T
ds
Since {T, N, B} forms a right-handed
system, a vector which is simultaneously
perpendicular to both B and T necessarily
means it is pointing in the N direction.
 dB
First, show ⊥B
ds
d  B⋅B d B dB dB
= ⋅BB⋅ =2 B⋅
ds ds ds ds
Independently, 2
d  B⋅B d  B  dB
= =2 B
ds ds ds
dB dB
2 B⋅ =2 B =0
ds ds
d B d∣B∣ dB
because = =0 ⇒ ⊥B
ds ds ds
 dB
Now show ds ⊥ T

By definition , B=T×N
dB dT dN
= ×N T × 
ds ds ds
dN
= N ×N T× 
ds
dN
=T ×
ds
 dB
⊥T
ds
dB dN
=T ×
ds ds
dB
⇒ ⊥T
ds
The above relation can be deduced immediately
(from the properties of cross product). Can you
see how?

If you want to prove it explicitly, this is how

dB dN dN
T⋅ =T⋅T × =T ×T ⋅ =0
ds ds ds
We have shown that dB/ds is simultaneously
perpendicular to both B and T. This means it is
pointing in the N direction, since {T, N, B}
forms a right-handed system.

dB
=−τN
ds
Don't simply prove it using brute force.
Prove it using the proof you have got in
(a) and (b), namely,

and use the following “ingredients”

{T, N, B} forms a right-handed system
N=B×T
 Chapter 4

Gradient, divergence and curl

Del operator converts a scalar field
into a vector field
 Directive unit vector

is
 For example, a=i
∂ϕ( x , y , z ) ∂ ϕ( x , y , z )
∇ ϕ(x , y , z)= i+ j
∂x ∂y
∂ ϕ(x , y , z )
+ k
∂y

∂ϕ( x , y , z ) Rate of change of φ

(∇ ϕ)⋅i=
∂x at (x,y,z) along the
direction i
 A scalar
A scalar operator
function
This is just an algebraic exercise.
But still you got to make sure you
know how to do it despite the
straightforwardness.
 Can you recognize these?

Of course, you have already seen them before

(apart from a constant) in at least two instances
 Can you recognize these?

r̂
= 2
r
ϕ (r): spherically symmetric gravitational/Coulomb
potential

: gravitational field / Coulomb field

 z dr
P

r is perpendicular to dr at
P on the surface x
0 Need to show ∇ ϕ is
y perpendicular to dr
Show is perpendicular to dr

since
Use the previous result.

Find at P(2,-2,3). It is a
vector normal to the surface at P.

The normal unit vector at P

can then be obtained via
Is there any other unit normal?
 See mathematica file chap4.nb for the looks of
these surfaces
The angle between the surfaces at the point is the
angle between the normals to the surfaces at the
point.

Simply: find the angle between these two

vectors
θ
See mathematica code, Ch4.nb, to visualise these
surfaces and the the normals
This is a simply a simple algebraic example to
show how divergence work on a vector field.
 Gauss's law in integral form
(Electrostatics)
ρ
∫Gaussian surface E⋅d A= ϵ0
Gauss law for electrostatic field

For Gaussian surface enclosing zero net charge,

Gauss law in the integral form reduce to

∫Gaussian surface E⋅d A=0

 Gauss's law in differential form
∫Gaussian surface E⋅d A=0
You will learn in later chapter that Gauss law can
be cast into an equivalent form
∫Gaussian surface E⋅d A=0 ∇⋅E=0

On the other hand, electric field E and the electric

potential ϕ are related via

E=−∇ ϕ
 Derivation of Laplace equation
2
∇ ϕ=0
● Taking the divergence of the electric field, we
can express the Gauss law in differential form
in terms of electric potential
E=−∇ ϕ
Taking divergence

∇⋅E=∇⋅(−∇ ϕ)
2
∇⋅E=−∇ ϕ=0
because ∇⋅E=0
 Solution to the Laplace equation

Laplace equation is just Gauss law stated in the

form of the Laplacian of the electric potential.
So, instead of dealing with electric field on the
Gaussian surface, you just simply need to solve
the Laplace equation for the potential field.
 Example: Application of divergence in
fluid dynamics
The flow of fluid is characterised by the velocity vector
field, v(x,y,z)
Calculate the loss in the volume element per unit time
 Continuity equation for
incompressible fluid
●Incompressible fluid: fluid that is
neither created nor destroyed at any
point.
Hence no 'volume loss' will occur to
an incompressible fluid.
Solenoidal vector
Hint: 1. Cast into a form that
explicitly contains

2. Use:

∇⋅(A×r )=r⋅(∇ × A)− A⋅(∇ ×r )

Note:
A is a general vector field;
r is position vector.
The curl of a position vector is zero
∇×r =0

∇×r =i[( ∂ ∂
) z−( ) y ]+
∂y ∂z
∂ ∂
− j [( ) z−( ) x ]+
∂x ∂z
∂
k [( ) y−( ∂ )x ]=0
∂x ∂y
∇⋅(A×r )=r⋅( ∇ × A)− A⋅(∇ ×r )
=r⋅(∇ × A)
Note: φ a general scalar field
Note: A a general vector field
Where will you see this equation?
When you learn magnetostatics.
Magnetic field B is express in
terms of vector potential via
B=∇ × A
The fact that there is no magnetic
monopole is represented by the
statement ∇⋅B=0

Hence the equation

follows.
 The curl of a vector field v
● This problem indicates that the curl of a vector
field has something to do with rotational
properties of the field.
● If the field v is that due to a moving fluid, for
example, then a paddle wheel placed at various
points in the field would tend to rotate in regions
where curl v ≠ 0 (vortex field)
● If curl v = 0 in the region there would be no
rotation and the field v is then called
irrotational.
 The curl of a vector field v
● curl v ≠ 0: v is a vortex field (it 'rotates')
● curl v = 0: v is an irrotational field (it does not
'rotate')
So that you can prove 29 using the vector
triple cross product result
Where would you encounter the
above identity?

When you learn Maxwell equations

in electrodynamics.
and
You start from the identity proven in the previous
slide and apply it on the E field:

A→E

Eq. (1)

because
Independently,

Eq. (2)
Combining Eq. (1) and Eq. (2),
The electric and magnetic fields, E and H,
propagates according to the wave equation –
prediction of electromagnetic wave propagating in
free space.
In other words, light is just electromagnetic wave.
 Chapter 5

Vector integration
 The trick is: do not integrate directly by brute force.
Instead, cast the integrant into the form of a
differentiation
d2 A d
∫ [ A× dt 2 ]dt=∫ [ dt (something)]dt
so that
d
∫ [ dt (something)] dt=something+ constant vector
 P2

y Curve C
r(u=u2)
P1
r(u=u1)
r(u) P
x
O
 Line integration as the limit of discrete
Ai sum Ai●∆ ri
i discrete variable
∆ ri y Curve C
ri ∆ ri=ri+1 - ri

P2 x P1
O

i=N P2
lim N  ∞ ∑ i=1 Ai⋅ r i =∫P A⋅d r =∫C A⋅d r
1
Path C, as defined by

When the parameter

t varies from 0 to 1,
a point P traverses
from (0,0,0) to
(1,1,1) along the
curve C

See mathematica
code, Ch5.nb
Path (b)
The path C is divided into three segments, each
is to be evaluated independently according to
the geometry of the line segment involved.
Note: Line integration of a generic vector from one
fixed point to another along two different paths are
generally not the same, as seen in previous
examples.

5
 The circular path with
P(x,y) radius r is
parametrised by the
parametric equations

x=cos t, y = sin t

When t varies from 0 to 2π, point

P would have traversed clockwise
by a full circle
P(x,y)

Since the circular path C is

located on the x-y plane,
there is no variation in the
z-variable, hence dz=0 in
dz=0 dr.
 x=cos t, dx =-sin t dt; y = sin t, dy=cos t dt

Setting the limit of integration from t=0 to t=2π

amounts to performing the line integration in the
anticlockwise (+ ve) direction.
If the limit of integration is instead set from
t=2π to t=0, that amounts to performing the
line integration in the clockwise (-ve)
direction. As result, the line integration
would result in a relative negative sign:

Clockwise , -ve direction

 Line integration is direction-
dependent

From the previous example, it is seen that in a

given line integration, the direction of integration
matters.

In most cases, integrating along opposite direction

results in a relative minus sign between two line
integrations (but this is not in general true).
 Line integration is direction-
dependent
So, which is the correct answer in the previous
example? The +ve one or the -ve one?

The correct statements are:

The work done is +18π if integrate along the circle

in the clockwise direction;

The work done is -18π if integrate along the circle

in the anticlockwise direction;
 P2 You are asked
to show that, if
Path 1
F=∇ ϕ
then
P1 Path 2

∫Path1 F⋅d r=∫Path2 F⋅d r

If F=∇ ϕ
Conservative field
Conversely, we can also show that if the line integration

∫C F⋅d r
is independent of the path C joining any two points,
there exist a function φ such that

F=∇ ϕ

φ is known as the 'scalar potential'

(see next slide for a proof)
Chap 4, page 61, Q8, part (b), (c)
 Conclusion of Q10
A conservative field is one which line integration
of a field is independent of path.

If a field can be expressed in terms of a

differentiable scalar function φ via
A=∇ ϕ
then A is conservative.

Conversely, if a field A is conservative, it can

always be expressed a gradient of some scalar
function,
A=∇ ϕ
it can be proven that

For details of the proof, read it yourself.

 (i.e., A is conservative)

Theorem No.1 is what we have just proven in

Q10.

Theorem No.2 can be proved using Theorem 1.

 Closed Line integration
Consider a line integration along a path C
Need to prove two statements:

If

If

then
 Proof of the first statement

The first part is proven. But

we still need to prove the
converse statement.
Proof of the converse statement
 Conclusion from
Q10, Q11, Q13
The following statements are equivalent:

F is conservative
Easiest to prove the statement by showing that
the curl of F is zero.
Integrate (1), (2) and (3) to obtain φ
+ constant
Another method
A better still method is to use the potential function

P2
=∫P d ϕ=ϕ(P2 )−ϕ(P1 )=Δϕ
1

=ϕ(3,1,4 )−ϕ(1,−2,1)=202
Work done by conservative field and
potential energy difference
For conservative fields, work done by the field
when moving a particle from point P1 to P2 is equal
to the change of potential energy between P1 to P2

r2
W (r 1 →r 2 )=∫r F(r )⋅d r =Δ ϕ=ϕ(r 2 )−ϕ(r 1)
1
 Have you seen any conservative
field before?
Of course yes.
GM m
Gravitational force, Fg (r )=− 2
r̂
r
GMm
Its corresponding potential function, U (r )=−
r
Both are related via Fg ( r )=∇ U (r )

Work done by gravitational force on a mass m

from r1 to r2
r2
W (r 1 →r 2 )=∫r Fg ( r )⋅d r=ΔU=U (r 2)−U (r 1 )
1
 Other conservative fields you have
seen before
Electrostatic field is also a conservative field.
qQ Qq
F(r )= r̂ U ( r )= F( r )=∇ U (r )
4 πϵr 2
4 πϵr
Elastic force of a spring,
k 2 d
F(x)=k x x̂ U (x )= x F(x)= U ( x )
2 dx

What else?
 Is frictional force conservative?
● No.
● Why?
● Despite the end points are the same, work
done by frictional force is path dependent.
● Hence, there is no corresponding “frictional
potential” from which work done can be
calculated as the difference of the “potential
function” between two points.
ZCT 211 Vector Analysis

Tutorial questions to submit. Take the questions from the textbook by Spiegel, Vector Analysis, Schaum
series.

Chapter 1 Vectors and Scalar

Q38, Q39, Q44, Q47, Q59, Q61, Q64(a)

Q66(c) modified:
Tutorial for Chapter 2
 Tutorial 3

Pass up personally
on
Monday class
12 pm, 27 Oct 2014
Tutorial for Chapter 3
 Tutorial 4

Pass up personally
on
Friday 10 pm class
7 Nov 2011
Tutorial for Chapter 4