Introduction to Tensors

Introduction

The origin of tensors…

Why we need tensors?
What tensors are actually?
The origin of tensors

▪ The word tensor has its root tensus in Latin, meaning stretch
or tension.
▪ J. W. Gibbs in (1884) developed the concept of dyadics (object
with two parts or elements) and polyadics (an object
comprising of many elements) which are actually tensors in
modern sense but only different by names.
The origin of tensors

▪ W. Voigt in (1898) introduced the term tensor in his study of

stress and strain of crystals.
▪ Later on many physicists including Albert Einstein and Marcel
Grossmann in (1913) used the Ricci’s concepts with the name
tensors instead of systems.
Why we need tensors?
▪ A scalar is a physical quantity that is represented by a
number at a particular point in space. For example, mass,
density and temperature.

▪ A vector is a tool to keep track of two pieces of information,

typically magnitude and direction, for a physical quantity. For
example position, force and velocity.
Why we need tensors?
What happens when we need to keep track of three pieces of information for a
given physical quantity? Say, three pieces of information are magnitude,
direction and plane. For example stress, strain and inertia tensors.
Why we need tensors?
In short:
A scalar is a quantity whose specification requires just one
number.
On the other hand, a vector is a quantity whose specification
requires three numbers, namely its components with respect to
some basis.
Scalars and vectors are both special cases of a more general term
called a tensor.
What is a tensor?

Tensor can be regarded as a general name of multi-way array

data.
Tensor Definition:
In an m dimensional space, a tensor of rank n is a mathematical
object that has n indices, 𝑚𝑛 components and obey certain
transformation rule.
Rank of a tensor
The rank of a tensor also known as order or degree of tensor is the
total number of free indices in a tensor.
For example,
1) m (scalar) tensor of rank zero

2) 𝑣𝛼 vector tensor of rank 1,

𝑣 𝛽 (vector) tensor of rank 1
3) 𝑇𝜇𝛿 (tensor) tensor of rank 2
Higher Dimensional Spaces
Dimensions:
The dimensions of a mathematical space (or object) is informally
defined as the minimum number of coordinates needed to
specify any point within it.
2-Dimensional space

A coordinate system in which the exact position of a point is

determined with two numerical values, is called the two
dimensional space.

In Euclidean geometry, two perpendicular number-lines are

intersected at their middle points for splitting the whole space
into four equal parts. The number lines are used as base line to
measure the distance of every point in two perpendicular
directions from their bisection point.
2-Dimensional Space
There are four fundamental concepts related to two dimensional
Cartesian coordinate system:
Coordinate axes, origin, coordinates of a point and quadrants.
3-Dimensional space
We can expand our 2-dimensional (xy) coordinate system into
a 3-dimensional coordinate system, using x, y, and z-axes. In 3-
dimensional space the exact position of a point is determined
with three numerical values, called the coordinates.
3-Dimensional space
The xy-plane is horizontal in diagram below and shaded green. It
can also be described using the equation z = 0,
The xz-plane is vertical and shaded pink above. This plane can be
described using the equation y=0.
The yz-plane is also vertical and shaded blue. The yz-plane can
be described using the equation x=0.
Higher Dimensional Spaces
More than three dimensional spaces are treated as higher
dimensional spaces.
For example, we can think of dimensions as ways in which we
can move up and down, left and right, forward and backwards,
which leads us to believe we live in a three-dimensional world.
Every point in this world is defined by three coordinates. We can't
imagine living in a four-dimensional world because we simply
can't conceive of another direction to move in.
Higher Dimensional Spaces
However, now imagine you are arranging to meet someone. Not
only do you need to give them the three coordinates of the place
you are meeting, you also need to tell them the time you are
meeting. Thus, time and space together form a four-dimensional
space: you need four numbers to specify a point.

Generally, a point in an n-dimensional space is given by

n numbers, or coordinates. Even though we can’t visualize higher
dimensional spaces and objects within them as geometric objects.
Algebra gives the tools to deal with these objects.
Applications of Tensors
Applications of tensors
Tensors have a wide range of applications in different fields.
Following are different fields of Mathematics and Physics where
tensors are used.
▪ Mechanics
Stress tensor, Strain tensor, Elasticity tensor, Inertia tensor
▪ Electromagnetics
Maxwell equations and related tensors
▪ Special theory of relativity
Metric tensor, four vector formalism
▪ General theory of relativity
Riemannian tensor, Ricci tensor, Einstein tensor
Applications of tensors

▪ Following are some applications of tensors in Engineering

• To represent mechanical stress as the stress
tensor in continuum mechanics
• To measure deformations through deformation tensor
and strain via strain tensor in continuum mechanics
• To represent diffusion as a tensor in diffusion tensor imaging.
• To describe moment of inertia as inertia tensor.
Applications of tensors
Tensor Flow
Tensor flow is an open-source end-to-end platform for creating
machine learning applications. It is a symbolic math library that
uses dataflow and differentiable programming to perform various
tasks focused on training and inference of deep neural networks.
How Tensor Flow Works

Tensor Flow enables you to build dataflow graphs and

structures to define how data moves through a graph by
taking inputs as a multi-dimensional array called Tensor.
It allows you to construct a flowchart of operations that
can be performed on these inputs, which goes at one end
and comes at the other end as output.
Applications of tensor flow
Currently, tensor flow is used in a variety of Google applications
like Google photos, Google translate, Gmail, search engine,
translation, image captioning or recommendations.
Free and Dummy Indices
Tensor Notation:

Tensor notation provides a convenient and unified system for

describing physical quantities.
In the most general representation, a tensor is denoted by a
symbol followed by a collection of subscripts (or superscripts).
For example

X i ,  ij , U l , Bklm
Subscript and Superscript:
A subscript is a character (such as a number or letter) that is written
slightly below and a superscript is written above the normal line of
type. It is usually smaller than the rest of the text.
Subscripts appear below the baseline, while superscripts are
above.

For example:
Ai , u j subscripts

k ij
B ,C superscripts

These subscripts and superscripts are also called indices.

Indices:
Indices assume values from an index set. They successively assume
each value from the given index set.
For example, if 𝑖 takes the values from 1 to 4, then 𝒂𝒊 denotes the
set (𝒂𝟏 , 𝒂𝟐 , 𝒂𝟑 , 𝒂𝟒 ) and we say that the index 𝑖 runs from 1 to 4.

Example:
a j ; j = 1, 2, 3
bk ; k = 0,1, 2, 3
Free and dummy Indices:
In the index notation, indices are categorized into two groups: free
indices and dummy indices.
▪ An index which may take on any particular value from the index
set is called a free index. For example
Ai ; i = 1, 2,3
Bij ; i = 1, 2,3& j = 1, 2,3
▪ A dummy index means summation. Dummy indices are repeated
indices in an expression. For example
 A ; i = 1, 2, 3
i
ii

 AB
i
i ij ; i = 1, 2, 3& j = 1, 2, 3
 Example 1:
1) Write down explicitly the equations represented
3
by the expression yi =  aij x j where i = 1, 2,3 and j = 1, 2,3
j =1

Solution: i) First differentiate the free and dummy indices in expression.

Here i is free index and j is dummy index.
ii) Then apply the sum over the dummy index.
yi = ai1 x1 + ai 2 x2 + ai 3 x3
iii) After that, put the values of free index one by one which
yield a set of equations.
y1 = a11 x1 + a12 x2 + a13 x3
y 2 = a21 x1 + a22 x2 + a23 x3
y3 = a31 x1 + a32 x2 + a33 x3
Example 2:
n n
Show that if i = 1, 2,3,..., n and j = 1, 2,3, ..., n then  ai xi =  a j x j .
i =1 j =1

Solution: First consider

n
L.H.S.=  ai xi = a1 x1 + a2 x 2 + a3 x3 + ... + an x n
i =1
n
R.H.S.=  a j x j = a1 x1 + a2 x 2 + a3 x 3 + ... + an x n
j =1

Note: So any dummy index can be replaced by other index ranging

the same numbers.
Example 3: Consider the expression Tij =  Aijm xm +  Bip C jp
m p

1) Which are the free indices?

2) Which are the dummy indices?
3) How many terms are on the right hand side when m and p has
variation 1,2.
4) How many equations are described by the given
expression when i and j has variation 1,2?
Example 3: Consider the expression Tij =  Aijm xm +  Bip C jp
m p

3) How many terms are on the right hand side when m and p has
variation 1,2.
4) How many equations are described by the given
expression when i and j has variation 1,2?
Example 3: Consider the expression Tij =  Aijm xm +  Bip C jp
m p

3) How many terms are on the right hand side when m and p has
variation 1,2.
4) How many equations are described by the given
expression when i and j has variation 1,2?
Tensor Definition:
In simple words, tensors are simply arrays of numbers
or functions that transform according to certain rules
under a change of coordinates.
OR
In an m dimensional space, a tensor of rank n is a
mathematical object that has n indices, 𝑚𝑛 components
and obey certain transformation rule.
Rank/order of a tensor
The rank of a tensor is the number of indices required to uniquely
select each element of the tensor. Rank is also known as order, degree.
In simple words, It is the total number of free indices in a tensor.
For example,
1) m (scalar) tensor of rank zero

2) 𝑣𝛼 vector tensor of rank 1,

𝑣 𝛽 (vector) tensor of rank 1
3) 𝑇𝜇𝛿 (tensor) tensor of rank 2

4)
A (tensor) tensor of rank 1
Einstein Summation Convention
Einstein Summation Convention:

This convention was introduced with Einstein’s words:

"I have made a great discovery in mathematics; I have suppressed
the summation sign every time that the summation must be made
over an index which occurs twice..."

Einstein summation is a notational convention for simplifying

expressions including summations of vectors, matrices and tensors.
Following are the rules of Einstein summation notation.
1. Repeated indices are summed over.
n
ai ai =  ai ai
i =1
n
aik aij =  aik aij
i =1
2. Each index can appear at most twice in any term.

3. Each term must contain identical non-repeated indices.

So,

M ij v j =  M ij v j
j

is a valid expression whereas

M ij u j v j + wi and Tijk uk + M ip
are not valid expressions.
Question: Check which of the following is a valid expression.
1) ams = bm (cr − d r )
not valid
2) ams = bm (cs − d s )
valid
3) ti = a ji n j
valid
4) ti = a ji ni
not valid
5)  ij = xi xij
not valid
6) bij c j = 3
not valid
Example 1:Expand the summation, ci ( xi + yi ) for n = 4

ci (x i + yi ) = c1 (x1 + y1 )
+ c2 (x 2 + y2 )
+ c3 (x 3 + y3 )
+ c4 (x 4 + y4 )
Double Sums:
▪ An expression can involve more than one summation index.
▪ For example aij xi y j indicates a summation taking place on
both i and j simultaneously.
▪ If an expression has two summation indices varying from 1 to
𝑛, there will be a total of 𝑛2 terms in the sum. If there are three
summation indices, there will be 𝑛3 terms in the sum and so
on.
Example 2

Simplify aij xi y j , where i = 1, 2,3 and j = 1, 2,3.

Solution: First sum over i
aij xi y j = a1 j x1 y j + a2 j x2 y j + a3 j x3 y j
Then sum over j
aij xi y j = a11 x1 y1 + a12 x1 y2 + a13 x1 y3
+ a21 x2 y1 + a22 x2 y2 + a23 x2 y3
+ a31 x3 y1 + a32 x3 y2 + a33 x3 y3
Example 3: Write down the equations for n=2 corresponding to
the expression:𝑦𝑖 = 𝑐𝑖𝑟 𝑎𝑟𝑠 𝑥𝑠
Solution:
Substitutions:
Suppose it is required to substitute yi = aij x j in the equation Q = bij yi x j
it would lead to Q = bij aij x j x j which is an invalid indicial
expression.
The correct method is that first we identify any dummy indices
in the expression to be substituted that coincide with indices
occurring in the main expression. Changing these dummy indices
to characters not found in the main expression one can proceed
substitution in the usual fashion.
Example 4
If yi = aij x j , express the quadratic form Q = gij yi y j
in terms of x-variable.
Solution:
First replace j with k in yi = aij x j for yi : yi = aik xk
and i with j replace j with s in yi = aij x j for y j : y j = a js xs
Then by substitution it becomes
Q = gij aik xk a js xs = g ij aik a js xk xs
Q = hrs xk xs ; hrs = gij aik a js
Example 5: Express 𝑏 𝑖𝑗 𝑦𝑖 𝑦𝑗 in terms of 𝑥 variable, if
𝑗
𝑦𝑖 = 𝑐𝑖𝑗 𝑥𝑗 and 𝑏 𝑖𝑗 𝑐𝑖𝑘 = 𝛿𝑘 .
Solution:
Kronecker Delta Symbol
Kronecker Delta
The Kronecker delta is defined as
1; i = j
 ij =  i = 
j ij
=
0; i  j
Clearly  ij =  ji for all i, j.
The Kronecker-𝛿 can be thought of as the identity matrix, for
example for 𝑖 = 1,2,3 and 𝑗 = 1,2,3
1 0 0

 ij =  0 1 0 
 0 0 1 
Example: If i = 1, 2,3 and j = 1, 2,3 simplify  ij xi x j .
Solution:
 ij xi x j = 1 j x1 x j +  2 j x2 x j +  3 j x3 x j
= 11 x1 x1 + 12 x1 x2 + 13 x1 x3
+  21 x2 x1 +  22 x2 x2 +  23 x2 x3
+  31 x3 x1 +  32 x3 x2 +  33 x3 x3
= (1) x1 x1 + (0) x1 x2 + (0) x1 x3
+ (0) x2 x1 + (1) x2 x2 + ( 0 ) x2 x3
+ (0) x3 x1 + (0) x3 x2 + (1) x3 x3
 ij xi x j = x1 x1 + x2 x2 + x3 x3 = xi xi
Hence in general
 ij xi x j = xi xi
 jr air xi = aij xi
Kronecker delta is a much used symbol in tensor calculus. It has
the effect of annihilating the off diagonal terms.
Example: For arbitrary n, evaluate  ii .
Example: For n = 3, evaluate  ij ij .
Example: For arbitrary n, evaluate  ij c .
j
k ik
Permutation symbol
The Permutation Symbol or Levi-Civita Symbol
In three dimensions, the permutation symbol is a three index
object also called the Levi-Civita symbol and alternating tensor. It
is defined by

0 for i = j , j = k or k = i

 ijk =  1 for (i, j , k )  (1, 2, 3), (2, 3,1), (3,1, 2)

 −1 for (i, j , k )  (1, 3, 2), (3, 2,1), (2,1, 3)

OR 1
 ijk = (i − j )( j − k )(k − i )
2
For the indices (i, j, k) in εijk, the values 1, 2, 3 occurring in
the cyclic order (1, 2, 3) correspond to ε = +1, while occurring in
the reverse cyclic order correspond to ε = −1, otherwise
ε = 0.
Levi-Civita Symbol

Example: The permutation symbol satisfies :

i)  ij  ijk = 0
ii )  ipq jpq = 2 ij
iii )  ijk  ijk = 6
where  ij is the Kronecker delta.
i)  ij ijk = 0
ii )  ipq jpq = 2 ij
iii )  ijk  ijk = 6
Tensor Notation for Matrices
Tensor Notation for Matrices
In the matrix notation [𝑎𝑖𝑗 ]𝑚𝑛 , the first subscript i, tells what row
the number 𝑎𝑖𝑗 lies in and the second j designates the column, m
denotes the number of rows and n the number of columns.
Examples:

 c11 c12 c13 

cij = c21 c22 c23 
 c31 c32 c33 
 d11 d12 d13 d14 
d ij = 
 d 21 d 22 d 23 d 24 
Basic Formulas:
The essential formulas involving matrices, vectors and
determinants will be written in terms of summation convention.
▪ Matrix Multiplication
If 𝐴 = [𝑎𝑖𝑗 ]𝑚𝑛 and B = [𝑏𝑖𝑗 ]𝑛𝑘 , then

𝐴𝐵 = [𝑎𝑖𝑟 𝑏𝑟𝑗 ]𝑚𝑘

▪ Identity Matrix
In terms of the kronecker delta, the identity matrix of order n is

𝐼 = [𝛿𝑖𝑗 ]𝑛𝑛
Which has the property
𝐴 = 𝐴𝐼 = 𝐼𝐴 = [𝑎𝑖𝑗 𝛿𝑗𝑖 ]𝑛𝑛
For any square matrix of order n.
▪ Inverse of a square matrix
A square matrix 𝐴 = [𝑎𝑖𝑗 ]𝑛𝑛 is invertible if there exists a unique
matrix B= [𝑏𝑖𝑗 ]𝑛𝑛 called the inverse of A, such that
𝐴𝐵 = 𝐵𝐴 = 𝐼
In terms of components:
𝑎𝑖𝑟 𝑏𝑟𝑗 = 𝑏𝑖𝑟 𝑎𝑟𝑗 = 𝛿𝑖𝑗
▪ Transpose of a Matrix
The transpose of a matrix in index notation is defined as

AT = [aij ]Tmn = [a t ij ]nm , where a t ij = a ji

A matrix A is orthogonal if 𝐴𝑇 = 𝐴−1

Example: Display explicitly the matrices
(a) [bij ]42 (b) [h ij ]33
 Prove by using tensor notation and product
Example:
rule for matrices that (AB)T = BT AT .
Coordinate Transformations
2-Dimensional Coordinate Transformations
A two dimensional coordinate transformation is a mapping of the
form
𝑇 𝑢, 𝑣 = (𝑥 𝑢, 𝑣 , 𝑦(𝑢, 𝑣))
The transformation T maps a set S in the uv-plane to a set T(S) in
the xy-plane.
Example:
If P has the coordinates (1,2) in uv-plane then find the
coordinates in xy-plane if: 𝑥 = 𝑢 + 𝑣 and 𝑦 = 2𝑢 − 𝑣.
Example:
Transform the line 𝑃𝑄 where P has coordinates (1, −1) and Q
has coordinates (0,1) from uv-plane to xy-plane if: 𝑥 = 𝑢 + 𝑣 and
𝑦 = 2𝑢 − 𝑣.
Some Well-Known Transformations
Coordinate Transformations
Translation

A translation slides an object a fixed distance in a given direction.

The original object and its translation have the same shape and
size, and they face in the same direction. To translate a shape every
point of the shape must move the same distance and in the same
direction.
For example the transformation
(𝑥, 𝑦) → (𝑥 + 5, 𝑦 + 3)
moves the parallelogram to the right 5 units and up 3 units.
Rotation
A rotation is a transformation where a figure is turned around a fixed point to
create an image. The two dimensions rotation matrix is
cos𝜃 −sin𝜃
𝑅=
sin𝜃 cos𝜃
To perform the rotation on a plane point with coordinates (𝑥, 𝑦), it should be
written as a column vector, and multiplied by the matrix R:
Scaling
To change the size of an object, scaling transformation is used. In the scaling
process, you either expand or compress the dimensions of the object. Scaling
can be achieved by multiplying the original coordinates of the object with the
scaling factor to get the desired result.
Reflection
A reflection is a transformation that turns a figure into its mirror image by
flipping it over a line. Reflection is the mirror image of original object. In
other words, we can say that it is a rotation operation with 180°. In reflection
transformation, the size of the object does not change.
Shear
A transformation that slants the shape of an object is called the shear
transformation. There are two shear transformations X-Shear and Y-Shear. One
shifts X coordinate values and other shifts Y coordinate values. However; in
both the cases only one coordinate changes its coordinates and other preserves
its values. Shearing is also termed as Skewing.
A shear parallel to x-axis results in 𝑥 ′ = 𝑥 + 𝛼𝑦 and 𝑦 ′ = 𝑦

𝑥′ 1 𝛼 𝑥
′ = 𝑦
𝑦 0 1

Similarly a shear parallel to y-axis results in 𝑥 ′ = 𝑥 and 𝑦 ′ = 𝑦 + 𝛼𝑥

𝑥′ 1 0 𝑥
=
𝑦′ α 1 𝑦
Applications of Coordinate
Transformations
 Applications of coordinate transformations
Transformations can represent geometric operations, which are caused by
movement or action, as well as changes of coordinates, which are caused by
changes of interpretation. Many common spatial transformations, including
translations, rotations, and scaling are represented by matrix operations.

Here we will discuss some applications of transformations.

▪ Camera image processing
▪ Robotics
▪ Computer graphics
Camera image processing
Consider the three coordinate frames:

▪ World coordinate frame:

These are 3D coordinates fixed in the world, say with respect to one corner
of the room.
▪ Camera coordinate frame:
These are 3D coordinates fixed in the camera. The origin of the camera
coordinates is at the center of projection of the camera. The z-axis is taken to
be the optical axis of the camera (with points in front of the camera in the
positive z direction).
▪ Image coordinate frame:
▪ Coordinate transformations:
Robotics:
For the purpose of controlling a robot, it is important to know the kinematics where
transformations between the coordinate frames are performed. These frames are
attached to different robot links of the robot. Vectors and matrices are used in robotics
to represent 2D and 3D positions, directions, rigid body motion, and coordinate
transformations.
Computer Graphics:

Computer graphics deals with generating images with the aid of computers.
Today, computer graphics is a core technology in digital photography, film,
video games, cell phone and computer displays, and many specialized
applications.
A transformation is used to change the object’s
▪ Position (translation)
▪ Size (scaling)
▪ Orientation (rotation)
▪ Shapes (shear)
Coordinate transformations
Position vector and point coordinates:
We will discuss the coordinates and position vector of a point in a three
dimensional Cartesian coordinate system. The unit vector
j , kˆalong 𝑥, 𝑦, 𝑧 axes are represented by eˆ1 , eˆ2 , eˆ3 in tensor
iˆ, ˆ
notation or ei , i = 1, 2,3 .
Similarly, the coordinates of a point (𝑥, 𝑦, 𝑧) are represented as:
Coordinates of 𝑃: 𝑥, 𝑦, 𝑧 = 𝑥1 , 𝑥2 , 𝑥3 = 𝑥𝑖
and the position vector is represented as:

x = xiˆ + yjˆ + zkˆ

x = x1eˆ1 + x2eˆ2 + x3eˆ3
x = xi eˆi
Transformation of Cartesian Coordinate Systems
Consider two Cartesian coordinate systems, in an arbitrary
orientation relative to each other. We label one system by a tilde,
to distinguish one from another, and call it the tilde coordinate
system. The relative position of two coordinate systems in two
dimensions is given in figure.
The origin of the tilde coordinate system has the coordinates w.r.t
non-tilde system
O : ( x ,  y ,  z )
So, its position vector can be written as
 =  x eˆx +  y eˆy +  z eˆz = i ei

The basis vectors eˆi of the tilde coordinate system can be written
in terms of the no-tilde system as
eˆx = 11eˆx +  21eˆy +  31eˆz =  i1eˆi ,
eˆy = 12 eˆx +  22 eˆy +  32 eˆz =  i 2 eˆi ,
eˆz = 13eˆx +  23eˆy +  33eˆz =  i 3eˆi ,
Implies eˆ j =  ij eˆi .
▪ The transformation between no-tilde and tilde coordinate
systems is completely described by i and  ij .
▪ The  ij are called the transformation coefficients or the
transformation matrix.
Properties of the Transformation Coefficients
▪ Let  i be the coordinates of the vector eˆ j , i.e.,
eˆ j =  i eˆi
Comparing this expression with
eˆ j =  ij eˆi
we have
 ij =  i

Implies that  ij is the ith-component of the vector eˆ j with respect

to basis eˆi .
▪ Another interpretation of  ij is as follows. As
eˆ j =  ij eˆi ,
eˆ j . eˆk =  ij eˆi . eˆk ,
eˆ j . eˆk =  ij ik =  kj .
Since eˆ j . eˆk is the cosine of the angle between eˆ j and eˆk ,
we get
 ij = cos(eˆi , eˆ j ).
Where the  ij are the so called direction cosines.
▪  ij is not the same as  ji.
For example ˆ
 ij = cos(eˆi , e j ) and
 ji = cos(eˆ j , eˆi )
• Since both the eˆi and the eˆi are pairwise perpendicular unit
vectors, they satisfy the relations

eˆi . eˆ j =  ij and eˆi . eˆ j =  ij

This gives  ij = eˆi . eˆ j =  mi eˆm . nj eˆn ,
=  mi nj eˆm . eˆn ,
=  mi nj mn ,
 ij =  mi mj .
Hence the transformation coefficients form an orthogonal
matrix and
ik jk = ij , kikj = ij
These are called orthogonality relations.
Where orthogonal matrices satisfy the relation

det  ij  = 1
 Transformation Law for Basis Vectors

Since we have:
𝑒መ𝑗ǁ = 𝛼𝑖𝑗 𝑒Ƹ𝑖 .
Multiply the above equation with 𝛼𝑘𝑗 :
𝛼𝑘𝑗 𝑒መ𝑗ǁ = 𝛼𝑘𝑗 𝛼𝑖𝑗 𝑒Ƹ𝑖
𝛼𝑘𝑗 𝑒መ𝑗ǁ = 𝛿𝑘𝑖 𝑒Ƹ𝑖 = 𝑒Ƹ𝑘
After renaming the indices, we obtain the following transformation law for
basis vectors:

eˆ i =  ij eˆ j ,
eˆi =  ji êi
Example:
Coordinate Transformations
 Transformation law for point coordinates

The position vector to a point P can be expressed as

֊
𝑥 𝑖 = 𝑥𝑖 𝑒Ƹ𝑖
֊
𝑥෤ 𝑖 = 𝑥෤𝑖 𝑒መǁ𝑖
And following is the relation between them
֊
𝑥 = 𝛽 +֊
֊ 𝑥෤

Implies
֊
𝑥 = 𝛽𝑖 𝑒Ƹ𝑖 + 𝑥෤𝑖 𝛼𝑘𝑖 𝑒Ƹ𝑘
𝑥𝑖 𝑒Ƹ𝑖 = 𝛽𝑖 𝑒Ƹ𝑖 + 𝑥෤𝑖 𝛼𝑘𝑖 𝑒Ƹ𝑘

𝑥𝑖 𝑒Ƹ𝑖 = 𝛽𝑖 𝑒Ƹ𝑖 + 𝑥෤𝑗 𝛼𝑖𝑗 𝑒Ƹ𝑖

𝑥𝑖 𝑒Ƹ𝑖 = (𝛽𝑖 + 𝑥෤𝑗 𝛼𝑖𝑗 )𝑒Ƹ𝑖
Since the representation of a position vector with respect to a basis is unique,
we have
𝑥𝑖 = 𝛽𝑖 + 𝑥෤𝑗 𝛼𝑖𝑗
Similarly, the inverse relation is

𝑥෤𝑖 = 𝛼𝑗𝑖 𝑥𝑗 − 𝛼𝑗𝑖 𝛽𝑗

 The Transformation Law for Vector Coordinates
A vector 𝐴,റ in a Cartesian coordinate system with basis vectors 𝑒Ƹ𝑖 , can be
written as
֊
𝐴 = 𝐴1 𝑒1Ƹ + 𝐴2 𝑒Ƹ2 + 𝐴3 𝑒Ƹ3 = 𝐴𝑖 𝑒Ƹ𝑖

We can obtain the transformation law for the coordinates of a vector from the
transformation law for point coordinates. From the figure

A = y − x or A = y − x
so A = y − x implies
A i eˆi = ( yi − xi )eˆi
and A = y − x
A i eˆi = ( yi − xi )eˆi
So we obtain
Ai eˆi = ( yi − xi )eˆi
implies Ai = yi − xi
and similarly
Ai = yi − xi
After using the transformation law for coordinates of points we obtain

𝐴𝑖 = 𝛼𝑖𝑗 𝐴ሚ𝑗
𝑎𝑛𝑑 𝐴ሚ 𝑖 = 𝛼𝑗𝑖 𝐴𝑗
Thus we say that a vector is a quantity which in a coordinate system can be
represented by
֊
𝐴 = 𝐴1 𝑒1Ƹ + 𝐴2 𝑒Ƹ2 + 𝐴3 𝑒Ƹ3 = 𝐴𝑖 𝑒Ƹ𝑖

And whose coordinates transform according to following rule

𝐴𝑖 = 𝛼𝑖𝑗 𝐴ሚ𝑗
𝑎𝑛𝑑 𝐴ሚ 𝑖 = 𝛼𝑗𝑖 𝐴𝑗
Curvilinear Coordinate Systems
Curvilinear Coordinate System

▪ In Curvilinear Coordinate system, the coordinate lines may be curved and

this is not allowed in Cartesian Coordinate system.
▪ The basis set in Curvilinear coordinate system may not be orthogonal
everywhere in space i.e. the basis vectors don’t have to be at 90 degrees to
each other, and they don’t have a magnitude of 1.
Orthogonal Curvilinear Coordinate System

A curvilinear coordinate system is called orthogonal when the basis vectors are
always orthogonal.

Coordinate Basis Vectors Example

Lines
Straight Orthogonal Cartesian coordinate systems

Straight Not Orthogonal Skew coordinate systems

Curved Orthogonal Plane polar coordinate system, Cylindrical coordinate

system, Spherical coordinate system
Curved Not orthogonal Elliptical cylindrical coordinates
Transformation of Coordinates
Let the rectangular coordinates (𝑥, 𝑦, 𝑧) of any point can be expressed as the
function of (𝑢1 , 𝑢2 , 𝑢3 ) so that
x = x(u1 , u2 , u3 )
y = y (u1 , u2 , u3 )
z = z (u1 , u2 , u3 )
and
u1 = u1 ( x, y, z )
u2 = u 2 ( x, y , z )
u3 = u3 ( x, y, z )

Given a point 𝑃 with rectangular coordinates (𝑥, 𝑦, 𝑧), we can associate a unique
set of coordinates (𝑢1 , 𝑢2 , 𝑢3 ) called the curvilinear coordinates of the point 𝑃.
Hence any point 𝑃 can then be defined in space not only by rectangular
coordinates (𝑥, 𝑦, 𝑧), but also by curvilinear coordinates (𝑢1 , 𝑢2 , 𝑢3 ).
Unit vectors
Let 𝑟റ = 𝑥 𝑖+𝑦 Ƹ 𝑘෠
Ƹ 𝑗+z
be the position vector of a point P. Then using the transformation between
cartesian and curvilinear coordinates we have
𝑟റ = 𝑟(𝑢
റ 1 , 𝑢2 , 𝑢3 )

A tangent vector to the 𝑢1 curve at P (for which 𝑢2 and 𝑢3 are constants) is

𝜕𝑟റ
. Then a unit tangent vector in this direction is
𝜕𝑢1
𝜕𝑟
𝜕𝑢1
𝑒1Ƹ = 𝜕𝑟
𝜕𝑢1

Similarly if 𝑒Ƹ2 and 𝑒Ƹ3 are unit tangent vectors to the 𝑢2 and 𝑢3 curves at P,
respectively, then
𝜕𝑟 𝜕𝑟
𝜕𝑢2 𝜕𝑢3
𝑒Ƹ2 = 𝜕𝑟
, 𝑒Ƹ3 = 𝜕𝑟
𝜕𝑢2 𝜕𝑢3
Scale factors

From the unit tangents along 𝑢1 , 𝑢2 and 𝑢3 curve we have

𝜕 𝑟റ 𝜕 𝑟റ 𝜕 𝑟റ
= 𝑒1Ƹ =ℎ1 𝑒1Ƹ where ℎ1 =
𝜕𝑢1 𝜕𝑢1 𝜕𝑢1
And similarly
𝜕 𝑟റ 𝜕 𝑟റ 𝜕 𝑟റ
= 𝑒Ƹ2 =ℎ2 𝑒Ƹ2 where ℎ2 =
𝜕𝑢2 𝜕𝑢2 𝜕𝑢2
𝜕 𝑟റ 𝜕 𝑟റ 𝜕 𝑟റ
= 𝑒Ƹ3 =ℎ3 𝑒Ƹ3 where ℎ3 =
𝜕𝑢3 𝜕𝑢3 𝜕𝑢3
The quantities ℎ1 , ℎ2 and ℎ3 are called scale factors. The scale
factor gives a measure of how a change in the coordinate,
changes the position of a point.
Unit normal vectors:

𝛻𝑢1 is a vector at P normal to the surface 𝑢1 = 𝑐1 , a unit vector

in this direction is given by
𝛻𝑢
𝐸෠1 = 1
𝛻𝑢1
Similarly, the unit vectors
𝛻𝑢
𝐸෠2 = 2 and
𝛻𝑢2
𝛻𝑢3
෠
𝐸3 =
𝛻𝑢3
at P are normal to the surfaces 𝑢2 = 𝑐2 and 𝑢3 = 𝑐3 , respectively.
▪ Thus at each point P of a curvilinear coordinate system there
exist two sets of unit vectors 𝑒1Ƹ , 𝑒Ƹ2, 𝑒Ƹ3 tangent to the coordinate
curves and 𝐸෠1 , 𝐸෠2, 𝐸෠3 normal to the coordinate surface.
▪ The two sets become identical if and only if the curvilinear
system is orthogonal.
▪ Both sets are analogous to 𝑖,Ƹ 𝑗,Ƹ 𝑘෠ unit vectors in rectangular
Cartesian coordinates but are unlike in the way that they may
change directions from point to point.
▪ These tangent vector and normal vectors constitute reciprocal
system of vectors. (A vector having the same direction as that of a
given vector but magnitude equal to the reciprocal of the given vector is
known as the reciprocal of vector.)
Orthogonal curvilinear coordinate system
If the coordinate curves intersect at right angles, the curvilinear coordinate
system is called orthogonal. The 𝑢1 , 𝑢2 𝑎𝑛𝑑 𝑢3 coordinate curves of an
orthogonal curvilinear system are similar to the 𝑥, 𝑦 𝑎𝑛𝑑 𝑧 coordinate axes
of a rectangular Cartesian system. In an orthogonal curvilinear coordinate
system, the unit vectors 𝑒1Ƹ , 𝑒Ƹ2 , 𝑒Ƹ3 are mutually orthogonal.

eˆ1.eˆ1 = eˆ2 .eˆ2 = eˆ3 .eˆ3 = 1, and eˆ1.eˆ2 = eˆ2 .eˆ3 = eˆ3 .eˆ1 = 0
 If the system is right-handed then

eˆ1  eˆ2 = −eˆ2  eˆ1 = eˆ3 , eˆ2  eˆ3 = eˆ3  eˆ2 = eˆ1 , eˆ3  eˆ1 = −eˆ1  eˆ3 = eˆ2
If the system is left-handed then
eˆ1  eˆ2 = −eˆ2  eˆ1 = −eˆ3 , eˆ2  eˆ3 = eˆ3  eˆ2 = −eˆ1 , eˆ3  eˆ1 = −eˆ1  eˆ3 = −eˆ2
Curvilinear Coordinate Systems
A vector in curvilinear coordinate system:
A vector 𝐴റ can be represented in terms of the unit basis vectors 𝑒1Ƹ , 𝑒Ƹ2 , 𝑒Ƹ3 or 𝐸෠1
𝐸෠2 , 𝐸෠3 in the form
֊
𝐴 = 𝐴1 𝑒1Ƹ + 𝐴2 𝑒Ƹ2 + 𝐴3 𝑒Ƹ3
֊
𝐴 = 𝑎 𝐸෠ + 𝑎 𝐸෠ + 𝑎 𝐸෠
1 1 2 2 3 3
Where 𝐴1 , 𝐴2 , 𝐴3 and 𝑎1 , 𝑎2 , 𝑎3 are the respective components of vector 𝐴റ in
each system.
𝜕𝑟റ 𝜕𝑟റ 𝜕𝑟റ
We can also represent 𝐴റ in terms of , , or 𝛻𝑢1 , 𝛻𝑢2 , 𝛻𝑢3 as
𝜕𝑢1 𝜕𝑢2 𝜕𝑢3

֊ 𝜕𝑟റ 𝜕 𝑟റ 𝜕 𝑟റ
𝐴 = 𝐶1 + 𝐶2 + 𝐶3
𝜕𝑢1 𝜕𝑢2 𝜕𝑢3
֊
𝐴 = 𝑐 𝛻𝑢 + 𝑐 𝛻𝑢 + 𝑐 𝛻𝑢
1 1 2 2 3 3

Where 𝐶1 , 𝐶2 , 𝐶3 are called the contravariant components of the vector 𝐴റ and

റ
𝑐1 , 𝑐2 , 𝑐3 are called the covariant components of 𝐴.
Arc length in general curvilinear coordinates
The square of the element of arc length in general curvilinear
coordinates can be expressed by

ds 2 =  hi h j ( eˆi .eˆ j )dui du j

3 3

i =1 j =1

r = r (u1 , u2 , u3 )
r r r
dr = du1 + du2 + du3
u1 u2 u3
dr = h1du1eˆ1 + h2 du2eˆ2 + h3du3eˆ3
ds 2 = dr .dr
Arc length element in orthogonal curvilinear coordinates

The arc length ds is the length of the infinitesimal vector 𝑑 𝑟റ :

( ds ) = dr .dr
2

In curvilinear coordinates, we have

r = r (u1 , u2 , u3 )
r r r
dr = du1 + du2 + du3
u1 u2 u3
dr = h1du1eˆ1 + h2 du2eˆ2 + h3du3eˆ3
ds 2 = h12 du12 + h22 du22 + h32 du32
Area element in an orthogonal curvilinear system:
Along a 𝑢1 curve, 𝑢2 𝑎𝑛𝑑 𝑢3 are constants so that 𝑑 𝑟റ = ℎ1 𝑑𝑢1 𝑒1Ƹ . Then the
differential of arc length 𝑑𝑠1 along 𝑢1 at P is ℎ1 𝑑𝑢1 .
The area elements are given by

dA1 = (h2 du2 eˆ2 )  (h3du3eˆ3 )

dA1 = h2 h3 eˆ2  eˆ3 du2 du3 = h2 h3du2 du3
As eˆ2  eˆ3 = eˆ1 = 1
similarly
dA2 = (h1du1eˆ1 )  (h3du3eˆ3 ) = h1h3du1du3
dA3 = (h1du1eˆ1 )  (h2 du2eˆ2 ) = h1h2 du1du2
Volume element:
We know that the absolute value of the scalar triple product gives the
volume of the parallelepiped.

dv = (h1du1eˆ1 ).(h2 du2eˆ2 )  (h3du3eˆ3 )

dv = h1h2 h3du1du2 du3 eˆ1.eˆ2  eˆ3 = h1h2 h3du1du2 du3
As eˆ1.eˆ2  eˆ3 = 1
Jacobian of transformation
If 𝑢1 , 𝑢2 and 𝑢3 are the orthogonal curvilinear coordinates and

x = x(u1 , u2 , u3 ), y = y (u1 , u2 , u3 ), z = z (u1 , u2 , u3 )

x, y , z
then the Jacobian of 𝑥, 𝑦, 𝑧 is denoted by J or J ( ) is defined as
u1 , u2 , u3

x y z
u1 u1 u1
x, y , z  ( x, y , z ) x y z
J = J( )= =
u1 , u2 , u3  ( u1 , u2 , u3 ) u2 u2 u2
x y z
u3 u3 u3
 From the definition of a scalar triple product, the above determinant
can be written as

x, y , z  x y z   x y z   x y z 
J( )= i+ j+ k  . i+ j+ k  i+ j+ k
u1 , u2 , u3  u1 u1 u1   u2 u2 u2 
  3u u3 u3 

x, y , z r r r
J( )= .  = h1 eˆ1 .h2 eˆ 2  h3 eˆ 3
u1 , u2 , u3 u1 u2 u3
x, y , z
J( ) = h1h2 h3 (eˆ1 .eˆ 2  eˆ 3 )
u1 , u2 , u3
x, y , z
J( ) = h1h2 h3 ( since eˆ1 .eˆ 2  eˆ 3 = 1)
u1 , u2 , u3
𝜕𝑟റ 𝜕𝑟റ 𝜕𝑟റ
If the Jacobian equals to zero identically then 𝜕𝑢 , 𝜕𝑢 , 𝜕𝑢 are coplanar vectors
1 2 3
and the linear coordinate transformation breaks down, i.e., there is a relation
between 𝑥, 𝑦, 𝑧 having the form 𝐹 𝑥, 𝑦, 𝑧 = 0. We shall therefore require the
Jacobian to be different from zero.
Curvilinear Coordinate Systems
Example: Find the scale factors, arc length and Jacobian in rectangular
Cartesian coordinates.
Example: Consider the curvilinear coordinates system defined by
x = u3 , y = eu2 cos u1 , z = eu2 sin u1
Find the unit vectors 𝑒1Ƹ , 𝑒Ƹ2 , 𝑒Ƹ3 and show that the system is orthogonal. Compute the
scales factors ℎ1 , ℎ2 , 𝑎𝑛𝑑 ℎ3 .
Example: Check that the following system of elliptical cylindrical coordinates are
orthogonal or not?
֊
𝑟 = 𝑎𝜌cos𝜃 𝑖Ƹ + 𝑏𝜌sin𝜃 𝑗Ƹ + 𝑧𝑘෠ ; 𝑎 ≠ 𝑏
 Differential Operators in
Orthogonal Curvilinear Coordinate
Systems
Nabla operator
The differential operator nabla often appear in vector analysis. In
the space of three variables it is defined as
  
= i+ j+ k
x y z

▪ Gradient
▪ Divergence
▪ Curl
▪ Laplacian
Gradient of a scalar field in curvilnear coordinates
Let 𝑢1 ,𝑢2 and 𝑢3 are the curvilinear coordinates.

Let  = f1eˆ1 + f 2eˆ2 + f 3eˆ3

Where 𝑓1 , 𝑓2 𝑎𝑛𝑑 𝑓3 are to be determined.

Since dr = r du1 + r du2 + r du3

u1 u2 u3
dr = h1du1eˆ1 + h2 du2eˆ2 + h3du3eˆ3
we have
d =  .dr = h1 f1du1 + h2 f 2 du2 + h3 f 3du3
  
d = du1 + du2 + du3
u1 u2 u3
 1  1  1 
f1 = , f2 = , f3 =
h1 u1 h2 u2 h3 u3
Then
1  1  1 
 = eˆ1 + eˆ2 + eˆ3
h1 u1 h2 u2 h3 u3
Expression for Divergence of a vector field
Consider
. A = .( A1eˆ1 + A2 eˆ2 + A3eˆ3 )
. A = .( A1eˆ1 ) + .( A2 eˆ2 ) + .( A3eˆ3 )
since eˆ1 , eˆ2 , and eˆ3 form a right handed system, therefore
eˆ1 = eˆ2  eˆ3 = h2 h3u2  u3
eˆ2 = eˆ3  eˆ1 = h3h1u3  u1
eˆ3 = eˆ1  eˆ2 = h1h2u1  u2
Now .( A1eˆ1 ) = . ( A1h2 h3u2  u3 )
=  ( A1h2 h3 ) .u2  u3 + A1h2 h3. ( u2  u3 )
 . ( A  B ) = B. (   A ) − A. (   B ) and    = 0
 . ( u2  u3 ) = u3 .(  u2 ) − u2 .(  u3 ) = 0
 eˆ2 eˆ3
 .( A1eˆ1 ) =  ( A1h2 h3 ) .  +0
h2 h3
eˆ
=  ( A1h2 h3 ) . 1
h2 h3
1  1  1  eˆ
=[ ( A1h2 h3 ) eˆ1 + ( A1h2 h3 ) eˆ2 + ( A1h2 h3 ) eˆ3 ]. 1
h1 u1 h2 u2 h3 u3 h2 h3
1 
.( A1eˆ1 ) = ( A1h2 h3 )
h1h2 h3 u1
similarly
1 
.( A2 eˆ2 ) = ( A2 h3h1 )
h1h2 h3 u2
1 
and .( A3eˆ3 ) = ( A3h1h2 )
h1h2 h3 u3
Hence,
1     
. A =  ( A h h ) + ( A h h ) + ( A )
3 1 2 
h h
  
1 2 3 2 3 1
h1h2 h3  1u u 2 u 3 
Expression of curl

Consider
  A =   ( A1eˆ1 + A2 eˆ2 + A3eˆ3 )
=   ( A1eˆ1 ) +   ( A2 eˆ2 ) +   ( A3eˆ3 )
  ( A1eˆ1 ) =   ( A1h1u1 ) (As, eˆ1 = h1u1 )
=  ( A1h1 )  u1 + A1h1  u1
eˆ1
=  ( A1h1 )  + 0 (As,   u1 = 0)
h1
1  1  1   eˆ1
= ( A1h1 ) eˆ1 + ( A1h1 ) eˆ2 + ( A1h1 ) eˆ3  
 h1 u1 h2 u2 h3 u3  h1
1  1 
  ( A1eˆ1 ) = ( 1 1) 2
A h ˆ
e − ( A1h1 ) eˆ3
h3h1 u3 h1h2 u2
similarly
1  1 
  ( A2 eˆ2 ) = ( A2 h2 ) eˆ3 − ( A2 h2 ) eˆ1
h1h2 u1 h2 h3 u3
1  1 
  ( A3eˆ3 ) = ( A3h3 ) eˆ1 − ( A3h3 ) eˆ2
h2 h3 u2 h3h1 u1
Hence,
 
eˆ1   eˆ2    
 A =  ( 3 3)
A h − ( 2 2 )
A h +  ( 1 1)
A h − ( 3 3 )
A h

 2
h2 h3 u u 3  3 1 3
h h u u 1 
eˆ    
+ 3  ( A2 h2 ) − ( A1h1 ) 
h1h2  u1 u2 
This can be write as
h1eˆ1 h2 eˆ2 h3eˆ3
1   
 A =
h1h2 h3 u1 u2 u3
A1h1 A2 h2 A3h3
Expression of Laplacian
Consider
1  1  1 
 = eˆ1 + eˆ2 + eˆ3 . and
h1 u1 h2 u2 h3 u3
1   1  1  
. A =  ( A1h2 h3 ) + ( A2 h3h1 ) + ( A3h1h2 )
 u1
h1h2 h3 h1h2 h3 u2 h1h2 h3 u3 
1  1  1 
Let , A =  then A1 = , A2 = , A3 =
h1 u1 h2 u2 h3 u3
and . A = . =  2
1   h2 h3   h3h1   h1h2  
 =
2
 ( )+ ( )+ ( )
h1h2 h3 
 1 1
u h u1 u 2 h2 u 2 u3 h3 u3 
 Differential Operators in
Orthogonal Curvilinear Coordinate
Systems
Example: Consider the curvilinear coordinate system defined by

x = u3 , y = eu2 cos u1 , z = eu2 sin u1

Find ∇𝜓 and ∇2 𝜓, If 𝜓=𝑢12 + 𝑢22 + 𝑢32 .

  
= 2u1 , = 2u2 , = 2u3
u1 u2 u3
In an orthogonal coordinate system
1  1  1 
 = eˆ1 + eˆ2 + eˆ3
h1 u1 h2 u2 h3 u3
= e − u2 (2u1 )eˆ1 + e − u2 (2u2 )eˆ2 + (2u3 )eˆ3
= 2e − u2 (u1eˆ1 + u2eˆ2 ) + (2u3 )eˆ3
and
1   h2 h3   h3h1   h1h2  
 =
2
 ( )+ ( )+ ( )
h1h2 h3  u1 h1 u1 u2 h2 u2 u3 h3 u3 
   e u2    e u2   u2 u2 
= u2 u2
1

(e )(e )(1)  u1  e
 u2 (2u1 )  +  u2 (2u2 )  + ( e e (2u3 ) ) 
 u2  e  u3 
−2 u2     2 u2 
=e  (2u1 ) + (2u2 ) + ( e (2u3 ) ) 
 u1 u2 u3 
= e −2u2  2 + 2 + 2e 2u2  = e −2u2  4 + 2e 2u2 
Example. Find ∇. 𝐴റ and ∇ × 𝐴റ in the system defined in previous example for
the vector field
𝐴റ = 𝑢3 𝑒1Ƹ − 𝑢2 𝑒Ƹ3 .
1     
. A =  (h2 h3 A1 ) + (h1h3 A2 ) + (h1h2 A3 ) 
h1h2 h3
 1u  u 2 u 3 
1     
= u2 u2  ( u (1) e u2
) + ((0)(1) e u2
) + ( −u ( e u2
)( e u2
)) 
(e )(e )(1)  u1 u2 u3
3 2

1     3u 2  1     
=  (u e u2
) + (0) − ( e )  =  (0) + (0) − (0) 
e 2u2  u1 u2 u3   
3 2 u2
 e  1u u 2 u 3 
1
. A = 2u2  (0) + (0) − (0) 
e
. A = 0
and
h1eˆ1 h2 eˆ2 h3eˆ3
1   
 A =
h1h2 h3 u1 u2 u3
A1h1 A2 h2 A3h3
eu2 eˆ1 eu2 eˆ2 eˆ3
1   
= u2 u2
(e )(e )(1) u1 u2 u3
u3eu2 0 − e u2
1  u2    u2 
= 2u2 e eˆ1 (−e − 0) + e eˆ2
u2 u2
(u3e − 0) + eˆ3 (0 −
u2
(u3e ) 
e  u2 u3 u2 
1
= 2u2 eu2 eˆ1 (−eu2 ) + eu2 eˆ2 (eu2 ) − (u3eˆ3 ) 
e
1
  A = u2  −eu2 eˆ1 + eu2 eˆ2 − u3eˆ3 
e
Example: Find the expressions of gradient, divergence and curl in parabolic
cylindrical coordinates defined by
Example: Using the expressions of the last example evaluate
the gradient of scalar field
Example. Find ∇. 𝐴റ and ∇ × 𝐴റ in parabolic cylindrical coordinates
system for the vector field
𝐴റ = 𝑢3 𝑒1Ƹ − 𝑢2 𝑒Ƹ3 .
Polar Coordinate System
Polar coordinate system
The polar coordinate system is a two-dimensional coordinate system in which
each point on a plane is determined by a distance “r” from the pole or reference
point and an angle “θ”.

▪ The reference point is called pole.

▪ The distance 𝑟 from the pole is called the radial coordinate and the angle is
called the angular coordinate.
The coordinate lines for r are circles, those for theta are lines. The
circles always intersect the lines at right angles.
Polar coordinates to rectangular coordinates

Following are the conversion formulas from polar to rectangular coordinates.

𝑥 = 𝑟𝑐𝑜𝑠𝜃

𝑦 = 𝑟𝑠𝑖𝑛𝜃

and

r 2 = x2 + y 2  r = x2 + y 2

Tan = y   = Tan −1 y
x x
Example No 1:
𝜋
Convert the polar coordinates (3, ) as rectangular coordinates.
2
x = r cos 
= 3cos 
2
= 3(0) = 0
x=0

y = r sin 
= 3sin 
2
= 3(1) = 3
y =3

The rectangular coordinates are (0,3).

Example No 2:
Convert the rectangular coordinates (3,3) to polar coordinate.

r 2 = x2 + y 2
r 2 = (3) 2 + (3) 2  r2 = 9 + 9
r = 18  r =3 2

y
Tan  =
x
3
Tan  =  Tan  = 1
3

 = Tan −1 (1)  =
4
𝜋
Hence the polar coordinates are(3 22, ).
4
Example No 3

Convert the polar equation 4𝑟 cos 𝜃 + 𝑟 sin 𝜃 = 8 into a rectangular

equation that expresses 𝑦 in terms of 𝑥.
Solution: Using

𝑥 = 𝑟𝑐𝑜𝑠𝜃
𝑦 = 𝑟𝑠𝑖𝑛𝜃

4𝑟 cos 𝜃 + 𝑟 sin 𝜃 = 8
4 𝑥 + 𝑦 = 8
4𝑥 + 𝑦 = 8
𝑦 = 8 – 4𝑥
Unit Vectors in Polar Coordinates:
The position vector in plane polar coordinates is:
𝑟റ = 𝑥 𝑖Ƹ + 𝑦𝑗Ƹ
Using the transformation 𝑥 = 𝑟 cos 𝜃, 𝑦 = 𝑟 sin 𝜃
𝑟റ = 𝑟 cos 𝜃 𝑖Ƹ + 𝑟 sin 𝜃 𝑗Ƹ
𝜕𝑟റ
=cos 𝜃 𝑖Ƹ + sin 𝜃 𝑗Ƹ
𝜕𝑟
𝜕𝑟റ
=−𝑟 sin 𝜃 𝑖Ƹ + 𝑟 cos 𝜃 𝑗Ƹ
𝜕𝜃
𝜕𝑟
𝜕𝑟
𝑒1Ƹ = 𝜕𝑟
=cos 𝜃 𝑖Ƹ + sin 𝜃 𝑗Ƹ
𝜕𝑟

𝜕𝑟
𝜕𝜃 Ƹ
−𝑟 sin 𝜃 𝑖+𝑟 cos 𝜃 𝑗Ƹ
𝑒Ƹ2 = 𝜕𝑟
= = −sin 𝜃 𝑖Ƹ + cos 𝜃 𝑗Ƹ
𝑟
𝜕𝜃

The scale factors are:

𝜕𝑟റ 𝜕𝑟റ
ℎ1 = =1 and ℎ2 = =𝑟
𝜕𝑟 𝜕𝜃
Cylindrical Polar Coordinates
Cylindrical Polar Coordinates
Cylindrical coordinates simply combine the polar coordinates in the 𝑥𝑦-plane with the
usual 𝑧 coordinate of Cartesian coordinate system. Hence cylindrical polar coordinates
are represented by (𝑟, 𝜃, 𝑧).

Let 𝑃(𝑥, 𝑦, 𝑧) be any point whose projection on the 𝑥𝑦 − 𝑝𝑙𝑎𝑛𝑒 is 𝑄(𝑥, 𝑦). Then its
cylindrical coordinates are 𝑟, 𝜃, 𝑧 .
Transformation Equations from Cylindrical coordinates to
Cartesian coordinates:
From this figure the transformation equations expressing the rectangular
coordinates in terms of cylindrical coordinate:

x = r cos 
y = r sin 
z=z Where 𝑟 ≥ 0, 0≤ 𝜃 ≤ 2𝜋, and -∞ < 𝑧 < ∞.
Transformation Equations from Cartesian coordinates to Cylindrical
coordinates:
squaring and adding,
x = r cos 
y = r sin 
we have
x 2 + y 2 = r 2 (cos 2  + sin 2  ) = r 2
or r = x2 + y 2
Moreover:
y r cos 
= = tan 
x r sin 
y
or  = tan −1
x
and z = z
Example. Convert the cylindrical coordinates 4, 𝜋, 1 in
rectangular coordinates.
Tangent vectors in cylindrical coordinate system:
The position vector of any point 𝑃 in cylindrical polar coordinates is

r = xi + y j + zk
r = r cos  i + r sin  j + zk
The tangent vectors in the direction of 𝑟, 𝜃 𝑎𝑛𝑑 𝑧 respectively, are given by

֊
𝑟 = 𝑥 𝑖Ƹ + 𝑦𝑗Ƹ + 𝑧𝑘෠
𝜕֊𝑟
= cos𝜃𝑖Ƹ + sin𝜃𝑗Ƹ
𝜕𝑟
𝜕֊
𝑟
= −𝑟sin𝜃𝑖Ƹ + 𝑟cos𝜃 𝑗Ƹ
𝜕𝜃
𝜕֊ 𝑟
= 𝑘෠
𝜕𝑧
Unit tangent vectors in cylindrical coordinate system:

The unit vectors in these direction of r ,  and z are given by

r
cos  i + sin  j
eˆr = r = = cos  i + sin  j
r cos  + sin 
2 2

r
r
−r sin  i + r cos  j
eˆ =  = = − sin  i + cos  j
r r sin  + r cos 
2 2 2 2


r
eˆz = z = k
r
z
In matrix notaion, we have

 eˆr   cos  sin  0  i 

eˆ  =  − sin    
   cos  0  j 
 eˆz   0  
0 1   k 
 
Scale factors:
The scale factor for the cylindrical coordinates are given by

r
h1 = hr = = cos 2  + sin 2  = 1
r
r
h2 = h = = (−r sin  ) 2 + (r cos  ) 2

= r cos 2  + sin 2  = r
r
h3 = hz = = k =1
z
hence thescale factor are:
h1 = hr = 1 h2 = h = r h3 = hz = 1
Question: In cylindrical polar coordinates, show that.

eˆr eˆ eˆ

= 0, r = eˆ , r = 0
r  z
eˆ eˆ eˆ
= 0,  = −eˆr ,  = 0
r  z
eˆz eˆz eˆz
= 0, = 0, =0
r  z
Question: Prove that in cylindrical polar coordinates
d . d . d
eˆr =  eˆ  , eˆ  = −  eˆr , eˆz = 0
dt dt dt

Where dot denote differentiation w.r.t time t.

eˆr = cos  i + sin  j , eˆ = − sin  i + cos  j, eˆz = k

d d d d .
eˆr = − sin  i + cos  j = (− sin  i + cos  j ) =  eˆ ,
dt dt dt dt
d d d d .
eˆ = − cos  i − sin  j = −(cos  i + sin  j ) = −  eˆr
dt dt dt dt
d d
eˆz = k = 0
dt dt
Arc length element:

In cylindrical polar coordinates, the elements of arc length are determined from

u1 = r , u2 =  , u3 = z , h1 = 1, h2 = r , h3 = 1
(ds) 2 = h12 (du1 ) 2 + h22 (du2 ) 2 + h32 (du3 ) 2
In cylindrical polar coorrdinates, these becomes
(ds) 2 = (1) 2 (dr ) 2 + (r ) 2 (d ) 2 + (1) 2 (dz ) 2
(ds) 2 = (dr ) 2 + r 2 (d ) 2 + (dz ) 2
Area elements
we know that the elements of area in an orthogonal curvilinear coordinates are:
dA1 = h2 h3du2 du3 , dA2 = h1h3du1du3 dA3 = h1h2 du1du2
In cylindrical polar coordinates, we have
dA1 = (r )(1)d dz = rd dz
dA2 = (1)(1)drdz = drdz
dA3 = (1)(r )drd = rdrd

Volume element

dv = h1h2 h3du1du2 du3

In cylindrical polar coordinate, we have
dv = (1)(r )(1)drd dz = rdrd dz
Expression for Jacobian in cylindrical polar coordinates:

We know that the Jacobian in orthogonal curvilinear coordinates

𝑢1 , 𝑢2 , 𝑢3 is given by

x, y , z
J = J( ) = h1h2 h3
u1 , u2 , u3
h1 = 1, h2 = r , h3 = 1 and u1 = r , u2 =  , u3 = z
x, y , z
J( ) = (1)( r )(1) = r
r , , z
Cylindrical Coordinate System
 𝜕 𝜕 𝜕
Example: Find the expressions of , , in cylindrical coordinate system.
𝜕𝑥 𝜕𝑦 𝜕𝑧
x = r cos  , y = r sin  , z=z and
y
r = x 2 + y 2 ,  = tan −1 ( ), z = z
x
r x r cos 
= = = cos 
x x2 + y 2 r

𝜕𝑟 𝑦𝑟sin𝜃
= = = sin𝜃
𝜕𝑦 2
𝑥 +𝑦 2 𝑟
𝜕𝑟
=0
𝜕𝑧
𝜕𝜃 𝑦 𝑟sin𝜃 sin𝜃
=− 2 = 2 =−
𝜕𝑥 𝑥 + 𝑦2 𝑟 𝑟
𝜕𝜃 𝑥 𝑟cos𝜃 cos𝜃
= = =
𝜕𝑦 𝑥 2 + 𝑦 2 𝑟2 𝑟
𝜕𝜃
=0
𝜕𝑧
𝜕𝑧 𝜕𝑧 𝜕𝑧
= 0, = 0, =1
𝜕𝑥 𝜕𝑦 𝜕𝑧
Now by the chain rule, we have
  r    z  sin   
= + + = cos  − + (0)
x r x  x z x r r  z
 sin  
= cos  −
r r 
  r    z  cos   
= + + = sin  + + (0)
y r y  y z y r r  z
 cos  
= sin  +
r r 
  r    z   
= + + = ( 0) + ( 0) + (1)
z r z  z z z r  z

=
z
Expression for gradient, divergence, curl, and Laplacian in cylindrical polar
coordinates:
Cylindrical polar coordinates are (𝑟, 𝜃, 𝑧).

h1 = 1, h2 = r , h3 = 1,

Gradient:

In orthogonal curvilinear coordinates, we have

1  1  1 
 = eˆ1 + eˆ2 + eˆ3
h1 u1 h2 u2 h3 u3
In cylindrical polar coordinates, this becomes
1  1  1 
 = eˆr + eˆ + eˆz
1 r r  1 z
 1  
 = eˆr + eˆ + eˆz
r r  z
Divergence:

In orthogonal curvilinear coordinates, we have

1     
.A=  ( A1h 2 h 3 ) + ( A 2 h 3h1 ) + ( A3h1h 2 )
h 1h 2 h 3 u1 u 2 u 3 
In cylindrical polar coordinates, we have
1    
. A = ( ( r )(1) Ar) + ( (1)(1) A) + ( (1)( r ) Az )
(1)(r )(1)  r  z 
1  A  
. A =  ( rAr ) +  + ( rAz ) 
r  r  z 
Curl:

In orthogonal curvilinear coordinates, we have

h1eˆ1 h2 eˆ2 h3eˆ3
1   
 A =
h1h2 h3 u1 u2 u3
A1h1 A2 h2 A3h3
In cylindrical polar coordinates, we have
eˆr reˆ eˆz
1   
 A =
r u1 u2 u3
Ar rA Az
1  A    A A    Ar  
  A =  z − ( rA )  eˆr +  r r − r z ˆ +
  
e ( rA ) −  eˆz 
r   z   z r   r z  
Laplacian:

In orthogonal curvilinear coordinates, we have

1   h2 h3   h3h1   h1h2  
 =
2
 ( )+ ( )+ ( )
h1h2 h3  u1 h1 u1 u2 h2 u2 u3 h3 u3 
In cylindrical polar coordinates,it becomes
1   (r )(1)h3   (1)(1)   (1)(r )  
 =
2
 ( )+ ( )+ ( )
(1)(r )(1)  r (1) r  (r )  z (1) z 
1   1  2  2
 =
2
(r )+ 2 + 2
r r r r  2
z