Advanced Mathematics for Engineers

Chapter 5
Vector and Tensor Analysis
 Contents
1. Scalar and Vector fields
2. Introduction to Tensors

CH 5 Vec. and Ten

2
 Scalar and Vector Fields
 A vector function is called a field. i.e., If a function has more
than one independent variables, then it is called a field.
 A scalar-valued field is a scalar field, and a vector-valued
field is called a vector field.
 Example:
A temperature field of a certain surface is a scalar field.
A distance function in space is a scalar field.
A velocity function of a moving object in space is a vector
field.

CH 5 Vec. and Ten

3
 Gradient of f: grad f or f
 Let f be a real variable in three dimensional scalar field and
assume that it is differentiable w.r.t each of the its variables.
The gradient of f, denoted by grad f or f is given by
f f f
grad f  f  i j k
x y z
  
where nabla ,   i  j k
 Example: x y z
Let f (x, y, z) = 2x + xy - yz2 be a scalar field.
  
Then f x, y, z   2  y i  x  z 2 j  2 yz k
(2 + y, x - z2, 2yz) is the gradient and it is a vector field.
 Hence the gradient of a scalar field is a vector field.
 If f  p   V  p  , then f (p) is called a potential function of V(p).

CH 5 Vec. and Ten

4
 Divergence of V: div V
 Consider a vector field V in cartesian coordinate system.
 If V(x, y, z) = v1(x, y, z)i+v2(x, y, z)j+v3(x, y, z) k, then the
function
v1 v2 v3
div V   V   
x y z
is called the divergence of V and is a scalar field.
 Example: Let V(x, y, z) = 3xyi - 2yzj + xz2k.
 Then div V = 3y - 2z + 2xz, [ε R].
Note that    
j  k   v1i  v2 j  v3k 

div V   V   i 
 x y z 
 The physical significance of div V at a point P is that, it
describes the outflow of V per unit volume at P.

CH 5 Vec. and Ten

5
 Rotation of V: Rot V or Curl V
 Let V(x, y, z) = v1(x, y, z)i+v2(x, y, z)j+v3(x, y, z) k, be a
differentiable vector field in Cartesian coordinate. The vector
function (field) curl (or rotation) of V is given by
i j k
  
Rot V    V 
x y z
v1 v2 v3
 v v   v v   v v 
  3  2 i   1  3  j   2  1 k
 y z   z x   x y 

 Rot V(P) is twice the angular velocity of the flowing fluid

V(P) at point P.

CH 5 Vec. and Ten

6
 Bases of Vector Spaces and Linear Transformation
 Let W be a vector space over the field, say, K (could be Real or
Complex). Recall that
 The set of vectors {v1, v2, …, vm} in W is said to be linearly
independent (LI) if for any scalars α1,α2, …, αm ε K
m

 v i i  0  i  0 i  1, 2, ... m. m
i 1
Otherwise (i.e., if there is at least one αi ≠ 0, with

i 1
vi i 0
)
the set of vectors is called linearly dependent (LD).
 The maximal set of linearly independent set of vectors in W is
called a basis for W. OR a set B = {v1, v2, …, vm} is a basis for W
if B is a LI set and every vector v ε W can be written as linear
combination of vectors in B

CH 5 Vec. and Ten

7
 Example
1. The set {(1, 0), (0, 1)} is a basis for 2D Cartesian, called
the standard basis.
2. The set {(1, 1); (-1, 1)} is also a basis for 2D Cartesian
(Verify!)
3. The set B = {(1, 0, 0), (0, 1, 0), (0, 0, 1)} is the standard
basis for 3D Cartesian.
4. The set B' = {(1, 1, 0), (0, -1, 1), (0, 0, 1)} is also a basis
for 3D Cartesian. (Verify!)
 Remark
 A basis for any vector space is not unique !

 Exercise: Calculate the value of a for which the vectors

(1,1,0) (a,1,1), and (1,a,0) form a basis.
CH 5 Vec. and Ten
8
 Component Vectors
 Let B = {e1, e2, …, en} form a basis for the N-D space. Then
for any x ε RN, the expression x = α1e1 + α2e2 + … + αnen is
unique.
 In this case the scalars α1, α2, … and αn are called the
components of x w.r.t. the basis vectors e1, e2 …, and en
respectively and hence x = (α1, α2, …, αn ).
 Example: If x = (a, b, c) ε R3, then
x = (a, b, c) = a(1, 0, 0) + b(0, 1, 0) + c(0, 0, 1) = ae1 +
be2 + ce3
 Hence the numbers a, b, c are components of x w.r.t the basis
vectors e1, e2, e3 respectively.

CH 5 Vec. and Ten

9
  Example. Let x = (2, 3) ε R2. Then 2 and 3 are components of
x w.r.t the standard basis of R2.
 Moreover, since x = (2, 3) = 2.5(1, 1) + 0.5(-1, 1), the
numbers 2.5 and 0.5 are components of x w.r.t the basis B =
{(1, 1), (-1, 1)}. i.e.
x = (2.5,0.5) w.r.t B.
 Definition
 A basis B = {v1, v2, …, vm} of a vector space V is said to be
Orthogonal if every pair of vectors in B are mutually
perpendicular. i.e. vi . vj = 0 for all i ≠ j .
Orthonormal if B is orthogonal and each of the vectors in
B are unit vectors. i.e. the magnitude being 1.
CH 5 Vec. and Ten
10
  If B = {e1, e2, …, en} is an Orthonormal basis for a vector
space Rn, then the components of any vector x ε Rn with
respect to the basis B can be found by
 i  x  ei , i  1, 2, ... , n.
 A function whose arguments and values are all vectors is
known as a transformation. A transformation T from a
vector space V1 into V2 (i.e., T: V1 → V2) is called a linear
transformation if
(Additivity) T(x + y) = T(x) + T(y) for all vectors x, y ε
V1 and
(Linearity) T(αx) = αT(x) for all vector x ε V1 and any
scalar α ε K.
CH 5 Vec. and Ten
11
  If we know how a linear function transforms the base vectors,
then we can determine the transformation of any arbitrary
vector x by the function f .
 To see this, let us denote
e1  f e1 , e2  f e2 ,... , en  f en 
Then
n
aij  ei  e j iff ei   aij ei
i 1

CH 5 Vec. and Ten

12
  Now consider an arbitrary vector x ε Rn. Then x can be
written uniquely as x = α1e1 + α2e2 + … + αnen . Thus

f ( x)  M 1 ,  2 , ... ,  n 
T

 where M = (f (e1), f (e2), … , f (en)) is an n x n matrix with

each f (ei ) is its column vector.
 The matrix M is called the matrix representation of f relative
1
to the bases B = {e1, e2,…, en} and B'  e , e , ... , e
2 n 

CH 5 Vec. and Ten

13
  Example
 Let B1 = {e1, e2} = {(1, 0)T , (0, 1)T} and
 B'  e1 , e2  = {(1, 1)T , (-1, 1)T}
e1  f (e1 ) and e2  f (e2 )
 Then we have f (e1) = (1, 1)T and f (e2) = (-1, 1)T.
 Thus for any vector x ε R2 we have
1  1 1  1  1
f ( x)       M   
1 1 
 2  1 1 
 Every linear transformation L of Rn into Rm can be
represented in terms of an m x n matrix M and conversely.

CH 5 Vec. and Ten

14
 2. Introduction to Tensors
 Mathematically: Tensors are arrays of numbers, or functions, that
transform according to certain rules under change of coordinates.
 In physics and engineering: Tensors characterize the properties
of a physical system.
 Definition: Tensors are arrays of numbers (or functions), that the
physical quantities they represent do not change under any
transformation.
 A scalar has no direction. Thus it represents a quantity without a
direction information.
 A vector is a single quantity with a magnitude and one direction.
 There are also other physical quantities which can not be described
with only one direction.
For instance, a stress on an elastic spherical membrane.
CH 5 Vec. and Ten
15
  Definition: The order (or rank) of a tensor is the number of
independent directions required to uniquely specify the physical
quantity it represents.
 Describing Tensors in terms of their magnitude and directions
will be very difficult (even impossible) when the order is greater
than 1. →So, we employ component wise description of a Tensor.
 A tensor may consist of a single number: which is referred to as a
tensor of order zero.
 Vectors containing n components in n-dimensional space, is a
tensor of order one.
 Square Matrices containing n components in n-dimensional
space, is a tensor of order two.
 To define tensors of higher orders we need the Operations called
Tensor Product and Contraction.
CH 5 Vec. and Ten
16
 Properties of Tensor Product and Contraction
 Tensors can be multiplied by other tensors to form new tensors
 Tensor product is not commutative in general (unless one of them
is a zero order tensor). Similarly, a contraction is not
commutative in general (except that both of them are of rank 1).
 The rank (order) of the new tensor formed by the product of two
other tensors is the sum of the ranks of the individual tensors.
 The rank of a new tensor formed by the inner product
(contraction) of two other tensors is the sum of their individual
ranks minus 2.
 A tensor of rank r in n-dimensional space has nr components.

CH 5 Vec. and Ten

17
 How to identify a certain quantity as a tensor?
 Tensors are arrays of numbers (or functions), that the physical
objects or quantities they represent do not change under any
transformation.
 Example (Scalars)
 A quantity m representing the mass of an object does not
change under any transformation. Hence m is a Tensor of order
0.
 But let F represent the frequency of light emanating from a
monochromatic source at a point P. If observers at reference
points Q and R measure the frequency of the light, they get a
different value. Hence though F is a scalar, it is not a tensor of
order zero, as it varies under transformation of reference points.

CH 5 Vec. and Ten

18
  Example (Vectors)
 Consider a vector V in n-dimensional space whose tail is at the
point A and head is at point B. Then the coordinates A and B
may vary under coordinate transformation (or if measured at
different reference points), but the vector V*= B* - A* is the
same as V = B – A. Hence V is a tensor of order 1.
 However, if V represents a position vector of a point P in one
coordinate space, say K, i.e., V = OP, then V is not invariant
under coordinate transformation, as V* = O*P is not the same
as V. Hence, position vectors are vectors but NOT tensors, as
they vary with coordinate transformation.

CH 5 Vec. and Ten

19
 Notes
 All scalars are NOT tensors; but all tensors of rank 0 are scalars.
 All vectors are NOT tensors; but all tensors of rank 1 are vectors.
 All matrices are NOT tensors; but all tensors of rank 2 are matrices.
 ...
 Just like vectors, scalars or matrices, Tensors are also single
entities.
 Vectors can be well specified either by using its components or
using its magnitude and a direction. However, a tensor in general, is
specified only by its components. But these components depend on
the system of references or coordinates (or coordinate bases).
 To recognize whether the set of components form a tensor, we must
test it by the rules for the transformation of components.

CH 5 Vec. and Ten

20
 Tensor analysis: Generalized coordinate transformation
 Let q1, q2 and q3 … are generalized coordinates that need not be
orthogonal.
 For 3D system
x = x (q1, q2, q3); y = y(q1, q2, q3); z = z(q1, q2, q3);
Where x, y and z are the ordinary rectangular coordinate
system.
 The relation b/n the two coordinate systems are formally expressed
as q1 q1 q1
dq1  dx  dy  dz  q1 q1 q1 
x y z  dq1   x y z   
dx
  
q q q    q q2 q2   
dq2  2 dx  2 dy  2 dz  dq2    2
x y z  dy 
   x y z   
q q q  dq   q3 q3 q3   dz 
dq3  3 dx  3 dy  3 dz
x y z  3   
CH 5 Vec. and Ten  x y z 
21
  In tensor analysis the Cartesian coordinate (x, y, z) will be written
as (x1, x2, x3) and the generalized coordinates (q1, q2 and q3) are
written as (q1, q2, q3)  q1 q1 q1  1
 dq   1
1
  dx 
   x x x   
2 3

 2   q 2 q 2 q 2   2 
dq    1 3 
dx ...... (a )
   x x 2 x  

 3   q 3 q 3 q  dx 3 
3 

 dq   1 3   
 x x 2 x 
 For function like ϕ, the transformation function will be
  x1  x 2  x 3   x1
x 2 x 3 
 1 1 2 1 3 1     
q 1
x q x q x q   q1   q1    
q1 q1   x1 
 x1  x 2  x 3     1

 1 2  2 2  3 2      x x 2 x 3    
2
 2 ...... (b)
q 2
x q x q x q   q 2   q 2 q 2 q   x 
    1 3    
  x1  x 2  x 3   3   x x 2 x  3 
 1 3 2 3 3 3 
q x q x q x q   q   q 3 3  x 
3
 q 3 q 
CH 5 Vec. and Ten
22
  Comparing the coefficient matrix in a and b, the matrix transformation in
b is the inverse of the matrix transformation of a.
 Therefore there are two types of transformation of vectors:
Components which transform using a are called Contravariant
tensor components of the vector
Components which transform using b are called Covariant
tensor components of the vector.
 Einstein notations:
Contravariant tensors are usually written in superscripts,
Covariant tensors are written in subscripts.
A velocity v in α coordinate system has its Contravariant
component vα = dα/dt and Covariant component vα = ∂ϕ/∂α

CH 5 Vec. and Ten

23
  Contravariant components of v will be given by (v1, v2, v3) in the
Cartesian coordinate system (x1, x2, x3) and the covariant
components in the same system are given by (v1, v2, v3)
 Similarly referred to the coordinate system (q1, q2, q3) the
Contravariant and Covariant components of v, respectively are
(v1', v2', v3') and (v'1, v'2, v'3)
 The Contravariant transformation (Eq. a) for a vector v may be
written as:
   3 
1 1 1
q q q
v  1v  2v  3v 
1' 1 2

x x x 
 q 2
 q 2
q 2
 3
q i
v 2 '  1 v1  2 v 2  3 v 3   v i '   j v j ... (c)
x x x  j 1 x

   3 
3 3 3
q q q
v  1v  2v  3v 
3' 1 2

x x x 
CH 5 Vec. and Ten
24
  x1 x 2 x 3 
 The Covariant  v'1   q1 q1

q1   v1 
transformation (Eq.     
   x1 x 2 3 
x   3
x j
b) for a vector v v ' 2    2 v  v'i   i v j ... (d )
2  2 
may be written as:    q q 2 q    j 1 q

v'   x1 x 2 x 3  v 
 3  3  3 
 q q 3 q 
3

Symmetric and skew-symmetric Tensors

 Given any tensor Aij, we can write:
 Aij = 0.5(Aij + Aji) + 0.5 (Aij – Aji)
 If Bij = 0.5(Aij + Aji) and Cij = 0.5 (Aij – Aji)
 It is easily seen that Bij = Bji and Cij = - Cji
 Bij and Cij are called symmetric and skew-symmetric tensors,
respectively.
 In general Cij = 0, if i = j.
CH 5 Vec. and Ten
25
 Metric Tensor
 The position vector r = x1i + x2j + x3k, locates a point at which the
curvilinear coordinates are erected. Since q1, q2, and q3 are
functions of x1, x2, and x3, we can write the differential change dr
as: 3
r  3 r   3 r 
dr   dq i
 ds  dr  dr  
  dq i

  
  dq j


i 1 q  i 1 q   j 1 q
i i j

3 3
r r
ds   i dq j
dq i
 g dq j
dq i

  j ij
i 1 j 1 q q
 gij is called the Covariant metric tensor. It is symmetric i.e. gij = gji
 Note x1 x1 x 2 x 2 x 3 x 3
g ij   
q q
i j
q q
i j
q i q j
 Metric tensors are usually used to convert Covariant in to
Contravariant components and vice versa.

CH 5 Vec. and Ten

26
 Reciprocal Bases and Covectors
 Given any vector A, let i1, i2, i3 be three orthonormal vectors.

A   A  i1 i1   A  i2 i2   A  i3 i3

Then

 Suppose e1, e2 and e3 be orthogonal but not necessarily unit

basis vectors in R3. Then e e e
eˆ1  1
, eˆ2  2
, eˆ3  3
,
e1 e2 e3
 are Orthonormal bases vectors in R3.
 Therefore,
A  ei
A  A e1  A e2  A e3 where A  2
1 2 3 i

ei
CH 5 Vec. and Ten
27
  Definition: Two bases e1, e2, e3 and e1, e2, e3 in R3 are said to
be reciprocal if they satisfy the condition
0, if i  k
ei  ek  
1, if i  k
 i.e., Two vectors whose indices have different values are
perpendicular.
 Moreover, since
j
 j

ei e cos ei , e  1  0
 
 cos ei , e j  0
 Each vector of one basis makes an acute angle (including
right angle) with the vector of the other basis whose index
has the same value.

CH 5 Vec. and Ten

28
  Let V represents the volume of the parallelepiped spanned by the
vectors e1, e2,e3 and let V' be the volume of the parallelepiped
spanned by the reciprocal vectors e1, e2,e3.

V  e1  e2  e3  and v'  e e  e 1

 2 3

 From the equation in the previous slide we can see that
1
e 
j


ei cos ei , e j  (reciprocal )
 In general, given a basis e1, e2, e3, to construct a reciprocal basis,
e1, e2, e3 we use: e  e
e e j k
e  and ei  i j k
i j k

ei  e j  ek  e  e e  
 where the order of i , j , k is a cyclic permutation of 1, 2, 3.

CH 5 Vec. and Ten

29
  The reciprocals of the standard unit vectors i, j, k are the vectors
themselves. i.e., if e1 = i; e2 = j; e3 = k, then e1 = i; e2 = j; e3 = k.
 Two reciprocal bases are either both right-handed or both left-
handed.
 Now suppose A be any vector in R3, whose components w.r.t. the
non-coplanar vectors e1, e2, e3, are respectively, A1, A2 and A3. i.e.,
A = A1e1 + A2e2 + A3e3
 For the corresponding reciprocal basis vectors e1, e2, e3, we have,
 3 k  j
A  e    A ek   e  A j e j  e j  A j , j  1,2,3.
j

 k 1 
A  e2  e3 
eg . A  A  e 
1 1

e1  e2  e3 
  
Thus A  A  e1 e1  A  e 2 e2  A  e3 e3   
CH 5 Vec. and Ten
30
  A vector A can be expanded as
 3
i 
A  A ei   A ei  where Ai  A  ei
i

 i 1 

 with respect to basis e1, e2, e3.

 Or  3 
A  Ai e   Ai e  where Ai  A  ei
i i

 i 1 
 with respect to the reciprocal basis e1, e2, e3
 Definition: The numbers Ai are called the contra-variant
components of A and the numbers Ai are called the covariant
components of A.
CH 5 Vec. and Ten
31
 Relation between Covariant and Contravariant
components
 From the two equations in the above slides, we have:

A  ei  Ak ei  ek  and A  ei  Ak ei  e k  
Denote : ei  ek  g ik  g ki ; e  e  g  g
i k ik ki

0 if i  k
And ei  e  g  
k k

1 if i  k
i

Thus Ai  g ik A ; A  g Ak
k i ik

• gik is called the metric tensor. It is symmetric i.e. gik = gki

CH 5 Vec. and Ten

32
  Ex.9: Consider triple of vectors in R3, u = (1, 0, 0); v = (1, 1, 0);
w = (1, 1, 1)
 Can this set of vectors be a basis (yes/no)?
 If yes then
 Compute the reciprocal basis
 Determine the Contravariant components of the vector (1, 2, 3)
 Solution: For basis vectors need first be linearly independent; thus
 α1u + α2v + α3w = 0, iff α1 = α2 = α3 = 0
1  1   2  1   3  1  0 

1  0   2  1   3  1  0   1   2   3  0
1  0   2  0   3 1  0
 Thus the given vectors are linearly independent.

CH 5 Vec. and Ten

33
  Reciprocal let u  e1 , v  e2 , w  e3
vector: e2  e3 e3  e1 e1  e2
e 
1
;e 
2
and e 
3

e1  e2  e3  e1  e2  e3  e1  e2  e3 

i j k
e2  e3  1 1 0  i  j  0k  1,1,0 
1 1 1
e1  e2  e3   (1,0,0)  (1,1,0)  1 1  0  (1)  0  0  1
i j k
e3  e1  1 1 1  0 i  j  k  0,1,1
1 0 0
i j k
e1  e2  1 0 0  0 i  0  j  k  0,0,1
CH 5 Vec. and Ten 1 1 0
34
 e1 
e2  e3

1,1,0  1,1,0; e 2  e3  e1  0,1,1  0,1,1
e1  e2  e3  1 e1  e2  e3  1

e 
3 e1  e2

0,0,1
 0,0,1
e1  e2  e3  1

 Determine the Contravariant components of the vector B = (1, 2,

3)
 B ' = B.e = (1,2,3).(1,-1,0)/√2 = -1/√2
1 1

2 2
 B ' = B.e = (1,2,3).(0,1,-1) = -1/√2

 B ' = B.e = (1,2,3).(0,0,1) = 3/√1

3 3

 The Contravariant components of B are thus (-1/√2,-1/√2,3)

CH 5 Vec. and Ten

35
  Ex. 10: Let q1, q2, q3 be coordinates related to rectangular coordinates x1,
x2, x3 with Orthonormal basis i1, i2, i3 by the formulae q1 = x1 + x2; q2 =
x1 - x2; q3 = 2x3:
a) Find the basis vectors corresponding to the vectors q1, q2, q3;
b) Find the metric tensor gik;
c) Find the Covariant and Contravariant components of the
vectors B = 2i1 - 3i3
 Solution: We are given that the bases vector as
q1 = x1 + x2; e1 = (1, 1, 0)


q2 = x1 - x2; e2 = (1, -1, 0)



q3 = 2x3; e3 = (0, 0, 2)


CH 5 Vec. and Ten

36
  The basis vectors e1, e2, and e3 are computed as
i1 i2 i3
e 2  e3  1  1 0  2i1  2i2  0 i3   2,2,0
0 0 2
 
e1  e 2  e3  (1,1,0)  (2,2,0)  4
e 2  e3
e1  1 2 3 
 2,2,0   1 , 1 ,0 
 
e  e e 4  2 2 
i1 i2 i3
e e 3 1
e2  1 2 3 but e 3  e1  0 0 2  2i1  2i2  0 i3   2,2,0 
e  e e  1 1 0

 e2 
 2,2,0   1 1 
 , ,0  
4 2 2 
CH 5 Vec. and Ten
37
 i1 i2 i3
e e1 2
e3  1 2 3 but e1  e 2  1 1 0  0i1  0i2  2i3  0,0,2

e  e e  1 1 0

 e2 
0,0,2 
 0,0,
1
 
4  2
 b. find the metric tensor gik = ei.ek
1 1  1 1  1 1 1  1 1 
g11  e1  e1   , ,0    , ,0   g 21  e2  e1   , ,0    , ,0   0
2 2  2 2  2 2 2  2 2 
1 1  1 1  1 1  1 1  1
g12  e1  e2   , ,0    , ,0   0 g 22  e2  e2   , ,0    , ,0  
2 2  2 2  2 2  2 2  2
1 1   1 1 1   1
g13  e1  e3   , ,0    0,0,   0 g 23  e2  e3   , ,0    0,0,   0
2 2   2 2 2   2

CH 5 Vec. and Ten

38
  1 1 1   1 1 1 
g 31  e3  e1   0,0,    , ,0   0; g 32  e3  e2   0,0,    , ,0   0
 2 2 2   2 2 2 
 1  1 1 1 2 0 0
g 33  e3  e3   0,0,    0,0,  
 2  2 2 g ik   0 1 2 0 
 0 0 1 2
 C. find the Covariant and the Contravariant components of the
tensor B = 2i1 – 3i3
 The Contravariant component of B is given by B = B.e , where B
k k

= (2, 0, -3) and ek as given by e1 = (1, 1, 0)/√2, e2 = (1, -1, 0)/√2

and e3 = (0, 0, 2)/√4
 The Covariant component of B is given by Bk = B.ek, where B =
(2, 0, -3) and ek as derived in section a. e1 = (0.5, 0.5, 0)/√0.5, e2 =
(0.5, -0.5, 0)/√0.5, e3 = (0, 0, 0.5)/√0.25
CH 5 Vec. and Ten
39
 1 1 
B1  B  e1  2,0,3   , ,0  0.5 
1
2 2  0.5
1 1 
B2  B  e2  2,0,3   , ,0  0.5 
1
2 2  0.5
 1
B3  B  e3  2,0,3   0,0,  0.25  3
 2
B 1  B  e 1  2,0,3  1,1,0  2 2
B 2  B  e 2  2,0,3  1,1,0  2 2
B 3  B  e 3  2,0,3  0,0,2 2  3

CH 5 Vec. and Ten

40