Covariant, Contravariant Vectors and Tensors

Summation Convention :
A polynomial of degree n in the variable x with constant coefficients
can be n
a0 + a1 x + a2 x2 + + an xn = ak xk
X

k=0

A summation convention (some times called also as the Einstein

summation convention) enables a shorter notation to be adopted as
simply ak xk , according to which when a dummy index, a subscript
or a superscript is repeated in a given term, then that index is to
be summed over from 0 or 1 to n. The repeated dummy index is
also called sometimes as an umbral index.

Einstein proposed the convention that repeated indices imply that

the summation is to be done. The value of the Einstein conven-
tion is that it applies to vector spaces, tensor products and duality.

Contravariant and covariant components of a vector

Let A~ be a given vector w.r.to two general curvilinear coordinate
systems (u1 , u2 , u3 ) and (u1 , u
2 , u
3 ). Suppose the transformation
equations from a rectangular Cartesian coordinate system (x, y, z)
to the systems (u1 , u2 , u3 ) and ( u1 , u
2 , u
3 ) are given by
x = x1 (u1 , u2 , u3 ), y = y1 (u1 , u2 , u3 ), z = z1 (u1 , u2 , u3 ),
x = x2 (
u1 , u
2 , u
3 ), y = y2 (
u1 , u
2 , u
3 ), z = z2 (
u1 , u
2 , u
3 ). (1)
Then there exists a transformation directly from (u1 , u2 , u3 ) system
to the (
u1 , u
2 , u
3 ) system defined by
u1 = u1 (
u1 , u
2 , u
3 ), u2 = u2 (
u1 , u
2 , u
3 ), u3 = u3 (
u1 , u
2 , u
3 ), (2)
and conversely,
u
1 = u
1 (u1 , u2 , u3 ), u
2 = u
2 (u1 , u2 , u3 ), u 3 (u1 , u2 , u3 ). (20 )
3 = u
From (1),
~r ~r ~r
d~r = du1 + du2 + du3 =
~ 1 du1 +
~ 2 du2 +
~ 3 du3
u1 u2 u3

1
 ~r ~r ~r
d~r = du1 + du2 + du3 =
1 d
u1 +
2 d
u2 +
3 d
u3
u1 u2 u3
Then

1 du1 + 2 du2 + 3 du3 =

1 d
u1 +
2 d
u2 +
3 d
u3 (3)

From (2),
u1 u1 u1
du1 = d
u1 + d
u2 + d
u3
u1 u2 u3
u2 u2 u2
du2 = d
u1 + d
u2 + d
u3
u1 u2 u3
u3 u3 u3
du3 = d
u1 + d
u2 + d
u3
u1 u2 u3
Substituting into (3) and equating the coefficients of d
u1 , d
u2 , d
u3 ,
one obtains:
u1 u2 u3

1 = 1 + 2 + 3

u1
u1
u1
u1 u2 u3

2 = 1 + 2 + 3 (4)

u2
u2
u2
u1 u2 u3

3 = 1 + 2 + 3

u3
u3
u3
Now A~ can be expressed in the two coordinate systems as:

1
A = C1 1 + C2 2 + C3 3 and A = C 2
1 + C 3
2 + C 3 (5)

where C1 , C2 , C3 and C1 , C2 , C3 are the contravariant and covariant

components of A in the two systems. Substituting, (4) into (5)
1
C1 1 + C2 2 + C3 3 = C 2
1 + C 3
2 + C 3
u1 u1 u1
1 ( 1 ( u2 )+C 2 ( u2 )+C
3 ( u2 )) 2 +
(C )+C2 ( )+C3 ( )) 1 +(C

u1
u2 u3
u1
u2
u3
u3 u3 u3
(C1 ( ) + C2 ( ) + C3 ( )) 3
u1 u2 u3
Then
C1 = (C 1 ( u1 ) + C 2 ( u1 ) + C 3 ( u1 ))
u1 u2
u3

2
 1 ( u2 2 ( u2 ) + C
3 ( u2 ))
C2 = (C )+C (6)

u1
u2
u3

C3 = (C 1 ( u3 ) + C
2 ( u3 ) + C
3 ( u3 )).

u1
u2
u3
Or, in shorter notation

Cp = C 2 ( up ) + C
1 ( up ) + C 3 ( up ), p = 1, 2, 3 (7)

u1
u2
u3
In an even shorter notation (7) can be written as:
3
q up ,
X
Cp = C p = 1, 2, 3 (8)
q=1
uq

Similarly, by interchanging the coordinates

3
p =
X
up
C Cq , p = 1, 2, 3 (9)
q=1
uq

The above results leads one to adopt the following definition: If

three quantities C1 , C2 , C3 of a coordinate system (u1 , u2 , u3 ) are re-
lated to three other quantities C1 , C2 , C3 of another coordinate sys-
tem (u1 , u2 , u3 ) by the transformation equations (6), (7), (8) or (9),
then the quantities are called components of a contravariant vector,
or a contravariant tensor of the first rank.

Contravariant Tensor
A contravariant tensor is a tensor having specific transformation
properties. To examine the transformation properties of a con-
travariant tensor, first consider a tensor of rank 1 (a vector):
dr = dx1 x
1 + dx2 x
2 + dx3 x
3 (1)
for which
x0i
dx0i = dxj . (2)
xj
Now let Ai = dxi , then any set of quantities Aj which transform
according to
x0i
A0i = Aj , (3)
xj

3
or, defining
x0i
aij = , (4)
xj
according to
A0i = aij Aj (5)
is a contravariant tensor. Contravariant tensors are indicated with
raised indices, i.e., a .
Covariant tensors are a type of tensor with differing transformation
properties, denoted a . However, in three-dimensional Euclidean
space,
xj x0i
= = aij (6)
x0i xj
for i, j = 1, 2, 3, meaning that contravariant and covariant tensors
are equivalent. Such tensors are known as Cartesian tensor. The
two types of tensors do differ in higher dimensions, however. Con-
travariant four-vectors satisfy

a = a , (7)

where is a Lorentz tensor. To turn a covariant tensor a into a

contravariant tensor a (index raising), use the metric tensor g
to write
g a = a . (8)
Covariant and contravariant indices can be used simultaneously in
a mixed tensor.

Kronecker delta function

The numerical value of the Kronecker delta function is:
(
1 if a = b
ab =
0 6 b.
if a =

This delta function named after Leopold Kronecker, is a function of

two variables, usually positive integers. It is also written as ij and
it is a piecewise function of the variables i and j. The Kronecker
delta appears naturally in many areas of Mathematics, Physics and

4
Engineering.

In Linear Algebra, the n n identity matrix I has entries equal

to the Kronecker delta Iij = ij , where i, j take rational values
1, 2, ..., n and the inner product of two vectors can be written as:
n
~a ~b =
X
ai bj ij .
i,j=1

The following equations are satisfied:

X X X
aj ij = ai , ai ij = aj , ik kj = ij .
j i k

Another useful representation is the following form:

N
1 X k
nm = exp{2i (n m)}
N k=1 N

The Kronecker delta ab is actually a tensor, as it transforms ac-

cording to the rules for such a tensor.

The rule of a contravariant tensor of rank - 1 (a vector) is:

x0a i
T 0a = T (1)
xi
and for a covariant vector :

xi
Ta0 = Ti (2)
x0a
For a mixed tensor, one multiplies by the right combination of
derivatives to effect a coordinate transformation.

Note that a superscript is used to indicate contravariant compo-

nents and a subscript is used to indicate covariant components.

5
 Contravariant, Covariant and Mixed tensors
If N 2 quantities Aqs in a coordinate system (x! , x2 , ..., xN ) are related
to N 2 other quantities Apr in another coordinate system ( x1 , x2 , ..., xN )
by the transformation equations
N X
N
xp xr qs
Apr =
X
q s
A , p, r = 1, 2, ..., N
s=1 q=1 x x

xp xr qs
Apr = A , p, r = 1, 2, ..., N
xq xs
by the adopted (Einstein summation)conventions. They are called
the contravariant components of the components of a tensor of the
second rank (or rank 2).

The N 2 quantities Aqs are called covariant components of a tensor

of the second rank if
xq xs
Apr = Aqs .
xp xr
Similarly, the N 2 quantities Aqs are called components of a mixed
tensor of rank 2 if
xp xs q
Apr = A
xq xr s
A tensor Aqst
kl is a mixed tensor of rank 5, which is a contravariant
tensor of order 3 and a covariant tensor of order 2, if they transform
according to the relations :

xp xr xm xk xl qst
Aprm
ij = A .
xq xs xt xi xj kl
Scaras or Invariants : Suppose is a function of the coordinates xk ,

and denote the functional value under a transformation to a new
set of coordinates xk . The is called a scalar or invariant w.r.to
the coordinate transformation if = . Such a scalar or invariant
is also called as a tensor of rank zero.

6
Symmetric and Skew-symmetric tensors: A tensor is called sym-
metric w.r.to two contravariant or two covariant indices if its com-
ponents remain unaltered upon interchange of the indices. Thus,
if
Ampr
qs = Apmr
qs or Ampr
qs = Ampr
sq ,

the tensor is symmetric w.r.to any two contravariant or any two

covariant indices, it is called symmetric.

A tensor is called skew-symmetric w.r.to two contravariant or co-

variant indices if its components change sign upon interchange of
the indices. Thus if
Ampr
qs = Apmr
qs

the tensor A is skew-symmetric in m and p. If a tensor is skew-

symmetric w.r.to any any two contravariant and any two covariant
inidices it is called skew-symmetric.

Fundamental operations with tensors

1. Additon: The sum of two or more tensors of the same rank

and type (i.e. the same number of contravariant and covariant
indices)is also a tensor of the same rank and type. Thus, if
Amp
q and Bqmp are tensors, then

Cqmp = Amp mp
q + Bq

is also a tensor. Addition of tensors is commutative and as-

sociative.

2. Subtraction: The difference of two tensors of the same rank

and type is also a tensor of the same rank and type. Thus, if
Amp
q and Bqmp are tensors, then

Dqmp = Amp mp
q Bq

is also a tensor.

7
3. Outer multiplication : The product of two tensors is a tensor
whose rank is the sum of the ranks of the given tensors. This
product which involves ordinary multiplication of the compo-
nents of the tensor is called the outer product. For example,
Thus, if Amp
q and Bqmp are tensors, then

Apr m
q Bs
prm
= Cqs

is the outer product of Aprq and Bsm . However, it is to be

noted that not every tensor can be written as a product of
two tensors of lower rank. Because of this reason, division of
tensors is not always possible.

4. Contraction : If one contravariant and one covariant index of

a tensor are set equal, the result indicates that a summation
over the equal indices is to be taken according to the summa-
tion convention. This resulting sum is a tensor of rank two
less than that of the original tensor. The process is called
contraction. For example, in the tensor of rank 5, Ampr qs set
mpr mp
r = s to obtain Aqr = Bq a tensor of rank 3. Further, by
setting p = q, one obtains Bpmp = C m , a tensor of rank 1 (or a
vector).

5. Inner Multiplication: By the process of outer multiplication

of two tensors followed by a contraction, we obtain a new ten-
sor called an inner product of the given tensors. This process
is called inner multiplication. For example, given the tensors,
Amp r mp r
q , Bst , the outer product is Aq Bst . Letting q = r, we
obtain the inner product Amp r B r
st . Letting q = r and p = s,
mp r
another inner product Ar Bpt is obtained. Inner and outer
multiplication of tensors is commutative and associative.

6. Quotient law: Suppose it is not known whether a quantity X

is a tensor or not. If an inner product of X with an arbitrary
tensor is itself a tensor, then X is also a tensor. This is called
the quotient law.

8
 MATRICES
A matrix of order m n is an array of quantities apq , called its
elements, arranged in m rows and n columns and generally denoted
by

a11 a12 a1n a11 a12 a1n

a21 a22 a2n


a21 a22 a2n

.. .. .. or .. .. ..
. . . . . .


am1 am2 amn am1 am2 amn

or in abbreviated notation as (apq ) or [apq ], p = 1, 2, . . . m; q = 1, 2, . . . n.

If m = n the matrix is a square matrix of order m m;

if m = 1 it is a row matrix or a row vector;
if n = 1 it is a column matrix or a column vector.

The diagonal of a square matrix containing the elements a11 , a22 , . . .,

amm is called the principal or main diagonal. A square matrix
whose elements are equal to one in the principal diagonal and zero
elsewhere is called a Unit Matrix and it denoted by I. A null matrix,
denoted by O, is a matrix all of whose elements are zero.

Matrix Algebra
If A = (apq ) and B(bpq ) are matrices having the same order m n,
then

1. A = B if and only if apq = bpq .

2. The sum S and difference D are the matrices defined by

S = A + B = (apq + bpq ), D = A B = (apq bpq )

3. The product P = AB is defined only when the number n of

column in A equals the number of rows in B and is then given
by
n
X
P = AB = (apq )(bpq ) = (apr brq ) = apr brq
r=1

9
 by the summation convention. Matrices whose product is de-
fined are called conformable.
In general, multiplication of matrices is not commutative i.e.AB 6=
BA. However, the associative law for multiplication holds, i.e.
A(BC) = (AB)C provided the matrices are conformable. Also
the distributive laws hold, i.e. A(B + C) = AB + AC, and
(A + B)C = AC + BC.

4. The determinant of a square matrix A = (apq ) is denoted by

|A|, or det A, or |A|, or det(apq ). If P = AB then |P | = |A||B|.

5. The inverse of the square matrix A is denoted by A1 , such

that AA1 = I, where I is the unit matrix. A necessary
and sufficient condition that A1 exists is that det A 6= 0. If
det A = 0, then A is called as a singular matrix.

6. The product of a scalar by a matrix A = (apq ), is denoted

by A, is the matrix (apq ) where each of the elements of A
is multiplied by the scalar .

7. The transpose of a matrix A is a matrix AT which is formed

from A by interchanging its rows and columns. Thus, if A =
(apq ), then AT = (aqp ).
E.T. Bell (1933) defined the reciprocal of a matrix A by writing
the defining n diagonals for an n n square matrix :

a11 a12 a13 a1n
a21 a22 a23 a2n



A =
a31 a32 a33 a3n

.. .. .. .
. . . ..



an1 an2 an3 ann

and writing these diagonals as rows to construct a Reciprocal Array

AR as:
a11 a22 a33 an1,n1 ann
a21 a32 a43 an,n1 a1n


AR = a31 a42 a53 a1,n1 a2n
. .. .. ..
.

. . . .

an1 a12 a23 an2,n1 an1,n

10
 n n
aR
Y Y
If xi = aij and yi = ij then aii = gcd(xi , yi ),
j=1 j=1

where the greatest common divisor (g.c.d) is a condition on the

elements of the array and its reciprocal.

The Line element and the metric tensor

By definition, if (x, y, z) are the Cartesian coordinates of a point

P, and (u1 , u2 , u3 ) are the Curvilinear coordinates of the same point
P, then
x = x(u1 , u2 , u3 ), y = y(u1 , u2 , u3 ) z = z(u1 , u2 , u3 ) (1)
and if (1) can be solved for (u1 , u2 , u3 ) in terms of (x, y, z), then
u1 = u1 (x, y, z), u2 = u2 (x, y, z), u3 = u3 (x, y, z), eqno(2)
Eqs. (1) and (2) define a transformation of coordinates.

u1 = c1 , u2 = c2 , u3 = c3
are coordinate surfaces. The intersection of each pair of these
surfaces give rise to coordinate curves or lines. If the coordinate
surfaces intersect at right angles, then the coordinate system is
called Orthogonal.
Let ~r = xi + yj + z k be the position vector of the point P. Then
~
r
~r = ~r(u1 , u2 , u3 ) and u 1
is a tangent vector to the curve at P, with
u2 and u3 constant. The unit tangent vector in this direction is:
~r ~r ~r ~r
e1 = /| | , so that = h1 e1 , h1 = | |
u1 u1 u1 u1
~r ~r ~r ~r
e2 = /| | , so that = h2 e2 , h2 = | |
u2 u2 u2 u2
~r ~r ~r ~r
e3 = /| | , so that = h3 e3 , h3 = | |
u3 u3 u3 u3

where h1 , h, h3 are called scale factors, and

e1 , e2 , e3 are called unit vectors in the directions of increasing
u1 , u2 , u3 .

11
5u1 at a point P is a vector perpendicular to the surface u1 = c1 ,
so that a unit vector in this direction is
5u1 5u2 5u3
E1 = ; E2 = ; E3 = .
|5u1 | |5u2 | |5u3 |

e1 , e2 , e3 are tangents to the coordinate curves and

E1 , E2 , E3 are normals to these coordinate surfaces.

NOTE: If the curvilinear coordinate system is orthogonal,then the

tangents and the normals to the surfaces become identical, i.e.

e1 , e2 , e3 = E1 , E2 , E3 .
~ can be written in terms of these base vectors as:
Any vector A
~ = A1
A e 1 + A2
e2 + A e
e3

and
~ = a1 E
A 1 + a2 E
2 + a3 E
3
And it follows by definition that

~ = C1 ~r + C2 ~r + C3 ~r = C1
A 1 + C2
2 + C3
3 ,
u1 u2 u3
~
r
where p = u p
, p = 1, 2, 3 and c1 , c2 , c3 are the contravariant compo-
~
nents of A. Similarly,
~ = c1 5u1 + c2 5u2 + c3 5u3 = c1 1 + c2 2 + c3 3 ,
A

where p = 5p , p = 1, 2, 3 and c1 , c2 , c3 are the covariant components

~ If u1 , u2 , u3 is an orthogonal coordinate system, then the
of A.
vector differential operator Del, written 5 is

e1
e2
e3
5 + +
h1 u1 h2 u2 h3 u3
where
e1 |e1 |
5u1 = ; |5u1 | = = h1
1
h1 h1

12
Similarly,
|5u2 | = h1
2 ; |5u3 | = h1
3 ;

and
5up ep
Ep = = = ep .
|5up | hp h1
p

The Jacobian of the transformation of (x, y, z) with respect to (u1 , u2 , u3 )

is  
!  x y z 
 u1 u1 u1
x, y, z (x, y, z)

 x y z

J = =  u2 u2 u2

u1 , u2 , u3 (u1 , u2 , u3 )  x y z


u3 u3 u3
 

x y z x y z x y z
= ( i+ j+ k)( i+ j+ k)( i+ j+ k)
u1 u1 u1 u2 u2 u2 u3 u3 u3
r r r
= = h1 h2 h3 e1 e2 e3 = h1 h2 h3 .
u1 u2 u3
Therefore, the volume element is given by

dV = dx dy dz
!
x, y, z
= h1 h2 h3 du1 du2 du3 = J du1 du2 du3 .
u1 , u2 , u3
Arc length and Metric form

If ~r = ~r(u1 , u2 , u3 ) then

~r ~r ~r
d~r = du1 + du2 + du3
u1 u2 u3
= h1 e1 du1 + h2 e2 du2 + h3 e3 du3
and the square of the arc length ds2 is given by

ds2 = d~r d~r = h21 du21 + h22 du22 + h23 du23

for orthogonal systems,since

e1 e2 = e2 e3 = e3 e1 = 0.

13
For general curvilinear coordinates:

d~r =
1 du1 +
2 du2 +
3 du3 .

ds2 = d~r d~r = 1 du21 +

1 1 1
2 du1 du2 + 3 du1 du3 +
2
1 du2 du1 + 2 du22 +
2 2
3 du2 du3 +
3
3
1 du3 du1 + 3 du23
3
2 du3 du2 +
3 X
3
ds2 =
X
gpq dup duq
p=1 q=1

where gpq = p
q . This is called the fundamental quadratic form or
metric form. gpq are called metric coefficients and are symmetric,
i.e. gpq = gqp . Such spaces are called 3-dimensional Eucledian spaces.

For an orthogonal coordinate system:

2
gpq = 0, f or p 6= q, and g11 = h21 , g22
2
= h22 , g33
2
= h23 .

Rectangiular Coordinate System

h1 = h2 = h3 = 1
e1 = i, e2 = j, e3 = k
u1 = x, u2 = y, u3 = z.

Spherical coordinates

x = r sin cos , y = r sin sin , z = r cos

r 0, 0 , 0 2
~r = r sin cos i + r sin sin j + r cos k
u1 = r, u2 = , u3 =
~r ~r ~r
h1 = | | = 1, h2 = | | = r, h3 = | | = r sin
u1 u2 u3

14
Therefore,

dV = dx dy dz = h1 h2 h3 dr d d = r2 sin d d

EXERCISE : Find er , e , e , derive the expressions for the Del

operator and for its square.

Generalization
A generalization to an N dimensional space with coordinates (x1 , x2 , . . . , xN )
is immediate. Define the line element ds in this space to be given
by the quadratic form ( called as metric form, or metric):
N X
N
ds2 = gpq dxp dxq
X

p=1 q=1

or, using the summation convention, simply

ds2 = gpq dxp dxq

In the special case where there exists a transformation of coordi-

nates xj to xk such that the metric form is transformed into

x1 )2 + (d
(d x2 )2 + + (d
xN ) or xk d
d xk ,

then the space is called an N dimensional Eucledian space. In the

general case, this space is called as Riemannian.

The quantities gpq are the components of a covariant tensor of rank

2, called the metric tensor or fundamental tensor. This can always
(and will always be ) chosen to be symmetric.

For, let
= ajk Aj Ak .
Then it is also true that

= akj Ak Aj = akj Aj Ak .

Thus 2 = ajk Aj Ak + akj Aj Ak = (ajk + akj )Aj Ak

and
1
= (ajk + akj )Aj Ak = bjk Aj Ak ,
2
15
where
1
bjk = (ajk + akj ) = bkj
2
is symmetric. Let

= ds2 , Aj = dxj Ak = dxk ,

then it follows that gjk can be chosen to be symmetric. Also, since

ds2 is an invariant,
xj p xk q xj xk p q
xp d
gpq d xq = gjk dxj dxk = gjk d
x d
x = g jk d
x d
x
xp xq xp xq
Then
xj
gpq = gjk
xq
and gjk si a symmetric tensor of rank 2, called the metric tensor.

Conjugate or Reciprocal tensors

Let g = |gpq | denote the determinant with elements gpq and
6 0. Define gpq by
suppose g =
cof actor of gpq
gpq =
g
Then g pq is symmetric contravariant tensor of rank 2 called the
conjugate or reciprocal tensor gpq .

Note: One can expand the determinant of gjk in terms of the

elements of any row and its cofactors. For instance, when the
expansion is in terms of the elements of its second row and their
corresponding cofactors:
 

 g11 g12 g13 

g = det 

g21 g22 g23 
 = g21 G(2, 1) + g22 G(2, 2) + g23 G(2, 3)

 g31 g32 g33 

by the elementary principle of determinants and where the cofac-

tors are denoted by:
   
2+1
 g12 g13 
2+2
 g11 g13 
G(2, 1) = (1) G(2, 2) = (1)
   
g32 g33 g31 g33
   
   

16
and  
2+3
 g11 g12 
G(2, 3) = (1)
 
g31 g32
 
 

If any two rows of the determinant g are equal then the value of
that determinant is zero and this property is expressed as:
gjk G(p, k) = 0, if j 6= p.
For, if the last two rows in g are identical:
 

 g11 g12 g13 

g = det  g21 g22 g23 

 
 g21 g22 g23 

then it follows by expanding according to the elements of the last

row
g21 G(1, 3) + g22 G(3, 2) + g23 G(3, 3) = 0.
If by definition
G(j, k)
g jk =
g
then the property discussed earlier:
g21 G(2, 1) + g22 G(2, 2) + g23 G(2, 3) = g
becomes
g21 g 21 g + g22 g 22 g + g23 g 23 g = g g2k g 2k = 1
which for g = |gjk | =
6 0 can be written as:
gjk g pk = jp
in terms of the Kronecker delta function.

Note: The notation g jk is a contravariant tensor of rank 2. This

can be proved as follows:

Since gjk is symmetric, it follows that g jk = G(j, k)/g is also symmet-

ric. If B p is another contravariant vector, Bq = gpq B p is an arbitrary
covariant vector. Multiplying by g jq ,
g jq Bq = g jq gpq B p = pj B p = B j

17
or
g jq Bq = B j
Since Bq is an arbitrary vector, g jq is a contravariant tensor of rank
2, by application of the quotient law. The tensor g jk is called the
conjugate metric tensor.

Associated Tensors
Given a tensor one can derive another tensors by raising or lowring
indices. For instance, given the tensor Apq we obtain by raising the
index p, the tensor

Ap.q = g rp Arq , Apq = g rp g sq Ars , Ap.rs = g rq Apq

..s

Aqm.tk
..n = g pk gsn g rn Aq.st
.r..p

For rectangular coordinates gpq = 1 if p 6= q, so that Ap = Ap ,

which explains why no distinction is made between contravariant
and covariant components of a vector in vector analysis.

Length of a vector. Angle between vectors

The quantity Ap Bp , which is the inner product of Ap and Bp , is a
scalar analogous to the scalar product om rectangular coordinates.
Define the length L of the vector Ap or Ap as

L2 = Ap Ap = g pq Ap Aq = gpq Ap Aq

One can therefore define the angle between Ap and Bp as:

Ap Bp
cos = q .
(Ap Ap )(B p Bp )

Physical components of a vector

The physical components of a vector Ap or Ap denoted by Au , Av
and Aw are the projections of the vector on the tangents to the co-
ordinate curves and are given in the case of orthogonal coordinates
as
A1
Au = g11 A1 = ,
g11

18
 A2
Av = g22 A2 = ,
g22
A3
Aw = g33 A3 = .
g33
Similarly, the physical components of a tensor Apq or Apq are given
by
A11
Auu = g11 A11 =
g11
A12
Auv = g11 g22 A12 =
g11 g22
A13
Auw = g11 g33 A13 = , etc.
g11 g33

19
 Christoffels symbols
The Christoffel symbols are named for Elwin Bruno Christoffel
(1829 - 1900). In mathematical sciences, the Christoffel symbols
are an array of numbers describing a metric connection. The met-
ric connection is a specialization of the affine connection to surfaces
or other manifolds endowed with a metric, allowing distances to be
measured on that surface. In differential geometry, an affine con-
nection can be defined without any reference to a metric, and many
additional concepts follow: parallel transport, covariant deriva-
tives, geodesics, etc. also do not require the concept of a metric.

However, when a metric is available, these concepts can be directly

tied to the shape of the manifold itself; that shape is determined
by how the tangent space is attached to the cotangent space by the
metric tensor.

Abstractly, one would say that the manifold has an associated

(orthonormal) frame bundle, with each frame being a possi-
ble choice of a coordinate frame. An invariant metric implies
that the structure group of the frame bundle is the orthogonal
group SO(m,n). As a consequence, such a manifold is necessarily
a (pseudo-)Riemannian manifold.

The Christoffel symbols provide a concrete representation of the

connection of (pseudo-)Riemannian geometry in terms of coordi-
nates on the manifold.

In general, there are an infinite number of metric connections for a

given metric tensor; however, there is one, unique connection, the
Levi-Civita connection, that is free of any torsion. It is very com-
mon in physics and general relativity to work almost exclusively
with the Levi-Civita connection, by working in coordinate frames
(called holonomic coordinates) where the torsion vanishes.

At each point of the underlying n-dimensional manifold, for any

local coordinate system around that point, the Christoffel symbols
are denoted ijk for i, j, k = 1, 2, . . . , n. Each entry of this n n n

20
array is a real number. Under linear coordinate transformations
on the manifold, the Christoffel symbols transform like the com-
ponents of a tensor, but under general coordinate transformations
(diffeomorphisms) they do not. Most of the algebraic properties of
the Christoffel symbols follow from their relationship to the affine
connection; only a few follow from the fact that the structure group
is the orthogonal group SO(m, n) (or, the Lorentz group SO(3, 1) for
general relativity).

Christoffel symbols are used for performing practical calculations.

For example, the Riemann curvature tensor can be expressed en-
tirely in terms of the Christoffel symbols and their first partial
derivatives. In general relativity, the connection plays the role of
the gravitational force field with the corresponding gravitational
potential being the metric tensor. When the coordinate system
and the metric tensor share some symmetry, many of the ijk are
zero.

The symbols
1 gpr gqr gpq
[pq, r] = ( + )
2 xq xp xr
( )
s
= g sr [pq, r]
pq
are called the Christoffel symbols of the first
( and ) second kind,
s
respectively. Other symbols used instead of are [pq, s] and
pq
spq . The latter symbol suggests however a tensor character, which
is not true in general.

Summary
Definition: A tensor is an array of contravariant and covariant
klm
components, Tnp (functions of xi ) which transform into a similar
0 0 0
array of components Tnk0 pl 0m (functions of xi ) as follows:
klm n0 p0 k0 l0 m0
Tnp = Akk0 All0 Am
m0 Bn Bp Tn0 p0

21
where the transformation A, and its covariant counterpart B,are
linear general transformations on the coordinates.

Note: The most common coordinate transformations in three di-

mensional Eucledian space is a rotation from one Cartesian co-
ordinate system to another. In relativistic problems, A is gener-
ally a Lorentz transformation function (x, y, z, t) system to another
(x0 , y 0 , z 0 , t0 ). A tensor is defined in terms of a transformation A.
Examples: A strain tensor w.r.to relations between Cartesian sys-
tems:
1 j k
Skj = [ rk r j ]
2 x x
where rj (~ r2 r~1 ).x~j and the ~ri are position vectors.

22