Scribd
Upload
75 views6 pages

App Cart TEN

The document summarizes key concepts in Cartesian tensors: (1) Cartesian tensors transform according to specific rules under coordinate transformations, allowing physical laws to be expressed invariantly. (2) Vectors transform according to their direction cosines between coordinate systems. (3) Higher order tensors like second rank tensors transform based on the product of the direction cosines for each index. (4) Operations like addition, multiplication, and contraction of tensors results in a tensor of a corresponding higher or lower rank.

Uploaded by

jmScri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
75 views6 pages

App Cart TEN

The document summarizes key concepts in Cartesian tensors: (1) Cartesian tensors transform according to specific rules under coordinate transformations, allowing physical laws to be expressed invariantly. (2) Vectors transform according to their direction cosines between coordinate systems. (3) Higher order tensors like second rank tensors transform based on the product of the direction cosines for each index. (4) Operations like addition, multiplication, and contraction of tensors results in a tensor of a corresponding higher or lower rank.

Uploaded by

jmScri
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

1

Lecture Notes on Fluid Dynamics


(1.63J/2.21J)
by Chiang C. Mei, February 6, 2007

A.1

Appendix on Cartesian tensors

[Ref 1] : H Jeffreys, Cartesian Tensors;


[Ref 2] : Y. C. Fung, Foundations of Solid Mechanics.
A true law of physics must satisfy the following basic requirements on dimensions : (a)
two physical quantities are not equal unless they are of the same dimension, and (b) a physical
equation must be invariant under a change of fundamental units. Equally fundamental, a
true physical law must be free of the frame of reference, and should remain valid in any
coordinate system.The mathematical tool that can express physical laws in invariant forms
is Tensor Analysis.
Let us first introduce the index notation. A group of three variables x, y, z can be denoted
alterntively by xi , i = 1, 2, 3 where x1 = x, x2 = y, x3 = z. Thus the equation of a plane in
x, y, z space can be written as
a1 x + a2 y + a3 z = p
or,
a1 x 1 + a2 x 2 + a3 x 3 = p
or

3
X

ai x i = p

(A.1.1)

i=1

As a simplification, we add a rule due to A. Einstein: If an index is repeated once (and only
once) in a term, summation over the full range of that index is implied but the summation
P
sign i is omitted. For example the previous equation for a plane is now written as ai xi = p.
Since tensor analysis is motivated by coordinate transformation, let us look at a transformation law for rotation.

A.1.1

Transformation of a vector

Refering to Figure A.1.1, let {x , y } axes be different from {x, y} by a simple rotation. Then
~ in the two coordinate systems are related by
the compoents of a vector A
Ax = Ax cos(x , x) + Ay cos(x , y)
Ay = Ax cos(y , x) + Ay cos(y , y)

(A.1.1)

where (x , y) denotes the angle between the x and y axies. Using the index notation, it can
also be written
A1 = A1 cos(x1 , x1 ) + A2 cos(x1 , x2 )
A2 = A1 cos(x2 , x1 ) + A2 cos(x2 , x2 )

(A.1.2)

Figure A.1.1: Two coordinate systems related by rotation


In three-dimensions, the components of a vector are transformed under rotation as follows
A1 = A1 cos(x1 , x1 ) + A2 cos(x1 , x2 ) + A3 cos(x1 , x3 )
A2 = A1 cos(x2 , x1 ) + A2 cos(x2 , x2 ) + A3 cos(x2 , x3 )
A3 = A1 cos(x3 , x1 ) + A2 cos(x3 , x2 ) + A3 cos(x3 , x3 )

(A.1.3)

Let us introduce the shorthand notation


Cik cos(xi , xk )

(A.1.4)

then , in the summation convention, (A.1.3) can be written in matrix form


Ai =

Cik Ak = Cik Ak

(A.1.5)

where

C11 C12 C13

[Cik ] = C21 C22 C23


C.
C31 C32 C33

(A.1.6)

is the transformation matrix. Some properties of the matrix Cik are derived below:
(a) Orthogonality of Cik . Since the length of vector A must be invariant, i.e., the same
in both coordinate systems,
Ai Ai = Ai Ai .

3
In view of (A.1.5)
Ai Ai = Cik Ak Cij Aj = Cik Cij Ak Aj = Aj Aj .
hence,
Cik Cij =

1, k = j
0, k =
6 j

(A.1.7)

Introducing the Kronecker delta


kj =

1, k = j
0, k =
6 j

we have,
Cik Cij = kj

(A.1.8)

Notice that the summation is performed over the first indices. This property is called the
orthogonality of the transformation matrix C, which is a generalization of a similar concept in
vectors. Equation (A.1.8) for k, j, = 1, 2, 3 represents six constraints among nine components
of Cij , hence only three components of Cij are independent.
~ in both systems S and system
(b) The inverse transformation: Consider the vector A
S which are related by rotation. We may write
Ai = Ci A ,

Ci = cos (xi , x ) .

Note that by definition


Ci = cos (xi , x ) ,

Ci = cos (x , xi )

Since the angles are the same in magnitude, we have


Ci = Ci

(A.1.9)

Clearly,

= ij ,
Ckj
Cki

hence, it follows from (A.1.9) that


Cik Cjk = ij

(A.1.10)

Comparing (A.1.8) and (A.1.10), we note that summation in the latter is performed over the
second indices.
Remark: A general transformation is equal to a translation plus a rotation, but a vector
is not affected by translation at all since only the end points matter.

A.1.2

Definition of a cartesian tensor

A tensor T of rank r is an array of components denoted by Tijk...m with r indices ijk . . . m.


In three dimensional space T has 3r components. The defining property of a cartesian tensor
is the following law : From coordinate system S to S by a rotation, the components of a
tensor transform according to

Tijk...m
= Cis Cjt Cku Cmv Tstu...v .

(A.1.11)

As special cases, a scalar is a zero-th rank tensor T = T . A vector is a first rank


tensor which is transformed according to Ti = Cij Tj . A second rank tensor is transformed
according to Tij = Cis Cjt Tst .
Problem: Show that ij is a second-rank tensor.
Hint: Consider Cis Cjt st and use the definition of Knonecker delta.
Remark: General non-cartesian tensors are defined by more general laws of transformation.

A.1.3

The Quotient Law

A set of 3r numbers form the components of a tensor of rank r, if and only if its scalar
product with another arbitrary tensor is again a tensor. This is called the quotient law and
can be used as a litmus test whether a set of numbers form a tensor.
We only give a one-way proof for a third rank tensor. Consider a set of numbers Aijk .
Let be the components of an arbitrary vector. Then, if
Ajk = Bjk
is a tensor component, it must obey the transformation law of a second-order tensor, i.e.,

Bik
(= Aik ) = Ci Ckm Bm (= Ci Ckm Am ).

But,

= C = C
.

hence,
Aik = (Ci Ckm C Am ) .
Since is arbitrary
Aik = Ci Ckm C Am .
it follows that Am is a third rank tensor.

A.1.4

Tensor Algebra

(a) Addition. The sum of two tensors of equal rank is another tensor of the same rank. Let
us give the proof for second-rank tensors only.
Given two tensors Aij and Bij , we define the sum in S by
Eij = Aij + Bij .
In S system we have, by definition,
Eij = Aij + Bij = Ci Cjm Am + Ci Cjm Bm
= Ci Cjn (Am + Bm ) = Ci Cjn Em ,
hence, Eij is a tensor of second rank after using linearlity.
(b) Multiplication. (A tensor of rank b) times (a tensor of rank c) = a tensor of rank
b + c with 3b+c components
Eij...krs...t = Aij...k Brs...t .
We only give the proof for the special case Eijrs = Aij Brs . Define

Eijrs
= Aij Brs
in S
= (Cik Cj Ak ) (Crm Csn Bmn )
= Cik Cj Crm Csn Ak Bmn = Cik Cj Crm Csn Ekmn ,

hence, E is a tensor of fourth rank.


(c) Contraction : If any pair of indices of of an r-th rank tensor with are set equal and
summed over the range 12,3, the result is a tensor of rank r 2.
Consider Aijk , which obeys Arstu = Cri Csj Ctk Cu Aijk . Let s = t and sum over the index
s,
Arssu = Cri (Csj Csk ) Cu Aijk
= Cri jk Cu Aijk
= Cri Cu Aijj .
This is precisely the transformation law of a second rank tensor. Hence, Arssu is a second
rank tensor.
~ B,
~ which is known
Note: The scalar product Ai Bi = D is a special case of contraction. A
as a dyad, is a seocnd-order tensor = (AB)ij = Ai Bj . Contraction makes it a zero-th order
tensor, i.e., a scalar. Thus a scalar product is the result of multiplication and contraction.

A.1.5

Tensor Calculus

(a) Differentiation A tensor whose components are functions of spatial coordinates x1 , x2 , x3


in a region R forms a tensor field.

6
a.1) Gradient: Taking the gradient of a tensor of rank r gives rise to a tensor of rank
r + 1.
Let (x1 , x2 , x3 ) be a scalar (tensor of rank 0). Is Vi = /xi a tensor of rank 1 (a
vector)?
Since ~x = (xi ) is a vector, it transforms like one: xi = Cik xk . Now,
Vi

xj

=
= Cji
=

xi
xj xi
xj

=
C
by(A.1.8)
xj ij
= Cij Vj
by definition.

Hence, Vi is a vector component, and the gradient of a scalar is a vector. In other words,
the gradient of a tensor of rank zero is a tensor of rank 1.
In general
Rij...k
Tij...k =
.
(A.1.12)
x
(a.2) Divergence is contraction in differentiation.
Taking the divergence of a tensor of rank r gives rise to a tensor of rank r 1.
Consider = vi /xi . Is a scalar? Lets check the transformation law.


Cij Vj xk
Vi
Vi xk
=
=
=
xi
xk xi
xk
xi

Vj
Vj
Cki = Cij Cik
= Cij

xk
xk
V
Vk
= jk j =
= .
xk
xk
Hence, is a scalar and = .
Problem: Prove that the strain components
eij =

ui uj
+
xj
xi

ij =

ui
uj

xi
xj

and the quantities

are components of tensors.

You might also like