Chapter 2

Cartesian tensor and matrix calculus

Chapter Outline
2.1 Cartesian tensor 43 2.2.3 Derivatives with matrices 50
2.1.1 Vector 43 2.2.4 Identities 51
2.1.2 Summation convention 44 2.3 Exercise problems 51
2.1.3 Kronecker delta 44 2.3.1 Problem 1: prove the following formulations 52
2.1.4 Permutation symbol 45 2.3.2 Problem 2: prove the identity of three arbitrary
2.1.5 e 2 δ Identity 45 vectors 53
2.1.6 Differentiation of a function f(x1, x2, x3) 45 2.3.3 Problem 3: express the constitutive equation
2.1.7 Transformation of coordinates 46 in a tensor form 53
2.1.8 Tensor 46 2.3.4 Problem 4: write the tensor equation in a coordinate
2.1.9 Quotient rule 47 (xyz) form 54
2.1.10 Index forms of some important variables 47 2.3.5 Problem 5: prove the following identities using
2.1.11 Two primary identities 48 index notations 54
2.2 Matrix calculus 48 2.3.6 Problem 6: prove eijkaiajbk 5 0 for nonzero
2.2.1 Types of matrix derivatives 49 vectors a and b 55
2.2.2 Derivatives with vectors 50

In this chapter, some preliminary knowledge on Cartesian tensor analysis and matrix calculus used throughout the book
is briefly discussed. For tensor knowledge, the main reference book referred to in this chapter is Fung (1977), and its
further extension is Fung and Tong (2001). Further readings on tensors with applications in mechanics can be found in
the historical books by Michal (1947), Sokolnikoff (1958), Aris (1962), Brillouin (1964), as well as the books in
Chinese by Huang et al. (1986) and Guo (1988). For matrix calculus, the books by Fan and Zhang (1990) and by Golub
and Van Loan (1996) can provide more detailed information. Readers who are familiar with tensor analysis and matrix
calculus may not need to read this chapter.

2.1 Cartesian tensor

2.1.1 Vector
We choose a Cartesian coordinate system o 2 x1 x2 x3 as shown in Fig. 2.1. A vector A can be represented as
A 5 A1 g 1 1 A 2 g 2 1 A3 g 3 ; (2.1)
where Ai , (i 5 1,2,3) denotes the three components of this vector, and gi represents the three unit vectors along three
coordinate axes, respectively. The subscript i (or other subscripts denoted by English letters) is an index that has a value
of 1, 2, or 3. This index is called a free index. Therefore, the vector A given in Eq. (2.1) can be written as Ai or Aj using
a component form. These two notations with different free indexes give the same physical concept denoting the three
components of the same vector under the Cartesian coordinate system o 2 x1 x2 x3 .

Fluid
Solid Interaction Dynamics. DOI: https://doi.org/10.1016/B978-0-12-819352-5.00002-1

© 2019 Higher Education Press. Published by Elsevier Inc. All rights reserved. 43
44 Fluid
Solid Interaction Dynamics

x3
A3
A

g3
A2
o x2
g2
g1
A1

x1
FIGURE 2.1 Cartesian coordinate system with its base vectors and a vector A.

2.1.2 Summation convention

Eq. (2.1) can be further expressed by
X
3
A5 A i g i 5 Ai g i 5 Aj g j : (2.2)
i51

Here, the convention is as follows: the repetition of an index in a term will denote a summation with respect to that
index over its range. For example, a dot product of two vectors A and B can be denoted by
X
3
AUB 5 Ai Bi 5 Ai Bi 5 Aj B j ; (2.3)
i51

and the square of the length of the line element is

X
3
ds2 5 dxi dxi 5 dxi dxi 5 dxj dxj : (2.4)
i51

2.1.3 Kronecker delta

We define that a notation

1; i 5 j
δij 5 ; (2.5)
0; i 6¼ j
which is called the Kronecker delta and represents a unit matrix with three lines and three columns. Therefore, we have
the following identities:
δij 5 δji ;
δii 5 δ11 1 δ22 1 δ33 5 3;
8 9
< A1 ;
> i51 >=
δij Aj 5 δi1 A1 1 δi2 A2 1 δi3 A3 5 A2 ; i 5 2 5 Ai ; (2.6)
>
: >
;
A3 ; i53
δij Bj 5 Bi ;
δij δjk 5 δik ;
so that Eq. (2.4) can be rewritten in the form
ds2 5 δij dxi dxj : (2.7)
 Cartesian tensor and matrix calculus Chapter | 2 45

2.1.4 Permutation symbol

The permutation symbol is defined by
8
>
> 1; ijk 5 123; 231; 312;
<
eijk 5 2 1 ijk 5 213; 132; 321; (2.8)
>
>
:
0; ijk 5 others:
Based on this definition, the permutation symbol has the following characteristics:
1. It vanishes whenever the values of any two indexes coincide.
2. It equals 1 when the subscripts permute like 1, 2, 3, and it equals 21 otherwise.
3. It is antisymmetrical to any two indexes, such as e123 5 2 e213 .
Using this notation, we can calculate the determinant of a matrix aij by the following equation:
 
aij  5 erst ar1 as2 at3 5 erst a1r a2s a3t : (2.9)
The cross-multiplication of two vectors ai and bj can be calculated by
 
 g1 g2 g3 
 
 
ða 3 bÞ 5  a1 a2 a3  5 erst gr as bt ;

  (2.10)
b b b 
1 2 3
ða 3 bÞr 5 erst as bt :
The volume of a parallelepiped having any three vectors ai , bj , and cj , as edged, is given by
 
 c1 c2 c3 
 
 
cUða 3 bÞ 5  a1 a2 a3  5 erst cr as bt : (2.11)
 
b b b 
1 2 3

From this equation, it follows that

 
1 0 0
 
   
g1 U g2 3 g3 5  0 1 0  5 1 5 e123 ;

  (2.12)
0 0 1
 
gi U gj 3 gk 5 eijk :
As an application, we can write the moment Ki , about the origin of the coordinate system, of a force fk acting at a
point xj in the following tensor form:
Ki 5 eijk xj fk : (2.13)

2.1.5 e 2 δ Identity
eijk eirs 5 δjr δks 2 δjs δkr (2.14)

2.1.6 Differentiation of a function f(x1, x2, x3)

@f
df ðx1 ; x2 ; x3 Þ 5 dxi 5 f;i dxi (2.15)
@xi
46 Fluid
Solid Interaction Dynamics

2.1.7 Transformation of coordinates

As shown in Fig. 2.2, the coordinate system O 2 x~1 x~2 is obtained from O 2 x1 x2 by rotating an angle θ in the counter-
clockwise direction about origin O. For a point p, the coordinates under these two systems satisfy the following
relationships:
" # " #" #
x1 cosθ 2sinθ x~1
5 ;
x2 sinθ cosθ x~2
" # " #" # (2.16)
x~1 cosθ sinθ x1
5 :
x~2 2sinθ cosθ x2
These two equations can be represented as the following index forms:
x~i 5 β ij xj ; xi 5 β ji x~j ; (2.17)
where
@x~i  
β ij 5 5 cos x~i ; xj ;
@xj
(2.18)
@xi    T
β ji 5 5 cos x~j ; xi 5 β ij :
@x~j

From Eqs. (2.16) and (2.17), we can obtain that

 T  21
β ji 5 β ij 5 β ij ; (2.19)
which implies that β ij is an orthogonal matrix. For an orthogonal matrix, we have
βUβ21 5 βUβT 5 I;
@x~i @xl @x~i
β il β jl 5 5 5 δij ;
@xl @x~j @x~j (2.20)
@x~l @xj @xj
β li β lj 5 5 5 δij :
@xi @x~l @xi

2.1.8 Tensor
A set of quantities is called a scalar, a vector, or a tensor depending on how the components of the set are defined
in the variables x1 , x2 , and x3 and on how they are transformed when the variables x1 , x2 , and x3 are changed to
x~1 , x~2 , and x~3 .

~
x2 p
x2
~
x1

θ
x1
o
FIGURE 2.2 Rotation of coordinates.
 Cartesian tensor and matrix calculus Chapter | 2 47

Scalar
A single component φ
Transformation φ~ 5 φ
A tensor of rank 0
Vector
Three components ai
Transformation a~ i 5 β ij aj
A tensor of rank 1
Stress
Nine components τ ij
Transformation τ~ ij 5 β ir β js τ rs
A tensor of rank 2

The important property of tensor fields is this: if a tensor equation. can be established in one coordinate system,
then it must hold for all coordinate systems.

2.1.9 Quotient rule

Consider a set of functions Aði; j; kÞ; ði; j; k 5 1; 2; . . .; nÞ; for which we do not know whether it is a tensor. Suppose we
know the nature of the product of Aði; j; kÞ with an arbitrary tensor; then we can determine whether the function Aði; j; kÞ
is a tensor without applying the law of transformation directly.
For example, let γ i be a vector, a tensor of rank 1, and the product
X
n
Aði; j; kÞγ i 5 AðI; j; kÞγ I 5 Ajk (2.21)
I51

yields a tensor Ajk . Then we can demonstrate that Aði; j; kÞ is a tensor of the type Aijk .

Proof: Since Ajk is a tensor, it is transformed into x~ coordinates as

 
A~ði; j; kÞγ~ i 5 A~jk 5 β jr β ks Ars 5 β jr β ks Aðm; r; sÞγ m : (2.22)
However, the tensor γ m 5 β im γ~ i , when substituted into Eq. (2.22), gives
 
A~ði; j; kÞ 2 β im β jr β ks Aðm; r; sÞ γ~ i 5 0: (2.23)
Now, γ i is an arbitrary vector, so that the quantity within the bracket must vanish, and we obtain
~ j; kÞ 5 β im β jr β ks Aðm; r; sÞ;
Aði; (2.24)
which is precisely the law of transformation of the tensor Aijk . This demonstration can be generalized to higher-order
tensors.

2.1.10 Index forms of some important variables

Using the index notation and defining an operator r 5 gi @=@xi , we can express the variables often used in continuum
mechanics in their index forms in Table 2.1.
The vector identity
Δv 5 grad div v 2 curl curl v (2.25a)
can be expressed in the index form
vi;jj 5 vj;ji 2 eijk ekst vt;sj ; (2.25b)
which is easily demonstrated by using the e 2 δ identity.
48 Fluid
Solid Interaction Dynamics

TABLE 2.1 The vector and index notations of some variables in mechanics.

Name Vector notation Index notation Its rank

Vector v vi 1
Dot, scalar, or inner product λ 5 uUv λ 5 ui vi 0
Cross or vector product w5u3v wi 5 eijk uj vk 1

Gradient of scalar field grad φ 5 rφ @φ=@xi 5 φ;i 1

Vector gradient grad v 5 rv @vi =@xj 5 vi;j 2

Divergence div v 5 rUv @vi =@xi 5 vi;i 0

Curl curl v 5 r 3 v eijk @vk =@xj 5 eijk vk;j 1

Laplacian r2 v 5 rUrv 5 Δv @2 vj =ð@xi @xi Þ 5 xj;ii 1

2.1.11 Two primary identities

For the application in Chapter 3, Fundamentals of continuum mechanics, to derive the deformation of continuums, we
demonstrate the following two identities:
Identity I: Assuming that B is an arbitrary tensor of rank 2 and that a; b; c are three arbitrary vectors, we have the
following identity:

ðBUaÞ 3 ðBUbÞUðBUcÞ 5 J B a 3 bUc; (2.26)

in which J B 5 jBj, the determinant of the tensor B.

Proof: Using Eqs. (2.9)
(2.12), we can obtain that

     
ðBUaÞ 3 ðBUbÞUðBUcÞ 5 Bij aj gi 3 Brs bs gr U Blm cm gl
 
5 Bij Brs Blm aj bs cm gi 3 gr Ugl 5 elir Bij Brs Blm aj bs cm (2.27)

5 emjs J B aj bs cm 5 J B a 3 bUc:

Identity II: With the same assumptions for the identity I, we have the second identity

ðBUaÞ 3 ðBUbÞ 5 J B B2T Uða 3 bÞ: (2.28)

Proof: Based on the dot multiplication rules, we can rewrite Eq. (2.26) in the form

ðBUaÞ 3 ðBUbÞUðBUcÞ 5 BT U½ðBUaÞ 3 ðBUbÞ Uc 5 J B a 3 bUc; (2.29)

in which the transpose of tensor B is introduced since the dot multiplication BUc is done by the second index of B.
Now, because vector c is arbitrary, premultiplying the inverse tensor B2T on both sides of Eq. (2.29) gives Eq. (2.28).

2.2 Matrix calculus

Numerical equations are normally expressed in matrix form, based on which derivative of a matrix with respect to
another vector is needed during the mathematical derivation. Therefore, the matrix derivative is a convenient notation
for keeping track of partial derivatives while doing calculations. Here, we present the information on matrix calculus
adopted in this book.
 Cartesian tensor and matrix calculus Chapter | 2 49

2.2.1 Types of matrix derivatives

We use a letter, such as x or y, to denote a scalar, a bold letter to indicate a vector, and a bold capital letter for a matrix.
The six types of derivatives that can be most neatly organized in matrix form are presented in Table 2.2.
We could also consider the derivatives of a vector with respect to a matrix or any of the unfilled cells in Table 2.3.
However, these derivatives are most naturally organized in tensor form, as discussed in Section 2.1, so they do not fit
neatly into a matrix.
Two competing notational conventions split the field of matrix calculus into two separate groups, which can be
distinguished by whether they write the derivative of a scalar of size 1 with respect to a vector of size n as a 1 3 n row
vector or as an n 3 1 column vector. The former is called as the numerator layout convention, or the Jacobian
formulation, while the latter is referred to as the denominator layout convention, or the Hessian formulation. Generally,
the derivative of a vector y with respect to a vector x may be denoted in the following formulation:
dy
Numerator layout: ;
dxT
(2.30a)
dyT
Denominator layout: :
dx

From this definition, we obtain that

0 1T
dx dxT @ dy T
A 5 dy ;
T
5I5 ; T
dx dx dx dx
(2.30b)
dxT dx
6¼ I 6¼ :
dxT dx

Serious mistakes can result when combining results from different authors without carefully verifying that
compatible notation has been used. Therefore great care should be taken to ensure notational consistency.

TABLE 2.2 Types of matrix derivatives.

Types Scalar Vector Matrix

Scalar @y=@x @y=@x @Y=@x
Vector @y=@x @y=@x
Matrix @y=@X

TABLE 2.3 Identities of vector by vector @y=@x.

Derivatives Numerator layout @y=@xT Denominator layout @yT =@x

@a=@x 0 0
@x=@x @x=@xT 5 I @xT =@x 5 I
@ðAxÞ=@x A AT
@ðbuÞ=@x b@u=@xT 1 u@b=@xT b@uT =@x 1 @b=@xuT
@ðAuÞ=@x A@u=@xT @uT =@xAT
@½gðuÞ=@x ð@g=@uT Þð@u=@xT Þ ð@uT =@xÞð@gT =@uÞ
du ð@u=@xT Þd x d xT ð@uT =@xÞ
50 Fluid
Solid Interaction Dynamics

2.2.2 Derivatives with vectors

2.2.2.1 Vector by scalar
The derivative of a vector
 T
y 5 y1 y2 ? ym (2.31a)
by a scalar x is written in numerator layout notation as

@y  T
5 @y1 =@x @y2 =@x ? @ym =@x ; (2.31b)
@x
which is known as the tangent vector of the vector y in vector calculus. For example, if x denotes the time t,
Eq. (2.31b) gives the tangent vector of the curves generated by the position vector y.

2.2.2.2 Scalar by vector

The derivative of a scalar y by a vector
 T
x 5 x1 x2 ? xn (2.32a)
is written in denominator layout notation as a column vector,
@y  T
5 @y=@x1 @y=@x2 ? @y=@xn ; (2.32b)
@x
and as a row vector in numerator layout notation
@y  
5 @y=@x1 @y=@x2 ? @y=@xn : (2.32c)
@x T

In vector calculus, Eq. (2.32b) gives the gradient vector of a scalar field yðxÞ.

2.2.2.3 Vector by vector

The derivative of a vector function y in Eq. (2.31a) with respect to a vector x in Eq. (2.32a) is written in numerator
layout notation as
2 3
@y1 =@x1 @y1 =@x2 ? @y1 =@xn
6 7
@y 6 @y2 =@x1 @y2 =@x2 ? @y2 =@xn 7
6 7
56 7; (2.33)
@xT 6 ^ ^ & ^ 7
4 5
@ym =@x1 @ym =@x2 ? @ym =@xn
which is called the Jacobian matrix when the vector x is the position vector in a space.

2.2.3 Derivatives with matrices

2.2.3.1 Matrix by scalar
The derivative of a matrix Y by a scalar x is known as the tangent matrix and is given in numerator layout notation by
2 3
@Y11 =@x @Y12 =@x ? @Y1n =@x
6 7
@Y 6 6 @Y21 =@x @Y22 =@x ? @Y2n =@x 7
7
56 7: (2.34)
@x 6 ^ ^ & ^ 7
4 5
@Yn1 =@x @Yn2 =@x ? @Ynn =@x
 Cartesian tensor and matrix calculus Chapter | 2 51

2.2.3.2 Scalar by matrix

The derivative of a scalar function yðXÞ with respect to an m 3 n matrix X is given in numerator layout notation by
2 3
@y=@x11 @y=@x21 ? @y=@xm1
6 7
@y 6 @y=@x12 @y=@x22 ? @y=@xm2 7
5 6 7; (2.35)
@XT 6 4 ^ ^ & ^ 7
5
@y=@x1n @y=@x2n ? @y=@xmn
of which each row is a 1 3 m vector. Important examples of scalar functions of matrices are the trace and determinant
of a matrix.

2.2.3.3 Matrix by matrix

In numerator layout, the derivative of an α 3 β matrix function FðXÞ with respect to an n 3 m matrix X is an m 3 n
matrix whose entries are an α 3 β matrix, i.e.,
2 3
@F=@x11 @F=@x21 ? @F=@xn1
6 7
@F 6 @F=@x12 @F=@x22 ? @F=@xn2 7
6 7:
T 56 7 (2.36)
@X 4 ^ ^ & ^ 5
@F=@x1m @F=@x2m ? @F=@xnm

2.2.4 Identities
2.2.4.1 Vector by vector
We assume that the vector a and the matrix A denote a constant vector and a constant matrix, respectively, and that
bðxÞ is a scalar function of vector x, as well as two vector functions uðxÞ; g½uðxÞ. According to these derivative
definitions, it is not difficult to demonstrate that the following identities are valid.

2.2.4.2 Scalar by vector

Many scalar functions of vectors or matrix are often used in numerical equations, for example, the inner products of
two vectors, the quadratic forms, and the norm of a vector. For those scalar functions, the derivative identities with
respect to scalar shown in Table 2.4 are valid.
It is not difficult to demonstrate the identities given in Tables 2.3 and 2.4 by using the definitions shown in
Eqs. (2.30a) and (2.30b). For example,
   T
@ xT Ax xT A@x @ xT Ax
5 1
@xT @xT @xT
(2.37)
xT A@x xT AT @x    
5 1 5x A1A I5x A1A :
T T T T
@xT @xT

Here, the first term denotes the derivative of column vector x, while the second term is the derivative of row vector
xT that needs to transpose to x in order to use Eq. (2.30b).

2.3 Exercise problems

In order for readers, especially students, who are not familiar with tensor operations to learn and use this mathematical
tool as used in this book, we present the following exercise problems and their solutions.
52 Fluid
Solid Interaction Dynamics

TABLE 2.4 Identities of scalar a by vector.

Expressions of scalar a Numerator layout @a=@xT Denominator layout @aT =@x

Inner product uT @v=@xT 1vT @u=@xT @uT =@xv 1@vT =@xu
a 5 uUv 5 uT v
Inner product 2xT 2x
a 5 xUx 5 xT x
Inner product aT a
a 5 aUx 5 aT x
Quadratic form uT A@v=@xT 1 vT AT @u=@xT @uT =@xAv 1 @vT =@xAT u
a 5 uUAv 5 uT Av
Quadratic form xT ðA 1 AT Þ ðA 1 AT Þx
a 5 xUAx 5 xT Ax
Quadratic form xT ðabT 1 baT Þ ðabT 1 baT Þx
a 5 aT xxT b
Norm ðx2aÞT ðx 2 aÞ
a 5 :x 2 a: :x 2 a: :x 2 a:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
5 ðx2aÞT ðx 2 aÞ

2.3.1 Problem 1: prove the following formulations

1. δii 5 3
Solution: This problem uses the summation convention and the definition of the Kronecker delta, so that

δii 5 δ11 1 δ22 1 δ33 5 1 1 1 1 1 5 3: (2.38a)

2. δij δij 5 3
Solution: Use the rule of Kronecker delta given by Eq. (2.6), which shows that, for any variable Ai with a subindex
the same as the one of δij , the result of Ai δij equals the replacement of the same subindex i of the variable by another
subindex j of δij . We then obtain

δij δij 5 δii 5 δjj 5 3: (2.38b)

3. eijk ejki 5 6
Solution: Using the properties of the permutation symbol for the shift of the subindex and the e 2 δ identity given in
Eq. (2.14), we obtain

eijk ejki 5 eijk eijk 5 δjj δkk 2 δjk δkj

(2.38c)
5 δjj δkk 2 δjj 5 3 3 3 2 3 5 6:

4. eijk Aj Ak 5 0
Solution: This problem involves the cross-multiplication of a vector A with itself, so that the result vanishes, i.e.,

eijk Aj Ak 5 ðA 3 AÞi 5 0: (2.39a)

 Cartesian tensor and matrix calculus Chapter | 2 53

However, here we wish to use the tensor operation to give the demonstration as follows:
eijk Aj Ak 5 2 eikj Aj Ak 5 2 eikj Ak Aj 5 2 eijk Aj Ak ;
(2.39b)
2eijk Aj Ak 5 0; eijk Aj Ak 5 0;
where the first equal sign results in a shift of subindexes j and k of the permutation symbol eijk , the second results from
the exchange rule of multiplication Aj Ak , and the third is obtained by exchanging the repeating indexes jj and kk
according to the summation convention rule.
5. δij δjk 5 δik
Solution: This can be immediately proved by Eq. (2.6). Furthermore, the summation convention rule can be used
to give the details, i.e.,

0; i 6¼ k;
δij δjk 5 δi1 δ1k 1 δi2 δ2k 1 δi3 δ3k 5 5 δik : (2.40)
1; i 5 k;
6. eijk δjk 5 0
Solution: This can be proved by two methods. One is just to use Eq. (2.6) and the definition of the permutation
symbol, by which the components with two same indexes vanish, i.e.,
eijk δjk 5 eijj 5 eikk 5 0: (2.41a)
Another way is the same as the one used in Eq. (2.39b) and considering δjk 5 δkj , i.e.,
eijk δjk 5 2 eikj δjk 5 2 eijk δkj 5 2 eijk δjk ; eijk δjk 5 0: (2.41b)

2.3.2 Problem 2: prove the identity of three arbitrary vectors

A 3 ðB 3 CÞ 5 ðAUCÞB 2 ðAUBÞC
Solution: Using Eq. (2.10) for the cross-multiplication of two vectors and e 2 δ identity given in Eq. (2.14), we can
obtain the component i of the left-side vector:
  
A 3 ðB 3 CÞ i 5 eijk Aj eklm Bl Cm 5 δil δjm 2 δim δjl Aj Bl Cm

5 δil δjm Aj Bl Cm 2 δim δjl Aj Bl Cm (2.42)


5 ðAm Cm ÞBi 2 ðAl Bl ÞCi 5 ðAUCÞB2ðAUBÞC i :

2.3.3 Problem 3: express the constitutive equation in a tensor form

1 ð1 1 νÞ
exx 5 ½σxx 2 νðσyy 1 σzz Þ; exy 5 σxy ;
E E
1 ð1 1 νÞ
eyy 5 ½σyy 2 νðσzz 1 σxx Þ; eyz 5 σyz ;
E E
1 ð1 1 νÞ
ezz 5 ½σzz 2 νðσxx 1 σyy Þ; ezx 5 σzx :
E E

Solution: We define Θ 5 σxx 1 σyy 1 σzz and use subindexes 1, 2, and 3 to represent x, y, and z, respectively, so that
we can express the constitutive equation in tensor form:
ð1 1 νÞ ν
eij 5 σij 2 Θδij : (2.43)
E E
54 Fluid
Solid Interaction Dynamics

2.3.4 Problem 4: write the tensor equation in a coordinate (xyz) form

1 @2 u i
G ui;kk 1 uk;ki 1 fi 5 ρ 2
1 2 2ν @t
Solution: This equation in tensor form represents three equations in the coordinate system. For index i taking values
1, 2, and 3 representing x, y, and z, respectively, we obtain the following three equations using the summation
convention:
0 1
1 @Θ A 1 fx 5 ρ @ ux ;
2
G@Δux 1
1 2 2ν @x @t2
0 1
1 @Θ A 1 fy 5 ρ @ uy ;
2
G@Δuy 1
1 2 2ν @y @t2
(2.44)
0 1
1 @ΘA @ uz 2
G@Δuz 1 1 fz 5 ρ 2 ;
1 2 2ν @z @t

@ux @uy @uz

Θ 5 rUu 5 1 1 :
@x @y @z

2.3.5 Problem 5: prove the following identities using index notations

1. curl curl v 5 grad div v 2 Δv
2. divðr n rÞ 5 ðn 1 3Þr n
3. curlðr n rÞ 5 0
4. Δðr n Þ 5 nðn 1 1Þr n22
Here, r 2 5 xj xj and r 5 xj gj . Other notations in the preceding equations are defined in Table 2.1.
Solution: Using the derivative
@r 2 2r@r 2xj @xj
5 5 5 2xj δji 5 2xi ;
@xi @xi @xi
@r xi
5 ;
@xi r

we can derive the following solutions:

ðcurl curl vÞi 5 ðr 3 r 3 vÞi 5 eijk ðr 3 vÞk;j
 
5 eijk eklm vm;lj 5 δil δjm 2 δim δjl vm;lj
(2.45a)
5 vm;mi 2 vi;jj 5 ðrUvÞ;i 2 Δvi
5 ðgrad div vÞi 2 ðΔvÞi :
@ðr n xi Þ nr n21 xi @r r n @xi
divðr n rÞ 5 5 1
@xi @xi @xi
nr n21 xi @r r n @xi nr n21 xi xi r n @xi (2.45b)
5 1 5 1
@xi @xi r @xi
5 nr n 1 3r n 5 ðn 1 3Þr n :
 Cartesian tensor and matrix calculus Chapter | 2 55

@r n @xk
ðcurlðr n rÞÞi 5 eijk ðr n xk Þ;j 5 eijk xk 1 eijk r n
@xj @xj
xj (2.45c)
5 eijk nr n21 xk 1 eijk r n δjk 5 nnr n22 eijk xj xk 1 eijj r n
r
5 0 1 0 5 0:
0 1
@ @  
@nr n21 A 5 @ nr n22 xi
@r
2 n
r
Δðr n Þ 5 5
@xi @xi @xi @xi @xi
@xi (2.45d)
5 nðn 2 2Þr n24 xi xi 1 nr n22
@xi
5 nðn 2 2Þr n22 1 3nr n22 5 nðn 1 1Þr n22 :

2.3.6 Problem 6: prove eijkaiajbk 5 0 for nonzero vectors a and b

Solution: Using Eq. (2.9) for the value of determinant, we obtain
 
 a1 a1 b1 
 
 
eijk ai aj bk 5  a2 a2 b2  5 0; (2.46)
 
a a b 
3 3 2

since the two column vectors in it are the same.