TENSOR ALGEBRA

Multimedia Course on Continuum Mechanics

 Overview
 Introduction to tensors Lecture 1

 Indicial or (Index) notation Lecture 2

 Vector Operations Lecture 3 Lecture 4

 Tensor Operations Lecture 5

 Differential Operators Lecture 6

 Integral Theorems
Lecture 7
 References

2
 Introduction

SCALAR ρ , θ , ...
v
VECTOR v, f , ...

MATRIX σ , ε, ...

? C, ...

4
 Concept of Tensor
 A TENSOR is an algebraic entity with various components
which generalizes the concepts of scalar, vector and matrix.

 Many physical quantities are mathematically represented as tensors.

 Tensors are independent of any reference system but, by need, are

commonly represented in one by means of their “component matrices”.

 The components of a tensor will depend on the reference system

chosen and will vary with it.

5
 Order of a Tensor
 The order of a tensor is given by the number of indexes
needed to specify without ambiguity a component of a tensor.

a  Scalar: zero dimension α = 3.14  1.2 

 
a , a  Vector: 1 dimension vi =  0.3 
 0.8 
A, A  2nd order: 2 dimensions    0.1 0 1.3 
 
Eij =  0 2.4 0.5 
A , A 3rd order: 3 dimensions  1.3 0.5 5.8 
 
A , A  4th order …

6
 Cartesian Coordinate System
 Given an orthonormal basis formed by three mutually
perpendicular unit vectors:
eˆ1 ⊥ eˆ 2 , eˆ 2 ⊥ eˆ 3 , eˆ 3 ⊥ eˆ1
Where:

=eˆ1 1=
, eˆ 2 1=
, eˆ 3 1

 Note that

1 if i = j 
=
eˆ i ⋅ eˆ j  =  δ ij
0 if i ≠ j 

7
 Indicial or (Index) Notation
Tensor Algebra

10
 Tensor Bases – VECTOR
 A vector v can be written as a unique linear combination of the
three vector basis eˆ i for i ∈ {1, 2,3} .
v
v = v1eˆ1 + v 2eˆ 2 + v3eˆ 3
v3
 In matrix notation:
 v1 
[ v ] =  v2  v1
 v3  v2
 In index notation:

v = ∑ vi eˆ i tensor as a physical entity

i
[ v ]i = vi component i of the tensor in the
given basis i ∈ {1, 2,3}

11
 Tensor Bases – 2nd ORDER TENSOR
 A 2nd order tensor A can be written as a unique linear combination
of the nine dyads eˆ i ⊗ eˆ j ≡ eˆ i eˆ j for i, j ∈ {1, 2,3} .

= A11 ( eˆ1 ⊗ eˆ1 ) + A12 ( eˆ1 ⊗ eˆ 2 ) + A13 ( eˆ1 ⊗ eˆ 3 ) +

A
+ A21 ( eˆ 2 ⊗ eˆ1 ) + A22 ( eˆ 2 ⊗ eˆ 2 ) + A23 ( eˆ 2 ⊗ eˆ 3 ) +
+ A31 ( eˆ 3 ⊗ eˆ1 ) + A32 ( eˆ 3 ⊗ eˆ 2 ) + A33 ( eˆ 3 ⊗ eˆ 3 )

Alternatively, this could have been written as:

A = A11 eˆ1eˆ1 + A12 eˆ1eˆ 2 + A13 eˆ1eˆ 3 +

+ A21 eˆ 2eˆ1 + A22 eˆ 2eˆ 2 + A23 eˆ 2eˆ 3 +
+ A31 eˆ 3eˆ1 + A32 eˆ 3eˆ 2 + A33 eˆ 3eˆ 3

12
 Tensor Bases – 2nd ORDER TENSOR
= A11 ( eˆ1 ⊗ eˆ1 ) + A12 ( eˆ1 ⊗ eˆ 2 ) + A13 ( eˆ1 ⊗ eˆ 3 ) +
A
+ A21 ( eˆ 2 ⊗ eˆ1 ) + A22 ( eˆ 2 ⊗ eˆ 2 ) + A23 ( eˆ 2 ⊗ eˆ 3 ) +
+ A31 ( eˆ 3 ⊗ eˆ1 ) + A32 ( eˆ 3 ⊗ eˆ 2 ) + A33 ( eˆ 3 ⊗ eˆ 3 )
 In matrix notation:
 A11 A12 A13 
[ A ] =  A21 A22 A23 
 A31 A32 A33 
 In index notation:
=A ∑ Aij ( eˆ i ⊗ eˆ j )
ij
tensor as a
physical entity

[ A ]ij = Aij component ij of the tensor

in the given basis i, j ∈ {1, 2,3}

13
 Tensor Bases – 3rd ORDER TENSOR
 A 3rd order tensor A can be written as a unique linear combination
of the 27 tryads eˆ i ⊗ eˆ j ⊗ eˆ k ≡ eˆ i eˆ j eˆ k for i, j , k ∈ {1, 2,3}.
A A111 ( eˆ1 ⊗ eˆ1 ⊗ eˆ1 ) + A121 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ1 ) + A131 ( eˆ1 ⊗ eˆ 3 ⊗ eˆ1 ) +
=
+ A211 ( eˆ 2 ⊗ eˆ1 ⊗ eˆ1 ) + A221 ( eˆ 2 ⊗ eˆ 2 ⊗ eˆ1 ) + A231 ( eˆ 2 ⊗ eˆ 3 ⊗ eˆ1 ) +
+ A311 ( eˆ 3 ⊗ eˆ1 ⊗ eˆ1 ) + A321 ( eˆ 3 ⊗ eˆ 2 ⊗ eˆ1 ) + A331 ( eˆ 3 ⊗ eˆ 3 ⊗ eˆ1 ) +
+ A112 ( eˆ1 ⊗ eˆ1 ⊗ eˆ 2 ) + A122 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ 2 ) + ...
Alternatively, this could have been written as:

A = A111 eˆ1eˆ1eˆ1 + A121 eˆ1eˆ 2eˆ1 + A131 eˆ1eˆ 3eˆ1 +

+ A211 eˆ 2eˆ1eˆ1 + A221 eˆ 2eˆ 2eˆ1 + A231 eˆ 2eˆ 3eˆ1 +
+ A311 eˆ 3eˆ1eˆ1 + A321 eˆ 3eˆ 2eˆ1 + A331 eˆ 3eˆ 3eˆ1 +
+ A112 eˆ1eˆ1eˆ 2 + A122 eˆ1eˆ 2eˆ 2 + ...

14
 Tensor Bases – 3rd ORDER TENSOR

A A111 ( eˆ1 ⊗ eˆ1 ⊗ eˆ1 ) + A121 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ1 ) + A131 ( eˆ1 ⊗ eˆ 3 ⊗ eˆ1 ) +
=
+ A211 ( eˆ 2 ⊗ eˆ1 ⊗ eˆ1 ) + A221 ( eˆ 2 ⊗ eˆ 2 ⊗ eˆ1 ) + A231 ( eˆ 2 ⊗ eˆ 3 ⊗ eˆ1 ) +
A311 ( eˆ 3 ⊗ eˆ1 ⊗ eˆ1 ) + A321 ( eˆ 3 ⊗ eˆ 2 ⊗ eˆ1 ) + A331 ( eˆ 3 ⊗ eˆ 3 ⊗ eˆ1 ) +
+
+ A112 ( eˆ1 ⊗ eˆ1 ⊗ eˆ 2 ) + A122 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ 2 ) + ...
 In matrix notation:
 A  
  A112113 AA 123 A133  
A132 
 A
 111 A 122 A 
 AA213 AA223 AA233  
121 131

[A] =  A211212 A221222 A231232   
  AA313 AA323 AA333  
 A311312 A321322 A331332   
   

15
 Tensor Bases – 3rd ORDER TENSOR

A A111 ( eˆ1 ⊗ eˆ1 ⊗ eˆ1 ) + A121 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ1 ) + A131 ( eˆ1 ⊗ eˆ 3 ⊗ eˆ1 )
=
+ A211 ( eˆ 2 ⊗ eˆ1 ⊗ eˆ1 ) + A221 ( eˆ 2 ⊗ eˆ 2 ⊗ eˆ1 ) + A231 ( eˆ 2 ⊗ eˆ 3 ⊗ eˆ1 ) +
+ A311 ( eˆ 3 ⊗ eˆ1 ⊗ eˆ1 ) + A321 ( eˆ 3 ⊗ eˆ 2 ⊗ eˆ1 ) + A331 ( eˆ 3 ⊗ eˆ 3 ⊗ eˆ1 ) +
+ A112 ( eˆ1 ⊗ eˆ1 ⊗ eˆ 2 ) + A122 ( eˆ1 ⊗ eˆ 2 ⊗ eˆ 2 ) + ...
 In index notation:
=A ∑ A ( eˆ ⊗ eˆ =
ijk
ijk ⊗ eˆ )
i j k

= Aijk ( eˆ i ⊗ eˆ j ⊗ eˆ k ) ≡ Aijk eˆ i eˆ j eˆ k
tensor as a
physical entity

[A]ijk = Aijk component ijk of the tensor

in the given basis i, j , k ∈ {1, 2,3}

16
 Repeated-index (or Einstein’s) Notation
 The Einstein Summation Convention: repeated Roman indices are
summed over. 3
i is a mute
index
ai bi = ∑ ai bi = a1b1 + a2b2 + a3b3
i =1

i is a talking 3
index and j is a Aij b j = ∑ Aij b j = Ai1b1 + Ai 2b2 + Ai 3b3
mute index j =1
 A “MUTE” (or DUMMY) INDEX is an index that does not appear in a
monomial after the summation is carried out (it can be arbitrarily changed
of “name”).
 A “TALKING” INDEX is an index that is not repeated in the same
monomial and is transmitted outside of it (it cannot be arbitrarily changed
of “name”). REMARK
An index can only appear up
to two times in a monomial.

18
 Repeated-index (or Einstein’s) Notation

Rules of this notation:

1. Sum over all repeated indices.
2. Increment all unique indices fully at least once, covering all
combinations.

3. Increment repeated indices first.

4. A comma indicates differentiation, with respect to coordinate xi .
∂ui 3
∂ui ∂ 2ui 3
∂ 2ui ∂Aij 3 ∂A
u= = ∑ = ui , jj = ∑ 2 A= = ∑ ij

∂xi i =1 ∂xi ∂x j ∂x j j =1 ∂x j ∂x j j =1 ∂x j
i ,i ij , j

5. The number of talking indices indicates the order of the tensor result

19
 Kronecker Delta δ
 The Kronecker delta δij is defined as:
1 if i = j
δ ij = 
0 if i ≠ j
 Both i and j may take on any value in
 Only for the three possible cases where i = j is δij non-zero.

1 if= i j (δ= δ= δ= 1)

δ ij =  11 22 33

0 if i ≠ j (δ12 =
δ13 =
δ 21... = 0)
REMARK

δ ij = δ ji Following Einsten’s notation: δ ii = δ11 + δ 22 + δ 33 = 3
Kronecker delta serves as a replacement operator:
=δ ij u j u=
i , δ ij A jk Aik
20
 Levi-Civita Epsilon (permutation) e
 The Levi-Civita epsilon eijk is defined as:

 0 if there is a repeated index


eijk =
 +1 if ijk =
123, 231 or 312
 −1 if ijk =
213, 132 or 321


 3 indices 27 possible combinations.


eijk = −eikj
REMARK
The Levi-Civita symbol is also named
permutation or alternating symbol.

21
 Example
 Prove the following expression is true:
eijk eijk = 6

23
 Example - Solution
k =1 k =2 k =3
eijk eijk = e111e111 + e112e112 + e113e113 + j =1
=1
+ e121e121 + e122e122 + e123e123 + j = 2
i =1 = −1
+ e131e131 + e132e132 + e133e133 + j = 3
= −1
+ e211e211 + e212e212 + e213e213 +
i=2 + e221e221 + e222e222 + e223e223 +
=1
+ e231e231 + e232e232 + e233e233 +
=1
+ e311e311 + e312e312 + e313e313 +
= −1
i=3 + e321e321 + e322e322 + e323e323 +

+ e331e331 + e332e332 + e333e333 = 6

24
 Vector Operations
Tensor Algebra

25
 Vector Operations
 Sum and Subtraction. Parallelogram law.

a+b = b+a = c c=
i ai + bi
a−b =d d=
i ai − bi

 Scalar multiplication

αa =
b=α a1eˆ1 + α a2eˆ 2 + α a3eˆ 3 bi = α ai

26
 Vector Operations
 Scalar or dot product yields a scalar
where θ is the angle
u⋅v =u v cos θ between the vectors u and v

 In index notation:
 i =3

u ⋅ v= ui eˆ i ⋅ v j eˆ j= ui v j eˆ i ⋅ eˆ j= ui v jδ ij= ui v i = ∑ i i = [ u] [ v ]
T
u v
 i =1 
u v δ ij =  0 (i ≠ j )

= 1=
 Norm of a vector ( j i)

u =(u ⋅ u ) =( uiui )
12 12

u = u ⋅ u = ui eˆ i ⋅ u j eˆ j = ui u jδ ij = ui ui
2

27
 Vector Operations
 Some properties of the scalar or dot product

u ⋅ v = v ⋅u
u⋅0 = 0
u ⋅ (α v + β w )= α ( u ⋅ v ) + β ( u ⋅ w ) Linear operator

u ⋅u > 0 u≠0
=
u ⋅u 0 = u 0
u=⋅ v 0, u ≠ 0, v ≠ 0 u⊥v

28
 Vector Operations
 Vector product (or cross product ) yields another vector
c =a × b =−b × a
c = a b sin θ where θ is the angle
between the vectors a and b
0 ≤θ ≤π
 In index notation:
=
c c= ˆ eijk a j bk eˆ i
i ei ⇒ =
ci eijk a j bk i ∈ 
i =1 i=2 i=3
 eˆ1 eˆ 2 eˆ 3 
= det  a1 a2 a3 
symb
c=( a2b3 − a3b2 ) eˆ1 + ( a3b1 − a1b3 ) eˆ 2 + ( a1b2 − a2b1 ) eˆ 3
e a2b3 + e132 a3b2 e231 a3b1 + e213 a1b3 e312 a1b2 + e321 a2b1  b1 b2 b3 
123
     
=1 =−1 =1 = −1 =1 = −1

29
 Vector Operations
 Some properties of the vector or cross product

u × v =− ( v × u )
u ×=
v 0, u ≠ 0, v ≠ 0 u || v
u × ( a v + b w ) = au × v + b u × w Linear operator

30
 Vector Operations
 Tensor product (or open or dyadic product) of two vectors:
A = u ⊗ v ≡ uv
Also known as the dyad of the vectors u and v, which results in a 2nd
order tensor A.
 Deriving the tensor product along an orthonormal basis {êi}:
A = ( u ⊗ v ) = ( ui eˆ i ) ⊗ ( v j eˆ j ) = ui v j ( eˆ i ⊗ eˆ j ) = Aij ( eˆ i ⊗ eˆ j )

[ A ]ij =Aij =[ u ⊗ v ]ij =ui v j i, j ∈ 

 In matrix notation:
 u1   u1 v1 u1 v 2 u1 v 3   A11 A12 A13 
[u ⊗= v ] u2  [ v1
v ] [u ][= v 3 ] u2 v1 u2 v 2
= u2 v 3  =  A21 A23 
T
v2 A22
u3  u3 v1 u3 v 2 u3 v 3   A31 A32 A33 

31
 Vector Operations
 Some properties of the open product:
(u ⊗ v ) ≠ ( v ⊗ u )
( u ⊗ v ) ⋅ w =u ⊗ ( v ⋅ w ) =u ( v ⋅ w ) =( v ⋅ w ) u
u ⊗ (α v + β w ) = α u ⊗ v + β u ⊗ w Linear operator
( u ⊗ v )( w ⊗ x ) = ( u ⊗ x )( v ⋅ w )
u ⋅ ( v ⊗ w ) = (u ⋅ v ) ⊗ w = (u ⋅ v ) w = w (u ⋅ v )

32
 Example
 Prove the following property of the tensor product is true:
u ⋅ ( v ⊗ w ) = (u ⋅ v ) ⊗ w

33
 Tensor Operations
Tensor Algebra

39
 Tensor Operations
 Summation (only for equal order tensors)
A+B =B+A =C C=
ij Aij + Bij

 Scalar multiplication (scalar times tensor)

αA = C Cij = α Aij

40
 Tensor Operations
 Dot product (.) or single index contraction product

A ⋅ b =
c ci = Aij b j Index “j” disappears (index
2nd 1st 1st contraction)
order order order

A ⋅ b =C Cij = Aijk bk Index “k” disappears (index

2nd
contraction)
3rd 1st
order order order

A ⋅ B =
C Cik = Aij B jk Index “j” disappears (index
2nd 2nd 2nd contraction)
order order order
REMARK
A⋅B ≠ B⋅A A⋅A =
A2

41
 Tensor Operations
 Some properties:
A ⋅ (α b + β c )= α A ⋅ b + β A ⋅ c Linear operator

 2nd order unit (or identity) tensor

1⋅ u = u ⋅1 = u
1 = δ ij ei ⊗ e j = ei ⊗ ei 1 0 0 
 [1] = 0 1 0 
[1]ij = δ ij
 0 0 1 

42
 2nd Order Tensor Operations
 Some properties:

1⋅ A = A = A ⋅1
A ⋅ ( B + C) = A ⋅ B + A ⋅ C
A ⋅ ( B ⋅ C ) =( A ⋅ B ) ⋅ C = A ⋅ B ⋅ C
A⋅B ≠ B⋅A

43
 Example
 When does the relation n ⋅ T = T ⋅ n hold true ?

44
 2nd Order Tensor Operations
 Transpose
 A11 A12 A13 
[ A ] = A21 A22 A23  [ AT ]
 A11
A
A21
A22
A31 
A32 
(A )
T T
=A
 12 ( A ⋅ B ) =BT ⋅ AT
T

 A31 A32 A33   A13 A23 A33 

( ⊗ ) =⊗
T
u v v u
[ AT ]ij = A ji (α A + β B )
T
=α AT + β BT
 Trace yields a scalar
Tr ( A ) = Aii ( = A11 + A22 + A33 ) Tr ( a ⊗ b ) =Tr  ai b j  =ai bi =a ⋅ b
 Some properties:
Tr ( AT ) = Tr A Tr ( A + B ) = TrA + TrB
Tr (α A ) = α TrA Tr ( A ⋅ B )= Tr ( B ⋅ A )
48
 2nd Order Tensor Operations

 Double index contraction or double (vertical) dot product (:)

A : B = c c = Aij Bij Indices “i,j” disappear (double index
zero contraction)
2nd 2nd
order order order
(scalar)
A : B = c ci = Aijk B jk Indices “j,k” disappear (double index
contraction)
3rd 2nd 1st
order order order

A : B = C Cij =  ijkl Bkl Indices “k,l” disappear (double index

4th 2nd 2nd contraction)
order order order
 Indices contiguous to the double-dot (:) operator get vertically repeated
(contraction) and they disappear in the resulting tensor (4 order reduction of the sum
of orders).

49
 2nd Order Tensor Operations
 Some properties

A : B= Tr ( AT ⋅ B )= Tr ( BT ⋅ A )= Tr ( A ⋅ BT )= Tr ( B ⋅ AT )= B : A

1= =
: A Tr A A :1
A : ( B ⋅ C ) = ( BT ⋅ A ) : C = ( A ⋅ CT ) : B
A : ( u ⊗ v ) =u ⋅ ( A ⋅ v )
(u ⊗ v ) : ( w ⊗ x) = (u ⋅ w ) ⋅ ( v ⋅ x)

REMARK
A : B= C : B ⇒ A= C

50
 2nd Order Tensor Operations

 Double index contraction or double (horizontal) dot product (··)

A ⋅⋅ B = c c = Aij B ji Indices “i,j” disappear (double index
zero contraction)
2nd 2nd
order order order
(scalar)

A ⋅⋅ B = c ci = Aijk Bkj Indices “j,k” disappear (double index

contraction)
3rd 2nd 1st
order order order

A ⋅⋅ B = C Cij =  ijkl Blk Indices “k,l” disappear (double index

4th 2nd 2nd contraction)
order order order
 Indices contiguous to the double-dot (··) operator get horizontally repeated
(contraction) and they disappear in the resulting tensor (4 orders reduction of the sum
of orders).

51
 Tensor Operations
 Norm of a tensor is a non-negative real number defined by
(A : A) ( Aij Aij )
12
= = ≥0
12
A

A ⋅⋅ B= Tr ( A ⋅ B )= Tr ( B ⋅ A )= B ⋅⋅A

REMARK
1 ⋅⋅A= TrA= A ⋅⋅1
A : B ≠ A ⋅⋅B
Unless one of the two
tensors is symmetric.

52
 Example
 Prove that:

A : B Tr ( AT ⋅ B )
=

⋅⋅ B Tr ( A ⋅ B )
A=

53
 2nd Order Tensor Operations
 Determinant yields a scalar
 A11 A12 A13 
[ A ] det  A21 A23  eijk A1i A
1
= =
det A det A22 = = A
2 j 3k eijk epqr Api Aqj Ark
6
A A32 A33 
 Some properties:  31
REMARK
det ( A ⋅ B=
) det A ⋅ det B The tensor A is SINGULAR if and
det AT = det A only if det A = 0.
det (α A ) = α 3 det A A is NONSINGULAR if det A ≠ 0.
 Inverse
There exists a unique inverse A-1 of A when A is nonsingular, which
satisfies the reciprocal relation:

 A ⋅ A −1 ==
1 A −1 ⋅ A

 A=
−1
ik Akj A= −1
ik Akj δ ij i, j , k ∈ {1, 2,3}
55
 Example
 Prove that det A = eijk A1i A2 j A3.k

57
 Differential Operators
Tensor Algebra

64
 Differential Operators
 A differential operator is a mapping that transforms a field
v ( x ) , A ( x ) ... into another field by means of partial derivatives.
 The mapping is typically understood to be linear.
 Examples:
 Nabla operator
 Gradient
 Divergence
 Rotation
 …

65
 Nabla Operator
 The Nabla operator  is a differential operator
“symbolically” defined as:
symbolic
∂ symb. ∂
∇= = eˆ i
∂x ∂xi
 In Cartesian coordinates, it can be used as a (symbolic) vector
on its own:
 ∂ 
 
∂
 1x
symb. 
∂ 
[∇ ] = 
 ∂x2 
 ∂ 
 
∂
 3
x

66
 Gradient
 The gradient (or open product of Nabla) is a differential
operator defined as:
 Gradient of a scalar field Φ(x):
 Yields a vector
 symb.
∂ ∂Φ
[ ∇Φ ]i [
= ∇ ⊗ Φ ]i [ ]i
= ∇ Φ =
∂xi
Φ =
∂xi
i ∈ {1, 2,3}

 ∂Φ
∇Φ = [∇Φ ] eˆ = ∂Φ eˆ ∇Φ = eˆ i
 i i
∂xi
i
∂xi
 Gradient of a vector field v(x):
 Yields a 2nd order tensor
 symb.
∂ ∂v j
 [ ∇ ⊗ v ] = [ ∇ ] [ v ] = vj = i, j ∈ {1, 2,3}

ij i j
∂xi ∂xi
 ∂v j
∇v = ∇ ⊗ v = [∇ ⊗ v ] eˆ ⊗ eˆ = ∂v j =
∇v eˆ i ⊗ eˆ j
eˆ i ⊗ eˆ j ∂xi
 ij i j
∂xi
67
 Gradient
 Gradient of a 2nd order tensor field A(x):
 Yields a 3rd order tensor

 symb.
∂ ∂A jk
[∇A ]ijk = [∇ ⊗ A ]ijk = [∇ ]i [ A ] jk = A jk = i, j , k ∈ {1, 2,3}
 ∂xi ∂xi

∇A = ∇ ⊗ A = [∇ ⊗ A ] eˆ ⊗ eˆ ⊗ eˆ = ∂A jk eˆ ⊗ eˆ ⊗ eˆ
 ijk i j k
∂xi
i j k

∂A jk
=
∇A eˆ i ⊗ eˆ j ⊗ eˆ k
∂xi

68
 Divergence
 The divergence (or dot product of Nabla) is a differential
operator defined as :
 Divergence of a vector field v(x):
 Yields a scalar
symb.
∂ ∂vi ∂vi
∇ ⋅ v = [∇ ]i [ v ]i = vi = ∇⋅v =
∂xi ∂xi ∂xi
 Divergence of a 2nd order tensor A(x):
 Yields a vector
 symb.
∂ ∂A ij
 ∇ ⋅ A j = [∇ ]i [ A ]ij =
[ ] A ij = j ∈ {1, 2,3}
 ∂x i ∂xi

∇ ⋅ A = [∇ ⋅ A ] eˆ = ∂A ij eˆ ∂A ij
 j j
∂xi
j
∇⋅A = eˆ j
∂xi

69
 Divergence

 The divergence can only be performed on tensors of order 1 or

higher.

 If ∇ ⋅ v = 0, the vector field v ( x ) is said to be solenoid (or

divergence-free).

70
 Rotation
 The rotation or curl (or vector product of Nabla) is a differential
operator defined as:
 Rotation of a vector field v(x):
 Yields a vector
[ symb. symb.
∂ ∂ vk
 ∇ × v=]i e [ ∇ ] [ v ]
= e =
v e i ∈ {1, 2,3}
∂ ∂x j
ijk j k ijk k ijk
 x j
 ∂ vk
∇ × v = [∇ × v ]i eˆ = e ∂ v k eˆ ∇ × v = eijk eˆ i
 i ijk
∂x j
i
∂x j
 Rotation of a 2nd order tensor A(x):
 Yields a 2nd order tensor
[ symb.
∂ ∂ A kl
=
]
∇ × A il e= A kl eijk i, j , k ∈ {1, 2,3}
∂ ∂
ijk
 x j x j
 ∂ A kl
∇ × A = [∇ × A ]il eˆ ⊗ eˆ = e ∂ A kl eˆ ⊗ eˆ = ∇ × A eijk
∂x j
eˆ i ⊗ eˆ l
 i l ijk
∂x j
i l

71
 Rotation

 The rotation can only be performed on tensors of order 1 or

higher.

 If ∇ × v = 0 , the vector field v ( x ) is said to be irrotational (or

curl-free).

72
 Differential Operators - Summary

scalar field vector field 2nd order tensor

Φ(x) v(x) A(x)

[∇ ⊗ Φ ]i = [∇ ⊗ v ]ij = [∇ ⊗ A ]ijk =
GRADIENT ∂Φ ∂v ∂A jk
= [∇Φ ]i = [∇v ]ij = j
= [ ]ijk
=∇ A =
∂xi ∂xi ∂xi

∂vi ∂A ij
DIVERGENCE ∇⋅v = [∇ ⋅ A ] j =
∂xi ∂xi

ROTATION [∇ × v ]i = eijk ∂ v k [∇ × A ]il = eijk ∂ A kl

∂x j ∂x j

73
 Example
 Given the vector v =v ( x ) =x1 x2 x3eˆ1 + x1 x2eˆ 2 + x1eˆ 3
determine ∇ ⋅ v, ∇ × v, ∇v.

74
 Example - Solution
 x1 x2 x3 
v =v ( x ) =x1 x2 x3eˆ1 + x1 x2eˆ 2 + x1eˆ 3
[ v ] =  x1 x2 
 x1 
 Divergence:

∂vi
∇⋅v =
∂xi

∂vi ∂v1 ∂v 2 ∂v3

∇=
⋅v = + + = x2 x3 + x1
∂xi ∂x1 ∂x2 ∂x3

75
 Example - Solution
 Divergence:  x1 x2 x3 
∇⋅v =
∂vi [ v ] =  x1 x2 
∂xi
 x1 
 In matrix notation:
 x1 x2 x3  symb
T
 ∂ ∂ ∂  
,   x1 x2  =
symb symb
∇ ⋅ v = [∇ ] [ v ] =  ,
T

 ∂x1 ∂x2 ∂x3   x 

 1 
13
31
symb
∂ ∂ ∂ ∂ ( x1 x2 x3 ) ∂ ( x1 x2 ) ∂x1
= x1 x2 x3 + x1 x2 + x1 = + + = x2 x3 + x1
∂x1 ∂x2 ∂x3 ∂x1 ∂x2 ∂x3
11

76
 Integral Theorems
Tensor Algebra

81
 Divergence or Gauss Theorem
 Given a field A in a volume V with closed boundary surface
∂V and unit outward normal to the boundary n , the
Divergence (or Gauss) Theorem states:

∫ ∇ ⋅A dV
V
= ∫
∂V
n ⋅A dS

∫ A ⋅∇=
V
dV ∫
∂V
A ⋅ n dS
Where:
A represents either a vector field ( v(x) ) or a tensor field ( A(x) ).

82
 Generalized Divergence Theorem
 Given a field A in a volume V with closed boundary surface
∂V and unit outward normal to the boundary n , the
Generalized Divergence Theorem states:

∫ ∇ ∗ A dV
V
= ∫
∂V
n ∗A dS

∫ A ∗∇=
V
dV ∫
∂V
A ∗ n dS
Where:
 ∗ represents either the dot product ( · ), the cross product (  ) or the
tensor product (  ).
 A represents either a scalar field ( ϕ(x) ), a vector field ( v(x) ) or a
tensor field ( A(x) ).

83
 Example
 Use the Generalized Divergence Theorem to show that

∫S
xi n j dS = V δ ij
nj
where xi is the position vector of nj .

∫∂V
A ∗ n=
dS ∫ A ∗∇ dV
V

85
 References
Tensor Algebra

88
 References

 José Mª Goicolea, Mecánica de Medios Continuos: Resumen de Álgebra y

Cálculo Tensorial, UPM.

 Eduardo W. V. Chaves, Mecánica del Medio Continuo,Vol. 1 Conceptos

básicos, Capítulo 1: Tensores de Mecánica del Medio Continuo, CIMNE,
2007.

 L. E. Malvern. Introduction to the mechanics of a continuous medium.

Prentice-Hall, Englewood Clis, NJ, 1969.

 G. A. Holzapfel. Nonlinear solid mechanics: a continuum approach for

engineering. 2000.

89