CONTINUUM MECHANICS - Introduction to tensors

CONTINUUM MECHANICS
- Introduction to tensors

Attila KOSSA

Budapest University of Technology and Economics

Department of Applied Mechanics

2016.09.13.

1 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Geometrical meaning of the scalar (or dot) product

a · b = |a| |b| cosϕ (1)

where ϕ is the angle between the tips of a and b, whereas |a| and |b|
represent the length of a and b. Vectors a and b are orthogonal (or
perpendicular to each other) if their scalar product is zero, i.e. a·b = 0.
2
Obviously we can observe that a · a = |a| .

2 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Geometrical meaning of the cross (or vector) product

a × b = (|a| |b| sinϕ) e (2)

where e is a unit vector perpendicular to the plane spanned by vectors

a and b. Rotating a about e with positive angle ϕ carries a to b. a
and b are parallel if a × b = 0. It follows that a × b = −b × a.

3 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Projection
Let the projection of vector a along the direction designated by the
unit vector e be denoted by ae . Then

ae = (a · e) e (3)

4 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Cartesian basis
A Cartesian basis defined by three mutually perpendicular vectors, e1 ,
e2 and e3 , with the following properties:

e1 · e2 = 0, e1 · e3 = 0, e2 · e3 = 0, (4)

e1 × e2 = e3 , e2 × e3 = e1 , e3 × e1 = e2 . (5)

e1 , e2 and e3 are unit vectors. A Cartesian coordinate frame is de-

fined by its origin O together with the right-handed orthonormal basis
{e1 , e2 , e3 }.

5 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Component representation
Any vector a can be uniquely defined with the linear combination of
the basis vectors (e1 , e2 and e3 ) as
a = a1 e1 + a2 e2 + a3 e3 , (6)
where the components (a1 , a2 and a3 ) are real numbers. The compo-
nents of a along the bases are obtained by calculating the projections
a1 = a · e1 , a2 = a · e2 , a3 = a · e3 . (7)
Arranging the components into a 3 × 1 column matrix we arrive at the
matrix representation of vector a as
 
a1
[a] =  a2  . (8)
a3
Obviously, the components of a vector a in other Cartesian basis will
be different numbers.
6 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Index notation I
Consider the component representation of vector a:
3
X
a = a1 e1 + a2 e2 + a3 e3 = ai ei (9)
i=1

In order to abbreviate (or simplify) the expression we can adopt the

Einstein’s summation convention: if and index appears twice in a term,
then a sum must be applied over that index. Consequently, vector a
can be given as

3
X
a= ai ei = ai ei . (10)
i=1

7 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Index notation II

3
X
a= ai ei = ai ei . (11)
i=1

The index used to represent the sum is called dummy index. Replacing
the index i in the above expression does not affect the final result, thus
we can use any symbol:

ai ei = ab eb = aM eM = aβ eβ etc. (12)

Any other index in an equation is a free index.

8 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Kronecker delta symbol

The Kronecker delta symbol can be used to represents the components
of the 3 × 3 identity matrix [I] as


0 if i 6= j
δij = (13)
1 if i = j

Therefore the identity matrix can be written as

   
1 0 0 δ11 δ12 δ13
[I] =  0 1 0  =  δ21 δ22 δ23  . (14)
0 0 1 δ31 δ32 δ33
In addition, the Kronecker delta symbol represents the scalar product
of the orthonormal basis:
ei · ej = δij . (15)
9 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Permutation symbol I
The permutation symbol is also called as alternating symbol or Levi–
Civita symbol. It can be imagined as a symbol which represents 27
numbers (either 0, 1 or −1) depending on the value of the indices:


 1 for even permutation of ijk
ijk = −1 for odd permutation of ijk (16)
0 if there is a repeated index


Consequently

123 = 231 = 312 = 1, (17)

132 = 213 = 321 = −1, (18)

111 = 122 = 113 = ... = 0. (19)

10 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Permutation symbol II
The cross product of the basis vectors can be easily expressed using the
permutation symbol as

ei × ej = ijk ek , (20)

e1 × e2 = 123 e3 = e3 , (21)
e2 × e3 = 231 e1 = e1 , (22)
e3 × e1 = 312 e2 = e2 . (23)

11 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Scalar product
The scalar product of vectors a = ai ei and b = bj ej is calculated as

a · b = (ai ei ) · (bj ej ) = ai bj ei · ej = ai bj δij = ai bi , (24)

ai bi = a1 b1 + a2 b2 + a3 b3 . (25)
Observe the replacement property of δij : If δij appears in a term, where
i (or j) is a dummy index, then it can be changed to j (or i) and δij
can be removed from the term. For example:
ai bj δij = ai bi = aj bj , (26)
σab δbk = σak , (27)
cijk δjr δsk = cirk δsk = cirs . (28)
12 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Cross product I
The cross product of vectors a = ai ei and b = bj ej is calculated as

a × b = (ai ei ) × (bj ej ) = ai bj ei × ej = ai bj ijk ek , (29)

| {z }
ck

where (20) was applied. Using the summation convention it can be

clearly concluded that ai bj ijk is a quantity having only one index,
namely k. We can denote this new quantity with ck for simplicity,
which is nothing else just the component of the new vector resulting
from the cross product.

13 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Cross product II
Therefore

a × b = ai bj ijk ek = ck ek = c, (30)
| {z }
ck

where
c1 = ai bj ij1 = a2 b3 231 + a3 b2 321 = a2 b3 − a3 b2 , (31)
c2 = ai bj ij2 = a3 b1 312 + a1 b3 132 = a3 b1 − a1 b3 , (32)
c3 = ai bj ij3 = a1 b2 123 + a2 b1 213 = a1 b2 − a2 b1 , (33)
   
c1 a2 b3 − a3 b2
[c] =  c2  =  a3 b1 − a1 b3  . (34)
c3 a1 b2 − a2 b1

14 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Cross product III

We get the same result using the classical method to compute the cross
product:
 
 e1 e2 e3 
 
a × b =  a1 a2 a3  , (35)
 b1 b2 b3 

a × b = (a2 b3 − a3 b2 ) e1 + (a3 b1 − a1 b3 ) e2 + (a1 b2 − a2 b1 ) e3 . (36)

15 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Triple scalar product I

Geometrically, the triple scalar product between vectors a, b and c
defines the volume of a paralelepiped spanned by a, b and c forming a
right-handed system. Let

a = ai ei , b = bj e j , c = cm em . (37)

Then

(a × b) · c = (ai bj ijk ek ) · (cm em ) = ai bj ijk cm (ek · em ) , (38)

(a × b) · c = ai bj ijk cm δkm = ai bj ck ijk . (39)

It can be verified that

(a × b) · c = (b × c) · a = (c × a) · b. (40)
16 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Triple scalar product II

The triple scalar product can be calculated using determinant as
   
   a1 b1 c1 
  
(a × b) · c =  [a] , [b] , [c]  =  a2 b2 c2  , (41)
   a3 b3 c3 

(a × b) · c =a1 (b2 c3 − b3 c2 ) + b1 (a3 c2 − a2 c3 ) + c1 (a2 b3 − a3 b2 ) .

(42)

(a × b) · c =a1 bj ck 1jk + b1 ai ck i1k + c1 ai bj ij1 . (43)

17 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Triple vector product

Let
a = aq eq , b = bi e i , c = cj ej . (44)
Then the triple vector product is obtained as
a × (b × c) = (aq eq ) × (bi cj ijk ek ) = aq bi cj ijk (eq × ek ) , (45)

a × (b × c) = aq bi cj ijk qkp ep . (46)

| {z }
dp

Thus, the matrix representation of the resulting vector is

 
aq bi cj ijk qk1
[a × (b × c)] =  aq bi cj ijk qk2  = ... (47)
aq bi cj ijk qk3
18 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Vectors

Epsilon-delta identities
The following useful identities can be easily verified:
δaa = 3, (48)
abc abc = 6, (49)
abm adm = δac δbd − δad δbc , (50)
acd bcd = 2δab . (51)

19 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Definition
A second-order tensor σ can be imagined as a linear operator. Applying
σ on a vector n generates a new vector ρ:

ρ = σn, (52)

thus it defines a linear transformation. In hand-written notes we use

double underline to indicate second-order tensors. Thus, the expression
above can be written as

ρ = σ n. (53)

The second-order identity tensor I and the second order zero tensor 0
have the properties

In = n, 0n = 0. (54)

The projection (3) can be expressed using second-order tensor P: Act-

ing P on a generates a new vector ae .
20 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Representation in a coordinate frame

The second-order tensor σ has nine components in a given coordinate
frame {e1 , e2 , e3 }. The components σij are computed by

σij = ei · (σej ) . (55)

The matrix representation of σ in a given coordinate frame is

 
σ11 σ12 σ13
[σ] =  σ21 σ22 σ23  . (56)
σ31 σ32 σ33

21 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Dyadic product of two vectors

The matrix representation of the dyadic (or tensor or direct) product
of vector a and b is

 
a1 b1 a1 b2 a1 b3
[a ⊗ b] =  a2 b1 a2 b2 a2 b3  , (57)
a3 b1 a3 b2 a3 b3

 
a1 
T 
[a ⊗ b] = [a] [b] =  a2  b1 b2 b3 . (58)
a3
The ij-th component of the resulting second-order tensor is ai bj .
It can be seen that

(a ⊗ b) · c = (b · c) a. (59) 22 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Representation of second-order tensors with dyads

The second-order tensor A can be written as the linear combination of
the dyads formed by the basis vectors:

A = Aij ei ⊗ ej . (60)

Thus, the identity tensor can be written as

I = δij ei ⊗ ej (61)

The dyad a⊗b is a 2nd-order tensor, but not all 2nd-order tensor can
be written as a dyadic product of two vectors! In general, a 2nd-order
tensor has 9 components, whereas a dyad has only 6 components (2×3)

23 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Indical notation I
Consider the equation

a = b + Mc. (62)

Its matrix representation is

[a] = [b] + [M] [c] , (63)

      
a1 b1 M11 M12 M13 c1
 a2  =  b2  +  M21 M22 M23   c2  . (64)
a3 b3 M31 M32 M33 c3
It can be formulated using indical notation as

ai = bi + Mij cj . (65)

24 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Indical notation II
It should be observed that the same free index must appear in
every term of an equation.
The indical notation is an order-independent representation. In
matrix notation the order of the multiplication cannot be
changed, however, in the indical notation (using the summation
convention) the terms can be rearranged without altering the
result. Example: Ab 6= bA, but Aij bj = bj Aij .
“The essence of the Einstein summation notation is to create a
set of notational defaults so that the summation sign and the
range of the subscripts do not need to be written explicitly in each
expression.”
“It is a collection of time-saving conventions. After an initial
investment of time, it converts difficult problems into problems
with workable solutions. It does not make easy problem easier,
however.”

25 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Trace
The trace of the second-order tensor A is

trA = tr [A] = A11 + A22 + A33 = Aii . (66)

It is an invariant quantity.

26 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Determinant
The determinant of the second-order tensor A is

detA = det [A] , (67)

detA = A11 (A22 A33 − A23 A32 ) − A12 (A21 A33 − A23 A31 )
+ A13 (A21 A32 − A22 A31 ) , (68)
detA = ijk A1i A2j A3k , (69)
It is an invariant quantity. A is singular when detA = 0.
Useful relation

det (AB) =detAdetB, detAT =detA (70)

27 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Double contraction
The double contraction (or double-dot product) between 2nd-order ten-
sors A and B is defined as

A : B = Aij Bij = A11 B11 + A12 B12 + ..., (71)

   
A : B = tr AT B = tr BT A = tr ABT = tr BAT .



(72)
Thus, the trace of A can be written as

trA = I : A = (δij ei ⊗ ej ) : (Amn em ⊗ en ) = δij Amn δim δjn ij ,

(73)

trA = δij Aij = Aii = A11 + A22 + A33 . (74) 28 / 58

CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Norm
The norm of the second-order tensor A is calculated as

√
kAk = A : A ≥ 0. (75)

29 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Symmetric and skew-symmetric parts I

The following identity holds for the transpose of A:

(Au) ·v = uAT · v.


(76)

A can be decomposed into the sum of a symmetric and a skew-symmetric

parts as

A = Asymm + Askew , (77)

1
A + AT ,


Asymm = symm (A) = (78)
2
1
A − AT .


Askew = skew (A) = (79)
2

30 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Symmetric and skew-symmetric parts II

Thus

Asymm = ATsymm , Askew = −ATskew . (80)

(Asymm )ab = (Asymm )ba (81)

(Askew )ab = − (Askew )ba (82)
Symmetric part has 6, whereas the skew-symmetric part has 3 inde-
pendent components.

31 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Symmetric and skew-symmetric parts III

A skew-symmetric tensor W behaves like a vector. The following rela-
tion can be easily verified:

Wu = ω × u, (83)

   
0 −ω3 ω2 ω1
[W] =  ω3 0 −ω1  =⇒ [ω] =  ω2  , (84)
−ω2 ω1 0 ω3

1
|ω| = √ kWk . (85)
2

32 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Inverse
The inverse A−1 of A is defined as

AA−1 = A−1 A = I. (86)

A necessary and sufficient condition for the existence of A−1 is that

detA 6= 0.
For invertible tensors A and B:
−1
(AB) = B−1 A−1 , (87)
−1 1 −1
(kA) = A , (88)
k

−1
T
AT = A−1 = A−T , (89)
1
det A−1


= . (90)
detA
33 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Orthogonal tensor
A tensor Q is said to be orthogonal if

QT = Q−1 , (91)

QQT = QT Q = I (92)

proper orthogonal detQ = 1

inproper orthogonal detQ = −1

Proper orthogonal tensors represent rotation, whereas inproper orthog-

onal tensors represent reflection.
34 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Definiteness
For all v 6= 0:

Positive semi-definite v · Av ≥ 0
Positive definite v · Av > 0
Negative semi-definite v · Av ≤ 0
Negative definite v · Av < 0

35 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Change of basis I
Let the bases vectors of two Cartesian coordinate system (having the
same origin) be denoted by

e1 , e2 , e3 and e1 , e
e e2 , e
e3 . (93)

Then, a vector a can be written as

a = ai ei ≡ e
aj e
ej , (94)

where the components ai and e aj are obviously different. Denote Qij

the scalar products between the two bases as

Qij = ei · e
ej . (95)

36 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Change of basis II
Then

e1 = (e1 · e
e e1 ) e1 + (e2 · e
e1 ) e2 + (e3 · e
e1 ) e3 , (96)
e2 = (e1 · e
e e2 ) e1 + (e2 · e
e2 ) e2 + (e3 · e
e2 ) e3 , (97)
e3 = (e1 · e
e e3 ) e1 + (e2 · e
e3 ) e2 + (e3 · e
e3 ) e3 . (98)

Thus

ej = Qij ei
e and ei = Qij e
ej . (99)

Combining (94) and (99) gives

ai ei = e
aj Qij ei and ai Qij e
ej = e
aj e
ej . (100)

Thus

ai = Qij e
aj and aj = ai Qij .
e (101)
37 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Change of basis III

It can be more easily expressed using matrix notation as

T
[a] = [Q] [e
a] and [e
a] = [Q] [a] , (102)

where Q contains the angle cosines as

 
e1 · e
e2 e1 · e
e2 e1 · e
e3
[Q] =  e2 · e
e2 e2 · e
e2 e2 · e
e3  . (103)
e3 · e
e2 e3 · e
e2 e3 · e
e3

For 2nd-order tensor A:

h i h i
e [Q]T
[A] = [Q] A and e = [Q]T [A] [Q] .
A (104)

38 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Deviatoric and spherical parts

Every tensor A can be decomposed into a deviatoric and spherical part
as

A = dev (A) + sph (A) , (105)

where

 
1
sph (A) = pI = trA I, (106)
3
 
1
dev (A) = A − trA I. (107)
3

39 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Eigenvalues, eigenvectors I
An eigenpair of a 2nd-order tensor A mean a scalar λi and unit vector
ni satisfying

Ani = λi ni with i = 1, 2, 3. (108)

λi are the eigenvalues (or principal values), whereas ni denote the

normalized eigenvectors (principal directions, principal axes).

The eigenvalues are the roots of the characteristic cubic equation

det (A − λi I) = 0. (109)

40 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Eigenvalues, eigenvectors II
The eigenvectors are defined by the linear homogeneous equations

(A − λi I) ni = 0. (110)

Eigenvalues of symmetric A are reals.

Eigenvalues of positive definite symmetric A are strictly positive.
Eigenvectors of symmetric A form mutually orthogonal basis.

41 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Spectral decomposition
Any symmetric 2nd-order tensor A can be represented by its eigenval-
ues λi and eigenvectors ni as

3
X 3
X
A= λi (ni ⊗ ni ) = λi mi , (111)
i=1 i=1

where mi = ni ⊗ ni is the basis tensor (or projection tensor).

Matrix representation of A in the coordinate system formed by its
eigenvectors ni is
 
λ1 0 0
[A] =  0 λ2 0  . (112)
0 0 λ3

42 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Principal invariants
The principal scalar invariants of the 2nd-order tensor A are

I1 = trA = λ1 + λ2 + λ3 , (113)

1 2

(trA) − tr A2 = tr A−1 detA = λ1 λ2 +λ1 λ3 +λ2 λ3 ,


I2 =
2
(114)

I3 = detA = λ1 λ2 λ3 . (115)

43 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor algebra
Second-order tensors

Cayley-Hamilton theorem
The Cayley-Hamilton theorem states that the 2nd-order tensor A sat-
isfies its characteristic equation. Thus

A3 − I1 A2 + I2 A − I3 I = 0, (116)

where I1 , I2 and I3 are the principal invariants of A.

44 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Indical notation

Indical notation
Commas in the subscript mean that a partial derivative is to be applied.
The index after the comma represents partial derivatives with respect
to the default arguments, which are usually the coordinates x1 , x2 and
x3 .

∂ui
ui,j = . (117)
∂xj

Example:
∂ui ∂u1 ∂u2 ∂u3
ai ui,j = ai = a1 + a2 + a3 . (118)
∂xj ∂xj ∂xj ∂xj

45 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Nabla operator

Nabla operator
The nabla operator (or del operator or vector-differential operator) is
defined as

∂ (•)
∇ (•) = ei , (119)
∂xi

∂ (•) ∂ (•) ∂ (•)

∇ (•) = e1 + e2 + e3 . (120)
∂x1 ∂x2 ∂x3

46 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Gradient

Gradient of a scalar field

The gradient of the smooth scalar field T (x) (or T (xi ) or T (x1 , x2 , x3 ))
is the vector field

∂T ∂T ∂T ∂T
gradT = ∇T = ei = e1 + e2 + e3 (121)
∂xi ∂x1 ∂x2 ∂x3

Indical notation:
∂T
T,i = (122)
∂xi
The matrix representation of the resulting vector:
 
T,1
[gradT ] =  T,2  (123)
T,3
47 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Gradient

Directional derivative

T (x) = constant denotes level surface

The normal to the surface is gradT
gradT
The unit normal is n = |gradT |
The directional derivative of T at x in the direction of a
normalized vector u is (gradT ) · u
It takes the maximum (minimum) when u = n (u = −n)
The particular directional derivative (gradT ) · n = |gradT | is
called as normal derivative

48 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Gradient

Gradient of a vector field

The gradient of the vector field u (x) is the second-order tensor field

∂u ∂ui
gradu = ∇u = =u⊗∇= ei ⊗ ej (124)
∂x ∂xj

Matrix representation:
 
∂u1 ∂u1 ∂u1  
∂x1 ∂x2 ∂x3 u1,1 u1,2 u1,3
∂u2 ∂u2 ∂u2
[gradu] =   = u2,1 u2,2 u2,3  (125)
  
∂x1 ∂x2 ∂x3
∂u3 ∂u3 ∂u3 u3,1 u2,2 u3,3
∂x1 ∂x2 ∂x3

Transposed gradient:
∂ui
gradT u = ∇ ⊗ u = ej ⊗ ei (126)
∂xj
49 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Gradient

Gradient of a 2nd-order tensor field

The gradient of the second-order tensor field A is the third-order tensor
field

 
∂ (•)
gradA = A ⊗ ∇ = (Aij ei ⊗ ej ) ⊗ ek (127)
∂xk
∂Aij
= ei ⊗ ej ⊗ ek = Aij,k ei ⊗ ej ⊗ ek (128)
∂xk

Thus, the ijk-th component is

(gradA)ijk = Aij,k (129)

50 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Divergence

Divergence of a scalar field

Meaningless

51 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Divergence

Divergence of a vector field

The divergence of the vector field a = ai ei is the scalar field

 
∂ (•) ∂ai ∂ai
diva = ∇ · a = ej · (ai ei ) = δij = = ai,i (130)
∂xj ∂xj ∂xi

diva = tr (grada) (131)

If diva = 0 then a is said to be solenoidal (or divergence-free or incom-

pressible).

52 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Divergence

Divergence of a 2nd-order tensor field

The divergence of the 2nd-order tensor field σ is the vector field

 
∂ (•) ∂σij
divσ = σ · ∇ = (σij ei ⊗ ej ) · ek = δjk ei = σij,j ei
∂xk ∂xk
(132)

Thus, the matrix representation is

  ∂σ11 ∂σ12 ∂σ13

∂x1 + +

σ1j,j ∂x2 ∂x3
 ∂σ ∂σ22 ∂σ23
[divσ] = σ2j,j =  ∂x21 + + (133)
  
1 ∂x2 ∂x3 
σ3j,j ∂σ31 ∂σ32 ∂σ33
∂x1 + ∂x2 + ∂x3

53 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Curl

Curl of a scalar field

Meaningless

54 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Curl

Curl of a vector field

The curl of the vector field a = aj ej is the vector field

 
∂
curla = rota = ∇ × a = ei × (aj ej ) (134)
∂xi
∂aj
= ijk ek = aj,i ijk ek (135)
∂xi

 
a3,2 − a2,3
[curla] =  a1,3 − a3,1  (136)
a2,1 − a1,2
If curla = 0 then the vector field is irrotational (or conservative or
curl-free).
If a can be expressed as a = gradφ, where φ is the potential of a, then
a is irrotational, beacause of the identity curl (gradφ) = 0.
55 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Curl

Curl of a 2nd-order tensor field

The curl of the 2nd-order tensor field A = Aij ei ⊗ ej is the 2nd-order
tensor field

 
∂
curlA = ∇ × A = ek × (Aij ei ⊗ ej ) (137)
∂xk
∂Aij
= kim em ⊗ ej = Aij,k kim em ⊗ ej (138)
∂xk

 
Ai1,k ki1 Ai2,k ki1 Ai3,k ki1
[curlA] =  Ai1,k ki1 Ai2,k ki1 Ai3,k ki1  (139)
Ai1,k ki1 Ai2,k ki1 Ai3,k ki1
∂A31 ∂A21
where Ai1,k ki1 = A31,2 − A21,3 = ∂x2 − ∂x3 for instance.
56 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Laplacian

Laplacian of a scalar field

The Laplacian operator is defined as

   
∂ (•) ∂ (•)
4 (•) = ∇2 (•) = ∇ (•) · ∇ (•) = ei · ej (140)
∂xi ∂xj
∂ 2 (•) ∂ 2 (•) ∂ 2 (•)
4 (•) = ei · ej = δij = (141)
∂xi ∂xj ∂xi ∂xj ∂xi ∂xi
∂ 2 (•) ∂ 2 (•) ∂ 2 (•) ∂ 2 (•)
4 (•) = 2 = + + (142)
∂xi ∂x21 ∂x22 ∂x23

The Laplacian of a scalar field T is the scalar field

∂2T ∂2T ∂2T

4T = + + = T,ii (143)
∂x21 ∂x22 ∂x23

57 / 58
CONTINUUM MECHANICS - Introduction to tensors
Tensor calculus
Laplacian

Laplacian of a vector field

The Laplacian of a vector field u = ui ei is the vector field

∂ 2 (ui ei )
4u = = (ui,j ),j ei = ui,jj ei (144)
∂x2j
 2 2 2 
∂ u1
∂x21
+ ∂∂xu21 + ∂∂xu21
 ∂2u 2 2
∂ 2 u2 ∂ 2 u2 

[4u] =  2 + ∂x2 + ∂x2 (145)
2
 ∂x 1 2 2 
∂ 2 u3 ∂ 2 u3 ∂ 2 u3
∂x 2 + ∂x 2 + ∂x 2
1 2 2

58 / 58