DEPARTAMENTO DE MATEMÁTICA APLICADA

NOTES ON VECTOR CALCULUS

Alfredo Bermúdez de Castro

October 2006

UNIVERSIDADE DE SANTIAGO DE COMPOSTELA

Contents

1 Mathematical background 3
1.1 Vector and tensor algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.1 Vector space. Basis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.2 Inner product. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.3 Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.4 The Affine Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2 Vector and tensor analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.1 Differential operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.2 Curves and curvilinear integrals. . . . . . . . . . . . . . . . . . . . . . . . 13
1.2.3 Gauss’ and Green’s formulas. Stokes’ Theorem. . . . . . . . . . . . . . . . 14
1.2.4 Change of Variable in Integrals . . . . . . . . . . . . . . . . . . . . . . . . 15
1.2.5 The area enclosed by a plane curve . . . . . . . . . . . . . . . . . . . . . . 16
1.2.6 Localization Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.3 Differential Operators in Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.3.1 Cartesian Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.3.2 Cylindrical Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.3.3 Spherical Coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

2 Some Equations of Continuum Mechanics in Coordinates 25

2.1 Mass Conservation Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2 Motion Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3 Constitutive Law for Newtonian Viscous Fluids in Coordinates . . . . . . . . . . 27

1
2
Chapter 1

Mathematical background

1.1 Vector and tensor algebra

1.1.1 Vector space. Basis.

A real vector space is an algebraic structure consisting of a set V endowed with two opera-
tions. One of them is an internal operation denoted by + and satisfying the following properties:

i) Associativity.
a + (b + c) = (a + b) + c ∀ a, b, c.

ii) Existence of a neutral element 0 such that

a + 0 = 0 + a = a ∀ a.

iii) Existence of symmetric element: for each a there exist b such that,

a + b = b + a = 0.

iv) Commutativity.
a + b = b + a ∀ a, b.

Thus, V with the + operation is a commutative group.

The second operation is an external one, the product by real numbers, satisfying the following
properties:

v) λ(a + b) = λa + λb.

vi) λ(µa) = (λµ)a.

vii) (λ + µ)a = λa + µa.

viii) 1a = a.

3
 CHAPTER 1. MATHEMATICAL BACKGROUND
4

Elements of V are called vectors.

A set of vectors {ei }N

i=1 is said

• linearly independent if
N
X
λi ei = 0 ⇒ λi = 0 ∀i ∈ {1, . . . , N }, (1.1)
i=1

• a system of generators if
N
X
∀v ∈ V ∃λi , i = 1, . . . , N, such that v = λi ei , (1.2)
i=1

• a basis if it is both linearly independent and a system of generators.

If B = {e1 , . . . , eN } is a basis of V, then each v ∈ V can be written as a linear combination,

N
X
v= vi ei ,
i=1

in a unique way. The vi ∈ R, i = 1, . . . , N, are called coordinates of v with respect to basis B.

Let us consider two bases in V, {ei }N N
i=1 and {Ei }i=1 . In particular, vectors Ei can be written
as a linear combination of vectors {ei }N i=1 , namely,

N
X
El = Cli ei . (1.3)
i=1

Now, let v be any vector in V. Let us denote by vie and viE , i = 1 . . . N , the coordinates of v with
respect to the basis {ei }N N e
i=1 and {Ei }i=1 , respectively. If we denote by v the column vector of
E
the former and by v the column vector of the latter, it is easy to see that

v e = [C]t v E , (1.4)

where [C] is the matrix [C]ij = Cij with Cij given in (1.3).

1.1.2 Inner product.

A mapping ϕ : V × V −→ R is called inner product if the following properties hold:

1) ϕ is bilinear :

1.1) ϕ(λ1 v1 + λ2 v2 , w) = λ1 ϕ(v1 , w) + λ2 ϕ(v2 , w),

1.2) ϕ(v, λ1 w1 + λ2 w2 ) = λ1 ϕ(v, w1 ) + λ2 ϕ(v, w2 ).
 1.1. VECTOR AND TENSOR ALGEBRA
5

2) ϕ is symmetric:

ϕ(v, w) = ϕ(w, v).

3) ϕ is positive definite:

ϕ(v, v) > 0 ∀ v 6= 0.

Finite dimensional vector spaces with inner product are called Euclidean vector spaces.
For them we can define a norm by,
1
|v| = ϕ(v, v) 2 .

In what follows we write v · w instead of ϕ(v, w).

Two vectors v and w are said to be orthogonal if v · w = 0.
A basis B = {e1 , . . . , eN } is said to be orthogonal if ei · ej = 0 ∀ i 6= j and is said to be
orthonormal if
½
1 if i = j,
ei · ej = δij =
0 if i =
6 j.

If B is orthonormal, then the i-th coordinate of a vector v is given by,

vi = v · ei .

Moreover, the matrix to change coordinates with respect to two orthonormal basis is orthog-
onal, namely,
[C][C]t = [I], (1.5)
where [I] is the identity matrix.

1.1.3 Tensors

The endomorphisms of V, that is the linear mappings from V in V are called (second order)
tensors. We denote by Lin the set of tensors on a vector space V. In Lin we can define a vector
space structure by introducing the following operations:

i) Sum:
(S + T)(v) = Sv + Tv.

ii) Product by scalars:

(λS)(v) = λSv.
 CHAPTER 1. MATHEMATICAL BACKGROUND
6

The dimension of this vector space is N 2 .

Let a and b two vectors in V. The tensor,

(a ⊗ b)v := v · b a,
is called the tensor product of a and b. We notice that ⊗ is a bilinear mapping from V × V in
Lin.
If e is a unit vector then
(e ⊗ e) v = (v · e) e,
is the orthogonal projection of v on the straight line generated by e. Moreover,

(I − e ⊗ e)v = v − (v · e) e, (1.6)

is the projection of v on the orthogonal plane to e.

Let B = {e1 , . . . , eN } an orthonormal basis in V. The set of tensors {ei ⊗ej : 1 ≤ i, j ≤ N }
is a basis of Lin. More precisely, we have
N
X
S= Sij ei ⊗ ej ,
i,j=1

where

Sij = S ej · ei
are the coordinates of S with respect to basis B.
We notice that the coordinates of a tensor S can be arranged as a matrix,
 
S11 . . . S1N
 .. .. .. 
[S] =  . . . .
SN 1 . . . SN N
Actually, this matrix is the standard matrix associated with the endomorphism S for the basis
B.
For example, the coordinate matrix of tensor a ⊗ b with respect to basis B is the matrix
product  
a1
 a2 
 
 ..  (b1 b2 . . . bN ), (1.7)
 . 
aN
N
X N
X
where a = ai ei and b = bi ei .
i=1 i=1
Let us see how tensor coordinates change when we change basis in the vector space. Let
S ∈ Lin a tensor. Let us denote by Seij and SE ij , i, j = 1 . . . n, its coordinates with respect to basis
{ei }N
i=1 and {E } N
i i=1 , respectively. Then it is not difficult to see that the following relation holds:

[SE ] = [C][Se ][C]t . (1.8)

 1.1. VECTOR AND TENSOR ALGEBRA
7

In vector space Lin we can introduce another internal operation which is the mapping com-
position, namely, given two tensors S and T, ST is the tensor defined by

(ST)v = S(Tv).
It is easy to see that the matrix of coordinates of ST is the product of those corresponding to S
and T.
The transpose tensor of S is the unique tensor St satisfying,

Sa · b = a · St b ∀ a, b ∈ V.

A tensor is called symmetric if S = St and skew if S = −St . The subspace of symmetric

tensors will be denoted by Sym and that of skew tensors by Skw.

We can also define an inner product of tensors by,

S · T = tr(St T),

where tr denotes the trace operator which is the unique linear operator from Lin to R satisfying,

tr(a ⊗ b) = a · b.

We have,
 
N
X N
X N
X N
X
tr(S) = tr  Sij ei ⊗ ej  = Sij ei · ej = Sij δij = Sii ,
i,j=1 i,j=1 i,j=1 i=1

and hence
N
X
S·T= Sij Tij .
i,j=1

If T ∈ Sym and W ∈ Skw it is easy to show that T · W = 0.

Any tensor S is the sum of one symmetric tensor E and another one skew W. More precisely,
E and W are defined by
1¡ ¢
E= S + St , (1.9)
2
1¡ ¢
W= S − St . (1.10)
2
Tensor E is called the symmetric part of of S and W is called the skew part of S. Furthermore,
this decomposition is unique, more precisely,
⊥
M
Lin = Sym Skw

The following equalities can be easily proved:

 CHAPTER 1. MATHEMATICAL BACKGROUND
8

•
(a ⊗ b)t = (b ⊗ a) (1.11)

•
(a ⊗ b)(c ⊗ d) = (b · c)a ⊗ d (1.12)

•
R · (ST) = (St R) · T = (RTt ) · S ∀R, S, T ∈ Lin (1.13)

•
I · S = tr(S) (1.14)

•
S · (a ⊗ b) = a · Sb (1.15)

•
(a ⊗ b) · (c ⊗ d) = (a · c)(b · d) (1.16)

A tensor S is invertible if there exists a tensor S−1 , called the inverse of S, such that

SS−1 = S−1 S = I. (1.17)

A tensor Q is orthogonal if it preserves the inner products:

Qu · Qv = u · v. (1.18)

A necessary and sufficient condition for Q to be orthogonal is that

QQt = Qt Q = I, (1.19)

which is equivalent to
Qt = Q−1 . (1.20)
In particular, det Q must be 1 or −1, where det Q := det[Q].
A rotation is an orthogonal tensor with positive determinant. The set of rotations is a subset
of Lin which will be denoted by Orth+ .
A tensor S is positive definite if

v · Sv > 0 ∀v 6= 0. (1.21)

A real scalar λ is an eigenvalue of S if there exists a non-null vector u such that

Su = λu. (1.22)

In this case u is an eigenvector associated to λ. The set of vectors

Eλ = {u ∈ V : Su = λu} (1.23)

is a vector subspace of V called characteristic space or eigenspace of S corresponding to λ.

If this space has dimension M ≤ N , then M is called the geometric multiplicity of λ. The
spectrum of S is the set of its eigenvalues.
 1.1. VECTOR AND TENSOR ALGEBRA
9

It is immediate to see that λ ∈ R is an eigenvalue of tensor S if and only if it satisfies the

characteristic equation
det(S − λI) = 0. (1.24)
The multiplicity of λ as a root of this algebraic equation is called the algebraic multiplicity
of λ. The algebraic multiplicity of any eigenvalue is greater than or equal to the geometric one.
Now, let us assume that N = 3 for the rest of this paragraph. The principal invariants of
S are the three numbers

ı1 (S) = tr(S), (1.25)

1
ı2 (S) = [(tr(S))2 − tr(S2 )], (1.26)
2
1
ı3 (S) = det(S) = [(tr(S))3 + 2(tr(S3 )) − 3tr(S2 )tr(S)]. (1.27)
6
The set of principal invariants of S will be denoted by IS . Moreover, the characteristic equation
can be written as
det(S − λI) = −λ3 + ı1 (S)λ2 − ı2 (S)λ + ı3 (S). (1.28)
We have the following important results:

Proposition 1.1.1 1. The eigenvalues of a positive definite tensor are strictly positive.

2. Each symmetric tensor has exactly three eigenvalues accounting for their multiplicities as
roots of the characteristic equation (1.24) (called algebraic multiplicities).

3. The characteristic spaces of a symmetric tensor are mutually orthogonal.

Theorem 1.1.1 (Spectral theorem) Let S be a symmetric tensor. Then there is an orthonormal
basis in V consisting of eigenvectors of S. Furthermore:

1. S has three distinct eigenvalues if and only it has three eigenspaces which are orthogonal
lines through 0.

2. S has two distinct eigenvalues if and only if it has two eigenspaces which are a line through
0 and its orthogonal plane through 0.

3. S has one distinct eigenvalue if and only if it has one eigenspace (which is equal to V).

Moreover, when S is symmetric all the roots of the characteristic equation are real and the Vieta
relations lead to the following equalities:

ı1 (S) = λ1 + λ2 + λ3 , (1.29)
ı2 (S) = λ1 λ2 + λ2 λ3 + λ1 λ3 , (1.30)
ı3 (S) = λ1 λ2 λ3 . (1.31)

Let us assume that {e1 , e2 , e3 } is a positively oriented orthonormal basis in V. We define

the vector product of two vectors a and b by,
 CHAPTER 1. MATHEMATICAL BACKGROUND
10

3
X
a×b= Eijk aj bk ei .
i,j,k=1

where

 1 if ijk is an even permutation of 123,
Eijk = −1 if ijk is an odd permutation of 123,

0 otherwise .

One can prove the following equality:

a × (b × c) = (a · c)b − (a · b)c (1.32)

showing, in particular, that the cross product is not associative.

There is an isomorphism between vector space V and the linear space of skew tensors. It is
defined by,

V −→ Skw
w −→ W
with Wv = w × v ∀v ∈ V. Vector w is called the axial vector associated to W.

1.1.4 The Affine Space

Let V be a vector space. A N -dimensional affine space is a triple (E, V, +) consisting of a

set E, a N -dimensional vector space V and a mapping

+
E × V −→ E
(p, v) −→ p + v
satisfying the following property:
For any p, q ∈ E there exists a unique v ∈ V such that p + v = q.
This vector v is usually denoted by −
→ or q − p.
pq
If V is an Euclidean vector space then the corresponding affine space is called Euclidean
affine space.
In an Euclidean affine space we can define a distance by,

d
E × E −→ R
(p, q) −→ d(p, q) = |−
→
pq|.

Thus (E, d) is a metric space and hence a topological space.

In an Euclidean affine space, a cartesian coordinate frame is a couple (o, {ei , i = 1, . . . , N })

where o is any fixed point in E called the origin and {ei , i = 1, . . . , N } is an orthonormal basis
 1.2. VECTOR AND TENSOR ANALYSIS
11

of the Euclidean vector space V.

For any point p ∈ E its coordinates with respect to the above cartesian frame are the numbers
pi , i = 1, . . . , N such that
N
X
−
→
op = pi ei .
i=1

1.2 Vector and tensor analysis

A field is any mapping defined in a subset of E and valued in R, V or Lin, in which case it
is called scalar, vector or tensor field, respectively.

1.2.1 Differential operators

Let W be a normed vector space (in practice W = R, V, Lin). A mapping f : R ⊂ E → W

defined in an open set R of E is said differentiable at point x ∈ R if there exists a linear
mapping X : V −→ W such that,

f(x + h) − f(x) − X (h) = o(h) as h → 0,

which means

kf(x + h) − f(x) − X (h)kW

lim = 0.
h→0 khkV

If f is differentiable at x then linear mapping X is unique and it is called the differential of

f at x. Usually, it is denoted by Df(x).

The above definition can be extended to affine space-valued mappings. In this case the dif-
ferential is a linear mapping from V into itself satisfying

kf(x + h) − f(x) − X (h)kV

lim = 0.
h→0 khkV

We notice that the expression in the numerator makes sense because f(x + h) − f(x) ∈ V.
Examples:

1) Let ϕ be a scalar field which is differentiable at x. Then

Dϕ(x) : V −→ R.

and the unique vector ∇ϕ(x) ∈ V such that

Dϕ(x)a = ∇ϕ(x) · a ∀a ∈ V,
 CHAPTER 1. MATHEMATICAL BACKGROUND
12

is called the gradient of ϕ at x.

If ϕ is differentiable at all points in R we can define the vector field,

R⊂E −→ V
x −→ ∇ϕ(x)

which is called the gradient of ϕ.

2) Let us consider the case where W = V, i.e., u is a vector field. Then Du(x) : V −→ V, that
is, Du(x) ∈ Lin.

If u is differentiable at all points in R, we can define the tensor field,

R⊂E −→ Lin
x −→ Du(x)

which is called the gradient of vector field u.

We will use the notation ∇u or gradu instead of Du, to denote this differential operator.

Given a differentiable vector field u, the scalar field

div u : R ⊂ E −→ R
x −→ div u(x) =: tr(∇u(x))
is called the divergence of u.

Now, let S : R ⊂ E −→ Lin be a tensor field. We define the vector field div S, called
divergence of S as follows:

For any x ∈ R, div S(x) is the unique vector such that,

div S(x) · a = div(St a)(x) ∀ a ∈ V.

For a scalar field ϕ : R ⊂ E −→ R, its Laplacian is the scalar field

∆ϕ(x) = div(∇ϕ)(x).

For a vector field u : R ⊂ E −→ V, its Laplacian is the vector field

∆u(x) = div(∇u)(x).

From now on we assume that N = 3. Let u : R ⊂ E −→ V be a smooth vector field. The

curl of u at x is the unique vector, curl u(x), such that,
 1.2. VECTOR AND TENSOR ANALYSIS
13

curl u(x) × a = (∇u(x) − ∇ut (x))a ∀ a ∈ V.

The following equalities hold:

∇(ϕv) = ϕ∇v + v ⊗ ∇ϕ,

div(ϕv) = ϕ div v + v · ∇ϕ,
∇(v · w) = (∇w)t v + (∇v)t w,
div(v ⊗ w) = v div w + (∇v)w,
div(St v) = S · ∇v + v · div S,
div(ϕS) = ϕ div S + S∇ϕ,
div(v × w) = w · curl v − v · curl w,
div ∇vt = ∇(div v)
curl(ϕv) = ∇ϕ × v + ϕ curl v
curl(v × w) = ∇vw − w div v + v div w − ∇wv,
div curl v = 0,
curl ∇ϕ = 0,
∆v = ∇(div v) − curl(curl v),
∆(∇ϕ) = ∇(∆ϕ),
div(∆v) = ∆(div v),
curl(∆v) = ∆(curl v),
∆(ϕψ) = ϕ∆ψ + ψ∆ϕ + 2∇ϕ · ∇ψ.

1.2.2 Curves and curvilinear integrals.

A (regular) curve c in R is a class C1 mapping,

c : [0, 1] −→ R,

such that ċ(σ) 6= 0 ∀σ ∈ [0, 1], where the derivative ċ(σ) ∈ V is given by

c(σ + h) − c(σ)
ċ(σ) = lim .
h→0 h

The curve is closed if c(0) = c(1).

The length of c is the number
Z 1
length(c) = |ċ(σ)| dσ.
0
 CHAPTER 1. MATHEMATICAL BACKGROUND
14

Let v : R ⊂ E −→ V be a continuous vector field. The curvilinear integral of v along c is

defined by
Z Z 1
v(x) · dx = v(c(σ)) · ċ(σ) dσ.
c 0

Similarly, let S : R ⊂ E −→ Lin be a continuous tensor field. The curvilinear integral of

S along c is the vector defined by
Z Z 1
S(x)dx = S(c(σ))ċ(σ) dσ.
c 0

If v is the gradient of a scalar field ϕ, then

Z Z Z 1 Z 1
d
v(x) · dx = ∇ϕ(x) · dx = ∇ϕ(c(σ)) · ċ(σ)dσ = (ϕ ◦ c)(σ) dσ = ϕ(c(1)) − ϕ(c(0)).
c c 0 0 dσ
R
In particular c ∇ϕ(x) · dx = 0 whenever c is closed.
A subset R ⊂ E is simply connected if any closed curve in R can be continuously deformed
(more precisely, is homotopic) to a point without leaving R.
An open region (or domain) is any open connected subset of E. The closure of an open
region will be called closed region. We call regular region all closed region with smooth
boundary ∂R.
Let R be a closed region and Φ a field defined in R. We say that Φ is C1 (R) if it is continuously
differentiable in the interior of R and Φ and ∇Φ have continuous extensions to all of R.
We end this section on vector calculus by recalling some fundamental results.
Potential Theorem. Let v be a smooth vector field on an open or closed simply connected
region R and assume that,
curl v = 0.
Then there is a C2 scalar field ϕ : R −→ R such that,

v = ∇φ.

1.2.3 Gauss’ and Green’s formulas. Stokes’ Theorem.

Divergence Theorem. Let R be a bounded regular region and let ϕ : R −→ R, v : R −→ V

and S : R −→ Lin be smooth fields. Then,
R R
1. ∂R ϕndA = R ∇ϕ dV,
R R
2. ∂R v ⊗ ndA = R ∇v dV,
R R
3. ∂R v · ndA = R div v dV,
 1.2. VECTOR AND TENSOR ANALYSIS
15

R R
4. ∂R n
× vdA = R curl v dV,
R R
5. ∂R SndA = R div S dV,

where ∂R denotes the boundary of R and n is an outward unit normal vector to ∂R.

Green’s formulas. Let R be a bounded regular region and let ϕ, ψ : R −→ R, v, w :

R −→ V and S : R −→ Lin be smooth fields. Then,
Z Z Z
∂ϕ
1. ψ dA = ∇ϕ · ∇ψ dV + ∆ϕ ψ dV .
∂R ∂n R R
Z Z Z
2. Sn · v dA = S · ∇v dV + v · div S dV .
∂R R R
Z Z Z
t
3. (Sn) ⊗ v dA = S∇v dV + div S ⊗ v dV .
∂R R R
Z Z Z
4. v · nψ dA = div v ψ dV + v · ∇ψ dV .
∂R R R
Z Z Z
5. w · nv dA = div w v dV + ∇vw dV .
∂R R R
Z Z Z
6. v × w · n dA = w · curl v dV − v · curl w dV.
∂R R R

Stokes’ Theorem. Let v be a smooth vector field on an open set R ⊂ E. Let S be a smooth
surface in R bounded by c supposed to be a closed curve. At each point of S we choose a unit
normal vector n consistent with the direction of traversing c in that it causes a right-handed screw
to advance along n. Then,
Z Z
curl v · n dA = v · dx.
S c

1.2.4 Change of Variable in Integrals

Theorem 1.2.1 Let f be an injective smooth mapping, f : B ⊂ E → E, where B is an open

subset of the affine space E, and ϕ : f (B) → R a smooth scalar field. Let c (resp. S) be a smooth
curve (resp. surface) in B and P an open subset of B. Then we have
R R
1. f (c) ϕt dlx = c ϕ ◦ f F k dlp ,
R R
2. f (S) ϕn dAx = S ϕ ◦ f detF F−t m dAp ,
R R
3. f (c) ϕ dlx = cϕ ◦ f kFkk dlp ,
R R
4. f (S) ϕ dAx = S ϕ ◦ f detFkF−t mk dAp ,
 CHAPTER 1. MATHEMATICAL BACKGROUND
16

R R
5. f (P) ϕ dVx = P ϕ ◦ f detF dVp ,

where F denotes the gradient of f , k (resp. t) denotes a unit tangent vector to curve c (resp.
f (c)) and m (resp. n) denotes a unit normal vector to surface S (resp. f (S)).

1.2.5 The area enclosed by a plane curve

Let us consider a plane closed curve c which intersect itself at most at a finite number of
points. Let c(s), s ∈ [a, b], be a parametrization of c. We denote by R the open region enclosed
by this curve and by Ri , i = 1, · · · , N, its connected components. Each of them has a boundary
∂Ri which consists of pieces of c which are walked in the same sense. To each ∂Ri we associate
a sign si which is +1 if it is walked counter-clockwise, and −1 otherwise. Then the net area of
R is defined as
N
X
net area(R) = si area(Ri ).
i=1
The following formulas for the net area of R hold:
Proposition 1.2.1 We have
Z b Z b
dc2 dc1
net area(R) = c1 (s) (s) ds = − c2 (s) (s) ds. (1.33)
a ds a ds
Proof.- It is an immediate consequence of the Stokes Theorem (in fact, for plane curves it is the
Green’s Theorem). For a smooth plane vector field w defined in R it says that
Z Z
curl w · si e3 dA = w · dx. (1.34)
Ri ∂Ri

Let us take
w(x) = −x2 e1 + x1 e2 . (1.35)
Then curl w = 2e3 and (1.34) yields
Z
2 si area(Ri ) = w · dx, i = 1, · · · , N. (1.36)
∂Ri

Adding the previous equalities we obtain

N Z Z Z bµ ¶
1X 1 1 dc1 dc2
net area(R) = w · dx = w · dx = −c2 (s) (s) + c1 (s) (s) ds.
2 ∂Ri 2 c 2 a ds ds
i=1
(1.37)
Moreover, by integration by parts
Z b Z b Z b
dc1 dc2 dc2
−c2 (s) (s) ds = − (c1 (b)c2 (b) − c1 (a)c2 (a)) + c1 (s) (s) ds = c1 (s) (s) ds.
a ds a ds a ds
(1.38)
because c is closed.
The result follows immediately by using (1.38) in (1.37).
 1.3. DIFFERENTIAL OPERATORS IN COORDINATES
17

1.2.6 Localization Theorem

Theorem 1.2.2 Let Φ be a continuous scalar mapping defined in an open set R ⊂ E. If

Z
Φ dV = 0,
B

for every closed ball B ⊂ R, then Φ = 0 in R. Moreover, if Φ is scalar and

Z
Φ dV ≥ 0,
B

for every closed ball B ⊂ R, then Φ ≥ 0 in R.

1.3 Differential Operators in Coordinates

1.3.1 Cartesian Coordinates

Definition
f : Ω = R3 → R3 (1.39)
f (u1 , u2 , u3 ) = (u1 , u2 , u3 ) (1.40)
Basis
• Contravariant
g1 = e1 , g2 = e2 , g3 = e3 . (1.41)

• Covariant
g1 = e1 , g2 = e2 , g3 = e3 . (1.42)

• Physical
e1 , e2 , e3 . (1.43)

Christoffel symbols of the second kind

 
0 0 0
1
Γjk =  0 0 0 , (1.44)
0 0 0
 
0 0 0
2
Γjk =  0 0 0 , (1.45)
0 0 0
 
0 0 0
Γ3jk =  0 0 0 . (1.46)
0 0 0
Differential Operators in the Physical Basis
 CHAPTER 1. MATHEMATICAL BACKGROUND
18

• Gradient of scalar field

∂ϕ ∂ϕ ∂ϕ
∇ϕ = e1 + e2 + e3 , (1.47)
∂x1 ∂x2 ∂x3

• Gradient of a vector field

∂w1 ∂w1 ∂w1

∇w = e1 ⊗ e1 + e1 ⊗ e2 + e1 ⊗ e3
∂x1 ∂x2 ∂x3
∂w2 ∂w2 ∂w2
+ e2 ⊗ e1 + e2 ⊗ e2 + e2 ⊗ e3
∂x1 ∂x2 ∂x3
∂w3 ∂w3 ∂w3
+ e3 ⊗ e1 + e3 ⊗ e2 + e3 ⊗ e3 . (1.48)
∂x1 ∂x2 ∂x3

• Convective term
½ ¾
∂w1 ∂w1 ∂w1
(∇w)w = w1 + w2 + w3 e1
∂x1 ∂x2 ∂x3
½ ¾
∂w2 ∂w2 ∂w2
+ w1 + w2 + w3 e2
∂x1 ∂x2 ∂x3
½ ¾
∂w3 ∂w3 ∂w3
+ w1 + w2 + w3 e3 . (1.49)
∂x1 ∂x2 ∂x3

• Curl of a vector field

µ ¶
∂w3 ∂w2
curl w = − e1
∂x2 ∂x3
µ ¶
∂w1 ∂w3
+ − e2
∂x3 ∂x1
µ ¶
∂w2 ∂w1
+ − e3 . (1.50)
∂x1 ∂x2

• Divergence of a vector field

∂w1 ∂w2 ∂w3

div w = + + (1.51)
∂x1 ∂x2 ∂x3

• Divergence of a tensor field

· ¸
∂S11 ∂S12 ∂S13
div S = + + e1
∂x1 ∂x2 ∂x3
· ¸
∂S21 ∂S22 ∂S23
+ + + e2
∂x1 ∂x2 ∂x3
· ¸
∂S31 ∂S32 ∂S33
+ + + e3 . (1.52)
∂x1 ∂x2 ∂x3
 1.3. DIFFERENTIAL OPERATORS IN COORDINATES
19

• Laplacian of a scalar field

∂2ϕ ∂2ϕ ∂2ϕ

∆ϕ = + + . (1.53)
∂x21 ∂x22 ∂x23

• Laplacian of a vector field

∆w = ∆w1 e1 + ∆w2 e2 + ∆w3 e3 . (1.54)

Measure elements for integration

• Line

dx1 , dx2 , dx3 . (1.55)

• Surface

dx2 dx3 , dx1 dx3 , dx1 dx2 . (1.56)

• Volume

dx1 dx2 dx3 . (1.57)

1.3.2 Cylindrical Coordinates

Definition

f : Ω = (0, ∞) × (0, 2π) × (−∞, ∞) → R3 (1.58)

f (r, φ, z) = (r cos φ, r sin φ, z) (1.59)

Basis

• Contravariant 
 g1 (r, φ, z) = cos φe1 + sin φe2 ,
g2 (r, φ, z) = −r sin φe1 + r cos φe2 , (1.60)

g3 (r, φ, z) = e3 .

• Covariant  1
 g (r, φ, z) = cos φe1 + sin φe2 ,
g2 (r, φ, z) = − 1r sin φe1 + 1r cos φe2 , (1.61)
 3
g (r, φ, z) = e3 .

• Physical 
 er (r, φ, z) = cos φe1 + sin φe2 ,
e (r, φ, z) = − sin φe1 + cos φe2 , (1.62)
 φ
ez (r, φ, z) = e3 .
 CHAPTER 1. MATHEMATICAL BACKGROUND
20

x3

r
.x
z
x32

x1

Figure 1.1: Cylindrical coordinates.

Christoffel symbols of the second kind

 
0 0 0
Γ1jk =  0 −r 0  , (1.63)
0 0 0

 1

0 r 0
Γ2jk =  1
r 0 0 , (1.64)
0 0 0

 
0 0 0
Γ3jk =  0 0 0 . (1.65)
0 0 0

Differential operators in the physical basis

• Gradient of scalar field

∂ϕ 1 ∂ϕ ∂ϕ
∇ϕ = er + eφ + ez . (1.66)
∂r r ∂φ ∂z

• Gradient of a vector field

· ¸
∂wr 1 ∂wr 1 ∂wr
∇w = er ⊗ er + − wφ er ⊗ eφ + er ⊗ ez
∂r r ∂φ r ∂z
· ¸
∂wφ 1 ∂wφ 1 ∂wφ
+ eφ ⊗ er + + wr eφ ⊗ eφ + eφ ⊗ ez
∂r r ∂φ r ∂z
∂wz 1 ∂wz ∂wz
+ ez ⊗ er + ez ⊗ eφ + ez ⊗ ez . (1.67)
∂r r ∂φ ∂z
 1.3. DIFFERENTIAL OPERATORS IN COORDINATES
21

• Convective term
½ ¾
∂wr 1 ∂wr ∂wr (wφ )2
(∇w)w = wr + wφ + wz − er
∂r r ∂φ ∂z r
½ ¾
∂wφ 1 ∂wφ ∂wφ wr wφ
+ wr + wφ + wz + eφ
∂r r ∂φ ∂z r
½ ¾
∂wz 1 ∂wz ∂wz
+ wr + wφ + wz ez . (1.68)
∂r r ∂φ ∂z

• Curl of a vector field

µ ¶
1 ∂wz ∂wφ
curl w = − er
r ∂φ ∂z
µ ¶
∂wz ∂wr
+ − + eφ
∂r ∂z
µ ¶
1 ∂ 1 ∂wr
+ (rwφ ) − ez . (1.69)
r ∂r r ∂φ

• Divergence of a vector field

1 ∂ 1 ∂wφ ∂wz
div w = (rwr ) + + . (1.70)
r ∂r r ∂φ ∂z
• Divergence of a tensor field
· ¸
1 ∂ ∂Srφ ∂Srz
div S = (rSrr ) + +r − Sφφ er
r ∂r ∂φ ∂z
· ¸
∂Sφr 1 ∂Sφφ ∂Sφz 1 1
+ + + + Srφ + Sφr eφ
∂r r ∂φ ∂z r r
· ¸
1 ∂ ∂Szφ ∂Szz
+ (rSzr ) + +r ez . (1.71)
r ∂r ∂φ ∂z

• Laplacian of a scalar field

· µ ¶ µ ¶ µ ¶¸
1 ∂ ∂ϕ ∂ 1 ∂ϕ ∂ ∂ϕ
∆ϕ = r + + r
r ∂r ∂r ∂φ r ∂φ ∂z ∂z
µ ¶ 2 2
1 ∂ ∂ϕ 1 ∂ ϕ ∂ ϕ
= r + 2 2 + 2. (1.72)
r ∂r ∂r r ∂φ ∂z

• Laplacian of a vector field

· ¸ · ¸
1 2 ∂wφ 1 2 ∂wr
∆w = ∆wr − 2 wr − 2 er + ∆wφ − 2 wφ + 2 eφ + ∆wz ez .
r r ∂φ r r ∂φ
(1.73)

Measure elements for integration

 CHAPTER 1. MATHEMATICAL BACKGROUND
22

• Line
dr, r dφ, dz. (1.74)

• Surface
r dφdz, drdz, r drdφ. (1.75)

• Volume
r drdφdz. (1.76)

1.3.3 Spherical Coordinates

Definition
f : Ω = (0, ∞) × (0, π) × (0, 2π) → R3
(1.77)
f (r, θ, φ) = (r sin θ cos φ, r sin θ sin φ, r cos θ)

x3

.
r
x32
.
x1

Figure 1.2: Spherical coordinates.

Basis
• Contravariant

 g1 (r, θ, φ) = sin θ cos φe1 + sin θ sin φe2 + cos θe3 ,
g2 (r, θ, φ) = r cos θ cos φe1 + r cos θ sin φe2 − r sin θe3 , (1.78)

g3 (r, θ, φ) = −r sin θ sin φe1 + r sin θ cos φe2 .

• Covariant  1
 g (r, θ, φ) = sin θ cos φe1 + sin θ sin φe2 + cos θe3 ,
g2 (r, θ, φ) = 1r cos θ cos φe1 + 1r cos θ sin φe2 − 1r sin θe3 , (1.79)
 3 1 1
g (r, θ, φ) = − r sin θ sin φe1 + r sin θ cos φe2 .

• Physical 
 er (r, θ, φ) = sin θ cos φe1 + sin θ sin φe2 + cos θe3 ,
e (r, θ, φ) = cos θ cos φe1 + cos θ sin φe2 − sin θe3 , (1.80)
 θ
eφ (r, θ, φ) = − sin φe1 + cos φe2 .
 1.3. DIFFERENTIAL OPERATORS IN COORDINATES
23

Christoffel symbols of the second kind

 
0 0 0
Γ1jk =  0 −r 0 , (1.81)
2
0 0 −r sin θ
 
0 1r 0
Γ2jk =  1r 0 0 , (1.82)
1
0 0 − 2 sin 2θ
 1

0 0 r
Γ3jk =  0 0 cot θ  . (1.83)
1
r cot θ 0
Differential operators
• Gradient of a scalar field
∂ϕ 1 ∂ϕ 2 ∂ϕ 3 ∂ϕ 1 ∂ϕ 1 ∂ϕ
∇ϕ = g + g + g = er + eθ + eφ , (1.84)
∂r ∂θ ∂φ ∂r r ∂θ r sin θ ∂φ

• Gradient of a vector field

· ¸ · ¸
∂wr 1 ∂wr 1 1 ∂wr 1
∇w = er ⊗ er + − wθ er ⊗ eθ + − wφ er ⊗ eφ
∂r r ∂θ r r sin θ ∂φ r
· ¸ · ¸
∂wθ 1 ∂wθ 1 1 ∂wθ 1
+ eθ ⊗ er + + wr eθ ⊗ eθ + − cot θ wφ eθ ⊗ eφ
∂r r ∂θ r r sin θ ∂φ r
· ¸
∂wφ 1 ∂wφ 1 ∂wφ 1 1
+ eφ ⊗ er + eφ ⊗ eθ + + wr + cot θwθ eφ ⊗ eφ .
∂r r ∂θ r sin θ ∂φ r r
(1.85)

• Convective term
( )
∂wr 1 ∂wr wφ ∂wr wθ2 + wφ2
(∇w)w = wr + wθ + − er
∂r r ∂θ r sin θ ∂φ r
( )
∂wθ 1 ∂wθ wφ ∂wθ wr wθ wφ2 cot θ
+ wr + wθ + + − eθ
∂r r ∂θ r sin θ ∂φ r r
½ ¾
∂wφ 1 ∂wφ wφ ∂wφ wr wφ wθ wφ cot θ
+ wr + wθ + + + eφ . (1.86)
∂r r ∂θ r sin θ ∂φ r r

• Curl of a vector field

µ ¶
1 ∂ ∂wθ
curl w = (wφ sin θ) − er
r sin θ ∂θ ∂φ
µ ¶
1 ∂wr 1 ∂
+ − (rwφ ) eθ
r sin θ ∂φ r ∂r
µ ¶
1 ∂ ∂wr
+ (rwθ ) − eφ . (1.87)
r ∂r ∂θ
 CHAPTER 1. MATHEMATICAL BACKGROUND
24

• Divergence of a vector field

1 ∂ 2 1 ∂ 1 ∂wφ
div w = 2
(r wr ) + (sin θwθ ) + . (1.88)
r ∂r r sin θ ∂θ r sin θ ∂φ

• Divergence of a tensor field

· ¸
∂Srr 1 ∂Srθ 1 ∂Sφr 1
div S = + + + (2Srr − Sθθ − Sφφ + Srθ cot θ) er
∂r r ∂θ r sin θ ∂φ r
· ¸
∂Srθ 1 ∂Sθθ 1 ∂Sθφ 1
+ + + + ((Sθθ − Sφφ ) cot θ + 3Srθ ) eθ
∂r r ∂θ r sin θ ∂φ r
· ¸
∂Sφr 1 ∂Sθφ 1 ∂Sφφ
+ + + eφ . (1.89)
∂r r ∂θ r sin θ ∂φ

• Laplacian of a scalar field

· µ ¶ µ ¶ µ ¶¸
1 ∂ 2 ∂ϕ ∂ ∂ϕ ∂ 1 ∂ϕ
∆ϕ = 2 r sin θ + sin θ +
r sin θ ∂r ∂r ∂θ ∂θ ∂ϕ sin θ ∂φ
1 ∂ ∂ϕ 1 ∂ ∂ϕ 1 ∂2ϕ
= 2 (r2 )+ 2 (sin θ ) + 2 2 . (1.90)
r ∂r ∂r r sin θ ∂θ ∂θ r sin θ ∂φ2

• Laplacian of a vector field

· ¸
2 2 ∂wθ 2 2 ∂wφ
∆w = ∆wr − 2 wr − 2 − 2 wθ cot θ − 2 er
r r ∂θ r r sin θ ∂φ
· ¸
2 ∂wr 1 2 cos θ ∂wφ
+ ∆wθ + 2 − 2 2 wθ − 2 2 eθ
r ∂θ r sin θ r sin θ ∂φ
· ¸
1 2 ∂wr 2 cos θ ∂wθ
+ ∆wφ − 2 2 wφ + 2 + 2 2 eφ . (1.91)
r sin θ r sin θ ∂φ r sin θ ∂φ

where the Laplacian operators in the right-hand side are given by (1.90).
Measure elements for integration
• Line

dr, r dθ, r sin θ dφ (1.92)

• Surface

r2 sin θ dθdφ, r sin θ drdφ, r drdθ. (1.93)

• Volume

r2 sin θ drdθdφ. (1.94)

Chapter 2

Some Equations of Continuum

Mechanics in Coordinates

In this section we write some partial differential equations of continuum mechanics in carte-
sian, cylindrical, and spherical coordinates.

2.1 Mass Conservation Equation

It is given by
∂ρ
+ div(ρv) = 0, (2.1)
∂t
where ρ is density and v is velocity.

• Cartesian coordinates
∂ρ ∂(ρv1 ) ∂(ρv2 ) ∂(ρv3 )
+ + + = 0. (2.2)
∂t ∂x1 ∂x2 ∂x3
• Cylindrical coordinates
∂ρ 1 ∂ 1 ∂(ρvθ ) ∂(ρvz )
+ (rρvr ) + + = 0. (2.3)
∂t r ∂r r ∂θ ∂z

• Spherical coordinates

∂ρ 1 ∂ 1 ∂ 1 ∂(ρvφ )
+ 2 (r2 ρvr ) + (sin θρvθ ) + = 0. (2.4)
∂t r ∂r r sin θ ∂θ r sin θ ∂φ

2.2 Motion Equation

The general motion equation has the form

∂v
ρ( + (grad v)v) = div T + b, (2.5)
∂t
where T is the Cauchy stress tensor and b is the body force.

25
 CHAPTER 2. SOME EQUATIONS OF CONTINUUM MECHANICS IN COORDINATES
26

• Cartesian coordinates

½ ¾
∂v1 ∂v1 ∂v1 ∂v1
ρ + v1 + v2 + v3
∂t ∂x1 ∂x2 ∂x3
· ¸
∂T11 ∂T12 ∂T13
= + + + b1 , (2.6)
∂x1 ∂x2 ∂x3
½ ¾
∂v2 ∂v2 ∂v2 ∂v2
ρ + v1 + v2 + v3
∂t ∂x1 ∂x2 ∂x3
· ¸
∂T21 ∂T22 ∂T23
= + + + b2 , (2.7)
∂x1 ∂x2 ∂x3
½ ¾
∂v3 ∂v3 ∂v3 ∂v3
ρ + v1 + v2 + v3
∂t ∂x1 ∂x2 ∂x3
· ¸
∂T31 ∂T32 ∂T33
= + + + b3 . (2.8)
∂x1 ∂x2 ∂x3
(2.9)

• Cylindrical coordinates

½ ¾
∂vr ∂vr vθ ∂vr ∂vr vθ2
ρ + vr + + vz −
∂t ∂r r ∂θ ∂z r
· ¸
1 ∂ ∂Trθ ∂Trz 1
= (rTrr ) + +r − Tθθ + br , (2.10)
r ∂r ∂θ ∂z r
½ ¾
∂vθ ∂vθ vθ ∂vθ ∂vθ vr vθ
ρ + vr + + vz +
∂t ∂r r ∂θ ∂z r
· ¸
∂Tθr 1 ∂Tθθ ∂Tθz 1 1
= + + + Trθ + Tθr + bθ , (2.11)
∂r r ∂θ ∂z r r
½ ¾
∂vz ∂vz vθ ∂vz ∂vz
ρ + vr + + vz
∂t ∂r r ∂vθ ∂z
· ¸
1 ∂ ∂Tzθ ∂Tzz
= (rTzr ) + +r + bz . (2.12)
r ∂r ∂θ ∂z
 2.3. CONSTITUTIVE LAW FOR NEWTONIAN VISCOUS FLUIDS IN COORDINATES
27

• Spherical coordinates

½ )
∂vr ∂vr 1 ∂vr vφ ∂vr vθ2 + vφ2
ρ + vr + vθ + −
∂t ∂r r ∂θ r sin θ ∂φ r
· ¸
∂Trr 1 ∂Trθ 1 ∂Tφr 1
= + + + (2Trr − Tθθ − Tφφ + Trθ cot θ) + br ,
∂r r ∂θ r sin θ ∂φ r
(2.13)

½ )
∂vθ ∂vθ 1 ∂vθ vφ ∂vθ vr vθ vφ2 cot θ
ρ + vr + vθ + + −
∂t ∂r r ∂θ r sin θ ∂φ r r
· ¸
∂Trθ 1 ∂Tθθ 1 ∂Tθφ 1
= + + + ((Tθθ − Tφφ ) cot θ + 3Trθ ) + bθ ,
∂r r ∂θ r sin θ ∂φ r
(2.14)
½ ¾
∂vφ ∂vφ 1 ∂vφ vφ ∂vφ vr vφ vθ vφ cot θ
ρ + vr + vθ + + +
∂t ∂r r ∂θ r sin θ ∂φ r r
· ¸
∂Tφr 1 ∂Tθφ 1 ∂Tφφ
= + + + bφ .
∂r r ∂θ r sin θ ∂φ
(2.15)

2.3 Constitutive Law for Newtonian Viscous Fluids in Coordi-

nates

In terms of the stretching D, it is given by

T = 2ηD + ξ div vI, (2.16)

where

1
D = (grad v + grad vt ). (2.17)
2
 CHAPTER 2. SOME EQUATIONS OF CONTINUUM MECHANICS IN COORDINATES
28

• Cylindrical coordinates

∂v1
T11 = ξ div v + 2η , (2.18)
∂x1
∂v2
T22 = ξ div v + 2η , (2.19)
∂x2
∂v3
T33 = ξ div v + 2η , (2.20)
∂x3
µ ¶
∂v1 ∂v2
T12 = T21 = η + , (2.21)
∂x2 ∂x1
µ ¶
∂v1 ∂v3
T13 = T31 = η + , (2.22)
∂x3 ∂x1
µ ¶
∂v2 ∂v3
T23 = T32 = η + , (2.23)
∂x3 ∂x2

with

∂v1 ∂v2 ∂v3

div v = + + . (2.24)
∂x1 ∂x2 ∂x3

• Cylindrical coordinates

∂vr
Trr = ξ div v + 2η , (2.25)
µ ∂r¶
1 ∂vθ vr
Tθθ = ξ div v + 2η + , (2.26)
r ∂θ r
∂vz
Tzz = ξ div v + 2η , (2.27)
· ∂z¸
1 ∂vr ∂ ³ vθ ´
Trθ = Tθr = η +r , (2.28)
r ∂θ ∂r r
µ ¶
∂vr ∂vz
Trz = Tzr = η + , (2.29)
∂z ∂r
µ ¶
∂vθ 1 ∂vz
Tθz = Tzθ = η + , (2.30)
∂z r ∂θ

with

1 ∂ 1 ∂vθ ∂vz
div v = (rvr ) + + . (2.31)
r ∂r r ∂θ ∂z
 2.3. CONSTITUTIVE LAW FOR NEWTONIAN VISCOUS FLUIDS IN COORDINATES
29

• Spherical coordinates
∂vr
Trr = ξ div v + 2η , (2.32)
µ ∂r¶
1 ∂vθ vr
Tθθ = ξ div v + 2η + , (2.33)
r ∂θ r
µ ¶
1 ∂vφ vr vθ cot θ
Tφφ = ξ div v + 2η + + , (2.34)
r sin θ ∂φ r r
· ¸
1 ∂vr ∂ ³ vθ ´
Trθ = Tθr = η +r , (2.35)
r ∂θ ∂r r
· ¸
∂ ³ vφ ´ 1 ∂vr
Trφ = Tφr = η r + , (2.36)
∂r r r sin θ ∂φ
· ¸
1 ∂vθ sin θ ∂ ³ vφ ´
Tθφ = Tφθ = η + , (2.37)
r sin θ ∂φ r ∂θ sin θ

with
1 ∂ 2 1 ∂ 1 ∂vφ
div v = (r vr ) + (sin θvθ ) + . (2.38)
r2 ∂r r sin θ ∂θ r sin θ ∂φ