Chapter 1

Vector analysis; a mathematical appendix

1.1 Vector algebra

A vector is a quantity with magnitude and a direction relative to a specific coordinate frame.

Figure 1.1: The cartesian coordinate system; the unit vectors ex , ey , ez of the three directions obey ex × ey = ez

With respect to a fixed coordinate system a vector is written as

A = Ax ex + Ay ey + Az ez (1.1)

Consider a rotation of angle φ of the reference frame about the Ox axis, as in Fig. 1.2. the unit vectors change as

Figure 1.2: Rotation of the coordinate system by the angle φ about Ox

ex → ex′ = ex , ey → ey′ = ey cos φ + ez sin φ, ez → ez′ = −ey cos φ + ez cos φ, (1.2)

Matrix form
 ′     
ex ex 1 0 0
 ey′  = R(ex , φ)  ey  , R(ex , φ) =  0 cos φ sin φ  (1.3)
ez′ ez 0 − sin φ cos φ
Properties of the matrix R

1-1
 • det(R) = 1
• R tr = R −1
• R is an orthogonal matrix, Rij Rik = δjk
We use the alternative notation {x , y , z} → {1, 2, 3} and the dummy index summation convention. Transformation of the coordi-
nates:
 
r = xi ei → xj′ ej′ , ei = R −1 ej′ = Rji ej′ (1.4)
ij

Then
xi ei = xi Rji ej′ ≡ xj′ ej′ ⇒ xj′ = Rji xi , (1.5)
The inversion is the reversing of the direction of the reference frame axes.
ei → ei′ = −ei , xi → xi′ = −xi (1.6)
The most general orthogonal transformation
xi → xi′ Rij xj , i = 1, 2, 3 (1.7)
−1 tr
where R is an orthogonal matrix: R = R and det R = ±1. A real orthogonal matrix with det R = 1 provides a matrix
representation of a proper rotation. n. The most general rotation matrix represents a counterclockwise rotation by an angle θ about
a fixed axis that lies along the unit vector n. A real orthogonal matrix with det R = −1 is the matrix representation of an improper
rotation, i.e. a proper rotation followed by an inversion.
A vector is defined as a quantity with a magnitude and direction, whose components change according to (1.7) when a rotation
is performed.
A scalar is a quantity invariant under rotations.
A cartesian tensor of rank 2 is a set of 9 numbers tij which under a rotation transforms as
tij → tij′ = Rik Rjl tkl . (1.8)
n
In general a cartesian tensor of rank n is a set of 3 numbers transforming as
ti1 ...in → ti′1 ...in = Ri1 j1 . . . Rin jn tj1 ...jn . (1.9)
Properties of tensors
• If Aij and Bij are second rank tensors, and α, β are scalars, then Tij = αAij + βBij is also a second rank tensor
• If Tij is a tensor then Tii is a scalar.
• If Tijk...lmn... is a tensor of rank n then Sjk...mn... = Tijk...imn... is a tensor of rank n − 2.
• If a and b are vectors then the outer product Tij defined by Tij = ai bj is a tensor of rank 2.
The Kronecker δ symbol is an isotropic tensor of rank 2; its components are the same in any reference frame, i.e.
δij′ = Rik Rjl δkl = Rij Rjk = δij (1.10)
The Levi-Civita antisymmetric tensor

 +1 if (i, j, k) is an even permutation of (1,2,3)
ǫijk = −1 if (i, j, k) is an odd permutation of (1,2,3) (1.11)
0 otherwise


Properties of the Levi-Civita tensor

δil δim δin
ǫijk ǫlmn = δjl δjm δjn , ǫijk ǫlmk = δil δjm − δim δjl , ǫijk ǫljk = 2δil , ǫijk ǫijk = 6, (1.12)
δkl δkm δkn
ǫijk is an isotropic tensor of rank 3.
Any second rank tensor Tij can be uniquely expressed as the sum of a symmetric and an anti-symmetric tensor;
1 1
Tij = Sij + Aij , Sij =
(Tij + Tji ), Aij = (Tij − Tji ) (1.13)
2 2
Furthermore, any anti-symmetric tensor Aij can be expressed in terms of a vector ω, (sometimes known as the dual vector) such
that
Aij = ǫijk ωk . (1.14)

1-2
1.1.1 Operations with vectors
• addition

a = ai ei , b = bi ei , a + b = c = ci ei , with ci = ai + bi (1.15)

• multiplication by a scalar

a = ai ei , λa = c = ci ei , with ci = λai (1.16)

• dot (scalar) product

a = ai ei , b = bi ei , a · b = ai bj ei · ej = ai bj δij = ai bi (1.17)

The dot product of two vectors is a scalar

a · b = ai bi → ai′ · bi′ = Rik ak Ril bl = Rik Rkl−1 ak bl = δlk ak bl = ak bk (1.18)

The norm of a vector is defined as

√ √
|a| = a · a = ai ai ≥ 0 (1.19)

Equivalent definition of dot-product:

a · b = |a| |b| cos θab (1.20)

where θab is the angle between the two vectors.

• cross (vector) product

a × b = c, c1 = a2 b3 − a3 b2 , c2 = a3 b1 − a1 b3 , c3 = a1 b2 − a2 b1 (1.21)

With the Levi-Civita tensor

a × b = ei ǫijk aj bk (1.22)

Determinant form
e1 e2 e3
c = a × b = a1 a2 a3 = −b × a (1.23)
b1 b2 b3

A cross product of two vectors transforms under proper rotations as a vector but is invariant with respect to spatial inversion;
such an object is named pseudovector
• Scalar triple product (mixed product)

a1 a2 a3
a · (b × c) = ǫijk ai bj ck = b1 b2 b3 (1.24)
c1 c2 c3
a · (b × c) = c · (a × b) = b · (c × a) (1.25)

• Vector triple product

a × (b × c) = ei ǫijk aj ǫklm bl cm (1.26)

Using (1.12)

a × (b × c) = ei aj bl cm (δil δjm − δim δjl ) = b(a · c) − c(a · b) (1.27)

1-3
1.2 Scalar fields and vector fields
A scalar field is a function Φ which associates a scalar value to every point r in space.

Φ : R → R3 (1.28)
′ ′
Under a rotation Φ is invariant, i.e. Φ(r ) → Φ (r ) = Φ(r ), i.e. when an orthogonal transformation is performed the coordinates
x , y , z change according to (1.7) and also the functional dependence of Φ changes such that the numerical value of the new function
Φ ′ at the new coordinates x ′ , y ′ , z ′ is the same as the numerical value of the old function Φ at the old coordinates x , y , z.
A vector field is a vector-valued function of three-dimensional vectors

V : R3 → R 3 (1.29)

Under a rotation described by the matrix R a vector field V (r ) changes as

Vi (r ) → Vi′ (r ′ ) = Rij Vj (r ) (1.30)

If V (r ) is a vector field then the field lines are curves s(t) in space such that the vector field V is tangent to the field line in every
point. The equation of the field line is then

ds(t)
= V (s(t)) (1.31)
dt
or, equivalent,
dsx dsy dsz
= = (1.32)
Vx (s) Vy (s) Vz (s)

1.3 Differential calculus

1.3.1 The gradient operator
The directional derivative of a scalar field is
Φ(r + ǫn) − Φ(r )
∂n Φ(r ) = (1.33)
ǫ ǫ→0

The differential of a scalar field is given by

∂Φ
dΦ(r ) = dxi (1.34)
∂xi
Sometimes we can use the equivalent notation
∂
∂i ≡ (1.35)
∂xi
Define the two vectors

dr = ei dxi , ∇Φ(r ) = ei ∂i Φ(r ) (1.36)

Then the differential of Φ is a scalar product.

dΦ(r ) = ∇Φ(r ) · dr 3 (1.37)

∇Φ is named gradient of the field Φ. In order to prove that ∇Φ is a vector we consider the orthogonal transformation xi′ = Rij xj ;
then the
∂Φ(r ) ∂xj ∂Φ(r ) −1
∂i Φ(r ) → ∂i′ Φ ′ (r ′ ) = = (R )ji = Rij ∂j Φ(r ) (1.38)
∂xj ∂xi′ ∂xj

i.e. the gradient of a scalar field behaves as a vector field. Properties of the gradient

1-4
 • we can see that the directional derivative of the field Φ can be written as ∂n Φ = n · ∇Φ = |n| |∇Φ| cos θ = |∇Φ| cos θ; then
the directional derivative is maximum if θ = 0 i.e. if ∇ and n are parallel. So, the direction of the most rapid increase in Φ
is the gradient direction, and the magnitude of the gradient is the directional derivative of Φ in that direction. If ∇Φ = 0 at
(x , y , z), then dΦ = 0 for small displacements about the point (x , y , z). This is, then, a stationary point of the function
Φ(x , y , z).
• consider the isosurfaces of the scalar field Φ(r ) (i.e. the surfaces S defined by the condition Φ(r ) = C and a curve r (t) on
the surface S. Then the total derivative of Φ(r (t)) with respect to t vanishes, as Φ(r (t)) = C . By direct calculation

dΦ(r (t)) ∂r
= · ∇Φ(r (t)) = 0 (1.39)
dt ∂t
i.e. ∇Φ(r (t)) is orthogonal on any curve r (t) on the surface S. The gradient of a vector-field Φ(r ) is in any point r orthogonal
on the level surface of Φ(r ) passing through that point.

1.3.2 The divergence and the curl

The del ∇ vector operator can act on scaler fields, the result being the gradient as we saw before, or on vector fields:
• via the dot product: divergence

∇ · V (r ) = ∂i Vi (r ) (1.40)

The divergence is a measure of how much the vector V (r ) spreads out (diverges) from the point in question. The divergence
of a vector is a scalar.
• via the cross product: curl

∇ × V (r ) = ei ǫijk ∂j Vk (r ) (1.41)

The curl is a measure of how much the vector V (r ) swirls around the point in question. The curl of a vector is a pseudovector.
Identities with differential operators

∇(φ + χ) = ∇φ + ∇χ (1.42)
∇ · (A + B) = ∇ · A + ∇ · B (1.43)
∇ × (A + B) = ∇ × A + ∇ × B (1.44)
df
∇f (φ(r )) = ∇φ(r ) (1.45)
dφ(r )
∇(φχ) = (∇φ)χ + φ(∇χ) (1.46)
∇(A · B) = A × (∇ × B) + B × (∇ × A) + (A · ∇)B + (B · ∇)A (1.47)
∇ · (φA) = (∇φ) · A + φ(∇ · A) (1.48)
∇ · (A × B) = B · (∇ × A) − A · (∇ × B), (1.49)
∇ × (φA) = φ(∇ × A) − A × (∇φ), (1.50)
∇ × (A × B) = (B · ∇)A − (A · ∇)B + A(∇ · B) − B(∇ · A)., (1.51)

Second order derivatives

∇ · ∇A = ∇2 A ≡ ∆A Laplacian of A (1.52)
∇ × (∇ · A) = 0 (1.53)
∇ · (∇ × A) = 0 (1.54)
∇ × (∇ × A) = ∇(∇ · A) − ∆A (1.55)

Example of calculation; using the notation r = |r |

r 1 r r
∇r = , ∇ = − 3, ∇r n = n r n−1 (1.56)
r r r r

1-5
1.4 Integral calculus
Possible line integrals with vectors are
Z Z Z
φ(r )dl , F (r ) · dl , F (r ) × dl (1.57)
Γ Γ Γ

where Γ is a curve and dl is the infinitesimal displacement vector dl = ex dx + ey dy + ez dz.

Surface integrals
Z Z Z
φ(r )da, F (r ) · da, F (r ) × da, (1.58)
S S S

where da = nda with da the infinitesimal element of area and n an unit vector indicating the normal direction. Note that one needs
a convention for the direction of n. If the
R surface
H is closed one takes the direction “towards outside“.
If the curve/surface is closed then →
Volume integral of vectors
Z
F (r )dτ (1.59)
V

where dτ is the volume element dτ = dx dy dz.

• The Fundamental Theorem for Gradients: If φ is a scalar field then
Zb
(∇φ(r )) · dl = φ(b) − φ(a). (1.60)
a

Consequence:
The integral is then independent on the path taken from a to b
H
(∇φ(r )) · dl = 0.
• The Fundamental Theorem for Divergences (Gauss theorem):
Z I
(∇ · F (r ))dτ = F (r ) · da (1.61)
V S

The integral of the divergence of a vector field F over a volume V is equal to the surface integral of the vector field V on the
boundary S of V. The surface integral is called the flux of the vector field F through the closed surface S.
• The Fundamental Theorem for Curls (Stokes theorem):
Z I
(∇ × F (r )) · da = F (r ) · dl (1.62)
S Γ

The integral of the curl of a function over a surface S is equal to the integral of the function on the boundary Γ of the surface.
Consequence:
R
(∇ × F (r )) · da depends only on the boundary line, not on the particular surface used.
S
H
(∇ × F (r )) · da = 0 for any closed surface,
S

1.5 The Dirac δ ”function“

The 1-D δ function is defined as
Zb
dx δ(x )f (x ) = f (0), if a<0<b (1.63)
a

1-6
Representations of the δ function: as limits of ”normal“ functions; e.g.
 1
, |x | < k
δ(x ) = lim δk (x ), δk (x ) = 2k (1.64)
k→0 0, |x | > k

Transformation of the δ function

X δ(x − xi )
δ(f (x )) = ∂f
, where xi are the roots of f (x ) = 0 (1.65)
i ∂x

In the three dimensional case

δ(r ) = δ(x )δ(y )δ(z), (1.66)

In curvilinear coordinates
1
δ(r − r0 ) = δ(u − u0 )δ(v − v0 )δ(w − w0 ) (1.67)
hu hv hw

1
∆ = −4πδ(r − r0 ) (1.68)
|r − r0 |

1.6 The Helmholtz Theorem

If the divergence D(r ) and the curl C(r ) of a vector function F (r ) are specified, and if they both go to zero faster than 1/r 2 as
r → ∞ , and if F (r ) goes to zero as r → ∞, then F is given uniquely by

F (r ) = −∇U(r ) + ∇ × V (r ) (1.69)

where
D(r ′ )
Z
1
U(r ) = dτ ′ (1.70)
4π |r − r ′ |
C(r ′ )
Z
1
V (r ) = dτ ′ (1.71)
4π |r − r ′ |

• Curl-less (or "irrotational") fields. The following conditions are equivalent

Zb I
∇ × F (r ) = 0, F (r ) · l is independent of path, F (r ) · l = 0, F (r ) = −∇V (r ) (1.72)
a Γ

• Divergence-less (or "solenoidal") fields. The following conditions are equivalent

Z I
∇ · F (r ) = 0, F (r ) · da depends only on the boudary of S, F (r ) · da = 0, F (r ) = ∇ × A(r ) (1.73)
S S

1.7 Calculation of ∆ 1r
By direct calculation, for r 6= 0
1 r
∇ =− 3 (1.74)
r r
and
1 1 r
∆ = ∇ · ∇ = −∇ · 3 = 0 (1.75)
r r r

1-7
On the other hand, consider the integral over a sphere Sǫ of radius ǫ
Z Z
1 r
I = dτ ∆ = − dτ ∇ · 3 (1.76)
r r
Sǫ Sǫ

Using the Gauss theorem

Z Z
r r r
I=− · da = − · dΩ = −4π (1.77)
r3 r r
∂Sǫ Ω

In the previous equation we have denoted the surface of the sphere by ∂Sǫ and used the surface element expression da = r 2 r /r dΩ
with dΩ the element of solid angle. Note that the result is independent of ǫ which could be arbitrary small. The definition of the δ
function
Z
dτ δ(r ) = 1, if O ∈ V (1.78)
V

so
1 1
∆ = −4πδ(r ), ∆ = −4πδ(r − r0 ), (1.79)
r |r − r0 |

1.8 Bibliography
[1] David J. Griffiths, Introduction to Electrodynamics, fourth edition, Pearson 2013, Cap. 1

1-8