Summary of Results from Previous Courses

Grad, Div, Curl and the Laplacian in Cartesian Coordinates

In Cartesian coordinates, =


x
,

y
,

z

. For a scalar eld (x) and a vector eld

F(x) = (F
1
, F
2
, F
3
), we dene:
Gradient =

x
,

y
,

z

(grad Phi)
Divergence . F =
F
1
x
+
F
2
y
+
F
3
z
(div F)
Curl F =

F
3
y

F
2
z
,
F
1
z

F
3
x
,
F
2
x

F
1
y

(curl F)
Laplacian
2
= . () =

2

x
2
+

2

y
2
+

2

z
2
(del-squared Phi)
The normal to a surface (x) = constant is parallel to .
Grad, Div and the Laplacian in Polar Coordinates
Cylindrical Polars (r, , z)
When the components (F
1
, F
2
, F
3
) of F are measured in cylindrical polar coordinates,
=

r
,
1
r

,

z

. F =
1
r

r
(rF
1
) +
1
r
F
2

+
F
3
z

2
=
1
r

+
1
r
2

2
+

2

z
2
Note: the formulae for plane polar coordinates (r, ) are obtained by setting

z
= 0.
Spherical Polars (r, , )
When the components (F
1
, F
2
, F
3
) of F are measured in spherical polar coordinates,
=

r
,
1
r

,
1
r sin

. F =
1
r
2

r
(r
2
F
1
) +
1
r sin

(F
2
sin) +
1
r sin
F
3

2
=
1
r
2

r
2

+
1
r
2
sin

sin

+
1
r
2
sin
2

2
Mathematical Methods II
Natural Sciences Tripos Part IB
Divergence and Stokes Theorems
Divergence Theorem in 3D

V
. FdV =

S
F . ndS
where the surface S encloses a volume V and n is its outward-pointing normal.
Divergence Theorem in 2D

f
x
+
g
y

dxdy =

C
(f dy g dx)
where S is a plane region enclosed by a contour C traversed anti-clockwise. We can also
write the right-hand side as

C
F. ndl where F = (f, g) and n is the outward-pointing
normal on C.
Stokes Theorem

S
(F) . ndS =

C
F . dl
where the open surface S is bounded by a contour C, n is the normal to S and dl is a
line element taken anti-clockwise around C.
SturmLiouville Theory
A SturmLiouville equation in self-adjoint form

d
dx

p(x)
dy
dx

+ q(x)y = w(x)y
in an interval a < x < b, where neither p(x) nor w(x) vanish in the interval, and with
appropriate boundary conditions, has non-zero solutions only for certain values of ,
namely the eigenvalues
i
. The corresponding solutions y
i
(x) (the eigenfunctions) are
orthogonal for distinct eigenvalues:

b
a
wy
i
y
j
dx = 0, i = j.
Vectors and Matrices
Vector identities:
|u|
2
= u . u
u (v w) = (u . w)v (u . v)w
u . (v w) = v . (wu) = w. (u v)
() = +
(u . v) = u (v) + (u . )v +v (u) + (v . )u
. (u) = . u +u .
. (u v) = v . (u) u . (v)
(u) = u + u
(u v) = (. v)u u . v (. u)v +v . u

2
u = (. u) (u)
Mathematical Methods II
Natural Sciences Tripos Part IB
A matrix A is orthogonal if A
T
A = AA
T
= I where I is the identity matrix and A
T
is
the transpose of A. This is true if and only if the columns of A are mutually orthogonal
unit vectors; similarly for the rows. Then A
1
= A
T
. In 3D, an orthogonal matrix is
either a rotation, a reection, or a combination of the two.
x is an eigenvector of a symmetric matrix A with eigenvalue if Ax = x. The eigenval-
ues can be found by solving the equation det(AI) = 0. The three unit eigenvectors
are orthogonal (or in the case of repeated eigenvalues, can be chosen to be so). The
eigenvalues are also given by the stationary values of a
T
Aa/a
T
a over all possible vectors
a (or equivalently, the eigenvalues are given by the stationary values of a
T
Aa subject
to the constraint a
T
a = 1).
The determinant of a matrix is unchanged by adding a multiple of one row to a dierent
row, or by adding a multiple of one column to a dierent column. Swapping two rows
changes the sign of the determinant, as does swapping two columns. Multiplying a row,
or a column, by a constant factor multiplies the determinant by . If two rows, or
columns, are the same, then the determinant is zero. For any square matrices A and B,
det A
T
= det A and det AB = det Adet B.
Fourier Series
Any (well-behaved) function f(x) with period L may be represented as the innite sum
f(x) = A
0
+

n=1

A
n
cos
2nx
L
+ B
n
sin
2nx
L

where
A
0
=
1
L

L
0
f(x) dx, A
n
=
2
L

L
0
f(x) cos
2nx
L
dx, B
n
=
2
L

L
0
f(x) sin
2nx
L
dx.
A function f(x) which is dened only in the region 0 x L may be represented as
a full Fourier Series as above by rst turning it into a periodic function with period L;
or may alternatively be represented either by a Fourier cosine series, in which only the
cosine terms appear, or by a Fourier sine series, in which only the sine terms appear.
For a cosine series,
f(x) = A
0
+

n=1
A
n
cos
nx
L
where
A
0
=
1
L

L
0
f(x) dx and A
n
=
2
L

L
0
f(x) cos
nx
L
dx.
For a sine series,
f(x) =

n=1
B
n
sin
nx
L
where
B
n
=
2
L

L
0
f(x) sin
nx
L
dx.
Mathematical Methods II
Natural Sciences Tripos Part IB
Legendre Polynomials
Legendres equation for P(x) is
d
dx

(1 x
2
)
dP
dx

+ P = 0.
There are regular singular points at x = 1. A series solution may be sought about
x = 0; but the resulting series is ill-behaved (specically, P is singular at x = 1)
except when = n(n + 1) for some non-negative integer n. Then the series terminates
after a nite number of terms, and the solution is the Legendre polynomial P
n
(x) of
degree n. P
n
(x) is an even/odd function of x (i.e., contains only even/odd powers of x)
when n is even/odd respectively. It is normalised so that P
n
(1) = 1 (and therefore
P
n
(1) = (1)
n
). Legendre polynomials are orthogonal:

1
1
P
m
(x)P
n
(x) dx =

0 m = n,
2
2n+1
m = n.
They can be found explicitly using Rodrigues formula
P
n
(x) =
1
2
n
n!
d
n
dx
n

(x
2
1)
n

.
Taylors Theorem (complex version)
Any smooth complex function can be expressed as a power series about z = z
0
in the
form
f(z) =

n=0
a
n
(z z
0
)
n
where a
n
= f
(n)
(z
0
)/n!.
Fourier Transforms
For suitable functions f(x), the Fourier Transform is dened by

f(k) =

f(x)e
ikx
dx,
and the inversion formula is
f(x) =
1
2

f(k)e
ikx
dk.
The Fourier Transform of f

(x) is ik

f(k). The Fourier Transform of f(xa) for constant
a is e
ika

f(k). The convolution h = f g, dened by
h(y) =

f(x)g(y x) dx,
satises

h(k) =

f(k) g(k).
Mathematical Methods II
Natural Sciences Tripos Part IB