Methods Revision Notes

1 Fourier Series
For a 2L-periodic (continuous) function f (x) which satisfies the Dirichlet conditions:
∞ h
1 X  nπx   nπx i
f (x) = a0 + an cos + bn sin .
2 n=1
L L

Orthogonality relations:
2L
Z (
 nπx   mπx  Lδnm , nm 6= 0
sin sin dx = ,
0
L L 0, nm = 0
2L
Z (
 nπx   mπx  Lδnm , n 6= 0 or m 6= 0
cos cos dx = ,
0
L L 2L, n=m=0
Z 2L  nπx   mπx 
sin cos dx = 0 ∀n, m.
0
L L

In particular, for n 6= 0:
Z 2π Z 2π
2
cos (nx) dx = sin2 (nx) dx = π.
0 0

Complex form:
∞
inπx
X
f (x) = cn e L ,
n=−∞

where
an − ibn an + ibn a0
cn = , c−n = c∗n = for n > 0, c0 = ,
2 2 2
and the orthogonality relation is
Z 2L
inπx −imπx
e L e L dx = 2Lδnm .
0

Parseval’s theorem:
∞ ∞
Z 2L " #
2
X
2 a20 X 2 2
E= f (x) dx = 2L |cn | = L + (an + bn ) .
0 n=−∞
2 n=1

Half-range series: We can extend a function f (x) defined on [0, L] to (−L, L] and hence to
a 2L-periodic function as an odd function:
∞
X  nπx 
f (x) = bn sin ,
n=1
L

1
where Z L
2  nπx 
bn = f (x) sin dx,
L 0
L
or as an even function:
∞
a0 X  nπx 
f (x) = + an cos ,
2 n=1
L
where Z L
2  nπx 
an = f (x) cos dx.
L 0
L

2 Sturm-Liouville Theory
Strum-Liouville form of 2nd order ODE:

0
Ly = − py 0 + qy = λwy.

Properties of self-adjoint operators:

1. The eigenvalues are real and countably many.
2. The eigenfunctions are orthogonal with w in the integral, and can be normalised such
that Z b
w Ym Yn dx = δmn .
a

3. The eigenfunctions form a complete basis, i.e. any function f (x) satisfying the same
boundary conditions can be written as
∞
X ∞
X
f (x) = an yn (x) = An Yn (x),
n=1 n=1

and by orthogonality
Z b
An = f (x) w Yn dx.
a

4. Parseval’s theorem: Z b ∞
X
wf 2 (x) dx = A2n .
a n=1

5. The eigenfunction series representation minimises the mean squared error.

Example.  
L − λ̂w y = f (x) = wF (x).

Write
∞
X ∞
X
f (x) = wF (x) = w(x) An Yn , y= Bn Yn ,
n=1 n=1

then
An
Bn = .
λn − λ̂

2
3 Separation of Variables
Wave Equation
∂2
φ(x, t) = c2 ∇2 φ(x, t)
∂t2

Diffusion Equation
∂
θ(x, t) = D ∇2 θ(x, t)
∂t

Laplace’s Equation
∇2 ψ(x) = 0

Bessel’s Equation
r2 R00 + rR0 + (r2 − n2 )R = 0
Appears in:
• the wave equation in 2D polar coordinates,
• Laplace’s equation in cylindrical polar coordinates.

Bessel’s functions: Jk (r) is bounded, Yk (r) is unbounded at r = 0. Usually found in the

form Jn (knj r), where knj are the zeros of Jn .

Legendre’s Equation
0
(1 − x2 )P 0 + n(n + 1)P = 0


Appears in Laplace’s equation in spherical polars (for the θ coordinate - substitute x = cos θ).

Legendre Polynomials:
• Scaled such that Pn (1) = 1.
• Pn has n zeros in [−1, 1].
• Pn is even for even n, odd for odd n.
First five Legendre Polynomials:
• P0 (x) = 1,
• P1 (x) = x,
3x2 −1
• P2 (x) = 2 ,
5x3 −3x
• P3 (x) = 2 ,
35x4 −30x2 +3
• P4 (x) = 8 .

Orthogonality Relations:
Z 1
2
Pn (x)Pm (x) dx = δnm .
−1 2n + 1

3
4 Green’s Functions
Ly = f (x) + BC
Green’s function G(x; ξ) satisfies:

LG(x; ξ) = δ(x − ξ),

with homogeneous BC.

Gives solution y1 with homogeneous BC:

Z b
y1 (x) = G(x; ξ) f (ξ) dξ.
a
For inhomogeneous BC, find solution y2 such that

Ly2 = 0
with the appropriate BC, then y(x) = y1 (x) + y2 (x).

5 Fourier Transforms
Z ∞
fe(k) = f (x)e−ikx dx
−∞
Z ∞
1
f (x) = fe(k)eikx dk
2π −∞

Important Properties:
• If g(x) = f (x − λ), then ge(k) = e−ikλ fe(k).

• If g(x) = f 0 (x), then ge(k) = ik fe(k).

• fe(k)e ∗ g(k), where

g (k) = f]
Z ∞
f ∗ g(x) = f (u)g(x − u) du.
−∞

Important Transforms:

f (x) fe(k)

δ(x − a) e−ika

e−ika
H(x − a) πe−ika δ(k) +
ik

1 
δ(x + a) + δ(x − a) cos ka
2

4
 1 
δ(x + a) − δ(x − a) sin ka
2

1  sin ka
H(x + a) − H(x − a)
2 k

e−a|x| 1
, a>0
2a a2 + k2

xm eax 1
, <(a) < 0, x>0
m! (ik − a)m+1

Example.
y 00 (x) − a2 y(x) = f (x), a > 0, y → 0, y 0 → 0 as |x| → ∞.
Take FT:
fe
ye = − ,
a2 + k2
so by the convolution theorem:
Z ∞
1
y(x) = − f (u) e−a|x−u| du.
2a −∞

6 Linear first-order PDEs - method of characteristics

α(x, y) ux + β(x, y) uy = γ(x, y, u) + BC
Write x = x(s, ξ), y = y(s, ξ), u = u(s, ξ) by defining:
∂ ∂
x = α(x, y), y = β(x, y),
∂s ∂s
and parameterising the BC using ξ. ξ = ξ(x, y) defines a characteristic.

Then solve
∂
u = γ(s, ξ, u) + BC,
∂s
to get u(s, ξ) = u(x, y).

7 Classification of PDEs
a(x, y) uxx + 2b(x, y) uxy + c(x, y) uyy + d(x, y) ux + e(x, y) uy + f (x, y) u = F (x, y)
√
−b ± b2 − ac
λ± =
2
• If b2 > ac, the equation is hyperbolic (two real characteristics).
• If b2 = ac, the equation is parabolic (one real characteristic).
• If b2 < ac, the equation is elliptic (two complex characteristics).

5
Characteristics (for hyperbolic equations):

dx
ξ = Cξ ⇔ = −λ+ ,
dy

dx
η = Cη ⇔ = −λ− .
dy
The reparameterisation u(x, y) = u(ξ, η) gives the canonical form of the equation:

uξη + A(ξ, η) uξ + B(ξ, η) uη + C(ξ, η) u = D(ξ, η).

8 Heat & wave equations/Duhamel’s principle

8.1 1D Diffusion Equation with inhomogeneous BC
∂ ∂2
θ(x, t) = D 2 θ(x, t), θ(x, 0) = Θ(x)
∂t ∂x
Take FT wrt x, solve to get
2
θ(k,
e t) = Θ(k)
e e−Dk t .
Note the FT pair:

−x2
 
1 2
Sd (x, t) = √ exp ↔ Sed (k, t) = e−Dk t ,
4πDt 4Dt

so
Z ∞
θ(x, t) = Θ(u) Sd (x − u, t) du
−∞
Z ∞
(x − u)2
 
1
=√ Θ(u) exp − du.
4πDt −∞
4Dt
Sd (x, t) is the fundamental solution/source function for the diffusion equation. It is the
solution to the problem:

∂ ∂2
θ(x, t) = D 2 θ(x, t), θ(x, 0) = δ(x).
∂t ∂x

8.2 Forced 1D heat equation with homogeneous BC

∂ ∂2
θ(x, t) − D 2 θ(x, t) = f (x, t), θ(x, 0) = 0
∂t ∂x
Green’s function satisfies:

∂ ∂2
Gd (x, t; ξ, τ ) − D 2 Gd (x, t; ξ, τ ) = δ(x − ξ) δ(t − τ ), Gd (x, 0; ξ, τ ) = 0.
∂t ∂x
Take FT wrt x, solve to get

e d (k, t; ξ, τ ) = H(t − τ ) e−ikξ e−Dk2 (t−τ ) ,

G
hence

6
 h 2
i
H(t − τ ) exp −(x−ξ)
4D(t−τ )
Gd (x, t; ξ, τ ) = H(t − τ ) Sd (x − ξ, t − τ ) = p .
4πD(t − τ )

Then
Z ∞ Z ∞
θ(x, t) = Gd (x, t; ξ, τ ) f (ξ, τ ) dξdτ
0 −∞

∞
Z tZ
−(x − ξ)2
 
1 f (ξ, τ )
=√ √ exp dξdτ.
4πD t−τ 4D(t − τ )
0 −∞

8.3 Forced 1D wave equation with homogeneous BC

∂2 2 ∂
2
∂
2
y(x, t) − c 2
y(x, t) = f (x, t), y(x, 0) = y(x, 0) = 0
∂t ∂x ∂t
Find the Green’s function. Take FT wrt x, solve to get:

−ikξ
e w (k, t; ξ, τ ) = H(t − τ ) e
G
sin[kc(t − τ )]
,
kc

hence

H(t − τ ) 


Gw (x, t; ξ, τ ) = H x − ξ − c(t − τ ) − H x − ξ + c(t − τ )
2c


H(t − τ ) H c(t − τ ) − |x − ξ|
= .
2c

So
Z ∞ Z ∞
y(x, t) = Gw (x, t; ξ, τ ) f (ξ, τ ) dξdτ
0 −∞
Z tZ x+c(t−τ )
1
= f (ξ, τ ) dξdτ.
2c 0 x−c(t−τ )

8.4 Unforced 1D wave equation with inhomogeneous BC

∂2 2 ∂
2
∂
y(x, t) = c y(x, t), y(x, 0) = φ(x), y(x, 0) = ψ(x)
∂t2 ∂x2 ∂t
Source functions:
1 
S1 (x, t) = δ(x + ct) + δ(x − ct) for the y(x, 0) = φ(x) condition,
2

1  1

S2 (x, t) = H(x + ct) − H(x − ct) = H ct − |x| for the yt (x, 0) = ψ(x) condition.
2c 2c
Then

7
 Z ∞ Z ∞
y(x, t) = φ(u) S1 (x − u) du + ψ(u) S2 (x − u) du,
−∞ −∞

which gives d’Alembert’s Solution:

Z x+ct
1  1
y(x, t) = φ(x + ct) + φ(x − ct) + ψ(u) du.
2 2c x−ct

Note that

Gw (x, t; ξ, τ ) = H(t − τ ) S2 (x − ξ, t − τ ),
and that S1 is the solution to

∂2 2 ∂
2
∂
y(x, t) = c y(x, t), y(x, 0) = δ(x), y(x, 0) = 0,
∂t2 ∂x2 ∂t
while S2 is the solution to

∂2 ∂2 ∂
2
y(x, t) = c2 2 y(x, t), y(x, 0) = 0, y(x, 0) = δ(x).
∂t ∂x ∂t

9 Poisson’s Equation
∇2 u(x) = f (x) in D + BC on ∂D

9.1 Free-space Green’s function

∇2 Gf (r; r0 ) = δ(r − r0 )

2D:
log kr − r0 k
Gf 2 (r; r0 ) =
2π

3D:
−1
Gf 3 (r; r0 ) =
4π kr − r0 k

9.2 Green’s Identities

Green’s 2nd identity (3D):
Z Z  
∂ψ ∂φ
φ ∇2 ψ − ψ ∇2 φ dV =
 
φ −ψ dS.
V
∂n ∂n
S

Let φ = u, ψ = Gf 3 , and consider volume V , with a ball of radius  centered at r0 missing.

Apply Green’s second identity on V (with boundary S ∪ S) and take  → 0 to get:

8
Green’s 3rd identity (3D):

Z
Z  
∂ ∂
u(r0 ) = f (r) Gf 3 (r; r0 ) dV + u(r) Gf 3 (r; r0 ) − Gf 3 (r; r0 ) u(r) dS.
V ∂n ∂n
S

9.3 Dirichlet Green’s functions

A function G(r; r0 ) is a Dirichlet Green’s function for the Laplacian operator on some domain
D if:

1. G(r; r0 ) has continuous second derivatives, and ∇2 G(r; r0 ) = 0 on D\{r0 };

2. G(r; r0 ) = 0 on ∂D;

3. H(r; r0 ) = G(r; r0 ) − Gf (r; r0 ) is finite at r0 , has continuous second derivatives in D, and

∇2 H(r; r0 ) = 0 at r0 (hence in D).

The last condition implies that ∇2 G(r; r0 ) = δ(r − r0 ). Now consider Poisson’s equation with
Dirichlet boundary conditions:

∇2 u(r) = f (r) in D, u(r) = h(r) on ∂D.

Apply Green’s 2nd identity on φ = u, ψ = H, add to Green’s 3rd identity to get:

Z Z
∂
u(r0 ) = f (r) G(r; r0 ) dV + h(r) G(r; r0 ) dS.
D ∂D
∂n

9.4 Method of Images

For domains with boundaries, we usualy construct G(r; r0 ) using the method of images.

Example. D = {z > 0} ⊆ R3 , then

−1 1
G(x; x0 ) = + ,
4π kx − x0 k 4π x − x−


0

where x−
0 = (x0 , y0 , −z0 ).

Example. D = {kx − x0 k < a2 } ⊆ R3 , then

−1 a 1
G(x; x0 ) = + ,
4π kx − x0 k kx0 k 4π kx − x∗0 k
a2 x0
where x∗0 = 2.
kx0 k
Example. D = {x ≥ 0, y ≥ 0} ⊆ R2 , then
1
log (x − x0 )2 + (y − y0 )2 + log (x + x0 )2 + (y + y0 )2
   
G(x; x0 ) =
4π 
− log (x + x0 )2 + (y − y0 )2 − log (x − x0 )2 + (y − y0 )2 .