MAST20030 Differential Equations

1500
The University of Melbourne
School of Mathematics and Statistics
1000
Semester 2, 2023

y
Prepared by Jesse F. Collis and Antoinette Tordesillas 500

© These notes have been made in accordance with the provisions of

Section 113P of the Copyright Act 1968 for educational purposes.
These notes are for the use of students of the University of 0
Melbourne enrolled in MAST20030 Differential Equations. 0 100 200 300
x
Course outline

1. Linear ordinary differential equations (ODEs) and initial value problems (IVPs)

2. Power series solutions to ODEs

3. Laplace transforms

4. Fourier series

5. Boundary value problems (BVPs) and linear partial differential equations (PDEs)

6. Fourier transforms

2
Topic 6: Fourier Transforms

Recall the last slide of Topic 4: Fourier Series:

Instead of using sines and cosines, we could use complex exponentials. This leads to the
complex Fourier series
n=∞
󰁛 󰁝 L
1
s(x) = cn eiπnx/L , cn = f (x)e−iπnx/L dx.
n=−∞
2L −L

This is our starting point for Fourier transforms later in semester.

Can we use Fourier series for non-periodic unbounded functions?

3
Unbounded functions

◮ Let’s first rearrange

∞ 󰁝
1 󰁛 π 󰁥 ikn x L
nπ
f (x) = fn e where f󰁥n = f (x)e−ikn x dx and kn =
2π n=−∞ L −L L

π
◮ = kn+1 − kn = ∆kn
L
∞
1 󰁛
f (x) = ∆kn f󰁥(kn )eikn x
2π n=−∞

◮ Since we are investigating non-periodic unbounded functions, let’s take L → ∞.

Then ∆kn → 0 and the infinite sum becomes an integral.

4
The Fourier transform

Takes a suitable function, f (x) defined on x ∈ R and maps it to a new function, (F f )(k),
usually defined for k ∈ R.

󰁝 ∞ 󰁝 ∞
1
(F f )(k) = f󰁥(k) = f (x)e−ikx dx and f (x) = f󰁥(k)eikx dk
−∞ 2π −∞

◮ The Fourier transform is a linear transformation.

1. (F {f + g})(k) = (F f )(k) + (F g)(k)

2. (F {αf })(k) = α(F f )(k)

5
Theorem (Existence and inversion of the Fourier transform)
Let f : R → C be a piecewise continuous function with finitely many critical points
and finitely many discontinuities on every closed interval within its domain such that
󰁝 ∞
|f (x)| dx converges.
−∞
Then the Fourier transform of f (x) exists and,
1. if f (x) is continuous at x, then
󰀓 󰀔
F −1 f󰁥 (x) = f (x)

2. if f (x) is discontinuous at x, then

󰀓 󰀔 󰀕 󰀖
−1 󰁥 1
F f (x) = lim f (ξ) + lim f (ξ)
2 ξ→x− ξ→x+

6
Exercise 6.1
󰀫
1 |x| < 1
Calculate the Fourier transform of f (x) =
0 otherwise

7
Exercise 6.2
Calculate the Fourier transform of u′ (x)

8
Exercise 6.3
Solve the differential equation u′ (x) = δ−1 (x) − δ1 (x) on the domain x ∈ R subject to
the condition that the Fourier transform of u(x) exists.

9
Different definitions
Be careful because there are many different definitions in use.
Always check if you are going to use someone else’s result!

󰁝 ∞ 󰁝 ∞
1
f󰁥(k) = −ikx
f (x)e dx, f (x) = f󰁥(k)eikx dk.
−∞ 2π −∞
󰁝 ∞ 󰁝 ∞
󰁥 1
f (k) = f (x)e−ikx dx, f (x) = f󰁥(k)eikx dk.
2π −∞ −∞
󰁝 ∞
1 1
f󰁥(k) = √ f (x)e−ikx dx, f (x) = √ f󰁥(k)eikx dx.
2π 2π −∞
󰁝 ∞ 󰁝 ∞
f󰁥(k) = f (x)e−2πikx dx f (x) = f󰁥(k)e2πikx dk.
−∞ −∞

Often k (wavenumber) is replaced with p (momentum) and when we have functions of

time we usually use ω (angular frequency) or f (frequency).
10
Table of Fourier transforms

f (x) f󰁥(k) f (x) f󰁥(k)

δa (x), a ∈ R e−iak eiax , a ∈ R 2πδa (k)

2a 1 π −a|k|
e−a|x| , a > 0 ,a>0 e
a2 + k 2 a2 + x2 a
π
H−a (x) − Ha (x), a > 0 2a sinc ak sinc ax, a > 0 (H−a (k) − Ha (k))
󰁵 a
2 π − k2 π πk
e−ax , a > 0 e 4a sech ax, a ∕= 0 sech
a a 2a

󰀫
sin y
∗ y , y ∕= 0
sinc(y) =
1, y=0

11
Exercise 6.4
Use the table of Fourier transforms to find the Fourier transform of f (x) = sin x and the
inverse Fourier transform of g󰁥(k) = sin k.

12
Convolution

Theorem (The Fourier transform of a convolution)

We define the convolution of f (x) with g(x) as
󰁝 ∞
(f ∗ g)(x) = f (x − x′ )g(x′ )dx′
−∞

Then the Fourier transform of the convolution is

(F {f ∗ g})(k) = (F f )(k)(F g)(k)

Note this is different to the definition we used in Laplace transforms because we are
working on x ∈ R instead of t > 0.

13
Exercise 6.5
1 1
Calculate the convolution, (f ∗ g)(x), of f (x) = with g(x) = 2 where x ∈ R.
x2 +4 x +9

14
Shift theorems

1. (F {f (x − y)})(k) = e−iky (F f )(k), for y ∈ R.

2. (F {eiqx f })(k) = (F f )(k − q), for q ∈ R.

Examples

󰀓 󰀔 2ae−2ik
F e−a|x−2| (k) = 2
a + k2
󰀕 󰀖
π(k − 5)
F −1 sech (x) = e5ix sech x
2

15
Application to ODEs

We’ve already seen that the Fourier transform of f ′ (x) is ik f󰁥(k).

1. Differentiation w.r.t. x turns into multiplication by k
󰀕 󰀖
dn f
F n (k) = (ik)n (F f )(k)
dx

2. Multiplication by x turns into differentiation w.r.t. k

dn
(F xn f )(k) = in (F f )(k)
dk n

16
Airy’s equation revisited

Recall from Topic 2 (slide 16) that we found two power series solutions to

y ′′ (x) − xy(x) = 0

2.5

2.0 y1
y2
1.5

1.0

0.5

0.0

-0.5

-1.0

-20 -15 -10 -5 0

x

17
Exercise 6.6
Use the method of Fourier transforms to find the solution to Airy’s differential equation

y ′′ (x) − xy(x) = 0

subject to the condition that the Fourier transform of y(x) exists.

18
19
 0.4

0.2

y(x) 0.0

-0.2

-0.4

-20 -15 -10 -5 0 5 10

x

20
Application to solving PDEs
Exercise 6.7
Use the method of Fourier transforms to solve the 1D heat equation
∂u ∂2u
=α 2
∂t ∂x
on x ∈ R and t > 0 subject to the conditions that the Fourier transform of f (x, t) w.r.t.
x exists, and u(x, 0) = f (x).

21
22
Exercise 6.8
Using the result from the previous exercise, find the solution to the heat equation subject
to the initial condition u(x, 0) = δ(x).

23
 α=1

u(x,t)

x
t

24
Exercise 6.9
Just for fun, use the method of Fourier transforms combined with the method of Laplace
transforms to solve the IBVP in Exercise 6.8.

25
The wave equation
Exercise 6.10
Use Fourier transform methods to solve the 1D wave equation
∂2u 2
2∂ u
= c
∂t2 ∂x2
subject to the condition that the Fourier transform of u w.r.t. x exists, and with the
initial conditions
u(x, 0) = f (x), and ∂t u(x, 0) = 0

26
27
Exercise 6.11
Find the solution to the 1D wave equation subject to the condition that the Fourier
transform of u(x, t) w.r.t. x exists, and the initial conditions
1
u(x, 0) = e− 2 |x| and ∂t u(x, 0) = 0

28
 c=1 c=2

u(x,t) u(x,t)

t x t x

29