FOURIER ANALYSIS: LECTURE 13

12 Ordinary Di↵erential Equations

A powerful application of Fourier methods is in the solution of di↵erential equations. This is because
of the following identity for the FT of a derivative:
 p
⇥ (p) ⇤ df
F T f (x) = F T = (ik)p f˜(k) (12.134)
dxp

Thus applying a FT to terms involving derivatives replaces the di↵erential equation with an algebraic
equation for f˜, which may be easier to solve.
Let’s remind ourselves of the origin of this fundamental
R result. The simplest approach is to write
a function f (x) as a Fourier integral: f (x) = f˜(k) exp(ikx) dk/2⇡. Di↵erentiation with respect
R
to x can be taken inside the integral, so that df /dx = f˜(k) ik exp(ikx) dk/2⇡. From this we can
immediately recognise ik f˜(k) as the FT of df /dx. The same argument can be made with a Fourier
series.
We will illustrate the application of this result with the familiar example of the driven damped
simple harmonic oscillator.

12.1 The driven damped Simple Harmonic Oscillator

Probably the most familiar physical context for this equation is where we have a mass m attached to
a spring with a spring constant k, and which is also immersed in a viscous fluid that exerts a resistive
force proportional to the speed of the mass, with a constant of proportionality D. Imagine further
that the mass is driven by an external force F (t). The equation of motion for the displacement z(t)
is
mz̈ = kz Dż + F (t) (12.135)
(negative signs in kz Dż because both spring and drag oppose motion in the z direction). Now
define a characteristic frequency by !02 = k/m, and let = D/m. Then we can write the equation
in what will be our standard form:

z̈ + ż + !02 z = f (t), (12.136)

where f (t) = F (t)/m.

The identical equation arises in an LCR electrical circuit consisting of an inductor of inductance
L, a capacitor of capacitance C and a resistor of resistance R. If they are in series, then in the
simplest case of one of each in the circuit, the voltage across all three is the sum of the voltages
across each component. The voltage across R is IR, where I is the current; across the inductor it
is LdI/dt, and across the capacitor it is Q/C, where Q is the charge on the capacitor:

dI Q
V (t) = L + RI + . (12.137)
dt C

55
 Figure 12.14: A simple series LCR circuit.

Now, since the rate of change of charge on the capacitor is simply the current, dQ/dt = I, we can
di↵erentiate this equation, to get a second-order ODE for I:
d2 I dI I dV
L 2
+R + = . (12.138)
dt dt C dt
This is the same di↵erential equation as before, with
(z, , !02 , f ) ! (I, R/L, 1/LC, V̇ /L). (12.139)

12.1.1 Periodic driving force

The simplest case to consider is where the driving force oscillates at a single frequency, !. Let’s
look at how this is often solved, without using any Fourier terminology. We can always choose the
origin of time so that f (t) = A cos !t, so we want to solve z̈ + ż + !02 z = A cos !t. The normal
approach is to guess that z must respond at the same frequency, so that z = a cos !t + b sin !t.
Substituting this guess, we get
(!02 ! 2 )(a cos !t + b sin !t) + !( a sin !t + b cos !t) = A cos !t. (12.140)
In order for the lhs to be a pure cos, the sin coefficient must vanish: b(!02
! ) a ! = 0. Equating 2
2 2
cos coefficients then gives a(!0 ! ) + b ! = A. These equations can be written in matrix form:
✓ 2 ◆✓ ◆ ✓ ◆
(!0 ! 2 ) ! a A
2 2 = . (12.141)
! (!0 ! ) b 0
So inverting the matrix gives
✓ ◆ ✓ ◆✓ ◆
a 1 (!02 !2) ! A
= 2 2 . (12.142)
b ! + (!02 ! 2 )2 ! (!02 !2) 0
Hence the solution is
⇥ ⇤ 1 ⇥ ⇤
z = A (!02 ! 2 )2 + 2
!2 (!02 ! 2 ) cos !t + ! sin !t . (12.143)
Because of the occurrence of sin and cos terms, there is a phase shift with respect to the driving
term, so we must be able to write this as z = z0 cos(!t + ), and expressions for the amplitude and
phase could be obtained with a bit of trigonometric e↵ort. But there is an easier way.

56
12.1.2 Complex solution

Write the driving term as f = A exp(i!t). The normal justification for this complex approach is that
we will take the real part at the end. This is fair enough, but we will give a better justification later.
Note that the amplitude A could be complex, A = |A| exp(i ), so we can easily include a phase
in the input signal; in the real formalism, we chose the origin of time so that this phase vanished,
otherwise the algebra would have been even messier. If we now try a solution z = c exp(i!t), where
again c can include a phase, we get

! 2 c + i !c + !02 c = A, (12.144)

since the exp(i!t) factor on each side can be divided out. This gives us the solution for c immediately
with almost no work.
The result can be made a bit more intuitive by splitting the various factors into amplitudes and
phases. Let A = |A| exp(i ) and ( ! 2 + i ! + !02 ) = |B| exp(i↵), where
q
|B| = (!02 ! 2 )2 + 2 ! 2 (12.145)

and
tan ↵ = !/(!02 ! 2 ). (12.146)
Then we have simply
|A|
z(t) = exp[i(!t + ↵)], (12.147)
|B|
so the dynamical system returns the input oscillation, modified in amplitude by the factor 1/|B|
and lagging in phase by ↵. For small frequencies, this phase lag is very small; it becomes ⇡/2 when
! = !0 ; for larger !, the phase lag tends to ⇡.
The same equations can be obtained using the real approach, but it takes a great deal longer. Once
again, we see the advantage of the complex formalism.

12.2 Fourier approach

The above traditional approach works, but it can seem a little ad hoc. We guess form for the
solution (how did we know to make this guess?), and we have to remember to take the real part of a
complex calculation and throw away the imaginary part. But Fourier analysis allows us to be more
systematic. The key point is that the di↵erential equations of interest are linear in the unknown,
z, and in its derivatives. Since the FT is also a linear operation, we immediately get an equation
relating the Fourier transforms of z and f :

z̈ + ż + !02 z = f (t) ) ! 2 z̃(!) + i! z̃(!) + !02 z̃(!) = f˜(!), (12.148)

because the FT of ż(t) is i!z̃(w), and the FT of z̈(t) is ! 2 z̃(!). Thus

f˜(!)
z̃(!) = . (12.149)
!02 !2 + i !

This solution in Fourier space is general and works for any time-dependent force. Once we have a
specific form for the force, we can in principle use the Fourier expression to obtain the exact solution
for z(t), assuming we can do the necessary integrals.

57
As a simple example, consider a driving force that can be written as a single complex exponential:

f (t) = A exp(i⌦t). (12.150)

Fourier transforming, we get

Z 1
f˜(!) = Aei⌦t e i!t
dt = 2⇡A (⌦ !) = 2⇡A (! ⌦). (12.151)
1

Unsurprisingly, the result is a -function spike at the driving frequency. Since we know that z̃(!) =
f˜(!)/(!02 ! 2 + i !), we can now use the inverse FT to compute z(t):
Z 1
1 f˜(!)
z(t) = 2
ei!t d! (12.152)
2⇡ 1 !0 ! 2 + i
Z 1
(! ⌦)
= A 2
ei!t d!
1 !0 !2 + i !
exp(i⌦t) |A|
= A 2 2
= exp(i⌦t + ↵),
!0 ⌦ + i ⌦ |B|

where A = |A| exp(i ) and !02 ⌦2 + i ⌦ = |B| exp(i↵). This is just the answer we obtained by
taking the usual route of trying a solution proportional to exp(i⌦t) – but the nice thing is that the
inverse FT has produced this for us automatically, without needing to guess.
Finally, we can also clarify the common use of complex exponentials to represent real oscillations.
The traditional argument is that (as long as we deal with linear equations) the real and imaginary
parts process separately and so we can just take the real part at the end. But in Fourier analysis,
we have noted that real functions require the Hermitian symmetry f˜(!) = f˜⇤ ( !). If f (t) is to be
real, it therefore makes no sense to consider purely a signal at a single !: we must allow for the
negative-frequency part simultaneously. If there is to be a spike in f˜ at ! = +⌦, we therefore need
a corresponding spike at ! = ⌦:

f˜(!) = 2⇡A (! ⌦) + 2⇡A⇤ (! + ⌦). (12.153)

The inverse Fourier transform of this is just a real oscillation with arbitrary phase:

f (t) = A exp(i⌦t) + A⇤ exp( i⌦t) = 2|A| cos(⌦t + ), (12.154)

where A = |A| exp(i ). Notice that this f (t) is the sum of a complex exponential and its conjugate,
so we have
f (t) = 2Re[A exp(i⌦t)]. (12.155)
Thus the Hermitian symmetry between positive and negative frequencies ends up instructing us to
adopt exactly the traditional approach: solve the problem with f / exp(i⌦t) and take the real part
at the end.
Finally, then, the time-dependent solution when we insist on this real driving force of given frequency
comes simply from adding the previous solution to its complex conjugate:
|A| |A| |A|
z(t) = exp[i(⌦t + ↵)] + exp[ i(⌦t + ↵)] = 2 cos(⌦t + ↵). (12.156)
|B| |B| |B|

The factor 2 is unimportant; it also occurs in the definition of f (t) = 2|A| cos(⌦t + ), so it can be
absorbed in the definition of |A|.

58
FOURIER ANALYSIS: LECTURE 14

12.3 Periodic vs non-periodic forces

The Fourier Transform approach does not assume the driving force, or the response, to be periodic.
It’s worth contrasting explicitly this general approach with what we do when the force is periodic
with period T (i.e. f (t + T ) = f (t) for all t).
In this case, we expand both f and z as Fourier Series. If we use a sin+cos series, then
X1 X1
1
z(t) = a0 + an cos (!n t) + bn sin (!n t) (12.157)
2 n=1 n=1
X1 X1
1
f (t) = A0 + An cos (!n t) + Bn sin (!n t)
2 n=1 n=1

where !n = 2⇡n/T .
1
X 1
X
dz
= !n an sin (!n t) + n!bn cos (n!t) (12.158)
dt n=1 n=1
X1 X1
d2 z(t)
= !n2 an cos (!n t) !n2 bn sin (!n t)
dt2 n=1 n=1

Then the l.h.s. of the di↵erential equation becomes

X1
!02 ⇥ ⇤
z̈ + ż + !02 z = a0 + !n2 an + !n bn + !02 an cos(!n t) + !n2 bn !n bn + !02 bn sin(!n t) .
2 n=1
(12.159)
This Fourier series must match the one for f (t). Now, because of the orthogonality of the sin & cos
modes, we can extract from either side the coefficients corresponding to a given sin or cos mode,
which must then match on either side of the equation, mode-by-mode. In the traditional solution
of the problem, this same matching approach is taken, but without the formal justification from
orthogonality.
Similarly, for a complex series of a periodic force, we would write
X X
z(t) = cn exp(i!n t); f (t) = dn exp(i!n t), (12.160)
n n

giving X
z̈ + ż + !02 z = [(!02 ! 2 ) + i !n ]cn exp(i!n t). (12.161)
n

Again, using orthogonality to justify equating coefficients, we would deduce

cn = dn /[(!02 !n2 ) + i !n ], (12.162)

Which is the same relation we found between z̃ and f˜ when we took the direct Fourier Transform
of the di↵erential equation.

59
12.4 Complex impedance
In the previous lecture, we looked at using Fourier transforms to solve the di↵erential equation
for the damped harmonic oscillator. It’s illuminating to reconsider this analysis in the specific
context of the LCR circuit, where LI¨ + RI˙ + I/C = V̇ . Let the voltage be a complex oscillation,
V = Ṽ exp(i!t), so that the current is of the same form, I = I˜ exp(i!t), obeying
1
! 2 LI˜ + i!RI˜ + I˜ = i! Ṽ , (12.163)
C
Thus we see that the circuit obeys a form of Ohm’s law, but involving a complex impedance, Z:

˜ i
Ṽ = Z(!)I; Z(!) = R + i!L . (12.164)
!C
This is a very useful concept, as it immediately allows more complex circuits to be analysed, using
the standard rules for adding resistances in series or in parallel.
The frequency dependence of the impedance means that di↵erent kinds of LCR circuit have functions
as filters of the time-dependent current passing through them: di↵erent Fourier components (i.e.
di↵erent frequencies) can be enhanced or suppressed. For example, consider a resistor and inductor
in series:
˜ Ṽ (!)
I(!) = . (12.165)
R + i!L
For high frequencies, the current tends to zero; for ! ⌧ R/L, the output of the circuit (current
over voltage) tends to the constant value I(!)/V (!) = R. So this would be called a low-pass filter:
it only transmits low-frequency vibrations. Similarly, a resistor and capacitor in series gives

˜ Ṽ
I(!) = 1
. (12.166)
R + (i!C)

This acts as a high-pass filter, removing frequencies below about (RC) 1 . Note that the LR circuit
can also act in this way if we measure the voltage across the inductor, VL , rather than the current
passing through it:
˜ Ṽ Ṽ
ṼL (!) = i!LI(!) = i!L = . (12.167)
R + i!L 1 + R(i!L) 1
1
Finally, a full series LCR circuit is a band-pass filter, which removes frequencies below (RC) and
above R/L from the current.

12.5 Resonance
It is interesting to look at the solution to the damped harmonic oscillator in a bit more detail. If the
forcing term has a very high frequency (! !0 ) then |B| is large and the amplitude is suppressed
– the system cannot respond to being driven much faster than its natural oscillation frequency. In
fact the amplitude is greatest if ! is about !0 (if is small). We can di↵erentiate to show that the
maximum amplitude is reached at the resonant frequency:
q
! = !res = !02 2 /2. (12.168)

2
When is small, this is close to the natural frequency of the oscillator: !res ' !0 /4!02 .

60
The amplitude of the oscillation falls rapidly as we move away from resonance. To see this, write
|B|2 = (!02 ! 2 )2 + 2 ! 2 in terms of ! 2 = !res
2
+ x:

|B|2 = 2 2
!res 4
/4 + x2 . (12.169)

If we write ! = !res + ✏, then x ' 2!res ✏ to lowest order in ✏. If ⌧ !0 , we can now neglect the
di↵erence between !res and !0 , so that the amplitude of oscillation becomes approximated by the
following expression:
1 ( !0 ) 2
' . (12.170)
|B|2 (1 + 4✏2 / 2 )
This is a Lorentzian dependence of the square of the amplitude on frequency deviation from reso-
nance. The width of the resonance is set by the damping: moving a frequency ✏ = /2 away from
resonance halves the squared amplitude.

12.6 Transients
Finally, note that we can always add a solution to the homogeneous equation (i.e. where we set the
right hand side to zero). The final solution will be determined by the initial conditions (z and dz/dt).
This is because the equation is linear and we can superimpose di↵erent solutions. For this additional
solution (also called the complementary function), we try an oscillating solution z / exp(i!t).
Substituting in the damped oscillator di↵erential equation gives the auxiliary equation:

! 2 + i ! + !02 = 0, (12.171)
p
a quadratic equation with the solution ! = i /2 ± 2 /4 + ! 2 . There are two main cases to
0
consider:
p
(1) Underdamped: /2 < !0 . z = e t/2 (Aei⌦t + Be i⌦t ), where ⌦ = 2 /4 + ! 2 .
0
0 0
p
(2) Overdamped: /2 > !0 . z = e t/2 (Ae⌦ t + Be ⌦ t ), where ⌦0 = 2 /4 !02 .
0
So there is only oscillation if the damping is not too high. For very heavy damping, e t/2 Ae⌦ t
yields very nearly a time independent z, as is physically reasonable for an extremely viscous fluid.
But in all cases, the solution damps to zero as t ! 1. Therefore, if the initial conditions require
a component of the homogeneous solution, this only causes an initial transient, and the solution
settles down to the steady-state response, which is what was calculated earlier.
One mathematical complication arises with critical damping: /2 = !0 , so that the two roots
coincide at ⌦ = 0. The simplest way of seeing how to deal with this is to imagine that ⌦ is non-zero
0 0
but very small. Thus Ae⌦ t + Be ⌦ t ' (A + B) + (A B)⌦0 t. So the critically damped solution
is z = e t/2 (C + Dt) (you can check that this does solve the critically-damped equation exactly).
One might also consider generalising the equation so that or !02 are negative. This changes the
physical behaviour and interpretation (e.g. we will now get runaway solutions that increase with
time), but no new algebraic issues arise.
Whatever the form of the complementary function, it presents a problem for the Fourier solution of
the di↵erential equation, especially with a periodic driving force: in general the undriven motion of
the system will not share this periodicity, and hence it cannot be described by a Fourier series. As a
result, the Fourier solution is always zero if the Fourier components of the driving force vanish, even
though this is unphysical: an oscillator displaced from z = 0 will show motion even in the absence
of an applied force. For a proper treatment of this problem, we have to consider the boundary

61
conditions of the problem, which dictate the amount of the homogeneous solution to be added to
yield the complete solution.
When dealing with Fourier transforms, this step may seem inappropriate. The Fourier transform
describes non-periodic functions that stretch over an infinite range of time, so it may seem that
boundary conditions can only be set at t = 1. Physically, we would normally lack any reason for
a displacement in this limit, so the homogeneous solution would tend to be ignored – even though
it should be included as a matter of mathematical principle. In fact, we will eventually see that the
situation is not this simple, and that boundary conditions have a subtle role in yielding the correct
solution even when using Fourier methods.

12.7 Approach using convolution

The solution to di↵erential equations using Fourier methods automatically generates an answer in
the form of a product in Fourier space, so a little thought and a memory of the convolution theorem
shows that the result can be presented in term of a convolution.
Let’s illustrate this point with a simple example that bears some resemblance to the harmonic
oscillator:
d2 z
!02 z = f (t). (12.172)
dt2
Taking the FT gives
f˜(!)
! 2 z̃(!) !02 z̃(!) = f˜(!) ) z̃(!) = 2 , (12.173)
!0 + ! 2
with a solution Z 1 ˜
1 f (!) i!t
z(t) = 2
e d!. (12.174)
2⇡ 1 !0 + ! 2
What this says is that a single oscillating f (t), with amplitude a, will generate a response in
antiphase with the applied oscillation, with amplitude a/(!02 + ! 2 ). For the general case, we
superimpose oscillations of di↵erent frequency, which is what the inverse Fourier transform does for
us.
But now we see that the FT of z(t) is a product (in Fourier space), of f˜(!) and
1
g̃(!) ⌘ (12.175)
!02 + ! 2
hence the solution is a convolution in real (i.e. time) space:
Z 1
z(t) = f (t0 )g(t t0 ) dt0 . (12.176)
1

An exercise for you is to show that the FT of

!0 |t|
e
g(t) = (12.177)
2!0
is g̃(!) = 1/(!02 + ! 2 ), so we finally arrive at the steady solution for a driving force f (t):
Z 1
1 0
z(t) = f (t0 )e !0 |t t | dt0 . (12.178)
2!0 1
!0 |t t0 |
Note how we have used g(t t0 ) = e /2!0 here, as required for a convolution.

62
FOURIER ANALYSIS: LECTURE 15

13 Green’s functions

13.1 Response to an impulse

We have spent some time so far in applying Fourier methods to solution of di↵erential equations
such as the damped oscillator. These equations are all in the form of
Ly(t) = f (t), (13.179)
where L is a linear di↵erential operator. For the damped harmonic oscillator, L = (d2 /dt2 +
d/dt + !02 ). As we know, linearity is an important property because it allows superposition:
L(y1 + y2 ) = Ly1 + Ly2 . It is this property that lets us solve equations in general by the method of
particular integral plus complementary function: guess a solution that works for the given driving
term on the RHS, and then add any solution of the homogeneous equation Ly = 0; this is just
adding zero to each side, so the sum of the old and new y functions still solves the original equation.
In this part of the course, we focus on a very powerful technique for finding the solution to such
problems by considering a very simple form for the RHS: an impulse, where the force is concentrated
at a particular instant. A good example would be striking a bell with a hammer: the subsequent
ringing is the solution to the equation of motion. This impulse response function is also called a
Green’s function after George Green, who invented it in 1828 (note the apostrophe: this is not a
Green function). We have to specify the time at which we apply the impulse, T , so the applied
force is a delta-function centred at that time, and the Green’s function solves
LG(t, T ) = (t T ). (13.180)
Notice that the Green’s function is a function of t and of T separately, although in simple cases it
is also just a function of t T .
This may sound like a peculiar thing to do, but the Green’s function is everywhere in physics. An
example where we can use it without realising is in electrostatics, where the electrostatic potential
satisfies Poisson’s equation:
r2 = ⇢/✏0 , (13.181)
where ⇢ is the charge density. What is the Green’s function of this equation? It is the potential
due to a charge of value ✏0 at position vector q:
1
G(r, q) = . (13.182)
4⇡|r q|

13.2 Superimposing impulses

The reason it is so useful to know the Green’s function is that a general RHS can be thought of as
a superposition of impulses, just as a general charge density arises from summing individual point
charges. We have seen this viewpoint before in interpreting the sifting property of the delta-function:
Z
f (x) = f (q) (x q) dq. (13.183)

63
To repeat, we normally tend to think of this as involving a single spike located at q = x, which pulls
out the value of f at the location of this spike. But we can flip the viewpoint and think of (x q)
as specifying a spike at x = q, where now the integral covers all values of q: spikes everywhere.
Alternatively, consider the analogy with the inverse Fourier transform:
Z
f (x) = f˜(k)/2⇡ exp(ikx), dk. (13.184)

Here, we have basis functions exp(ikx), which we think of as functions of x with k as a parameter,
with expansion coefficients f˜(k)/2⇡. From this point of view, the sifting relation uses (x q) as
the basis function, with q as the parameter specifying where the spike is centred.
So if f (x) is a superposition of spikes, we only need to understand the response of a linear system to
one spike and then superposition of responses will give the general solution. To show this explicitly,
take LG(t, T ) = (t T ) and multiply both sides by f (T ) (which is a constant). But now integrate
both sides over T , noting that L can be taken outside the integral because it doesn’t depend on T :
Z Z
L G(t, T )f (T ) dT = (t T )f (T ) dT = f (t). (13.185)

The last step uses sifting to show that indeed adding up a set of impulses on the RHS, centred at
di↵ering values of T , has given us f (t). Therefore, the general solution is a superposition of the
di↵erent Green’s functions: Z
y(t) = G(t, T )f (T ) dT. (13.186)

This says that we apply a force f (T ) at time T , and the Green’s function tells us how to propagate
its e↵ect to some other time t (so the Green’s function is also known as a propagator).

13.2.1 Importance of boundary conditions

When solving di↵erential equations, the solution is not unique until we have applied some boundary
conditions. This means that the Green’s function that solves LG(t, T ) = (t T ) also depends
on the boundary conditions. This shows the importance of having boundary conditions that are
homogeneous: in the form of some linear constraint(s) being zero, such as y(a) = y(b) = 0, or y(a) =
ẏ(b) = 0. If such conditions apply to G(t, T ), then a solution that superimposes G(t, T ) for di↵erent
values of T will still satisfy the boundary condition. This would not be so for y(a) = y(b) = 1,
and the problem would have to be manipulated into one for which the boundary conditions were
homogeneous – by writing a di↵erential equation for z ⌘ y 1 in that case.

13.3 Finding the Green’s function

The above method is general, but to find the Green’s function it is easier to restrict the form of the
di↵erential equation. To emphasise that the method is not restricted to dependence on time, now
consider a spatial second-order di↵erential equation of the general form

d2 y dy
2
+ a1 (x) + a0 (x)y(x) = f (x). (13.187)
dx dx
Now, if we can solve for the complementary function (i.e. solve the equation for zero RHS), the
Green’s function can be obtained immediately. This is because a delta function vanishes almost

64
everywhere. So if we now put f (x) ! (x z), then the solution we seek is a solution of the
homogeneous equation everywhere except at x = z.
We split the range into two, x < z, and x > z. In each part, the r.h.s. is zero, so we need to solve
the homogeneous equation, subject to the boundary conditions at the edges. At x = z, we have to
be careful to match the solutions together. The function is infinite here, which tells us that the
first derivative must be discontinuous, so when we take the second derivative, it diverges. The first
derivative must change discontinuously by 1. To see this, integrate the equation between z ✏ and
z + ✏, and let ✏ ! 0:
Z z+✏ 2 Z z+✏ Z z+✏ Z z+✏
dy dy
2
dx + a1 (x) dx + a0 (x)dx = (x z)dx. (13.188)
z ✏ dx z ✏ dx z ✏ z ✏

The second and third terms vanish as ✏ ! 0, as the integrands are finite, and the r.h.s. integrates
to 1, so
dy dy
= 1. (13.189)
dx z+✏ dx z ✏

13.3.1 Example

Consider the di↵erential equation

d2 y
+y =x (13.190)
dx2
with boundary conditions y(0) = y(⇡/2) = 0.
The Green’s function is continuous at x = z, has a discontinuous derivative there, and satisfies the
same boundary conditions as y. From the properties of the Dirac delta function, except at x = z,
the Green’s function satisfies
d2 G(x, z)
+ G(x, z) = 0. (13.191)
dx2
(Strictly, we might want to make this a partial derivative, at fixed z. It is written this way so it
looks like the equation for y). This is a harmonic equation, with solution
⇢
A(z) sin x + B(z) cos x x < z
G(x, z) = (13.192)
C(z) sin x + D(z) cos x x > z.

We now have to adjust the four unknowns A, B, C, D to match the boundary conditions.
The boundary condition y = 0 at x = 0 means that B(z) = 0, and y = 0 at x = ⇡/2 implies that
C(z) = 0. Hence ⇢
A(z) sin x x < z
G(x, z) = (13.193)
D(z) cos x x > z.
Continuity of G implies that A(z) sin z = D(z) cos z and a discontinuity of 1 in the derivative implies
that D(z) sin z A(z) cos z = 1. We have 2 equations in two unknowns, so we can eliminate A or
D:
sin2 z cos z
A(z) A(z) cos z = 1 ) A(z) = 2 = cos z (13.194)
cos z sin z + cos2 z
and consequently D(z) = sin z. Hence the Green’s function is
⇢
cos z sin x x < z
G(x, z) = (13.195)
sin z cos x x > z

65
The solution for a driving term x on the r.h.s. is therefore (be careful here with which solution for
G to use: the first integral on the r.h.s. has x > z)
Z ⇡/2 Z x Z ⇡/2
y(x) = z G(x, z) dz = cos x z sin z dz sin x z cos z dz. (13.196)
0 0 x

Integrating by parts,
1 ⇡
y(x) = (x cos x sin x) cos x (⇡ 2 cos x 2x sin x) sin x = x sin x. (13.197)
2 2

13.4 Summary
So to recap, the procedure is to find the Green’s function by

• replacing the driving term by a Dirac delta function

• solving the homogeneous equation either side of the impulse, with the same boundary condi-
tions e.g. G = 0 at two boundaries, or G = @G/@x = 0 at one boundary.

• Note the form of the solution will be the same for (e.g.) x < z and x > z, but the coefficients
(strictly, they are not constant coefficients, but rather functions of z) will di↵er either side of
x = z).

• matching the solutions at x = z (so G(x, z) is continuous).

• introducing a discontinuity of 1 in the first derivative @G(x, z)/@x at x = z

• integrating the Green’s function with the actual driving term to get the full solution.

66
FOURIER ANALYSIS: LECTURE 16

13.5 Example with boundary conditions at the same place/time

A mouse wishes to steal a piece of cheese from an ice-rink at the winter olympics. The cheese,
which has a mass of 1 kg, is conveniently sitting on a frictionless luge of negligible mass. The mouse
attaches a massless string and pulls, starting at t = 0. Unfortunately the mouse gets tired very
quickly, so the force exerted declines rapidly f (t) = e t . Find, using Green’s functions, the resulting
motion of the cheese, z(t) and its terminal speed.
The equation to be solved is
d2 z
= e t. (13.198)
dt2
Since the cheese is ‘sitting’ on the luge, we take the boundary conditions to be
dz
z = 0; = 0 at t = 0. (13.199)
dt
We can, of course, solve this equation very easily simply by integrating twice, and applying the
boundary conditions. As an exercise, we are going to solve it with Green’s functions. This also
makes the point that there is often more than one way to solve a problem.
For an impulse at T , the Green’s function satisfies
@G(t, T )
= (t T) (13.200)
@t2
so for t < T and t > T the equation to be solved is @ 2 G/@t2 = 0, which has solution
⇢
A(T )t + B(T ) t < T
G(t, T ) = (13.201)
C(T )t + D(T ) t > T

Now, we apply the same boundary conditions. G(t = 0) = 0 ) B = 0. The derivative G0 (t = 0) =

0 ) A = 0, so G(t, T ) = 0 for t < T . This makes sense when one thinks about it. We are applying
an impulse at time T , so until the impulse is delivered, the cheese remains at rest.
Continuity of G at t = T implies
C(T )T + D(T ) = 0, (13.202)
and a discontinuity of 1 in the derivative at T implies that

C(T ) A(T ) = 1. (13.203)

Hence C = 1 and D = T and the Green’s function is

⇢
0 t<T
G(t, T ) = (13.204)
t T t>T

The full solution is then Z 1

z(t) = G(t, T )f (T )dT (13.205)
0

67
 T
where f (T ) = e . Hence
Z t Z 1
z(t) = G(t, T )f (T )dT + G(t, T )f (T )dT. (13.206)
0 t

The second integral vanishes, because G = 0 for t < T , so

Z t ⇢ Z t
T
⇥ ⇤
T t
⇥ ⇤
T t T
z(t) = (t T )e dT = t e 0
Te 0
+ e dT (13.207)
0 0

which gives the motion as

z(t) = t 1 + e t. (13.208)
We can check that z(0) = 0, that z 0 (0) = 0, and that z 00 (t) = e t . The final speed is z 0 (t ! 1) = 1.
Note that this technique can solve for an arbitrary driving term, obtaining the solution as an
integral. This can be very useful, even if the integral cannot be done analytically, as a numerical
solution may still be useful.

13.6 Causality
The above examples showed how the boundary conditions influence the Green’s function. If we are
thinking about di↵erential equations in time, there will often be a di↵erent boundary condition,
which is set by causality. For example, write the first equation we considered in a form that
emphasises that it is a harmonic oscillator:

G̈(t, T ) + !02 G(t, T ) = (t T ). (13.209)

Since the system clearly cannot respond before it is hit, the boundary condition for such applications
would be expected on physical grounds to be

G(t, T ) = 0 (t < T ). (13.210)

Whether or not such behaviour is achieved depends on the boundary conditions. Our first example
did not satisfy this criterion, because the boundary conditions were of the form y(a) = y(b) = 0.
This clearly presents a problem if T is between the points a and b: it’s as if the system knows
when we will strike the bell, or how hard, in order that the response as some future time t = b
will vanish. In contrast, our second example with boundary conditions at a single point ended up
yielding causal behaviour automatically, without having to put it in by hand.
The causal Green’s function is particularly easy to find, because we only need to think about the
behaviour at t > T . Here, the solution of the homogeneous equation is A sin !0 t + B cos !0 t, which
must vanish at t = T . Therefore it can be written as G(t, T ) = A sin[!0 (t T )]. The derivative
must be unity at t = T , so the causal Green’s function for the undamped harmonic oscillator is
1
G(t, T ) = sin[!0 (t T )]. (13.211)
!0

13.6.1 Comparison with direct Fourier solution

As a further example, we can revisit again the di↵erential equation with the opposite sign from the
oscillator:
d2 z
!02 z = f (t). (13.212)
dt2
68
We solved this above by taking the Fourier transform of each side, to obtain
Z
1 1
f˜(!) i!t
z(t) = 2
e d!. (13.213)
2⇡ 1 !0 + !2

We then showed that this is in the form of a convolution:

Z 1
1
z(t) = f (T )e !0 |t T|
dT. (13.214)
2!0 1

This looks rather similar to the solution in terms of the Green’s function, so can we say that
G(t, T ) = exp( !0 |t T |)/2!0 ? Direct di↵erentiation gives Ġ = ± exp( !0 |t T |)/2, with the +
sign for t > T and the sign for t < T , so it has the correct jump in derivative and hence satisfies
the equation for the Green’s function.
But this is a rather strange expression, since it is symmetric in time: a response at t can precede
T . The problem is that we have imposed no boundary conditions. If we insist on causality, then
G = 0 for t < T and G = A exp[!0 (t T )] + B exp[ !0 (t T )] for t > T . Clearly A = B, so
G = 2A sinh[!0 (t T )]. This now looks similar to the harmonic oscillator, and a unit step in Ġ at
t = T requires
1
G(t, T ) = sinh[!0 (t T )]. (13.215)
!0
So the correct solution for this problem will be
Z t
1
z(t) = f (T ) sinh[ !0 (t T )] dT. (13.216)
!0 1

Note the changed upper limit in the integral: forces applied in the future cannot a↵ect the solution
at time t. We see that the response, z(t), will diverge as t increases, which is physically reasonable:
the system has homogeneous modes that either grow or decline exponentially with time. Special
care with boundary conditions would be needed if we wanted to excite only the decaying solution
– in other words, this system is unstable.

69