Mathematical Methods in Chemistry (MATH0042) Page 44
2.4 The wave equation
The wave equation is another PDE of fundamental importance in mathematics,
physics, chemistry and the natural sciences in general. It describes the propagation
of waves (this includes mechanical waves - such as water waves or sound waves - as
well as electromagnetic waves). It is stated as
∂2
f (x, y, z, t) = c2 ∆f (x, y, z, t), (2.102)
∂t2
where f (x, y, z, t) is a function of space (x, y, z) and time t that generally represents
a displacement (e.g., the displacement of a point on a string from its resting position
as a wave passes along the string) and c is a constant that represents the propagation
speed of the wave.
2.4.1 1D wave equation: a vibrating string
To begin, we will consider the 1D wave equation for a function f (x, t) that depends
only on a single spatial variable, x. In particular, let us consider a wave travelling
along a piece of elastic string that lies in the (x, y)-plane with its ends fixed at the
points (0, 0) and (0, L) on the x-axis. Let f (x, t) denote the vertical displacement
(i.e., y-coordinate) at a time t of a point on the string that has horizontal coordinate
x. See Figure 2.3. (We assume that there is no horizontal displacement of any point
on the string.)
(x, f (x, t))
f (x, t)
0 L x
string
Figure 2.3: A vibrating piece of string.
One can show that f (x, t) satisfies the 1D wave equation:
∂ 2 f (x, t) 2
2 ∂ f (x, t)
= c . (2.103)
∂t2 ∂x2
Page 45
The fact that the ends of the string are fixed results in the boundary conditions
f (0, t) = f (L, t) = 0 for all t. (2.104)
Let us solve (2.103) - subject to (2.104) - for f (x, t) by the method of separation
of variables. Thus, we seek a solution of the form
f (x, t) = g(x)h(t), (2.105)
for some functions g(x) and h(t) to be determined. Substituting (2.105) into (2.103)
leads to
c2 h(t)g ′′ (x) = g(x)h′′ (t), (2.106)
which we can rearrange as
g ′′ (x) 1 h′′ (t)
= 2 . (2.107)
g(x) c h(t)
By our usual considerations, we may deduce that
g ′′ (x) 1 h′′ (t)
= 2 = γ. (2.108)
g(x) c h(t)
for some constant γ. (2.108) gives two ODEs: one for g(x) and one for h(t).
For g(x) we have
g ′′ (x) − γg(x) = 0. (2.109)
Furthermore, it follows from the boundary conditions (2.104) that we must have
g(0) = g(L) = 0. (2.110)
(2.109) is another equation of the type that we considered in section 1.1.2. The
nature of its solutions will differ depending on whether γ is zero, positive or negative.
We dismiss the possibility that γ = 0 since in this case g(x) = A0 x + B0 for some
constants A and B, and this cannot satisfy the boundary conditions (2.110) unless
A0 = B0 = 0 (a linear function of x only has one zero), and so no non-trivial
solutions for g(x) exist. If γ > 0, then
√ √
g(x) = Ae γx
+ Be− γx
, (2.111)
for some constants A and B. In this case, in order to satisfy the boundary conditions
(2.110) we must have
√ √
A + B = 0 and Ae γL
+ Be− γL
= 0. (2.112)
Equivalently, we must have
A = −B (2.113)
and √ √
Ae− γL
(e2 γL
− 1) = 0. (2.114)
√ √
2 γL − γL
But e − 1 ̸= 0 since γ, L ̸= 0, and e ̸= 0 (ex vanishes only as x → −∞),
hence we must have A = 0 and thus also B = 0. Hence no non-trivial solutions for
g(x) exist for γ > 0 either, and so we dismiss this possibility as well.
Mathematical Methods in Chemistry (MATH0042) Page 46
Finally, suppose that γ < 0, say γ = −λ2 for some constant λ. It follows that
g(x) = Aλ cos(λx) + Bλ sin(λx), (2.115)
for some constants Aλ and Bλ . Now, in order to satisfy the boundary conditions
(2.110), one may deduce that Aλ = 0 (so that g(0) = 0) and that λ = nπ L
for
2
n ∈ Z, ≥ 1 (so that then g(L) = 0 as well). Thus, provided γ = −(nπ/L) , where
n ∈ Z, ≥ 1, we have the following solution for g(x):
nπ
gn (x) = bn sin x , (2.116)
L
where bn is an arbitrary constant.
Then, our ODE for h(t) becomes:
cnπ 2
′′
h (t) + h(t) = 0, (2.117)
L
(where n ∈ Z, ≥ 1). The general solution of this is
cnπ cnπ
hn (t) = Cn cos t + Dn sin t , (2.118)
L L
for arbitrary constants Cn and Dn .
Hence, combining the above solutions for g(x) and h(t), we get the following
form for the solution for f (x, t):
∞
X cnπt cnπt nπx
f (x, t) = cn cos + dn sin sin , (2.119)
n=1
L L L
for arbitrary constants cn and dn .
2.4.2 2D wave equation: a circular drum
Next, we will consider the 2D wave equation. Consider a disc-shaped elastic mem-
brane. Suppose that the edge of the membrane is fixed along the circle in the
(x, y)-plane that is centred on the origin and of radius a. It is most convenient to
use plane polar coordinates (r, θ). Now f = f (r, θ, t) represents the vertical dis-
placement at time t of a point on the membrane with polar coordinates (r, θ). (We
assume that there is no horizontal displacement of any point on the membrane.) It
can be shown that f = f (r, θ, t) satisfies the 2D wave equation:
∂ 2 f (r, θ, t)
2
= c2 ∆f (r, θ, t). (2.120)
∂t
Recall that the Laplacian in terms of plane polar coordinates is given by
∂2 1 ∂ 1 ∂2
∆= + + . (2.121)
∂r2 r ∂r r2 ∂θ2
Using separation of variables, let us seek a solution for f (r, θ, t) of the form
f (r, θ, t) = u(r, θ)T (t) (2.122)
Page 47
for some functions T (t) and u(r, θ), to be determined. Note that u(r, θ) must satisfy
the boundary conditions
u(a, θ) = 0 and u(r, θ + 2π) = u(r, θ) for all θ ∈ [0, 2π). (2.123)
Substituting (2.122) into (2.120) gives
c2 T (t) ∆u(r, θ) = u(r, θ) T ′′ (t), (2.124)
which we can rearrange as
∆u(r, θ) T ′′ (t)
c2 = . (2.125)
u(r, θ) T (t)
Using the usual arguments, one may deduce that
∆u(r, θ) T ′′ (t)
c2 = = γ, (2.126)
u(r, θ) T (t)
for some constant γ. (2.126) provides two differential equations: an ODE for T (t)
and a PDE for u(r, θ).
For T (t) we have
T ′′ (t) − γT (t) = 0. (2.127)
We will seek solutions for T (t) that remain finite for all t and are also periodic (in
t). We thus dismiss the possibility that γ ≥ 0. (e.g., compare our analysis of the
possible solutions for g(x) to (2.109)). Thus we consider just γ < 0, say γ = −ω 2
for some constant ω. Then (2.127) becomes
T ′′ (t) + ω 2 T (t) = 0. (2.128)
This has solution
T (t) = A cos(ωt) + B sin(ωt), (2.129)
for arbitrary constants A and B. Note that, by using trigonometric identities, this
can also be written as
T (t) = C cos(ωt + δ), (2.130)
for arbitrary constants C and δ.
Now, for u(r, θ), introducing λ = ω/c, we have
∆u = −λ2 u. (2.131)
Equation (2.131) is called the Helmholtz equation. (The Helmholtz equation is an
eigenvalue problem for the Laplacian operator. In this case, −λ2 is the eigenvalue.
Solutions of this equation are called eigenfunctions of the Laplacian operator.)
To solve (2.131) we once again use separation of variables. We seek a solution
for u(r, θ) of the form
u(r, θ) = R(r) Θ(θ). (2.132)
Substituting (2.132) into (2.131) (recalling (2.121)) gives
d2 R 1 dR R d2 Θ
Θ + Θ + 2 2 + λ2 RΘ = 0, (2.133)
dr2 r dr r dθ
Mathematical Methods in Chemistry (MATH0042) Page 48
which can be rearranged into the form
1 d2 R 1 dR 1 d2 Θ
r2 2
+ r + λ 2 2
r = − 2
= ν 2. (2.134)
R dr R dr Θ dθ
Here, we fix the separation constant (ν 2 ) to be positive in order that Θ(θ) is periodic.
Indeed, the ODE for Θ(θ) is then
Θ′′ (θ) + ν 2 Θ(θ) = 0, (2.135)
which has solution
Θ(θ) = D cos(νθ) + E sin(νθ), (2.136)
for arbitrary constants D and E.
The equation for R(r) is the second order ODE with non-constant coefficients
r2 ′′ r ′
R + R + r 2 λ2 = ν 2 , (2.137)
R R
which can be rearranged into the form
r2 R′′ (r) + rR′ (r) + (λ2 r2 − ν 2 )R(r) = 0. (2.138)
Introducing x = λr and y(x) = R(r), (2.138) can be rewritten as
x2 y ′′ (x) + xy ′ (x) + (x2 − ν 2 )y(x) = 0. (2.139)
(2.139) is known as Bessel’s equation. Bessel’s equation is so important in applied
mathematics that we will devote the next section to its detailed study.
