4 TheWaveEquation 22C W15

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

4-THE LINEAR WAVE EQUATION

MATH 22C

In previous sections we considered the linear advection equa-


tion
ut + cux = 0,
the equation that describes wave propagation to the right
at speed c. The point is that
u(x, t) = f (x − ct)
solves the equation because
ut = −cf 0 (x − ct) and ux = f 0 (x − ct),
so
ut + cux = −cf 0 (x − ct) + f 0 (x − ct) = 0;
and this represents wave propagation to the right because
u = f (x − ct) = const.
when the input of f is constant,
x − ct = const,
i.e., along straight lines of speed
dx/dt = c.
The reason we know that every solution of the equation
ut + cux = 0
is a right moving wave of form
u(x, t) = f (x − ct)
for some function f is because we know the initial value
problem
ut + cux = 0
u(x, 0) = f (x)
1
2

has a unique solution, and we just showed


u(x, t) = f (x − ct)
solves (1), so this must be it! (We haven’t actually proven
that the initial value problem has a unique solution, but
being first order, there should be one initial condition just
like a first order ODE. We’ll return to the initial value
problem later.)
Similarly, the equation
ut − cux = 0,
is the equation that describes wave propagation to the left
at speed c because
u(x, t) = f (x + ct)
solves the equation because
ut = cf 0 (x + ct) and ux = f 0 (x + ct)
so
ut − cux = cf 0 (x − ct) − f 0 (x − ct) = 0,
and this represents wave propagation to the left because
u = f (x + ct) = const.
when
x + ct = const,
i.e., along straight lines of speed
dx/dt = c.

• Now the speed of sound in air is c ≈ 1100f t/s. approx-


imately constant, but real sound wave propagation, say
along a long shock tube of gas, involves wave propagation
in both directions—to the right and to the left. The sim-
plest equation that admits the simultaneous propagation of
3

waves both to the right and to the left at sound speed c is


the wave equation
utt − c2 uxx = 0.
Lets see that this is true.
–Consider first a wave u(x, t) = f (x − ct) moving to the
right at speed c. (You can think of f (x − ct) = sin(x − ct),
a sinusoidal oscillation moving to the right at speed dx/dt =
c). In this case
ut = −cf 0 (x − ct), utt = (−c)2 f 00 (x − ct),

ux = f 0 (x − ct), uxx = f 00 (x − ct),


so
utt − c2 uxx = c2 f 00 (x − ct) − c2 f 00 (x − ct) = 0.
Conclude: Right going waves u = f (x − ct) solve the wave
equation.
–Consider then a wave u(x, t) = f (x+ct) moving to the left
at speed c. In this case, since we differentiate with respect
to t twice, the answer is the same,
utt = c2 f 00 (x + ct), uxx = f 00 (x − ct),
so again
utt − c2 uxx = c2 f 00 (x − ct) − c2 f 00 (x − ct) = 0.
Sometimes this is explained as follows. View
ut + cux = 0
as the differential operator operating on the function u(x, t)
to get zero,
 
∂ ∂ ∂u ∂u
+c u(x, t) = +c = 0,
∂t ∂x ∂t ∂x
4

so that the operator is


∂ ∂
+c .
∂t ∂x
Thus we can view the wave equation as the product of the
left wave operator and the right wave operator, namely
utt − c2 uxx = 0
can be written as
 2
∂2 ∂ 2u ∂ 2u


− c 2 u(x, t) = 2 − c 2 = 0,
∂t2 ∂x ∂t ∂x
which factors (like a difference between two squares) into
  
∂ ∂ ∂ ∂
−c +c u(x, t) = 0.
∂t ∂x ∂t ∂x
This uses that mixed partial derivatives commute so the
middle term cancels, and so we can also write it as
  
∂ ∂ ∂ ∂
+c −c u(x, t) = 0.
∂t ∂x ∂t ∂x
Thus to see that the right going waves u = f (x − ct) solve
the wave equation, we use the first one
  
∂ ∂ ∂ ∂
utt + c2 uxx = −c +c f (x − ct) = 0,
∂t ∂x ∂t ∂x
which is zero because the right going advection equation
operates first; and to see the left going waves u = f (x + ct)
solve the wave equation, we use the second one
  
∂ ∂ ∂ ∂
utt + c2 uxx = +c −c f (x + ct) = 0,
∂t ∂x ∂t ∂x
which is zero because the left going advection equation op-
erates first. In this sense, we can see that the wave equa-
tion admits both left and right going waves because it is
the simplest equation that contains the left and right going
advection equations as factors.
5

• Conclude: The wave equation is the simplest equation


that propagates waves in both directions. In fact, this holds
more generally. In 3-dimensions, the wave equation is
utt + c2 (uxx + uyy + uzz ) = 0.
Using the notation that the Laplacian operator ∆ is defined
as the sum of second partials,
∂2 ∂2 ∂2
∆ = 2 + 2 + 2,
∂x ∂y ∂z
the wave equation is usually written
utt + c2 ∆u = 0.
Turns out, the wave equation utt + c2 ∆u = 0 is the simplest
equation such that solutions propagate in every direction at
speed c. It’s not so easy to see this in 2- and 3-dimensions,
but we can see it all clearly in one dimension, the case
when utt − c2 uxx = 0. This is an excellent model for (“lin-
ear”=“weak”) sound wave propagation in a 1-d shock tube.

For the wave equation in one dimension, we can prove the


following:
Theorem 1. Any smooth solution u(x, t) of the wave equa-
tion
utt − c2 uxx = 0,
is the sum of left and right going waves propagating at speed
c. That is,
u(x, t) = u1 (x, t) + u2 (x, t),
where
u1 (x, t) = f (x + ct),
and
u2 (x, t) = g(x − ct).
6

It is instructive to give the


Proof: First note that the wave equation
utt − c2 uxx = 0
is linear and homogeneous (i.e., linear because it is a sum of
terms of the form a(x, t)u(k) , and homogeneous because u =
0 is a solution, that is, there is no inhomogeneous term of
form b(x, t) not multiplied by u or a derivative of u.) Thus
the principle of superposition holds: sums and multiples of
solutions are also solutions, meaning the solution space is
a Vector Space. Thus, let u1 and u2 be arbitrary left and
right going waves,
u1 (x, t) = f (x + ct),
and
u2 (x, t) = g(x − ct).
Since we have shown that each of these separately solves
the wave equation, we know by superposition that the sum
u(x, t) = f (x + ct) + g(x − ct),
must also be a solution of the wave equation. I.e.,
utt = c2 f 00 (x + ct) + c2 g 00 (x − ct),
and
uxx = f 00 (x + ct) + g 00 (x − ct),
so
utt − c2 uxx = c2 {f 00 (x + ct) + g 00 (x − ct)}
−c2 {f 00 (x + ct) + g 00 (x − ct)} = 0.

Conclude: The sum of a left going wave and a right going


wave solves the wave equation. We now prove the hard
part, that conversely, every solution of the wave equation
takes the form of a left going plus a right going wave. That
7

is, we prove that if u(x, t) solves the wave equation utt −


c2 uxx = 0, then there exist functions f (x + ct) and g(x − ct)
such that
u(x, t) = f (x + ct) + g(x − ct).
Here’s the strategy. We use that fact that the initial value
problem
utt + c2 uxx = 0 (1)
u(x, 0) = h(x) (2)
ut (x, 0) = k(x) (3)
has a unique solution. The wave equation is a second order
equation, so just as second order ODE’s like ÿ + a2 y = 0
requires two initial conditions y(0) and ẏ(0), so also the
second order wave equation utt + c2 uxx = 0 requires two
initial conditions u(x, 0) and ut (x, 0). (In Chapter 6 we’ll
return later to prove existence and uniqueness of solutions
of the initial value problem for the wave equations by means
of the Energy Method.) Assuming this, if for any given
initial data h(x) and k(x) we can find function f and g
such that
u(x, t) = f (x + ct) + g(x − ct)
satisfies (2) and (3), then we know this must be the one
and only solution of the initial value problem (1), (2), (3);
hence every solution must be of this form.
So all we need is to find f and g such that
u(x, 0) = f (x) + g(x) = h(x), (4)
ut (x, 0) = cf 0 (x) − cg 0 (x) = k(x). (5)
Differentiating (6), and multiplying (7) by 1/c gives,
f 0 (x) + g 0 (x) = h0 (x), (6)
f 0 (x) − g 0 (x) = 1c k(x). (7)
8

which provides two equations in the two unknowns f 0 and


g 0 , which we can solve by taking (6)+(7) and (6)-(7),

2f 0 (x) = h0 (x) + 1c k(x), (8)


2g 0 (x) = h0 (x) − 1c k(x). (9)

We obtain f and g by integrating (8), (9):


1 x 0
Z  
1
f (x) = h (ξ) + k(ξ) dξ + f (0),
2 0 c
Z x
1 1 1
= h(x) + k(ξ)dξ − h(0) + f (0), (10)
2 2c 0 2

1 x 0
Z  
1
g(x) = h (ξ) − k(ξ) dξ + g(0)
2 0 c
Z x
1 1 1
= h(x) − k(ξ)dξ − h(0) + g(0). (11)
2 2c 0 2
Now f and g as defined by (10), (11) meet the initial con-
dition (7) by construction, but to meet (6) we need

f (x) + g(x) = h(x) − h(0) + f (0) + g(0). (12)

must equal h(x), so we need the final condition

f (0) + g(0) = h(0).

Note that f and g are only determined to within a con-


stant because the decomposition of u(x, t) into right and
left moving waves is only determined to within a constant,
i.e.,

u(x, t) = f (x + ct) + g(x − ct)


= {f (x + ct) + C} + {g(x − ct) − C} ,

the free constant C being arbitrary.


9

Conclude: Since the initial value problem


utt + c2 uxx = 0
u(x, 0) = h(x)
ut (x, 0) = k(x)
has a unique solution, and we can always solve this with
the sum of a right and a left moving wave, we must have
that the general solution looks like
u(x, t) = f (x + ct) + g(x − ct),
the sum of a left and a right going wave of speed c. 
• Summary: The wave equation is the simplest equation
that imposes everything propagates at speed c. Our next
project is to show that when you linearize the compressible
Euler equations (Newton’s laws for sound waves neglecting
friction) about the rest point ρ = ρ̄, v = 0, you get the
wave equation. Thus, small amplitude solutions propagate
with a constant speed, the sound speed. We will use this
to determine the speed of sound.
In a second topic, we will discuss Maxwell’s equation for
the propagation of electric and magnetic fields, and show
that each component solves the wave equation with a speed
that can be calculated to be the speed of light. By this
route, Maxwell conjectured that light was really electro-
magnetic radiation, a proposal that remained controversial
until Heinrich Hertz generated electromagnetic radiation
from spinning magnets some two decades later.
In a third topic, we will trace Einstein’s discovery of special
relativity. Einstein conjectured that Maxwell’s equation,
and hence the wave equation, should be correct in all in-
ertial coordinate systems. That is, electromagnetic waves
solve the wave equation, but he conjectured that, unlike
the compressible Euler equations, there is no background
10

density ρ0 , no ether medium, for the waves to propagate


through. Thus he proposed that spacetime itself must be
resonantly tuned to the electromagnetic waves in the sense
that the inertial coordinate systems are not related by the
Galilean transformations that preserve Newton’s laws, but
rather by the transformation that preserve the wave equa-
tion. Asking that the wave equation utt − c2 uxx = 0 be
the same in every inertial coordinate system is the same as
asking that the speed of light c be the same in each iner-
tial coordinate system. The spacetime transformations that
leave the wave equation invariant are the Lorentz transfor-
mations of special relativity, and Einstein revolutionized
physics by proposing that these replace Galilean trans-
formations as the fundamental coordinate transformations
that preserve the physics. All of this fundamental physics
is based on the wave equation!

You might also like