2018 Wave Equation IMSC

The Wave Equation
A Simple Presentation for Teachers of Engineering

Colleges
Phoolan Prasad
DEPARTMENT OF MATHEMATICS
INDIAN INSTITUTE OF SCIENCE, BANGALORE
Teachers Enrichment Workshop

for teachers of engineering colleges
November 26 - December 1, 2018

Institute of Mathematical Sciences, Chennai
(Sponsored by National Centre for Mathematics and IMSc)

Syllabus for Teaching of PDE
I have noticed that participants to this workshop are also teachers
teaching mathematics at M.Sc. level. There are also some research
students.
What should be suitable syllabus for teaching differential
equations?
There was a very detailed discussion on this in two discussion
meeting, sponsored by Indian Academy of Sciences and RMS in
February, 2018.
Report, syllabus and some lectures are available at
http://web-japps.ias.ac.in:8080/SEP/pdffiles/discussion meeting.pdf
33 professors from IISERS, IISc, IITs, Universities and
TIFR-CAM took part in the two meetings, spread over 6 days.
There were also lectures on some equations of mathematical
physics as mathematicians and as physicists see them.
The Wave Equation P. Prasad Department of Mathematics 2 / 48

Wave Equation
When we talk of waves, in general we think of waves goverened by

The Wave Equation.
But most of the waves are not governed by “the wave equation”.
We have already seen this in the lecture on “Single Conservation

Law”.
We not mentioned one of a very important wave equation

“Korteweg de Vries (KdV) Equation”, which led to some of the
most important developments in mathematics and physics in the
last century (with some finest applications to engineering).

KdV Equation and Solitons
KdV equation led to the concept of soliton.
See a popular, very simple and beatiful article by Alex Kasman

“A brief history of solitons and the KdV equation”
Current Science, Volume 115 - Issue 8 : 25 October 2018:
http://www.currentscience.ac.in/Volumes/115/08/1486.pdf
See also a review of Kasman’s book by PP in a forthcoming issue of

Current Science.

Simplest wave equation - A Part of the Wave
Equation Waves
ut + c ux = 0 , c = real constant (1)

Method of characteristics for first order PDE gives
u = f (x − ct), f : R → R (2)
When f ∈ C 1 (R) ⇒ Genuine solution
ct
t=0 t>0
Figure: Wave of translation. This figure does not reprent a genuine solution.
Why?

The One-dimensional Wave Equation
utt − c2 uxx = 0, c = constant > 0 (3)

∂ ∂ ∂ ∂
⇒ +c −c u=0
∂t ∂x ∂t ∂x
LHS has two units of simplest wave operators.
Set
ξ = x − ct, η = x + ct
∂2u (4)
(15) ⇒ =0
∂ξ∂η
Shows general solution (genuine solution)
u = f (x − ct) + g(x + ct) ; f, g ∈ C 2 (R) (5)
⇒ two waves of translations, one moves with velocity c and another

with velocity −c.
The One-dimensional Wave Equation contd..
The curves
x − ct = constant, x + ct = constant
are called characteristic curves
In any problem in physics, a PDE comes with additional

conditions to completely determine the solution.
Cauchy Problem - here as an initial value problem.
utt − c2 uxx = 0, (x, t) ∈ R2 (6)
u(x, 0) = u0 (x), ut (x, 0) = u1 (x) (7)

For a genuine solution, we need
u0 (x) ∈ C 2 (R), u1 (x) ∈ C 1 (R) (8)
Substitute u = f (x − ct) + g(x + ct) in (7), and after one
integration of the second result and solving for f and g, we get

1 1 x δ
Z
f (x) = u0 (x) + u1 (τ )dτ + (9)
2 2c 0 2c
1 1 x δ
Z
g(x) = u0 (x) − u1 (τ )dτ − (10)
2 2c 0 2c
where δ is the constant of integration.

Now we get, D’Alembert’s solution (year 1747)
1 x+ct
Z
1
u(x, t) = u0 (x + ct) + u0 (x − ct) + u1 (τ )dτ (11)
2 c x−ct
Shows:
We need u0 ∈ C 2 (R), u1 ∈ C 1 (R)
the (genuine) solution exists and is unique.
How uniqueness?

Solution depends continuouly on initial data. There are three steps:
1). Let ui (x) and ūi (x), i = 1, 2 agree every where except in the
interval [a, b] and let given ε
ε εc
|ū0 − u0 | ≤ , |ū1 − u0 | ≤ for x ∈ [a, b] (12)
4 2 + (b − a)
2). D’Alembert’s solution for u and ū gives
1
|ū(x, t)−u(x, t)| ≤ [|ū0 (x + ct) − u0 (x + ct)| + |ū0 (x − ct) − u0 (x − ct)|]
2 Z
1 x+ct

+ |ū1 (τ ) − u1 (τ )|dτ
c x−ct
⇒ 1 1

1 ε

ε 1 1 1

≤ ε+ ε+ (b − a) ≤ + +
2 4 4 2 + (b − a) 2 4 4 2
3). Thus we have shown that
|ū(x, t) − u(x, t)| ≤ ε, (13)

which proves the result.
10 /
The Wave Equation P. Prasad Department of Mathematics 48
Solution depends continuously on initial data up to

any time.
The initial value problem is well posed in the sense

of Hadamard.
Ill posed problems today are quite relevant but are

more difficult.
11 /
One-dimensional wave equation contd..
Figure: Domain of dependence of a point (x0 , t0 ) is [x0 − ct0 , x0 + ct0 ].
12 /
Figure: Domain of determinancy of an interval [a, b] on the x−axis

b−a
is the closed region= {(x, t) : a + ct ≤ x ≤ b − ct, 0 < t ≤ 2c }
13 /
Figure: Range of influence of a point x0 on the x−axis is

{(x, t) : x0 − ct ≤ x ≤ x0 + ct}
14 /
A solution of the wave equation represents a phenomenon

with finite speed of propagation.
This is an important property of all hyperbolic PDEs.
Inclusion of dispersion and diffusion in the equations make

the speed of propagation to be infinite.
15 /
/ C 2 (R)
D’Alembert’s solution as a generalized solution, when u0 ∈
/ C 1 (R).
and u1 ∈
Example: Solution of the one-dimensional wave equation with


2
x
 for x > 0
2
u0 (x) = −x for x < 0 (14)

0 for x = 0
u1 (x) = 0, x ∈ R.
Note, using [ ] as a symbol for jump from left to right, we find the
jump in the second derivative at x = 0 as
[u000 (0)] = u000 (0+) − u00 (0−) = 4. (15)
16 /
The D’Alembert’s solution gives

2 2 2
x + c t , x > ct

u(x, t) = 2ctx, −ct < x < ct (16)

−(x2 + c2 t2 ), −ct < x

u(x, t) ∈ C 2 (R × R+ ) everywhere except the characteristics through the

initial point of discontinuity:
[uxx (−ct, t)] = 2 = [uxx (ct, t)] (17)
Note that initial discontinuity breaks into discontinuities, each of the

half the magnitude 4.
17 /
Result on the propagation of discontinuities of second and higher

order derivatives shows that discontinuities in these derivatives
can exist only across characteristic curves.
What about discontinuities in the function u and its first

derivatives?
The answer lies in the definition of a weak or generalized solution.

Theorem on the next slide motivates a definition.
18 /
Theorem u(x, t) ∈ C 2 (R) is a genuine solution of the one-dimensional
wave equation iff
u(A) + u(C) = u(B) + u(D) (18)
where A, B, C, D are vertices of any characteristic parallelogram
Figure: A characteristic parallelogram

19 /
Proof: Hint
use u = f (x + ct) + g(x − ct) to derive (30).
Use finite Taylor expansion to derive the wave

equation from (30).
20 /
Problems and Methods for Engineering Teachers and
Students
So far we have described essential and basic part of 1-D wave
equation, which every one should know.
Teaching only methods of solutions for some problems

without basic concepts (like the above) will leave students ill
prepared, a common complain we receive from both academic
and industrial employers.
I shall just mention some common problems and methods of

solution. You will find them in all engineering mathematics
books or some methods of mathematical physics.
These problems are too many and I need not discuss them in
just two lectures.
21 /
Students
1. Solve the initial - boundary value problem

PDE:
utt − c2 uxx = 0; 0 < x < L, t > 0 (19)
Initial conditions:
u(x, 0) = u0 (x), ut (x, 0) = u1 (x), 0 < x < L (20)
Boundary conditions:
u(0, t) = h1 (t), u(L, t) = h1 (t), t > 0 (21)
This problem has been beautifully solved using (18) in [1] on page
108.
22 /
Students · · · conti.
But in most engineering books, for u0 (x) = 0 and u1 (x) = 0 it
is solve by method of separation of variables and then using
Fourier series, for example see [2] page 109. There are many
more problems for engineering courses in the book.
I shall suggest all pages 109-129 of [2] for further problems,
including solution of of the Neumann problem for the
inhomogeneous wave equation (Neumann was an engineer in
Germany): u − c2 u = g(x, t); 0 < x < L, t > 0. (22)
tt xx
Initial conditions:
u(x, 0) = u0 (x), ut (x, 0) = u1 (x), 0 < x < L. (23)
Boundary conditions (this makes it Neumann problem):
ux (0, t) = h1 (t), ux (L, t) = h1 (t), t > 0. (24)
In [2] you will also find proof of uniqueness of solution using

energy method, which is quite simple.
23 /
A Circle
A circle of radius r and centre (0, 0) is given by
x2 + y 2 = r2 , (25)
we write its parametric representation as
x = r cos θ, y = r sin θ, 0 ≤ θ < 2π. (26)
24 /
Circle and Unit Circle
Generally small increment δs and differential ds are different, and in

limit δs → ds. In this case δs is not an approximation. Element of its
arc length of the circle (26) of radius r is rdθ.
For unit circle, which is is a circle with radius 1 and centre origin (0, 0):
x2 + y 2 = 1 (27)
with parametric representation x = cos θ, y = sin θ.
Arc length element = dθ.
25 /
Coordinates on a Sphere
A sphere of radius r and centre (0, 0) is given by
x2 + y 2 + z 2 = r 2 , (28)
we write its parametric representation as
x = r sin θ cos φ, y = r sin θ sin φ, z = r cos θ, (29)
where 0 ≤ θ ≤ π, 0 ≤Inφ mathematics,
< 2π. θ = 0a at unit sphere
the north is pole.
the set Interpret
of these 3
points ofφdistance
inequalities and 1 equality. 1 from and
is longitude a fixed
|π central
− θ|Npoint
or |π − θ|S is
latitude.
26 /
Solid Angle & Unit Sphere
A surface element of a sphere is given by dA = r2 sin θdθdφ.
For unit sphere, r = 1 the surface element is sin θdθdφ, which called
solid angle and I denote it by dω.
dω = sin θdθdφ (30)
27 /
Solid Angle & Unit Sphere · · · cont.
Problem: Write the expression for dωm in m-D.
m
Answer: In m-D ωm = 2π 2 Γ( m2 ), where Γ is gamma
function.
28 /
IVP for the wave equation in multi-dimensions
Let me pass on to some basic results for the wave equation in

multi-space dimensions.
m
X ∂2
x = (x1 , x2 , · · · xm ), 4m =
i=1
∂x2i
utt − c2 4m u = 0, (x ∈ Rm , t > 0) (31)
u(x, 0) = u0 (x), x ∈ Rm (32)
ut (x, 0) = u1 (x), x ∈ Rm (33)
29 /
Mean Value of Function on Surface of Sphere in 3-D
We define
1
Z
(M (t)u)(x) = u(x + νct)dω, (34)
4π |ν |=1
and ω is the surface area of unit sphere in 3-D and ν is unit

normal to the surface of the sphere with centre at x and radius ct.
Note that u(x + νct) is the value of u at a point on the surface of

sphere with centre at x and of radius ct directed in direction ν
from its centre x.
Another expression M (t)u is

1
Z
M (t)u = u(x + y)dSy , (35)
4πc2 t2 |y−x|=ct
where y is a point on the surface of sphere with center x and

radius ct and dSy is its surface element.
30 /
The wave equation in multi-dimensions contd..
Theorem: If u0 (x) ∈ C 3 and u1 (x) ∈ C 2 , for x ∈ R3 , the solution
for the Cauchy problem (36)-(38) for m = 3 is
∂
u(x, t) = tM (t)u1 + [tM (t)u0 ] (36)
∂t
We omit the proof.
In 1-space-D, Z x+ct
1
tM (t)u1 = t u1 (τ )dτ, (37)
2ct x−ct
∂ ∂ 1 x+ct 1
Z
[tM (t)u0 ] = u0 (τ )dτ = {u0 (x + ct) − u0 (x − ct)} .
∂t ∂t 2c x−ct 2
(38)
Thus we have derived the D’Alembert’s solution (11) from
expression (36), which is also valid for space of any odd
dimensions.
31 /
The Wave Equation: Solution in (2n + 1)-D and 2n-D
Expression 9360 is valid also for any odd space dimensions
m = 2n + 1.
It can be derived by a beautiful method of spherical means
(Fritz John1 , 1955).
For an even space dimensions m = 2n, we deduce the solution

from the solution for m = 2n + 1 by Hadamard’s method of
descent.
————————————————————————————–
1. His popular book on PDE was based on his lectures at IISc. Later, when he went
back to Counrant Institute, New York, there was some correspondence with him and
he wrote that he remembered me as most active person during his lectures
1
1
32 /
The wave equation in multi-dimensions contd..
For m = 2, let x̄ = (x1 , x2 )

Theorem: If u0 ∈ C 3 (R2 ) and u1 ∈ C 2 (R2 ), then the solution of
(36)-(38) for m = 2 is
∂
u(x̄, t) = {M̄ (t)}u1 + ({M̄ (t)}u0 ) (39)
∂t
where
1 ui (ȳ)
Z

{M̄ (t)}ui (x) = dȳ (40)
2πc |ȳ−x̄|≤ct {c2 t2 − |ȳ − x̄|2 }1/2
Note that in (M (t)u)(x) the integration is over the surface of the sphere and

in {M̄ (t)}u (x) integration is over the closed region inside and on the
boundary of the circle.
33 /
Geometry the Method of Descent: sphere with
center x and radius ct, y is point on the surface
p 2
Distance of N from x and that of P from
√22 P̄ = c2 t2 − |x̄ − ȳ| .
c t −|x̄−ȳ |2
x = (x1 , x2 , x3 ), x̄ = (x1 , x2 ), cos θ = ct , dy1 dy2 = dS cos θ.
34 /
Domain of dependence
m=1:
Figure: Solution at (x0 , t0 ) depends on values of u0 at P1 (x0 − ct0 ) and

P2 (x0 − ct0 ) and u1 from P1 to P2 .
35 /
Domain of dependence contd..
m=2:
Figure: Solution at (x0 , t0 ) depends on the boundary of the circle

|x − x0 | = ct0 and its interior.
36 /
Domain of dependence contd..
m=3:
Figure: Solution at (x0 , t0 ) depends only on the values of u0 , u1 and the first
order derivatives of u0 on the surface S of the sphere.
37 /
Propagation of confined initial disturbances and reverberation
Initial data u0 (x) and u1 (x) such that
Supports of u0 and u1 ⊂ {x : |x| < δ} (41)
What is the meaning of support?
3-Space Dimensions
Figure: The solution at t > δ/c is non-zero only in a spherical shell of outer
and inner radii ct + δ and ct − δ.
38 /
Propagation of confined initial disturbances and
reverberation contd..
At a fixed point x, |x| > δ, u(x, t) 6= 0 for a time interval

|x|−δ
c
< t < |x|+δ
c
.
This shows that in 3-D transmission of signals in the form of

waves has sharp leading and trailing fronts.
39 /
2-Space Dimensions: Consider a point ȳ outside the confined
circular domain of radius δ, i.e. |ȳ| > δ.
|ȳ |−δ |ȳ |+δ

Figure: Signal first reaches ȳ at time c . After time c , all points of
|x̄| < δ are always in the domain of dependence of a point ȳ and ȳ keeps on
getting signal.
40 /
2-Space Dimensions. Geometrical interpretation.
Figure: Reverberation phenomenon in 2-space-D. A sudden flash of light from

infinite tube light reaches a point at distance d in time dc but the point
continues to get light all time.
41 /
1-Space Dimensions. It is a very interesting case.
Part 12 [u0 (x + ct) + u0 (x − ct)] shows that if u0 has compact

support and u1 (x) = 0 we have transmission of signals with sharp
leading and trailing edges.
1
R x+ct
The part 2c x−ct 1
u (τ )dτ shows that if u1 (x) 6= 0, the signal has
sharp leading edge but then it continues indefinitely ⇒
reverberetion.
42 /
EXERCISE
1. In the theory of acoustics (linearised theory of sound with small

disturbances about an equilibrium state) the velocity components
u and v, pressure p and density ρ satisfy the following equations :
∂ρ ∂u ∂v
+ ρ0 + ρ0 =0
∂t ∂x ∂y
∂u ∂p ∂v ∂p
ρ0 + = 0, ρ0 + = 0, p − ρ0 = a20 (ρ − ρ0 )
∂t ∂x ∂t ∂y
where the subscript 0 denotes the equilibrium state. Show that
each of the quantities u, v, p, ρ satisfy the wave equation with
velocity of propagation a0 . Find the solution in the
one-dimensional case (when v = 0 and all quantities are
independent of y) given that initially u(x, 0) = f (x), p(x) = 0.
43 /
EXERCISE contd..
Rb
2. Using an energy integral I(t) = a (u2x + u2t )dx, show that the
solution to the mixed initial and boundary value problem of the
one-dimensional wave equation
utt − c2 uxx = 0, a < x < b, t < 0
with u(x, 0) = ϕ(x), ut (x, 0) = ψ(x), u(a, t) = χ(t), u(b, t) = λ(t) is
unique.
3. If u(x, t) satisfies the wave equation

utt − c2 uxx = 0
show that Z
{uτ (ξ, τ )dξ + c2 uξ (ξ, τ )dτ } = 0
C
around any simple closed curve. Deduce that the solution which
satisfies the initial conditions u = u0 (x), ut = u1 (x) when t = 0, is given
by (2.8)
44 /
EXERCISE contd..
4. Find the solution of the equation utt − c2 uxx = 0, given that on

t = 0, u = sin πx/c when 0 ≤ x ≤ c, u = 0 when x > c and
x < 0, ut = 0 for all x. Examine the continuity of u and its
derivatives.
5. A flexible string of length l is fastened at the ends x = 0 and

x = l and is in equilibrium under a uniform tension. It is
displaced at x = 21 l to an elevation h and then released. Find
subsequent displacement for all times, assuming that the motion is
governed by the wave equation.
Hint : Use the method of separation of variables.
45 /
EXERCISE contd..
6. Find the deflection u(x, t) of a taut string which was at rest at
time t = 0, if it is fastened at the end point x = l and subjected at
the other end point x = 0 to a motion represented by
u(0, t) = f (t).
7. Determine the generalised or weak solution of the equation

1
uxx − utt = 0
c2
given that

 x2 + 5, x > 0
u(x, 0) = 5, x = 0
 −x2 + 5, x < 0
ut (x, 0) = 0
Verify that the discontinuity in the second order derivatives will
propagate along characteristics.
46 /
EXERCISE contd..
8. Determine the weak solution of the equation

1 2
uxx − 2
utt + ut − ux = 0
c c
given that ut (x, 0) = 0 and
 2
 x , x>0
u(x, 0) = 0, x = 0
−x2 , x < 0

Examine how the discontinuity in the second order

derivatives will propagate.
47 /
Phoolan Prasad and Renuka Ravindran, Partial Differential Equations,
Wiley Eastern Ltd and John Wiley & Sons, 1984.
Yehuda Pinchover and Jacob Rubinstein, Introduction to Partial

Differential Equations, Cambridge University Press, 2005.
48 /
Thank You!
48 /

2018 Wave Equation IMSC

Uploaded by

Copyright:

Available Formats

2018 Wave Equation IMSC

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2018 Wave Equation IMSC

Uploaded by

Copyright:

Available Formats

The Wave Equation

A Simple Presentation for Teachers of Engineering

Teachers Enrichment Workshop

November 26 - December 1, 2018

(Sponsored by National Centre for Mathematics and IMSc)

The Wave Equation P. Prasad Department of Mathematics 2 / 48

When we talk of waves, in general we think of waves goverened by

We have already seen this in the lecture on “Single Conservation

We not mentioned one of a very important wave equation

The Wave Equation P. Prasad Department of Mathematics 3 / 48

KdV equation led to the concept of soliton.

See a popular, very simple and beatiful article by Alex Kasman

See also a review of Kasman’s book by PP in a forthcoming issue of

The Wave Equation P. Prasad Department of Mathematics 4 / 48

ut + c ux = 0 , c = real constant (1)

The Wave Equation P. Prasad Department of Mathematics 5 / 48

utt − c2 uxx = 0, c = constant > 0 (3)

u = f (x − ct) + g(x + ct) ; f, g ∈ C 2 (R) (5)

⇒ two waves of translations, one moves with velocity c and another

are called characteristic curves

In any problem in physics, a PDE comes with additional

Cauchy Problem - here as an initial value problem.

utt − c2 uxx = 0, (x, t) ∈ R2 (6)

u(x, 0) = u0 (x), ut (x, 0) = u1 (x) (7)

The Wave Equation P. Prasad Department of Mathematics 7 / 48

For a genuine solution, we need

u0 (x) ∈ C 2 (R), u1 (x) ∈ C 1 (R) (8)

Substitute u = f (x − ct) + g(x + ct) in (7), and after one

integration of the second result and solving for f and g, we get

The Wave Equation P. Prasad Department of Mathematics 8 / 48

Now we get, D’Alembert’s solution (year 1747)

The Wave Equation P. Prasad Department of Mathematics 9 / 48

|ū(x, t) − u(x, t)| ≤ ε, (13)

Solution depends continuously on initial data up to

The initial value problem is well posed in the sense

Ill posed problems today are quite relevant but are

Figure: Domain of dependence of a point (x0 , t0 ) is [x0 − ct0 , x0 + ct0 ].

Figure: Domain of determinancy of an interval [a, b] on the x−axis

Figure: Range of influence of a point x0 on the x−axis is

A solution of the wave equation represents a phenomenon

This is an important property of all hyperbolic PDEs.

Inclusion of dispersion and diffusion in the equations make

Example: Solution of the one-dimensional wave equation with

u(x, t) ∈ C 2 (R × R+ ) everywhere except the characteristics through the

[uxx (−ct, t)] = 2 = [uxx (ct, t)] (17)

Note that initial discontinuity breaks into discontinuities, each of the

Result on the propagation of discontinuities of second and higher

What about discontinuities in the function u and its first

The answer lies in the definition of a weak or generalized solution.

Figure: A characteristic parallelogram

use u = f (x + ct) + g(x − ct) to derive (30).

Use finite Taylor expansion to derive the wave

Teaching only methods of solutions for some problems

I shall just mention some common problems and methods of

1. Solve the initial - boundary value problem

In [2] you will also find proof of uniqueness of solution using

Generally small increment δs and differential ds are different, and in

with parametric representation x = cos θ, y = sin θ.

Arc length element = dθ.

Problem: Write the expression for dωm in m-D.