Problems and Solutions

for
Partial Differential Equations
by
Willi-Hans Steeb
International School for Scientific Computing
at
University of Johannesburg, South Africa
Yorick Hardy
Department of Mathematical Sciences
at
University of South Africa, South Africa

Contents
1 Linear Partial Differential Equations

2 Nonlinear Partial Differential Equations

12

3 Lie Symmetry Methods

56

Bibliography

68

Index

69

vi

Chapter 1

Linear Partial Differential

Equations

Problem 1.
equation

Show that the fundamental solution of the drift diffusion

2u
u
u
=
+2
t
x2
x

is given by


1 (x x0 + 2t)2
u(x, t) = exp
.
4t
4t
Problem 2.

(i) Show that

Dxm (f 1) =

mf
.
xm

(1)

(ii) Show that

Dxm (f g) = (1)m Dxm (g f ).

(2)

Dxm (f f ) = 0,

(3)

(iii) Show that

Problem 3.

for m

odd

(i) Show that

Dxm Dtn (exp(k1 x 1 t) exp(k2 x 2 t)) =
1

2 Problems and Solutions

(k1 k2 )m (1 + 2 )n exp((k1 + k2 )x (1 + 2 )t)

(1)

This property is very useful in the calculation of soliton solutions.

(ii) Let P (Dt , Dx ) be a polynomial in Dt and Dx . Show that
P (Dx , Dt )(exp(k1 x 1 t) exp(k2 x 2 t)) =
P (k1 k2 , 1 + 2 )
P (Dx , Dt )(exp((k1 + k2 )x (1 + 2 )t) 1)
P (k1 + k2 , 1 2 )
Problem 4.
boundary

(2)

Consider a free particle in two dimensions confined by the

G := { (x, y) : |xy| = 1 }.

Solve the eigenvalue problem

+ k 2 = 0
where
k2 =

2mE
~2

with
G = 0.
Problem 5. Consider an electron of mass m confined to the x y plane
and a constant magnetic field B parallel to the z-axis, i.e.

0
B = 0 .
B
The Hamilton operator for this two-dimensional electron is given by
p + eA)2
1
= (
H
=
((
px + eAx )2 + (
py + eAy )2 )
2m
2m
where A is the vector potential with
B=A
and

,
py = i~ .
x
y
(i) Show that B can be obtained from

0
A = xB
0
px = i~

Linear Partial Differential Equations

or

yB
A = 0 .
0

(ii) Use the second choice for A to find the Hamilton operator H.
(iii) Show that
px ] = 0.
[H,
(iv) Let k = px /~. Make the ansatz for the wave function
(x, y) = eikx (y)
= E reduces to
and show that the eigenvalue equation H


mc2
~2 d 2
2
+

(y y0 ) (y) = E(y)
2m dy 2
2
where

eB
~k
,
y0 :=
.
m
eB
(v) Show that the eigenvalues are given by


1
En = n +
~c ,
n = 0, 1, 2, . . . .
2
c :=

Problem 6.

Consider the Schrodinger equation




1
i~
=
+ V (x) .
t
2m

Find the coupled system of partial differential equations for




:= ,
v := =
.

Problem 7.

Consider the conservation law

c(x, t) j(x, t)
+
=0
t
x

where

c(x, t)
.
x
The diffusion coefficient D(x) depends as follows on x
j(x, t) = D(x)

D(x) = D0 (1 + gx)

g 0.

4 Problems and Solutions

Thus dD(x)/dx = D0 g. Inserting the current j into the convervation law
we obtain a drift diffusion equation
c(x, t)
c(x, t)
2 c(x, t)
.
= D0 g
+ D0 (1 + gx)
t
x
x2
The initial condition for this partial differential equation is
c(x, t = 0) = M0 (x)
where denotes the delta function. The boundary conditions are
j(x = x0 , t) = j(x = +, t) = 0.
Solve the one-dimensional drift-diffusion partial differential equation for
these initial and boundary conditions using a product ansatz c(x, t) =
T (t)X(x).
Problem 8.

Consider the time-dependent Schrodinger equation



1

i = + V (x)
t
2

where
=

2
2
2
+
+
.
x21
x22
x23

Let

:= ||2 = ,

(1)


v := =

= =

x1

x2

x3

Show that the time-dependent Schrodinger equation can be written as the

system of partial differential equations (Madelung equations)



(v1 ) (v2 ) (v3 )

= (v) =
+
+
(2)
t
x1
x2
x3


(1/2 )
v
+ (v )v = V (x)
.
(3)
t
21/2
Problem 9.

Consider the time-dependent Schrodinger equation

i~

(x, t)
~2 2
=
(x, t) + V (x, t)(x, t).
t
2m

Consider the ansatz

(x, t) = (x, t) exp(imS(x, t)/~)

Linear Partial Differential Equations

where the functions and S are real. Find the partial differential equations
are and S.
= E of a particle
Problem 10. Consider the Schrodinger equation H
on the torus. A torus surface can be parametrized by the azimuthal angle
and its polar angle
x(, ) = (R + a cos()) cos()
y(, ) = (R + a cos()) sin()
z(, ) = a sin()
where R and a are the outer and inner radius of the torus, respectively such
that the ratio a/R lies between zero and one.

(i) Find H.
(ii) Apply the separation ansatz
(, ) = exp(im)()
where m is an integer.
Problem 11.

The linear one-dimensional diffusion equation is given by

2u
u
= D 2,
t
x

t 0,

< x <

where u(x, t) denotes the concentration at time t and position x R. D

is the diffusion constant which is assumed to be independent of x and t.
Given the initial condition c(x, 0) = f (x), x R the solution of the onedimensional diffusion equation is given by
Z
u(x, t) =
G(x, t|x0 , 0)f (x0 )dx0

where
0

G(x, t|x , t ) = p

(x x0 )2
exp
4D(t t0 )
4D(t t0 )
1


.

Here G(x, t|x0 , t0 ) is called the fundamental solution of the diffusion equation
obtained for the initial data (x x0 ) at t = t0 , where denotes the Dirac
delta function.
(i) Let u(x, 0) = f (x) = exp(x2 /(2)). Find u(x, t).
(ii) Let u(x, 0) = f (x) = exp(|x|/). Find u(x, t).
Problem 12. Let f : R3 R3 be a differentiable function. Consider the
first order vector partial differential equation
f = kf

6 Problems and Solutions

where k is a positive constant.
(i) Find ( f ).
(ii) Show that f = 0.
Problem 13.

Consider Maxwells equation

B=

1 E
,
c2 t

E=

B
,
t

divE = 0,

divB = 0

with B = 0 H.
(i) Assume that E2 = E3 = 0 and B1 = B3 = 0. Simplify Maxwells
equations.
(ii) Now assume that E1 and B2 only depends on x3 and t with
E1 (x3 , t) = f (t) sin(k3 x3 ),

B2 (x3 , t) = g(t) cos(k3 x3 )

where k3 is the third component of the wave vector k. Find the system of
ordinary differential equations for f (t) and g(t) and solve it and thus find
the dispersion relation. Note that

E3 /x2 E2 /x3
E = E1 /x3 E3 /x1 .
E2 /x1 E1 /x2

Problem 14.
tial equation

Let k be a constant. Show that the vector partial differen u = ku

has the general solution

u(x1 , x2 , x3 ) = (cv) +

1
(cv)
k

where c is a constant vector and v satisfies the partial differential equation

2 v + k 2 v = 0.

Problem 15.

Consider the partial differential equation of first order




+
+
f = 0.
x1
x2
x3

Show that f is of the form

f (x1 x2 , x2 x3 , x3 x1 ).

Linear Partial Differential Equations

Problem 16.

Consider the operators

Dx +

,
x

Dt =

+ u(x, t) .
t
x

Find the commutator

[Dx , Dt ]f (x, t)
Problem 17. Let C be a constant column vector in Rn and x be a
column vector in Rn . Show that
T

Cex

ex

where denotes the gradient.

Problem 18.
tion)

Consider the partial differential equation (Laplace equa2u 2u

+ 2 =0
x2
y

on

[0, 1] [0, 1]

with the boundary conditions

u(x, 0) = 1,

u(x, 1) = 2,

u(0, y) = 1,

Apply the central difference scheme

 2 
uj1,k 2uj,k + uj+1,k
u

,
x2 j,k
(x)2

2u
y 2

u(1, y) = 2.

j,k

uj,k1 2uj,k + uj,k+1

(y)2

and then solve the linear equation. Consider the cases x = y = 1/3 and
x = y = 1/4.
Problem 19.

Consider the partial differential equation

 2

2

+
+ k 2 = 0.
x2
y 2

(i) Apply the transformation

u(x, y) = x2 y 2 ,

v(x, y) = 2xy,

e
(x, y) = (u(x,
y), v(x, y)).

(ii) Then introduce polar coordinates u = r cos , v = r sin .

Problem 20.

Starting from Maxwells equations in vacuum show that

k E = B.

8 Problems and Solutions

Problem 21.

Show that the linear partial differential equation

2u 2u
2 =0
2
t

admits the solution u(t, ) = sin() sin(t). Does this solution satisfies the
boundary condition u(t, = ) = 0 and the initial condition u(t = 0, ) =
0?
Problem 22.

Consider the Hamilton operator

2

= ~
H
2

3
X
k
1 2
+ ((q2 q1 d)2 + (q3 q2 d)2 )
2
mj qj
2
j=1

where d is the distance between two adjacent atoms. Apply the linear
transformation
(q1 , q2 , q3 ) = (q2 q1 ) d
(q1 , q2 , q3 ) = (q3 q2 ) d
1
(m1 q1 + m2 q2 + m3 q3 )
X(q1 , q2 , q3 ) =
M
where M := m1 +m2 +m3 and show that the Hamilton operator decouples.
Problem 23.

Consider the Hamilton operator for three particles

 2


2
2
= 1
H
+
+
6g(x1 x2 )(x2 x3 )
2m x21
x22
x23

and the eigenvalue problem Hu(x

1 , x2 , x3 ) = Eu(x1 , x2 , x3 ). Apply the
transformation
r 

2 1
(x1 + x2 ) x3
y1 (x1 , x2 , x3 ) =
3 2
1
y2 (x1 , x2 , x3 ) = (x1 x2 )
2
1
y3 (x1 , x2 , x3 ) = (x1 + x2 + x3 )
3
u
e(y1 (x1 , x2 , x3 ), y2 (x1 , x2 , x3 ), y3 (x1 , x2 , x3 )) = u(x1 , x2 , x3 )
where y3 is the centre-of mass position of the three particles and y1 , y2 give
their relative positions up to constant factors. Find the Hamilton operator
for the new coordinates.
Problem 24.

Consider the four dimensional Laplace equation

2u 2u 2u 2u
+
+
+
= 0.
x21
x22
x23
x24

Linear Partial Differential Equations

Transform the equation into polar coordinates (r, , , )

x1 (r, , , ) = r sin(/2) sin(( )/2)
x2 (r, , , ) = r sin(/2) cos(( )/2)
x3 (r, , , ) = r cos(/2) sin(( + )/2)
x4 (r, , , ) = r cos(/2) cos(( + )/2)
u(x) = u
e(r(x), (x), (x), (x))
where x = (x1 , x2 , x3 , x4 ).
Problem 25.
equation

A nonrelativistic particle is described by the Schrodinger




2
p

+ V (x, t) (x, t) = i~
2m
t

:= i~. Write the wave function in polar form

where p
(x, t) = R(x, t) exp(iS(x, t)/~)
where R, S are real functions and R(x, t) 0. Give an interpretation of
= R2 .
Problem 26. The time-dependent Schrodinger equation for the onedimensional free particle case can be written in either position or momentum space as
~2 2 (x, t)
(x, t)

= i~
2m x2
t
(x, t)
p2
(p, t) = i~
.
2m
t
Consider the momentum space approach with the solution
(p, t) = 0 (p) exp(ip2 t/2m~)
where (p, 0) = 0 (p) is the initial momentum distribution. We define
x
:= i~(/p). Find h
xi(t).
Problem 27.

Consider the partial differential equation

2
(ln(det(In + tDf (x))) = 0
xt

where Df (x) (f : Rn Rn ) is the Jacobian matrix. Find the solution of

the initial value problem.

10 Problems and Solutions

Problem 28. The partial differential equation

2 
2 
2 
2 
2 
2
u
u
u
2u
2u
2u
+
+
x1
x2 x3
x2
x1 x3
x3
x1 x2


2
2
2
2
u
u u u
u
u u 2 u
2u
u u u
+
+
2
x1 x2 x1 x3 x2 x3
x1 x3 x1 x2 x2 x3
x2 x3 x1 x2 x1 x3
3
u u u
u
+4
=0
x1 x2 x3 x1 x2 x3
has some link to the Bateman equation. Find the Lie symmetries. There are
Lie-B
acklund symmetries? Is there a Legendre transformation to linearize
this partial differential equation?
Problem 29.

Consider the two-dimensional heat equation

2T
2T
T
=
+
2
t
x
y 2

Show that
 2
 2


1
x + (y 1)2
x + (y + 1)2
1
T (x, y, t) = exp
exp
4t
4t
t
t
satisfies this partial differential equation.
Problem 30.

Let  > 0. Consider the Fokker-Planck equation

u
2u

= 2 +
(xu)
t
x
x

with u 0 for all x.

(i) Find steady state solutions, i.e. find solutions of the ordinary differential
equation
2u

(xu) = 0.
 2+
x
x
(ii) Find time dependent solution of the inital value problem with
u(t = 0, x) = N0 (x 0 ) =

2
1
e(x0 ) /(20 )
20

with 0 0 the variance and 0 the mean. With = 0 we have the delta
function at 0.
Problem 31. Consider the Schrodinger equation

N
2
X
X

(xj xk ) u(x) = Eu(x)

2 + 2c
x
j
j=1
j<k

Linear Partial Differential Equations

11

describing a one-dimensional Bose gas with the -fnction repulsive interaction. Show that for N = 2 the eigenvalue problem can be solved with a
exponential ansatz.
Problem 32.

(i) Show that the power series

P (t, ) =

n n cos( nt)

n=0

satisfies the linear partial differential equation (diffusion type equation)

2 P (t, )
= P (t, )
t2

with boundary conditions

P (0, ) =

P (0, )
= 0,
t

n n = (),

n=0

P (t, 0) = (0).

(ii) Perform a Laplace transformation

Z
e
P (z, ) :=
ezt P (t, )dt
0

and show that Pe(z, ) obeys the differential equation

z 2 Pe(z, ) z() =

e
P (z, )

with boundary condition

(0)
Pe(z, 0) =
z
and
Z
Pe(z, ) =
0

exp(z 2 ln(/x))

(x)
dx.
x

Chapter 2

Nonlinear Partial
Differential Equations

Problem 1. Consider the system of quasi-linear partial differential equations

u v
v
u
+
+u
v
=0
t
y
x
x
u v
+
= 0.
y
x
Show that this system arise as compatibility conditions [L, M ] = 0 of
an overdetermined system of linear equations L = 0, M = 0, where
(x, y, t, is a function, is a spectral parameter, and the Lax pair is
given by
L=

Problem 2.

v
,
t
x
y

M=

+u
.
y
x
x

Find the traveling wave solution

u(x, t) = f (x ct)

c = constant

(1)

of the one-dimensional sine-Gordon equation

2u 2u
2 = sin u .
x2
t
12

(2)

Nonlinear Partial Differential Equations

13

Problem 3. Show that the nonlinear nondispersive part of the Kortewegde Vries equation
u
u
+ ( + u)
=0
(1)
t
x
possesses shock wave solutions that are intrinsically implicit
u(x, t) = f (x ( + u(x, t))t)

(2)

with the initial value problem u(x, t = 0) = f (x), where , R.

Problem 4.

Consider the Korteweg-de Vries-Burgers equation

3u
u
2u
u
+ a1 u
+ a2 2 + a3 3 = 0
t
x
x
x

(1)

where a1 , a2 and a3 are non-zero constants. It contains dispersive, dissipative and nonlinear terms.
(i) Find a solution of the form
u(x, t) =

b1
(1 + exp(b2 (x + b3 t + b4 )))2

(2)

where b1 , b2 , b3 and b4 are constants determined by a1 , a2 , a3 and a4 .

(ii) Study the case t and t .
Problem 5.

The Fisher equation is given by

u
2u
=
+ au(1 u)
t
x2

(1)

where the positive constant a is a measure of intensity of selection. (i)

Consider the substitution
v(x, (t)) =

1
u(x, t),
6

(t) = 5t.

(2)

(i) Show that (1) takes the form

5

v
2v
+ av(1 6v).
=

x2

(3)

(ii) Consider the following ansatz

v(x, t) :=

vj (x, t)jp (x, t)

(4)

j=0

is single-valued about the solution movable singular manifold = 0. This

means p is a positive integer, recursion relationships for vj are self-consistent,

14 Problems and Solutions

and the ansatz (4) has enough free functions in the sense of the CauchyKowalevskia theorem.
(iii) Try to truncate the expansion (4) with vj = 0 for j 2.
Problem 6.

Consider Fishers equation

2u
u
+ u(1 u).
=
t
x2

Show that
u(x, y, t) =

1 + exp

!!2
(x y/ 2) (5/ 6)t

is a traveling wave solution of this equation. Is the the solution an element

of L2 (R2 ) for a fixed t?
Problem 7.

Consider the nonlinear partial differential equation



2

2
+
x2

2

2
2
2
= 0.
2
y
x y xy

(1)

(i) Show that this equation has an implicit solution for of the form
x = X(, y, t) = f (, t)y + h(, t)

(2)

where f and h are arbitrary differentiable functions of and t.

(ii) Show that (1) may also be solved by means of a Legendre transformation.
Problem 8. The sine-Gordon equation is the equation of motion for a
theory of a single, dimensionless scalar field u, in one space and one time
dimension, whose dynamics is determined by the Lagrangian density
!
4
1 2
m

L = (ut c2 u2x ) +
cos
u .
(1)
2

m
Here c is a limiting velocity while m, , and are real parameters. ut and
ux are the partial derivatives of with respect to t and x, respectively. In
the terminology of quantum field theory, m is the mass associated with the
normal modes of the linearized theory, while /m2 is a dimensionless, coupling constant that measures the strength of the interaction between these
normal modes. In classical theory m is proportional to the characteristic
frequency of these normal modes.

Nonlinear Partial Differential Equations

15

(i) Let

x
t
,
t ,
u mu
m
m
and set c = 1. Show that then the Lagrangian density becomes
x

L=

m4
((u2t u2x ) + 2 cos u)
2

(2)

(3)

with the corresponding Hamiltonian density being given by

H=

m2 2
(u + u2x 2 cos u) + .
2 t

(4)

(ii) By choosing
m4
(5)

show that the minimum energy of the theory is made zero and (4) can be
written as
m4 2
(u + u2x + 2(1 cos u)).
(6)
H=
2 t
=

Problem 9. Find the solution to the nonlinear partial differential equation

 2 ! 2
 2 ! 2
u
u
u u 2 u
u
u
1
+2
1+
= 0 (1)
t
x2
x t xt
x
t2
which satisfies the initial conditions (Cauchy problem)
u(t = 0) = a(x),

u
(t = 0) = b(x).
t

(2)

Equation (1) can describe processes which develop in time, since it is of the
hyperbolic type if 1 + (u/x)2 (u/t)2 > 0. Show that the hyperbolic
condition for (1) implies for the initial conditions that
1 + a02 (x) b2 (x) > 0.

Problem 10.

(3)

Consider the Korteweg-de Vries equation

u
u 3 u
6u
+
= 0.
t
x x3

From (1) we can derive the iteration scheme



(j)
(j)
ut + uxxx
1
u(j+1) =
,
j = 0, 1, 2, . . .
(j)
6
ux

(1)

(2)

16 Problems and Solutions

where
u(j)
x :=

u(j)
.
x

(3)

(i) Let
u(0) (x, t) = ln(x ct).

(4)

Show that (4) converges within two steps to an exact solution

c
u(x, t) = + 2(x ct)2
6

(5)

of the Korteweg-de Vries equation (1).

(ii) Show that
u(0) (x, t) = (x ct)3

(6)

also converges to the solution (5).

(iii) Let
u(0) (x, t) = cos(a(x ct))k .

(7)

Show that within two steps of the iteration we arrive at

u(x, t) =

4 2 c
a + 2a2 tan2 (a(x ct)).
3
6

(8)

(iv) Show that by demanding that u 0 for |x| , provides a = i c/2

and (5) becomes the well known soliton solution


c
c
(x ct) .
(9)
u(x, t) = sech2
2
2

Problem 11. The Zakharov-Kuznetsov equation for ion acoustic waves

and solitons propagating along a very strong external and uniform magnetic
field is, for a two-component plasma
u
u

+u
+
(u) = 0
t
x x
:=

2
2
2
+ 2 + 2.
2
x
y
z

(1)
(2)

Here u is the normalized deviation of the ion density from the average.
Exact solitonlike solutions exist in one, two and three space dimensions.
They depend on the independent variables through the combination
x ct,

:= ((x ct)2 + y 2 )1/2 ,

respectively, where

r := ((x ct)2 + y 2 + z 2 )1/2 , (3)


u
u c
u=0
2

(4)

Nonlinear Partial Differential Equations

17

and
2
:=
,
x2

1
:=

1
:= 2
r r

r
r
2


(5)

for the three cases. (i) Show that for the one-dimensional case we obtain
the soliton solution
u(x, t) = 3c sech2 (c1/2 (x ct x0 )/2).

(6)

(ii) Show that the flat soliton (6) is unstable with respect to nonaligned
perturbations.
Problem 12.

Show that the nonlinear Schr

odinger equation,
i

w 2 w
+
+ 2|w|2 w = 0,
t
x2

has one-zone solutions

p
w(x, t) = f ((x, t)) exp(i(x, t)),
(x, t) := kx t,

= 1

(1)

= + h(),

(2)

:= x t

(3)

where f () and h() are elliptic functions.

Problem 13. We consider the diffusion equation with nonlinear quadratic
recombination
2u
2u
u
2u
= u2 + Dx 2 + Dy 2 + Dz 2
t
x
y
z

(1)

where Dx , Dy and Dz are constants. This equation is relevant for the

evolution of plasmas or of charge carries in solids, of generating functions.
Find the condition on c1 , c2 such that
u(x, y, z, t) =

6
c1
+
2
2
5 x /Dx + y /Dy + z 2 /Dz + c2 (t + t0 )

24
x2 /Dx + y 2 /Dy + z 2 /Dz
(x2 /Dx + y 2 /Dy + z 2 /Dz + c2 (t t0 ))2

(2)

satisfies (1).
Problem 14. The Burgers equation determines the motion of a pressureless fluid subjected to dissipation
u
u
2u
+u
=
.
t
x
x2

(1)

18 Problems and Solutions

(i) Show that any solution v of the diffusion equation
v
2v
=
t
x2

(2)

yields a solution of the Burgers equation via the Hopf-Cole transformation

u=

2 v
.
v x

(3)

(ii) Solving (1) with respect to u we can introduce the iteration formula
u(j+1) =

(j)

(j)
ux

(u(j)
xx ut ),

j = 0, 1, 2, . . .

(4)

where ut u/t etc.. Show that for u(0) = v n we find the sequence
u(1) = (n 1)

1 v
v x

(5)

2 v
fixed point
(6)
v x
Remark. Thus the Hopf-Cole transformation is an attractor (fixed point)
of the iteration (4).
u(2) =

Problem 15.

Consider the nonlinear diffusion equation



u

u
=
u
.
t
x
x

(i) Show that this equation admits a solution of the form

 
 2 !
x

1
x

1
for t > 0,   < 1
x
x1
x1
1
u(x, t) =
 

0
for t > 0,   > 1
x1

(1)

(2)

where
x1 := (6t)1/3

(3)

Remark. The solution exhibits a wave-like behaviour although it is not a

wave of constant shape. The leading edge of this wave, that is where u = 0,
is at x = x1 and the speed of propagation is proportional to t2/3 .
Problem 16. Show that the nonlinear partial differential equation (Fishers
equation)

 2

u + u = u2
(1)
x2
t

Nonlinear Partial Differential Equations

19

admits the travelling wave solution

1
u(x, t) =
4




2
1
5

1 tanh
x vt
.
2 6
6

(2)

Problem 17. Show that the coupled system of partial differential equations
2n 2n
2
2E
E
= nE,

=
(|E|2 )
+
(1)
i
2
2
2
t
x
t
x
x2
admits the solution


 
is
s2
2
E(x, t) = E0 (x st) exp
x+i
t
(2a)
2
4
n(x, t) =
where

22
cosh2 (x st x0 )

(2b)

p
2(1 s)2
E0 (x st) =
cosh((x st x0 ))

(2c)

The soliton solution represents a moving one-dimensional plasma density

well which Langmuir oscillations are locked.
Problem 18.

The Navier-Stokes equation is given by

u
1
= u u + 2 u p
t

where u denotes the solenoidal

u divu =

u1
u2
u3
+
+
=0
x
y
z

flow velocity field, and are the constant kinematic viscosity and fluid
density. Here p is the fluid pressure. We have
u1

u1
1
u1 x + u2 u
y + u3 z

u2
u2
2
u u := u1 u
x + u2 y + u3 z
u3
u3
3
u1 u
x + u2 y + u3 z
Find the time evolution of
u3
v := curlu u =

y
u1
z
u2
x

u2
z
u3
x
u1
y

20 Problems and Solutions

Problem 19.

Show that

can be linearized into

u
u
+u
=0
t
x

(1)

=x

(2)

with
t(x, t) = t,

Problem 20.

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

u
(
x(x, t), t(x, t)) = x.

(3)

Consider the partial differential equation

u
2u
= u2 2 .
t
x

(1)

Show that (1) linearizes under the transformation

Z x
1
x
(x, t) =
ds,
t(x, t) = t,
u
(
x(x, t), t(x, t)) = u(x, t).
u(s, t)
(2)
It is assumed that u and all its spatial derivatives vanish at .
Problem 21.

Show that the equation

2u
u
= u2 2
t
x

(1)

is transformed under the Cole-Hopf transformation

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

1 u
,
u x

t(x, t) = t,

x
(x, t) = x

(2)

into Burgers equation

u

=
t
x

Problem 22.



u

u
2 +
.
x

(3)

Show that the the equation

u
2u
u
= u2 2 + cu2
t
x
x

is linearized by the transformation

Z x
x
(x, t) =
u(s, t)1 ds,
t(x, t) = t,

(1)

u
(
x(x, t), t(x, t)) = u(x, t).

(2)

Nonlinear Partial Differential Equations

Problem 23.

Show that the equation

 2
u
3 u 3 u
3 2 u 3 u
+
=
+ u 2 + u2
3
t
x
2 x
2 x
4 x

21

(1)

can be derived from the linear equation

3

= 3
t
t

(2)

and the transformation

 Z
1
(x, t) = exp
2

Problem 24. Using the transformation

Z x
ds
x
(x, t) =
,
t(x, t) = t,
u(s, t)


u(s, t)ds .

(3)

u
(
x(x, t), t(x, t)) = u(x, t) (1)

the equation
3u
u
= u3 3
t
x
can be recast into the differential equation
!
 2
u
3u

1 u
1 u

2u

3 2
u
2 = 0.
t x
3
u
x
2 x

(2)

(3)

(ii) Show that using the Cole-Hopf transformation

u
(
x, t) =

x, t)
1 v(
v(
x, t) x

(4)

(3) can further be reduced to the modified Korteweg-de Vries equation in

v. Hint. We have
 
Z x
Z x
x

1
u(s, t)
3 u(s, t)
2 u 1 u
=
ds =
u(s, t)
ds = u 2
.
2
t
t
s3
x 2 x
u (s, t)

(5)
Problem 25.

Show that the transformation

u
u
(
x, t) =
,
x

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

t(x, t) = t

(1)

transforms the nonlinear heat equation

u
2u
= A(u) 2
t
x

(2)

22 Problems and Solutions

into
u

=u
2
t
x

Problem 26.

A(
x)
x


.

(3)

Consider the nonlinear partial differential equation

u f (u)(u)2 + a(x, t)u + b(x, t)

u
=0
t

(1)

where is the gradient operator in the variables x1 , . . . , xn , := ,

f (u) and b(x, t) are given functions, and a(x, t) is a given n-dimensional
vector. Show that the transformation

 Z s
Z u(x,t) 
f (z)dz
ds v(x, t) = 0.
(2)
exp
u0

u0

reduces (1) to the linear partial differential equation

v + a(x, t)v + b(x, t)

v
= 0.
t

(3)

Hint. From (2) we find

v
= exp
t

u(x,t)

f (z)dz
u0

u
t

(4a)

!
Z u(x,t)
v
u
= exp
f (z)dz
(4b)
x1
x
1
u0
!
!
2

Z u(x,t)
Z u(x,t)
2v
u
2u
=
exp

exp

.
f
(z)dz
f
(z)dz
f
(u)
x21
x21
x1
u0
u0
(4c)
Problem 27.

Consider the two-dimensional sine-Gordon equation

2u 2u 2u
+ 2 2 = sin u
x2
y
t

It may be regarded as describing solitary waves in a two-dimensional Josephson junction. The Lamb substitution is given by
u(x, y, t) = 4 tan1 [M (x, y, t)] .
Find the equation for M .

(1)

Nonlinear Partial Differential Equations

Problem 28.

23

We consider the nonlinear dAlembert equation

u = F (u)

(1)

where u = u(x), x = (x0 , x1 , . . . , xn ),

 :=

2
2
2

2
2
2
x0
x1
xn

(2)

and F (u) is an arbitrary differentiable function.

(i) Consider the transformation
u(x) = (w(x))

(3)

where w(x) and (w) are new unknown functions. Show that (1) takes the
form
d
d2
w +
w w = F ()
(4)
dw
dw2
where

2 
2

2
w
w
w
w w :=

.
(5)
x0
x1
xn
(ii) Show that (4) is equivalent to the following equation
!
!
2
n
Pn
d

d
P
d
d2
w
+
+ 2 (w w )+
F () = 0 (6)
dw
Pn
dw
dw2
dw Pn
where Pn (w) is an arbitrary polynomial of degree n in w, and = 1, 0, 1.
Moreover Pn dPn /dw.
(iii) Assume that satisfies
!
d2 d Pn

+
= F ().
(7)
dw2
dw Pn
Show that (6) takes the form
d
dw

Pn
w
Pn

!
+

d2
(w w ) = 0.
dw2

(8)

(iv) Show that a solution of the system

w =

Pn
,
Pn

w w =

(9)

is also a solution of (8), and in this way we obtain a solution of (1) provided
satisfies (7).

24 Problems and Solutions

Problem 29.
sion

Consider the nonlinear wave equation in one space dimen-

2u 2u

u + u3 = 0.
t2
x2
(i) Show that (1) can be derived from the Lagrangian density
L(ut , ux , u) =

1 2
(u u2x ) g(u)
2 t

(1)

(2)

with g(u) the potential function

g(u) =

1 2
1
(u 1)2 (u4 2u2 + 1).
4
4

(3)

(ii) Definition. A conservation law associated to (1) is an expression of

the form
T (u(x, t)) X(u(x, t))
+
=0
(4)
t
x
where T is the conserved density and X the conserved flux. T and X are
functionals of u and its derivatives.
(ii) Show that the quantities
 2  2 !
u
u
1
u u
1
+
+ (u2 1)2 0,
T2 :=
(5)
T1 :=
2
t
x
4
x t
are conserved densities of (1), provided that u2 1, u/t, and u/x tend
to zero sufficiently fast as |x| +.
Hint. The corresponding fluxes are
u u
X1 =
,
x t

1
X2 =
2

u
t

2


+

u
x

2 !

1
(u2 1)2 .
4

(6)

T1 and T2 correspond to the energy and momentum densities, respectively.

The Euler-Lagrange equation is given by




L

L
L
+

=0
t ut
x ux
u

(7)

where L is the Lagrange density (2).

Problem 30.

Consider the nonlinear partial differential equation

2u 2u

u + u3 = 0.
t2
x2

(1)

Nonlinear Partial Differential Equations

(i) Show that (1) admits the solution

uK (x vt) = tanh((/ 2)(x vt x0 )) = uK (x vt)

25

(2)

where
2 (1 v 2 )1

(3)

and x0 is a constant. These solutions are of kink (K) and antikink (K)
type traveling at constant velocity v. These solutions do not tend to zero at
infinity, but they do connect two minimum states of the potential 14 (u2 1)2 .
(ii) The energy and momentum densities T1 and T2 for the kink and antikink
are obtained by substituting (2) into (5) of problem 1. Show that
T1 (x, t) =

2
cosh4 ((/ 2)(x vt x0 ))
2

(4)

2v
(5)
cosh4 ((/ 2)(x vt x0 )).
2
(iii) Show that by integrating over x we find the energy and the momentum
Z +
Z +
4
4 v

E :=
T1 dx =
,
P :=
T2 dx = .
(6)
3
3
2
2

T2 (x, t) =

(iv) Show that the densities T1 and T2 are localized in space, in contrast
with uK and uK .
(v) We associate the mass
M 2 := E 2 P 2

(7)

to (1) and the energy center

Z

xT1 dx
Xc := Z
.
+
T1 dx

(8)

Show that the kink and antikink solutions of (6) take the values
M2 =

8
,
9

Xc = vt + x0 .

(9)

Problem 31. We say that a partial differential equation described by

the field u(x, y) is hodograph invariant if it does not change its form by the
hodograph transformations
x
(x, y) = u(x, y),

y(x, y) = y,

u
(
x(x, y), y(x, y)) = x.

(1)

26 Problems and Solutions

Show that the Monge-Ampere equation for the surface u(x, y) with a constant total curvature K
 2 2
 2  2 !
u
u
u
2u 2u

=K 1+
+
(2)
2
2
x y
xy
x
y
is hodograph invariant.
Problem 32. Show that the nonlinear equation of the Born-Infeld type
for a scalar field u is obtained by varying the Lagrangian
L(ux , uy , uz , ut ) = 1 (1 + u2x + u2y + u2z u2t )1/2

(1)

and has the following form


 2  2  2  2 !  2
u
u
u
u
u 2u 2u 2u
+ 2 + 2 2
1+
+
+

x
y
z
t
x2
y
z
t


u
x

2

2u

x2

u
y

2

2u

y 2

u
z

2

2u

z 2

u
t

2

2u 2u
2u
2
2
t
xy xy

u u 2 u
u u 2 u
u u 2 u
u u 2 u
u u 2 u
2
+2
+2
+2
= 0.
x x xz
y z yz
t x tx
t y ty
t z tz
(2)

Problem 33. The sine-Gordon equation is the equation of motion for a

theory of a single, dimensionless scalar field u, in one space and one time
dimension, whose dynamics is determined by the Lagrangian density
!

1 2
m4
2 2
L(ut , ux , u) = (ut c ux ) +
cos
u .
(1)
2

m
Here c is a limiting velocity while m, , and are real parameters. ut and
ux are the partial derivatives of u with respect to t and x, respectively. In
the terminology of quantum field theory, m is the mass associated with the
normal modes of the linearized theory, while /m2 is a dimensionless, coupling constant that measures the strength of the interaction between these
normal modes. In classical theory m is proportional to the characteristic
frequency of these normal modes. Let
x

x
,
m

t
,
m

mu
u

(2)

and set c = 1. (i) Show that then the Lagrangian density becomes
L(ut , ux , u) =

m4
((u2t u2x ) + 2 cos u)
2

(3)

Nonlinear Partial Differential Equations

27

with the corresponding Hamiltonian density being given by

m4 2
(u + u2x 2 cos u) + .
2 t

H=

(4)

(ii) Show that by choosing

= m4 /

(5)

the minimum energy of the theory is made zero and (4) can be written as
H=

m4 2
(u + u2x + 2(1 cos u)).
2 t

(6)

Problem 34. Show that the following theorem holds. The conservation
law

(T (u/x, u/t, u)) +

(F (u/x, u/t, u)) = 0
(1)
t
x
is transformed to the reciprocally associated conservation law

(T 0 (u/x0 , u/t0 , u)} +

(F 0 (u/x0 , u/t0 , u)} = 0
0
t
x0

(2)

by the reciprocal transformation

dx0 (x, t) = T dx F dt,

(3)

1
T (u/x, u/t, u)

(4a)

F (u/x, u/t, u)
.
T (/x; /t; u)

(4b)

T 0 (u/x0 , u/t0 , u) =
F 0 (u/x0 , /t0 , u) =

t0 (x, t) = t,

(ii) Show that

1
= 0 0,
x
T x

F0

= 0 0 + 0.
t
T x
t

(5)

Problem 35. Consider the coupled two-dimensional nonlinear Schr

odinger
equation
u
2u
2u
i
2 + 2 u uu 2wu = 0
(1a)
t
x
y

2w
2w
2
+ 2 + 2 (u u) = 0
2
x
y
x

(1b)

where , , are arbitrary constants. Consider the following transformation

x
(x, y, t) =

x
,
t

y(x, y, t) =

y
,
t

1
t(x, y, t) = ,
t

(2a)

28 Problems and Solutions

1
u(x, y, t) = exp
t

ix2
iy 2
+
4t
4t

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

(2b)

1
w(
x(x, y, t), y(x, y, t), t(x, y, t)).
(2c)
t2
Show that u
and w
satisfy the same equation with the subscripts x, y, t
replaced by x
, y, t.
w(x, y, t) =

Problem 36. An simplified analog of the Boltzmann equation is constructed as follows. It is one-dimensional and the velocities of the molecules
are allowed to take two discrete values, c, only. Thus the distribution
function in the Boltzmann equation, f (x, v, t), is replaced by two functions
u+ (x, t) and u (x, t) denoting the density of particles with velocity +c or
c, repsectively, at point x and time t. The gas is not confined, but x varies
over all points of the real line (, ). It is further assumed that there
are only two types of interaction, viz., two + particles go over into two
particles, and vice versa, the probalitity for both processes to occur within
one unit of time being the same number . Then the Boltzmann equation,
which in absence of external forces reads
Z
f (x, v, t)
+vf (x, v, t) = d3 v1 d|vv1 |(|vv1 |, )[f 0 f10 f f1 ] (1)
t
where f 0 is the final distribution, i.e., the distribution after a collision.
Under the assumption described above (1) translates into a system of two
equations
u+
u+
+c
= (u2 u2+ ),
t
x

u
u
c
= (u2+ u2 )
t
x

(2)

where c and are positive constants. This model is called the Carleman
model. The Carleman model is rather unphysical. However with its aid one
can prove almost all those results which one would like to obtain for the
Boltzmann equation itself as, for instance, the existence of solutions for a
wide class of initial conditions or a rigorous treatment of the hydrodynamic
limit. (i) Show that as for the Boltzmann equation, the H theorem holds
for the Caleman model: The quantity
Z
(u+ (x, t) ln u+ (x, t) + u (x, t) ln u (x, t))dx
(3)
never decreases in time. (ii) Show that there exists the following generalizations of the H theorem. Let f be concave function, defined on the half-line
(0, ) which is once continuously differentiable. Let
Z
Sf := (f (u+ ) + f (u ))dx
(4)

Nonlinear Partial Differential Equations

29

Show that

d
Sf (u) 0.
(5)
dt
Thus, not only does entropy never decrease, but the same is true for all
quasientropies.
(iii) Show that as a consequence, all Renyi entropies
Z

S := (1 )1 ln (u
(6)
+ (x, t) + u (x, t))dx
never decrease. In information-theretical language, this means that all sensible measures of the lack of information are nondecreasing. In other words,
information is lost, or chaos is approached, in the strongest possible way.
Problem 37. For the functions u, v : [0, ) R R we consider the
Cauchy problem
u u
+
= v 2 u2 ,
t
x
u(0, x) = u0 (x),

v
v

= u2 v 2
t
x

(1a)

v(0, x) = v0 (x).

(1b)

This is the Carleman model introduced above.

(i) Define
S := u + v,
D := u v

(2)

Show that S and D satisfy the system of partial differential equations

D
S
+
= 0,
t
x

D S
+
= 2DS
t
x

(3)

and the conditions u 0, v 0 take the form

S0

and S 2 D2 0.

(4)

(ii) Find explict solutions assuming that S and D are conjugate harmonic
functions.
Problem 38. The Carleman model as an approximation of the Boltzmann equation is given by the non-linear equations
u+
u+
+c
= (u2 u2+ ),
t
x
u
u
c
= (u2+ u2 ).
t
x
(i) Show that if u (x, 0) 0, then u (x, t) 0 for all t > 0.

(1a)
(1b)

30 Problems and Solutions

(ii) Show that

Z
(u+ + u )dx = 0.

(2)

The two properties have to hold, of course, in any model that is considered
to be of some physical relevance.
(iii) Let
Z
H

(u+ ln u+ + u ln u )dx

(3)

Show that the analogue of Boltzmanns H-function is given by

dH
0
dt

(4)

(u+ ln u+ + u ln u ) + c (u+ ln u+ u ln u ) 0.
t
x

(5)

(iv) Show that

u+ ln u+ + u ln u has to be interpreted as the negative entropy density

and c(u+ ln u+ u ln u ) as the negative entropy flux.
(v) We define
Z
Sf :=

(f (u+ ) + f (u ))dx

(6)

where f is a concave (or convex, respectively) function. In the special case

f (s) = s ln s
Sf is the expression for entropy i.e., H. Show that
dSf
0
dt
if f is convex. Show that
dSf
0
dt
if f is concave.
Problem 39.

The Broadwell model can then be written as

f+
1
f+
+
= (f02 f+ f )
t
x

f0
1
= (f+ f f02 )
t

f
f
1

= (f02 f+ f ).
t
x


(1a)
(1b)
(1c)

Nonlinear Partial Differential Equations

31

The parameter  can qualitatively be understood as the mean free path.

This system of equations serves as a model for the Boltzmann equation.
The limit  0 corresponds to a vanishing mean free parth and the fluid
regime, while  approaches free molecular flow. The locally conserved
spatial densities
:= f+ + 2f0 + f ,

u := f+ f

(2)

corresponding to mass and x momentum. Show that these are governed by

the local conservation laws

+
(u) = 0,
t
x

Problem 40.

(u)

+
(f+ + f ) = 0.
t
x

(3)

Consider the hyperbolic system of conservation laws

u
+
f (u) = 0
t
x

(1)

where f : Rm Rm is continuously differentiable. A convex function (u)

is called an entropy for (1) with entropy flux q(u) if

(u) +
q(u) = 0
t
x

(2)

holds identically for any smooth vector field u(x, t) which staisfies (1).
(i) Show that (2) follows from (1) if
m
X
fj
q
=
,
u
u
u
j
k
k
j=1

k = 1, . . . , m.

(3)

(ii) Show that for m = 1, every convex function (n) is an entropy for (1)
with entropy flux
Z u
q(u) =
0 ()df ().
(4)
0

Problem 41.

Consider the so-called sine-Hilbert equation

 
u
H
= sin u
t

where the integral operator H is defined by

Z
1
u(y, t)
Hu(x, t) := P
dy.

y x

(1)

(2)

32 Problems and Solutions

This is the so-called Hilbert transform. P denotes the Cauchy principal
value. Let f be a continuous function, except at the singularity c. Then
the Cauchy principal value is defined by
Z

Z

c

Z
f (x)dx +

f (x) := lim

0


f (x)dx .

(3)

c+

The Cauchy principal value can be found by applying the residue theorem.
Let


f (x, t)
u(x, t) := i ln
(4)
f (x, t)
where
f (x, t) :=

N
Y

(x xj (t))

(5a)

j=1

=xj (t) > 0

j = 1, 2, . . . , N,

xn 6= xm

for n 6= m.

(5b)

Here xj (t) are complex functions of t and denotes complex conjugation.

(i) Show that
 

u
= (ln(f f ))
(6)
H
t
t
follows from (2), (4) and (5).
(ii) Show that (1) is transformed into the form
N
X
j=1

N
dxj
1
dxj X
1
1
+
=

x xj (t) dt j=1 x xj (t) dt

2i

QN

j=1 (x xj (t))
QN
j=1 (x xj (t))

QN

j=1 (x xj (t))
QN

j=1 (x xj (t))

(7)
(iii) Show that (7) is equivalent to the equation
1

(f f ) = (f 2 f 2 ) = =(f 2 ).
t
2i

(8)

(iv) Show that by multiplying both sides of (7) by x xn (t) and then
putting x = xn , we obtain
QN

dxn
1
j=1 (xn (t) xj (t))
= QN
dt
2i j=1(j6=n) (xn (t) xj (t))

Problem 42.

n = 1, 2, . . . , N.

(9)

Consider the Fitzhugh-Nagumo equation

u
2u
=
+ u(1 u)(u a)
t
x2

(1)

!
.

Nonlinear Partial Differential Equations

33

where a is a constant. Without loss of generality we can set 1 a < 1.

Insert the ansatz
u(x, t) = f (x, t)w(z(x, t)) + g(x, t)

(2)

into (1) and require that w(z) satisfies an ordinary differential equation.
This is the so-called direct method and z is the so-called reduced variable.
Problem 43.

Consider the nonlinear Dirac equations of the form

i

3
X

=0

M + F ()
= 0.
x

(1)

The notation is the following

: R4 C4 ,

1
2

=
3
4

(2)

M is a positive constant,
:= ( 0 , ) 1 + 2 3 4

1
2
3
4

(3)

where (, ) is the usual scalar product in C4 and the s are the 4 4

matrices of the Pauli-Dirac representation, given by




I 0
0
k
0 :=
,
k :=
for k = 1, 2, 3
(4)
0 I
k 0
where the Pauli matrices are given by




0 1
0 i
1 :=
,
2 :=
,
1 0
i 0


3 :=

1
0

0
1


(5)

and F : R R models the nonlinear interaction. (i) Consider the ansatz

(standing waves, stationary states)
(t, x) = eit u(x)

(6)

where x0 = t and x = (x1 , x2 , x3 ). Show that u : R3 C4 satisfies the

equation
3
X

u M u + 0 u + F (
uu)u = 0.
(7)
i
k
xk
k=1

(ii) Let
 

0
v(r)

 1
 .
u(x) :=

cos
iw(r)
i
sin e

(8)

34 Problems and Solutions

Here r = |x| and (, ) are the angular parameters. Show that w and v
satisfy the nonautonomous planar dynamical system
dw 2w
+
= v(F (v 2 w2 ) (M ))
dr
r

(9a)

dv
= w(F (v 2 w2 ) (M + )).
dr

(9b)

(ii) Hint. Notice that

u = u1 u1 + u2 u2 u3 u3 u4 u4
u

(10)

u = v 2 w2 .
u

(11)

Inserting (8) yields

Problem 44. Consider a one-dimensional system to describe the electronbeam plasma system

ne
+
(ne (ue + V )) = 0
t
x

(1a)

ue
e
ue
1 pe
+ (ue + V )
=
E
t
x
me
me ne x

(1b)

E
E
+ (ue + V )
= 4eni (ue + V ).
(1c)
t
x
where ne is the density of the beam-electron fluids, me the beam-electron
mass, ue the bulk fluid velocity, p = Kb Te ne the particle fluid pressure, e the
charge on an electron and E = . We study the simplest case of uniform
plasma in the absence of an external electromagnetic field, and assume
that the positive ions are taken to form a fixed, neutralizing background
of uniform density ni = N0 = const throughout the present analysis. The
electrons move with a beam drift velocity V corresponding to the ions.
Assume that electrostatic perturbation is sinusoidal


n(t)
1
ne (x, t) = N0 1 +
(sin(kx) + cos(kx)) ,
n(t) N0
(2a)
N0
2
ue (x, t) = u(t)(cos(kx) sin(2kx)),

E(x, t) = E(t)(cos(kx) sin(kx)).

(2b)

Define
X(t) := n(t),

Y (t) :=

1
ku(t),
2

Z(t) :=

e k
E(t),
me p 2

q :=

Kb Te k 2
,
m e N0 2

p :=
(3)

4e2
.
me

Nonlinear Partial Differential Equations

35

Show that when (2) and (3) are substituted into (1a)-(1c), we obtain
dX
= kV X XY + 2N0 Y
dt
dY
= qX + Y 2 pZ
dt
dZ
= N0 Y Y Z + kV Z.
dt
Problem 45.

Consider the partial differential equation

u
u
3u 2u 4u
+
=0
+u
+ 3 +
t
x
x
x2
x4

(1)

subject to periodic boundary conditions in the interval [0, L], with initial
conditions u(x, 0) = u0 (x). We only consider solutions with zero spatial
average. We recall that for L 2 all initial conditions evolve into u(x, t) =
0. We expand the solution for u in the Fourier series
u(x, t) =

an (t) exp(ikn x)

(2)

n=

where kn := 2n/L and the expansion coefficients satisfy

an (t) = a
n (t).

(3)

Here a
denotes the complex conjugate of a. Since we choose solutions with
zero average we have a0 = 0. (i) Show that inserting the series expansion
(2) into (1) we obtain the following system for the time evolution of the
Fourier amplitudes

X
dan
1
+ (kn4 kn2 ikn3 )an + ikn
(am anm + a
m an+m ) = 0.
dt
2
m=0

(4)

(ii) Show that keeping only the first five modes we obtain the system
da1
+ (1 ik 3 )a1 + ik(
a1 a2 + a
2 a3 + a
3 a4 + a
4 a5 ) = 0
dt
da2
+ (2 8ik 3 )a2 + ik(a21 + 2
a1 a3 + 2
a2 a4 + 2
a3 a5 ) = 0
dt
da3
+ (3 27ik 3 )a3 + 3ik(a1 a2 + a
1 a4 + a
2 a5 ) = 0
dt
da4
+ (4 64ik 3 )a4 + 2ik(a22 + 2a1 a3 + 2
a1 a5 ) = 0
dt

(5a)
(5b)
(5c)
(5d)

36 Problems and Solutions

da5
+ (5 125ik 3 )a5 + 5ik(a1 a4 + a2 a3 ) = 0
dt
where

k := 2 ,
L

n := kn4 kn2 .

(5e)
(6)

Problem 46.

The modified Boussinesq-Oberbeck equations are given by



4

4 4
=
+ 4 (4)

t
x
x z
z x
"


 2  2 #

2R
=R
+4

+
+
+
t
x
x z
z x
Ti
z
x
z
(1)
where is the Prandtl number, R the Rayleigh number, the stream
function, and a function measuring the difference between the profile of
temperature and a profile linearly decreasing with height and time, namely
= T T0 +

 2
F t + Rz.
Ti

(2)

Let L be the horizontal extenion of the convection cells, l = H/L and

Rc = 4 (1+l2 )3 /l2 the critical Rayleigh number for the onset of convection.
One sets r = R/Rc . Suppose that the temperature T1 is fixed and that T0
increases. We define the dimensionless parameter
:=

T0 T1 1
.
T1 r

(3)

This quantity is a constant once the fluid and the experimental setting are
chosen. The dimensionless temperatures appearing in the last term of the
second of equations (1) are now expressed by
T0 = Rc

(r + 1)
,

T1 =

Rc
.

(4)

The boundary conditions are

= 4 = 0
= 0,

=0
z
Show that the equation
dX
= X + Y,
dt

at z = 0, H

at z = 0, H

at x = kL,

kZ

dY
= rX Y XZ + W
dt

(5)



Z
1+
r

Nonlinear Partial Differential Equations

dW
= W Y
dt

Z
1+
r


,

dZ
= bZ + XY
dt

37
(6)

can be derived from (1) and the boundary condition (5) via a Fourier expansion of the function (x, z, t) and (x, z, t).
Problem 47.

Consider the complex Ginzburg-Landau equation

2w
w
= (1 + ic1 ) 2 + w (a + ic2 )|w|2 w.
t
x

(1)

(i) Show that setting a = 1 in (1), and writing

w(x, t) = R(x, t) exp(i(x, t))

(2)

one obtains two real equations which, after suitable linear combination and
division by c21 , can be written as
!

2 !

2
2
2
2 R
2
+R
= (1 +  )

+  + ( +  )R R (3a)

t
t
x2
x



= (1 + 2 )
r2
(1 + (1 )R2 )R2 . (3b)
 R 2 + 2 R 2
2 t
t
x
x
Here we have introduced c2 := c1 and  := 1/c1 . (ii) Make an expansion
in  of the form
R := R0 + 2 R2 + ,
Problem 48.
equation,

:= 1 (1 + 2 1 + )

(4)

Consider the cubic nonlinear one-dimensional Schr

odinger

w 2 w
+
+ Qw|w|2 = 0
(1)
t
x2
where Q is a constant. (i) Show that a discretization with the periodic
boundary conditions wj+N wj is given by
i

wj+1 + wj1 2wj

dwj
(k)
+
+ Q|wj |2 wj = 0,
dt
h2

k = 1, 2, . . .

(2)

where
(a)

(1)

wj

:= wj

and

(2)

(b) wj

:=

1
(wj+1 + wj1 ).
2

(3)

(ii) Show that both schemes are of second-order accuracy. (iii) Show that
in case (2a) there are first integrals, the L2 norm,
I :=

N
1
X
j=0

|wj |2

(4)

38 Problems and Solutions

and the Hamilton function
H = i

N
1 
X
j=0


1
|wj+1 wj |2
4
Q|wj | .
h2
2

(5)

The Poisson brackets are the standard ones. Thus when N = 2 the system
is integrable. This system has been used as a model for a nonlinear dimer.
(iv) Show that the Hamilton function of scheme (2b) is given by (h = 1)
H = i

N
1
X

(wj (wj1 + wj+1 )

j=0

1
4
ln(1 + Qwj wj )
Q
2

(6)

together with the nonstandard Poisson brackets

1
{qm , pn } := (1 + Qqn pn )m,n
2

(7)

{qm , qn } = {pm , pn } = 0.

(8)

and

Problem 49.

Consider the sine-Gordon equation in 3 + 1 dimensions

 = sin

where
 :=

(1)

+ 2+ 2 2
2
x
y
z
t

(2)

and (x, y, z, t) is a real valued scalar field. The sinh-Gordon equation is

given by
 = sinh
(3)
Let

0 :=

1
0

0
1


,

1 :=

0
1

1
0


,

2 :=

0
i

i
0


,

3 :=

1 0
0 1
(4)

be the Pauli spin matrices and

a := 0

+ i1
+ i3
+ 2 .
x
y
z
t

(5)

Let a
denote complex conjugates. Let 0 2, 0 2, < <
and < < be arbitrary parameters. We set
U := exp(i1 exp(i2 e 1 )).

(6)

Nonlinear Partial Differential Equations

39

Assume that and are solutions of (1) and (2), respectively. Let
u := i.

(7)

is then given by
The B
acklund transformation B

i = B(,
, )

(8)

where B(,
, ) is the B
acklund transformation operator. The functions
and are related by
 
1
1
au = sin
u
U.
(9)
2
2
The B
acklund transformation works as follows. Let (respectively ) be a
solution of (1) (respectively 3)) then solve (9) for (respectively ). The
solution then solves (3) (respectively (1)).
(i) Show for any U such that
= U 1
U

(10)

all solutions of (9) must be of the form (i.e. plane travelling wave with
speed v less than one)
u(x, y, z, t) = f (),

:= kx + ly + mz t

(11)

where k, l, m, are real constants and

k 2 + l2 + m2 2 = 1,

v=

k2

2
.
+ l2 + m2

(ii) Thus show that we have a Backlund transformation between the ordinary differential equations
d2
= sin ,
d 2

d2
= sinh
d 2

defined by
du
= 2ei sin
d


1
u
.
2

Problem 50. The equation which describes small amplitude waves in a

dispersive medium with a slight deviation from one-dimensionality is



u
u 2 u
2u
4
+ 6u
+
3 2 = 0.
(1)
2
x
t
x x
y

40 Problems and Solutions

The + refers to the two-dimensional Korteweg-de Vries equation. The
refers to the two-dimensional Kadomtsev-Petviashvili equation. Let (formulation of the inverse scattering transform)
u(x, y, t) := 2

K(x, x; y, t)
x

(2)

where
Z

K(x, z; y, t) + F (x, z; y, t) +

K(x, s; y, t)F (s, z; y, t)ds = 0

(3)

and F satsifies the system of linear partial differential equations

3F
F
3F
+
+
= 0,
x3
z 3
t

2F
F
2F

+
=0
x2
z 2
y

(4)

where = 1 for the two-dimensional Korteweg de Vries equation and = i

for the Kadomtsev Petviashvili equation. Find solutions of the form
F (x, z; y, t) = (x; y, t)(z; y, t)

(5)

K(x, z; y, t) = L(x; y, t)(z; y, t).

(6)

and

Problem 51. Show that Hirotas operators Dxn (f g) and Dxm (f g) given
in (1) and (2) can be written as
Dxn (f

Dxm Dtn (f

g) =

n
X
(1)(nj) n! j f nj g
g) =
,
j!(n j)! xj xnj
j=0

n
m X
X
(1)(m+nji) m!
j=0 i=0

j!(m j)!

(3)

n!
i+j f n+mij g
.
i!(n i)! ti xj tni xmj
(4)

Problem 52.

Consider the Korteweg-de Vries equation

u 3 u
u
+ 6u
+
= 0.
t
x x3

(1)

(i) Consider the dependent variable transformation

u(x, t) = 2

2 ln f (x, t)
.
x2

(2)

Nonlinear Partial Differential Equations

41

Show that f satisfies the differential equation

2f
f 3 f
f f
4f
f
+3

+f 4 4
xt x t
x
x x3

2f
x2

2
= 0.

(3)

(ii) Show that this equation in f can be written in bilinear form

(Dx Dt + Dx4 )(f f ) = 0.

Problem 53.

(4)

The Sawada-Kotera equation is given by

u
3u 5u
u
u 2 u
+
15u
+
= 0.
+ 45u2
+ 15
t
x
x x2
x3
x5

(1)

Show that using the dependent variable transformation

u(x, t) = 2

2 ln f (x, t)
x2

(2)

and integrating once with respect to x we arrive at

(Dx Dt + Dx6 )(f f ) = 0.

Problem 54.

The Kadomtsev-Petviashvili equation is given by



u 3 u
u
2u
+ 6u
+
= 0.
+
3
x t
x x3
y 2

(3)

(1)

Show that using the transformation

u(x, t) = 2

2 ln f (x, t)
x2

(2)

the Kadomtsev-Petviashvili equation takes the form

(Dx Dt + Dx4 + 3Dy2 )(f f ) = 0.

Problem 55.

Consider the system of partial differential equations

u
u v
= u
+
t
x x
v
u
v
w
= v
+ cu
+
t
x
x
x
w
u
w
= w
+ u
t
x
x

(3)

42 Problems and Solutions

where , , , c, are constants. Show that the system admits the converved
densities
H0 = u
1
H1 = v + ( c)u2
2

 3
1
+
u
H2 = uv +
w+
c
,

2
3
H3 = w,
= .
Problem 56.
form

6=

The polytropic gas dynamics in 1 + 1 dimensions is of the

u

u
=u
+ C
t
x
x

=
(u)
t x

where x is the space coordinate, t is the (minus physical) time coordinate,

u is the velocity, the density and = 2. Here is the polytropic
exponent. The constant C can be removed by a rescaling of . Express this
system of partial differential equations applying the Riemann invariants
r1,2 (x, t) = u(x, t)

Problem 57.

2
(+1)/2 (x, t).
+1

6= 1.

The Korteweg-de Vries equation is given by

u
u 3 u
6u
+
= 0.
t
x x3
For a steady-state pulse solution we make the ansatz


c
b
2
(x ct)
u(x, t) = sech
2
2
where

1
.
cosh(y)
Find the condition on b and c such that this ansatz is a solution of the
Korteweg-de Vries equation.
sech(y) :=

Problem 58. Consider the Korteweg-de Vries equation and its solution
given in the previous problem. Show that
Z p
|u(x, t)|dx = .

Nonlinear Partial Differential Equations

43

Hint. We have
Z

Z
sech(s)ds

Problem 59.

1
ds = 2 arctan(es ).
cosh(s)

Consider the Korteweg-de Vries equation

u
u 3 u
= 0.
6u
+
t
x x3

Show that
u(x, t) = 12

4 cosh(2x 8t) + cosh(4x 64t) + 3

(3 cosh(x 28t) + cosh(3x 36t))2

is a solution of the Korteweg-de Vries equation. This is a so-called two

soliton solution.
Problem 60.

Consider the one-dimensional Euler equations

u
u 1 p
+u
+
=0
t
x x

u
+u
+
=0
t
x
x
p
u
p
+ p
+u
=0
t
x
x

where u(x, t) is the velocity field, (x, t) is the density field and p(x, t) is
the pressure field. Here t is the time, x is the space coordinate and is the
ratio of specific heats. Find the linearized equation around u
e, pe, e.
Problem 61. Consider the classical it Heisenberg ferromagnetic equation
S
2S
=S
t
x2
where S = (S1 , S2 , S3 )T , S12 +S22 +S32 = 1 and denotes the vector product
The natural boundary conditions are S(x, t) (0, 0, 1) as |x| .
(i) Find partial differential equation under the stereographic projection
S1 =

2u
,
Q

S2 =

where Q = 1 + u2 + v 2 .
(ii) Perform a Painleve test.

2v
,
Q

S3 =

1 + u2 + v 2
Q

44 Problems and Solutions

(iii) The Heisenberg ferromagnetic equation in the form given for u, v is
gauge equivalent to the one-dimensional Schrodinger equation
u 2 v
+ 2(u2 + v 2 )v = 0
+
t
x2
v 2 u
2(u2 + v 2 )u = 0.

t
x2
Both systems of differential equations arise as consistency conditions of a
system of linear partial differential equations

= U ,
x

=V
t

where = (1 , 2 )T and U and V are 2 2 matrices. The consistency

condition is given by
U
V

+ [U, V ] = 0.
t
x
Two systems of nonlinear partial differential equations that are integrable
if there is an invertible 2 2 matrix g which depends on x and t such that
U1 = gU2 g 1 +

g 1
g ,
x

V1 = gV2 g 1 +

g 1
g .
t

Are the resonances of two gauge equivalent systems the same?

Problem 62. The system of partial differential equations of the system
of chiral field equations can be written as
u u
+
u (Jv) = 0
t
x
v v

v (Ju) = 0
t
x
where u = (u1 , u2 , u3 )T , v = (v1 , v2 , v3 ), u2 = v2 = 1, J = diag(j1 , j2 , j3 )
is a 3 3 diagonal matrix and denotes the vector product. Consider the
linear mapping M : R3 so(3)

0
u3 u2
M (u) = u3
0
u1
u2 u1
0
where so(3) is the simple Lie algebra of the 3 3 skew-symmetric matrices.
Rewrite the system of partial differential equations using M (u).
Problem 63.

The Landau-Lifshitz equation

u
2u
=u
+ u (Kv)
t
x2

Nonlinear Partial Differential Equations

45

where u = (u1 , u2 , u3 )T , u2 = 1, K = diag(k1 , k2 , k3 ) is a 3 3 diagonal

matrix and denotes the vector product. Consider the linear mapping
M : R3 so(3)

0
u3 u2
M (u) = u3
0
u1
u2 u1
0
where so(3) is the simple Lie algebra of the 3 3 skew-symmetric matrices.
Rewrite the system using M (u).
Problem 64.

Consider the nonlinear partial differential equation

u
u
=u .
t
x

Let
u(x, t) =

t
.
x

Find the partial differential equation for .

Problem 65.

(i) Show that the Burgers equation

2u
u
u
=
+u
2
t
x
x

admits the Lax reprsentation




u
+
= ,
x 2

=
+u
t
x2
x

where u is a smooth function of x and t.

(ii) Show that
[T, K] = 0
also provides the Burgers equation, where
T :=

Problem 66.

2
+u
,
2
x
x t

K :=

u
+ .
x
2

Consider the Burgers equation

2u
u
u
=
+u .
t
x2
x

Consider the operator (so-called recursion operator)

R := D +

1 u 1 u
D +
2 x
2

46 Problems and Solutions

where
D :=

,
x

D1 f (x) :=

f (s)ds.

(i) Show that applying the recursion operator R to the right-hand side of
the Burgers equation results in the partial differential equation
3u 3 2u 3 2 2u
u
+ u 2+ u
=
.
t
x3
2 x
4 x
(ii) Show that this partial differential equation can also be derived from the
linear partial differential equation

3
=
t
x3
and the transformation
(x, t) = exp
Note that
D

 Z
1
2

u
x


u(s, t)ds .


= u.

Problem 67.
tions

Find the solution of the system of partial differential equa-

Problem 68.

Consider the Kortweg-de Vries equation

f
+ f 2 = 0,
x
g
+ f g = 0,
x

f
+ fg = 0
t
g
+ g 2 = 0.
t

u 3 u
u
6u
+
= 0.
t
x x3
Show that

2
log (x, t)
x2
is a solution of the Kortweg-de Vries equation, where


cj ck
3
3
(x, t) = det jk +
exp((j + k )t (j + k )x)
.
j + k
j,k=1,...,N
u(x, t) = 2

This is the so-called N -soliton solution.

Nonlinear Partial Differential Equations

Problem 69.

47

Consider the partial differential equation

u 2 u
+
+
t
x2

u
x

2
= 0.

Let v(x, t) = exp(u(x, t)). Find the partial differential equation for v.
Problem 70.
tion

Consider the one-dimensional nonlinear Schrodinger equai

1 2
+ ||2 = 0.
+
t
2 x2

Show that
(x, t) = 2sech(s) exp(i(s, t))
with
s := 2(x (t)),
(t) = 2t + 0 ,

(s, t) :=

s + (t)

(t) = 4(2 + 2 )t + 0 .

is a solution (so-called one-soliton solution) of the nonlinear Schrordinger

equation.
Problem 71.
tions

Consider the system of nonlinear partial differential equa-

2u 2u

u + u(u2 + v 2 ) = 0,
t2 x2
(i) Show that

2v 2v 


v + v(u2 + v 2 ) = 0.
t2 x2
2

u(x, t) = tanh(s/ 2),

v(x, t) = 0

is a solution, where s := (x ct).

(ii) Show that
p
p
u(x, t) = tanh( /2s),
v(x, t) = (1 )1/2 sech( /2s)
is a solution, where s := (x ct).
(iii) Show that
p
p
v(x, t) = (1 )1/2 sech( /2s)
u(x, t) = tanh( /2s),
is a solution, where s := (x ct).
Problem 72.
tion

Consider the one-dimensional nonlinear Schrodinger equai

u 2 u
+
+ 2|u|2 u = 0,
t
x2

< x <

48 Problems and Solutions

Show that
u(x, t) = 2 exp(i(x, t))sech((x, t))
where
(x, t) = 2(x + 2( 2 2 )t) + 0 ,

(x, t) = 2(x + 4t) + 0

is a solution of the one-dimensional nonlinear Schrodinger equation.

Problem 73. The Landau-Lifshitz equation describing nonlinear spin
waves in a ferromagnet is given by
2S
S
+ S JS
=S
t
x2
where
S = (S1 , S2 , S3 )T ,

S12 + S22 + S32 = 1

and J = diag(J1 , J2 , J3 ) is a constant 3 3 diagonal matrix. Show that

w
= Lw,
x1

w
= Mw
x2

with x1 = x, x2 = it and
L=

3
X

z S

=1

M =i

3
X

z S

,,=1

3
X
S

+ 2z1 z2 z3
z1 S
x
=1

provide a Lax pair. Here 1 , 2 , 3 are the Pauli spin matrices and the spectral parameters (z1 , z2 , z3 ) constitue an algebraic coordinate of an elliptic
curve defined by
z2 z2 =

Problem 74.

1
(J J ),
4

, = 1, 2, 3.

Consider the nonlinear partial differential equation

u
2u
u
=
+ 2u .
t
x2
x

Consider the generalized Hopf-Cole transformation

Z s

v(t, x) = w(t)u(t, x) exp
dsu(s, t)
0

Nonlinear Partial Differential Equations

with
u(t, x) =

49

v(t, x)
Rs
.
w(t) + 0 dsv(s, t)

Find the differential equations for v and w.

Problem 75.
tion

Consider the one-dimensional nonlinear Schrodinger equa-

u 2 u
+ 2c|u|2 = 0
+
t
x2
where x R and c = 1. Show that it admits the solution
   

1 x 2
A
ln(t)
u(x, t) = exp it
+ 2cA2
+
.
4 t
t
t
t
i

Problem 76. Let be a positive constant. The Kadomtsev Petviashvilli

equation is given by


u 3 u
2u
u
+ 6u
+
+

= 0.
x t
x x3
y 2
Consider the one-soliton solution
u(x, y, t) = 2k12 sech2 (k1 x + k2 y t)
where sech(x) 1/ cosh(x) 2/(ex + ex ). Find the dispersion relation
(k1 , k2 ).
Problem 77. Find the partial differential equation given by the condition


u
u/z1
det
= 0.
u/z2 2 u/z1 z2
Find a solution of the partial differential equation.
Problem 78. A one-dimensional Schrodinger equation with cubic nonlinearity is given by
i~

~2 2
=
g,
t
2m x2

:=

where g > 0.
(i) Show that the partial differential equation admits the (soliton) solution


 mv
 ~

m2 v 2
2u
s (x, t) = exp i
(x ut)
, 2 =
1

.
~
gm cosh((x vt))
~2
v

50 Problems and Solutions

The soliton moves with group velocity v. The phase velocity u must be
u < v/2.
(ii) Show that the partial differential equation is Galileo-invariant. This
means that any solution of partial differential equation can be mapped into
another solution via the Galileo boost



1
i
mV x V t
(t, x V t).
x x V t,
(x, t) exp
~
2
Show that the soliton can be brought to rest.
Problem 79.

The KP-equations are given by



2
u
u 3 u
2 u
=
3
+ 6u
+
x t
x x3
y 2

with = i and = 1. Show that this equation is an integrability condition on

2
L
+
+ u = 0
y
x2


Z x

u
u(x0 , y) 0
M
+ 4 3 + 6u
+3

dx + = 0.
t
x
x
x
y

Problem 80.

Show that the nonlinear partial differential equation

2u
u
= (1 + u2 ) 2 u
t
x

u
x

2

is transformated under the transformation u 7 u/ 1 + u2 into the nonlinear diffusion equation

u
2
=
(tanh1 u).
t
x2
Problem 81.

Consider the metric tensor field

g = g11 (x1 , x2 )dx1 dx1 +g12 (x1 , x2 )dx1 dx2 +g21 (x1 , x2 )dx2 dx1 +g22 (x1 , x2 )dx2 dx2
with g12 = g21 and the gjk are smooth functions of x1 , x2 . Let det(g)
g11 g22 g12 g21 6= 0 and

g11
g22
g12
2
R=
det g11 /x1 g22 /x1 g12 /x1 .
(det(g))2
g11 /x2 g22 /x2 g12 /x2

Nonlinear Partial Differential Equations

51

Find solutions of the partial differential equation R = 0. Find solutions of

the partial differential equation R = 1.
Problem 82.
tion

Consider the one-dimensional nonlinear Schrodinger equai

w 2 w
+ 2|w|2 w = 0.
+
t
x2

Consider the ansatz

w(x, t) = exp(it)u(x).
Find the ordinary differential equation for u.
Problem 83. Consider the system of partial differential differential equations

j1 0 0
S
2S
=S
+ S (JS), J = 0 j2 0
t
x2
0 0 j3
where S = (S1 , S2 , S3 )T and S12 +S22 +S32 = 1. Express the partial differental
equation using p(x, t) and p(x, t) given by
p
p
S1 = 1 p2 cos(q), S2 = 1 p2 sin(q), S3 = p.
Problem 84.

Consider the two-dimensional sine-Gordon equation

1 2u
2u 2u
+

= sin(u).
x21
x22
c2 t2

Let
u(x1 , x2 , t) = 4 arctan(v(x1 , x2 , t)).
Find the partial differential equation for v. Separate this partial differential equation into a linear part and nonlinear part. Solve these partial
differential equations to find solutions for the two-dimensional sine-Gordon
equation. Note that
sin(4) 4 sin() cos() 8 sin3 () cos()
sin(arctan()) =

,
1 + 2

cos(arctan()) =

and therefore
sin(4 arctan(v)) =

1
1 + 2

4v(1 v 2 )
.
(1 + v 2 )2

Furthermore
2
2v
arctan(v) =
x21
(1 + v 2 )2

v
x1

2
+

1 2v
.
1 + v 2 x21

52 Problems and Solutions

Problem 85.

(i) Find a non-zero vector field in R3 such that

V curlV = 0.

(ii) Find a non-zero vector field in R3 such that

V curl(V ) = 0.

Problem 86. Let f : R R be a smooth function. Consider the partial

differential equation
u f (u(x)) 3 u
= 0.
+
+
t
x
x3
Find solutions of the form u(x, t) = (x ct) (traveling wave solutions)
where is a smooth function. Integrate the obtained ordinary differential
equation.
Problem 87.
tion)

Consider the partial differential equation (Thomas equa-

2u
u
u u u
+a
+b
+
=0
xt
x
t
x t
where a, b are constants. Show that the equation can be linearized with the
transformation
u(x, t) = bx at + ln(v(x, t).
Problem 88.

The Korteweg-de Vries equation is given by

u
u 3 u
6u
+
= 0.
t
x x3

Setting u = v/x we obtain the equation

v 2 v
4v
2v
6
+
= 0.
2
xt
x x
x4
(i) Show that this equation can be dervied from the Lagrangian density
1
1
L = vx vt + (vx )3 + (vxx )2
2
2
where the Lagrange equation is given by






L

L
2
L
L
+

= 0.
t vt
x vx
x2 vxx
v

Nonlinear Partial Differential Equations

53

(ii) Show the Hamiltonian density is given by

H = vt

Problem 89.

L
1
L = (vx )3 (vxx )2 .
vt
2

Show that the Korteweg-de Vries equation

u
u 3 u
=0
6u
+
t
x x3

admits the solution

c
u(x, t) = sech2
2
Show that


c
(x ct) .
2

p
dx |u(x, t)| = .

Problem 90.

Consider the one-dimensional sine-Gordon equation

2u
2u
c2 2 + 2 sin(u) = 0.
2
t
x

Show that
u+ (x, t) = 4 arctan(exp(d1 (x vt X) cosh())
u (x, t) = 4 arctan(exp(d1 (x vt X) cosh())
are solutions of the one-dimensional sine-Gordon equation, where v =
c tanh(), d = c/. Discuss.
Problem 91. Let (s, t) be the curvature and (s, t) be the torsion with
s and t being the arclength and time, respectively. Consider the complex
valued function
Z s
w(s, t) = (s, t) exp(i
ds0 (s0 , t)).
0

Show that if the motion is described by

2
r
r 2 r = b
t
s
s
where b is the binormal unit vector, then w(s, t) satisfies the nonlinear
Schr
odinger equation
i

2
1

w + 2 w + |w|2 w = 0.
t
s
2

54 Problems and Solutions

Problem 92.
tion

Consider the first order system of partial differential equa-

u
u
+A
=0
t
x
where u = (u1 , u2 , u3 )T and the 3 3 matrix is given by

u1
u2
u3
u2 u1
0
u3
0
u1
where and are constants. Along a characteristic curve C : x = x(s), t =
t(s) for the system of partial differential equation one has
det(A I3 ) = 0,

x0 (s)
.
t0 (s)

Find the Riemann invariants.

Problem 93.

Consider the system of partial differential equations

V
V
+
= Z2 V W
t
x
W
W

= Z2 V W
t
x
Z
1
1
= Z2 + V W
t
2
2

Let N := V + W + 4Z and J := V W . Find N/t + J/x and

J/t + V /x + W/x.
Problem 94.

Consider the Navier-Stokes equations

v
1
+ (v )v = p + v.
t

(i) Show that in the limit 0 the Navier-Stokes equation are invariant
under the scaling transformation ( > 0)
r r,

v h v,

t 1h t.

(ii) Show that for finite one finds invariance of the Navier-Stokes equation
if 1+h .
Problem 95.

(i) Show that

u(x, t) =

k 2 /2
,
(cosh((kx t)/2))2

k3 =

Nonlinear Partial Differential Equations

is a solution (solitary wave solution) of the Korteweg de Vries equation
u
u 3 u
= 0.
6u
+
t
x x3
(ii) Let
u(x, t) = 2
Show that F (x, t) = 1 + ekxt .

2
ln(F (x, t)).
x2

55

Chapter 3

Lie Symmetry Methods

Problem 1.

Show that the partial differential equation

u
u
=u
x2
x1

(1)

is invariant under the transformation

x
1 (x1 , x2 , ) = x1 ,

x
2 (x1 , x2 , ) = x1 + x2

u
(
x1 (x1 , x2 ), x
2 (x1 , x2 ), ) =

u(x1 , x2 )
1 u(x1 , x2 )

where  is a real parameter.

Problem 2.

Consider the Korteweg-de Vries equation

u
u 3 u
+ 12u
+
= 0.
t
x x3

Show that the scaling

x0 = cx,

t0 = c3 t,

u(x, t) = c2 u0 (x0 , t0 )

leaves the Korteweg-de Vries equation invariant.

Problem 3.

Consider the partial differential equation

(u, (u)2 , u) = 0
56

(2a)
(2b)

Lie Symmetry Methods

57

where is an analytic function, u depends on x0 , x1 , . . . , xn and


2 
2

2
u u
u
u
u
u
2
u :=

, (u) :=

.
x0 x1
xn
x0
x1
xn
Show that the equation is invariant under the Poincare algebra

,
,...,
x0 x1
xn
x0

+ x1
, x0
+ x2
, . . . , x0
+ xn
x1
x0
x2
x0
xn
x0
xj

xk
,
xk
xk

j, k = 1, 2, . . . , n

j 6= k .

Problem 4. Consider the nonlinear partial differential equation (BornInfeld equation)

 2 ! 2
 2 ! 2
u
u
u u 2 u
u
u
1
+2
1+
= 0.
(1)
t
x2
x t xt
x
t2
Show that this equation admits the following seven Lie symmetry generators

,
x

Z1 =

Z2 =

,
t

Z3 =

+t ,
Z5 = u
x
t
x
x
u

Z6 = u + t ,
Z7 = x
+t +u
.
t
u
x
t
u
Z4 = x

Problem 5.

(2)

Consider the Harry-Dym equation

u
3u
u3 3 = 0.
t
x

(1)

(i) Find the Lie symmetry vector fields.

(ii) Compute the flow for one of the Lie symmetry vector fields.
Problem 6. Consider the Magneto-Hydro-Dynamics equations and carry
out the Lie symmetry analysis. We neglect dissipative effects, and thus
restrict the analysis to the ideal case. The equations are given by

+ (v ) + v = 0
t

(1a)

58 Problems and Solutions




v
1
+ (v )v + (p + H2 ) (H )H = 0
t
2
H
+ (v )H + H v (H )v = 0
t

H=0
 
p
+ (v )
=0

(1b)
(1c)
(1d)
(1e)

with pressure p, mass density , coefficient of viscosity , fluid velocity v

and magnetic field H.
Problem 7. The stimulated Raman scattering equations in a symmetric
form are given by
v2
= ia2 v3 v1 ,
x

v1
= ia1 v2 v3 ,
x

v3
= ia3 v1 v2 .
t

(1)

The ai are real coupling constants that can be normalized to 1 and we

have set v1 = iA1 , v2 = A2 , v3 = X, where A1 , A2 and X are the complex
pump, Stokes and material excitation wave envelopes, respectively. The
stars denote complex conjugation. Equations (1) are actually a special
degenerate case of the full three wave resonant interaction equations. Find
the similarity solutions using the similarity ansatz
s(x, t) := xt

similarity variable

(2)

and
v1 (x, t) = t

1+
2

1 (s) exp(i1 (s)),

v2 (x, t) = t

1+
2

exp(ialnt)2 (s) exp(i2 (s))

1 ialnx
e
3 (s) exp(i3 (s))
x
where i 0, 0 i < 2 and  = 1.
v3 (x, t) =

(3)

Problem 8.
The motion of an inviscid, incompressible-ideal fluid is
goverened by the system of equations
u
+ u u = p
t

(1)

u=0

(2)

first obtained by Euler. Here u = (u, v, w) are the components of the

three-dimensional velocity field and p the pressure of the fluid at a position
x = (x, y, z).

Lie Symmetry Methods

59

(i) Show that the Lie symmetry group of the Euler equations in three
dimensions is generated by the vector fields
a

=a

+ a0
a00 x ,
x
u
p
vc = c

s1 = x

+ c0
c00 z ,
z
w
p

+y
+z
+t ,
x
y
z
t

r2 = y

x
+v
u ,
x
y
u
v
rx = z

vb = b

s2 = t

+ b0
b00 y
y
v
p
v0 =

u
v
w
2p
t
u
v
w
p

ry = z

x
+w
u
x
z
u
w

y
+w
v
,
y
z
v
w

vq = q

(4)

where a, b, c, q are arbitrary functions of t.

(ii) Show that these vector fields exponentiate to familiar one-parameter
symmetry groups of the Euler equations. For instance, a linear combination
of the first three fields va + vb + vc generates the group transformations
1
(x, t, up) (x + a, t, u + a0 , p x a00 + 2 a a00 )
2
where  is the group parameter, and a := (a, b, c). These represent changes
to arbitrarily moving coordinate systems, and have the interesting consequence that for a fluid with no free surfaces, the only essential effect
of changing to a moving coordinate frame is to add an extra component,
namely,
1
x a + 2 a a00
2
to the resulting pressure.
(iii) Show that the group generated by v0 is that of time translations,
reflecting the time-independence of the system. (iv) Show that the next
two groups are scaling transformations
s1 : (x, tu, p) x, t, u, p)
s2 : (x, t, u, p) (x, t, 1 u, 2 p).
The vector fields rx , ry , r2 generate the orthogonal group SO(3) of simultaneous rotations of space and associated velocity field; e.g., rx is just an
infinitesimal rotation around the x axis. The final group indicates that
arbitrary functions of t can be added to the pressure.

60 Problems and Solutions

Problem 9.

Consider the nonlinear diffusion equation



u

1 u
=
.
t
x u2 x

(1)

Within the jet bundle formalism we consider instead of (1) the submanifold
u2x
uxx
+
2
=0
u2
u3

(2)

uxxx
6ux uxx
6u3x
+

= 0, . . .
u2
u3
u4

(3)

x = dux uxx dx utx dt, . . .

(4)

F (x, t, u, ux , uxx ) ut
and its differential consequences
Fx (x, t, u, ux , ) utx
with the contact forms
= du ux dx ut dt,

The non-linear partial differential equation (1) admits the Lie point symmetries (infinitesimal generator)
Xv = ux

,
u

Tv = ut

,
Vv = (xux + u) .
(5)
u
u
The subscript v denotes the vertical vector fields. The non-linear partial
differential equation (1) also admits Lie-B
acklund vector fields. The first in
the hierarchy is


uxxx
9ux uxx
12u3x
.
(6)
U=

+ 5
u3
u4
u
u
Sv = (xux 2tut )

Find a similarity solution using a linear combination of the Lie symmetry

vector field Tv and U .
Problem 10.

Consider the nonlinear partial differential equation



3u
u
u
+
u
+
c
=0
x3
t
x

where c is a constant. Show that this partial differential equation admits

the Lie symmetry vector field
V =t

+ ct
+u
.
t
x
u

Lie Symmetry Methods

61

First solve the first order autonomous system of differential equations (initial value problem) which belongs to the vector field V , i.e.
dt0
= t0 ,
d

dx0
= ct0 ,
d

du0
= u0 .
d

From this solution of the initial value problem find the transformation between the prime and unprime system. Then using differentiation and the
chain rule show that the prime and unprime system have the same form.
Problem 11.

Consider the nonlinear Schrodinger equation

i

w
+ w = F (w)
t

with a nonlinearity F : C C and a complex valued w(t, x1 , . . . , xn ). We

assume that for the given nonlinearity the energy remains bounded. We
also assume that there exists a C 1 -function G : R+ R with G(0) = 0
such that F (z) = G0 (|z|2 )z for all z C.
(i) Show that the nonlinear Schrodinger is translation invariant.
(ii) Show that it is phase invariant, i.e. w ei w.
(iii) Show that it is Galilean invariant
2

w(t, x) eivx/2 ei|v|

for any velocity v.
(iv) Show that the mass defined by
Z
M (w) =

t/4

w(t, x vt)

|w(t, x)|2 dx

Rn

is a conserved quantity.
(v) Show that the Hamilton density
Z
1
1
|w(t, x)|2 + G(|w(t, x)|2 )dx
H(w) =
2
2
n
R
is a conserved quantity.
Problem 12. For a barotropic fluid of index the Navier-Stokes equation
read, in one space dimension
c
c
( 1) v
+v
+
c
=0
t
x
2
x

continuity equation

v
v
2
c
2v
+v
+
c

=0
t
x ( 1) x x2

Eulers equation

(1a)
(1b)

62 Problems and Solutions

where v represents the fluids velocity and c the sound speed. For the class
of solutions characterized by a vanishing presure (i.e., c = 0), the above
system reduces to the Burgers equation. We assume for simplicity the
value = 3 in what follows.
(i) Show that the velocity potential exists, and it is given by the following
pair of equations

= v,
x

v
1
=
(v 2 + c2 ).
t
x 2

(2)

Hint. From (2) with

2
2
=
tx
xt
the Euler equation (1b) follows. Thus the condition of integrability of
precisely coincides with the Euler equation (1b).
(ii) Consider the similarity ansatz
x
v(x, t) = f (s) ,
t

c(x, t) = g(s)

x
t

(3)

where the similarity variable s is given by

x
s(x, t) := .
t

(4)

Show that the Navier-Stokes equation yields the following system of ordinary differential equations


df
1 dg
s g+s f
+ g(2f 1) = 0
(5a)
ds
2 ds
2

d2 f
df
+
2
ds
ds




dg
4
+ s(1 2f ) + 2f (1 f ) = 2g s + g .
s
ds

(5b)

(iii) Show that the continuity equation (1a) admits of a first integral, expressing the law of mass conservation, namely
s2 g(2f 1) = C

(6)

where C is a constant. (iv) Show that g can be eliminated, and we obtain

a second-order ordinary differential equation for the function f .
Problem 13.

The nonlinear partial differential equation

u
2u
+
+ u2 = 0
x1 x2
x1

(1)

Lie Symmetry Methods

63

describes the relaxation to a Maxwell distribution. The symmetry vector

fields are given by

ex2 u
.
x2
u
(2)
Construct a similarity ansatz from the symmetry vector field



+ c2
+ c3 x1
+u
(3)
Z = c1
x1
x2
x1
u
Z1 =

,
x1

Z2 =

,
x2

Z3 = x1

+u ,
x1
u

Z4 = ex2

and find the corresponding ordinary differential equation. Here c1 , c2 , c3

R.
Problem 14.

Consider the Kortweg de Vries equation

u 3 u
u
6u
+
= 0.
t
x x3

Insert the similarity solution

u(x, t) =

w(s)
,
(3t)2/3

s=

x
(3t)1/3

where s the similarity variable and find the ordinary differential equation.
Problem 15. Let u be the velocity field and p the pressure. Show that
the Navier-Stokes equations
u
+ (u )u u + p = 0,
t

divu = 0

admits the Lie symmetry vector fields

,
t

,
x1

,
x2

x3

+
, t
+
, t
+
x1
u1
x2
u2
x3
u3

x2
+ u1
u2
,
x1
x2
x1
u2
u1

x2
x3
+ u2
u3
,
x3
x2
u3
u2

x3
x1
+ u3
u1
x1
x3
u1
u3

2t + x1
+ x2
+ x3
u1
u2
u3
2p .
t
x1
x2
x3
u1
u2
u3
p
t

64 Problems and Solutions

Find the commutators of the vector
pfields and thus show that we have a
basis of a Lie algebra. Show that (x21 + x22 )/x3 is a similarity variable.
Do which vector fields does it belong.
Problem 16. The AB system in its canonical form is given by the coupled
system of partial differential equations
2Q
= QS,

S
1 |Q|2
=

where and are semi-characteristic coordinates, Q (complex valued) and

S are the wave amplitudes satisfying the normalization condition


 Q 2
2


  + S = 1 .
Show that these system of partial differential equations can be written as
a compatibility condition
2 1,2
2 1,2
=

of two linear systems of partial differential equations

Problem 17.

1
= F 1 + G2 ,

1
= A1 + B2

2
= H1 F 2 ,

2
= C1 A2 .

Show that the system of partial differential equations

u1 u2
=
t
y
8 u1
u2 1 3 u1
=
+ u1
3
t
3 y
3
y

admits the (scaling) Lie symmetry vector field

S = 2t

Problem 18.

y
+ 2u1
+ 3u2
.
t
y
u1
u2

Show that the partial differential equation

 2
u u
2u
=
t x
x2

Lie Symmetry Methods

65

admits the Lie symmetry vector field

V =

Problem 19.

xu
+t
.
t 2 x
u

Consider the Kuramoto-Sivashinsky equation

4u
u
u
2u
+u
+ a 2 + b 4 = 0.
t
x
x
x

Show that the equation is invariant under the Galilean transformation

(u, x, t) 7 (u + c, x ct, t) .

Problem 20. Find the Lie symmetry vector fields for the Monge-Ampere
equations
 2 2
2u 2u
u

= K
2
2
t x
xt
where K = +1, K = 0, K = 1.
Problem 21. Let u = (u1 , u2 , u3 )T , v = (v1 , v2 , v3 )T and u21 +u22 +u23 = 1,
v12 + v22 + v32 = 1. Let be the vector product. The O(3) chiral field
equations are the system of partial differential equations
u u
+
+ u Rv = 0,
t
x

v v

+ v Ru = 0
t
x

where R is a 3 3 diagonal matrix with non-negative entries. Show that

this system of partial differential equations admits a Lax pair (as 4 4
matrices) L, M , i.e. [L, M ] = 0.
Problem 22. Consider the Schrodinger-Newton equation (MKSA-system)

~2
+ U =
2m
U = 4Gm2 ||2

where m is the mass, G the gravitational constant and the energy eigenvalues. The normalization consition is
Z
|(x1 , x2 , x3 )|2 dx1 dx2 dx3 = 1 .
R3

Write the Schr

odinger-Newton equation in dimension less form. Then find
the Lie symmetries of this system of partial differential equation.

66 Problems and Solutions

Problem 23.

Show that the one-dimensional wave equation

2u
1 2u
=
2
2
c t
x2

is invariant under


x0 (x, t)
ct0 (x, t)


=

cosh()
sinh()

sinh()
cosh()



u0 (x0 (x, t), t0 (x, t)) = u(x, t).

x
ct

Lie Symmetry Methods

67

68

Bibliography

Bibliography
Books
Courant R. and Hilbert D.
Methods of Mathematical Physics, Volume 1, Wiley-VCH Verlag (2004)
Courant R. and Hilbert D.
Methods of Mathematical Physics, Volume 2, Wiley-VCH Verlag (2008)
DiBenedetto E.
Partial Differential Equations, second edition, Springer (2010)
John F.
Partial Differential Equations, Volume I, Springer 1982
Sneddon I.
Elements of Partial Differential Equations, McGraw-Hill (1957)
Steeb W.-H.
The Nonlinear Workbook, fourth edition, World Scientific Publishing, Singapore, 2008
Steeb W.-H.
Problems and Solutions in Theoretical and Mathematical Physics, Second
Edition, Volume I: Introductory Level
World Scientific Publishing, Singapore (2003)
Steeb W.-H.
Problems and Solutions in Theoretical and Mathematical Physics, Second
Edition, Volume II: Adanced Level
World Scientific Publishing, Singapore (2003)
Papers
Ouroushev D., Martinov N. and Grigorov A., An approach for solving
the two-dimensional sine-Gordon equation, J. Phys. A: Math. Gen. 24,
L527-528 (1991)

Index
Antikink, 25
Boltzmann equation, 28, 31
Born-Infeld equation, 57
Born-Infeld type equation, 26
Boussinesq-Oberbech equations, 36
Broadwell model, 30
Burgers equation, 20, 45
Carleman model, 28, 29
Cauchy principal value, 32
Central difference scheme, 7
Chiral field equations, 44
Cole-Hopf transformation, 20
Conservation law, 24, 27
Conserved density, 24
Conserved flux, 24
dAlembert equation, 23
Diffusion equation, 5
Direct method, 33
Dispersion relation, 6, 49
Drift diffusion equation, 1, 4
Electron-beam plasma system, 34
Energy center, 25
Entropy density, 30
Entropy flux, 30
Euler-Lagrange equation, 24
Fisher equation, 13
Fitzhugh-Nagum equation, 32
Fokker-Planck equation, 10
Galilean transformation, 65
Ginzburg-Landau equation, 37

H thoerm, 28
Harry-Dym equation, 57
Heisenberg ferromagnetic equation,
43
Hilbert transform, 32
Hodograph invariant, 25
Hodograph transfromation, 25
Hopf-Cole transformation, 18
Inverse scattering transform, 40
Kadomtsev Petviashvilli equation, 49
Kadomtsev-Petviashvili equation, 41
Kink, 25
KLamb substitution, 22
Korteweg-de Vries equation, 13, 15
Korteweg-de Vries-Burgers equation,
13
Landau-Lifshitz equation, 44
Langmuir oscillations, 19
Laplace equation, 7
Laplace transformation, 11
Lax pair, 12
Lie-Backlund vector fields, 60
Magneto-Hydro-Dynamics equations,
57
Maxwells equation, 6
Monge-Ampere equation, 26
Navier-Stokes equation, 19, 61
Nonlinear Dirac equation, 33
Nonlinear one-dimensional Schrodinger
equation, 37
Nonlinear Schrodinger equation, 17
69

70
Pauli matrices, 33
Pauli spin matrices, 38
periodic boundary conditions, 37
Polar form, 9
Polytropic gas dynamics, 42
Product ansatz, 4
Raman scattering equations, 58
Recursion operator, 45
Reduced variable, 33
Renyi entropies, 29
Residue theorem, 32
Riemann invariants, 42
Sawada-Kotera equation, 41
Shock wave solutions, 13
Sine-Gordon equation, 12, 14, 26, 38
Sine-Hilbert equation, 31
Standing waves, 33
Stationary states, 33
torus, 5
Traveling wave solution, 12
Two-dimensional nonlinear Schrodinger
equation, 27
Velocity potential, 62
Zakharov-Kuznetsov equation, 16

Index