Royal Holloway University of London Department of Physics

PH2130 Mathematical Methods 1

Solutions of Laplaces equation in 3d
Motivation
The general form of Laplaces equation is:
=
2
0 ;
it contains the laplacian, and nothing else. This section will examine the form of the
solutions of Laplaces equation in cartesian coordinates and in cylindrical and spherical
polar coordinates. Of course it is nice to know how to solve Laplaces equation in these
coordinate systems, particularly recalling that the choice of coordinate system is
generally determined by the symmetry of the boundary conditions.
But there is a much more important reason why these solutions are of interest. Similar
parts of different partial differential equations separate off in the same way. Thus, for
example, lets take the (time independent) Schrdinger equation with a spherically
symmetric potential. Now the angular part of the equation is all in the angular part of the
laplacian. So the angular part of this Schrdinger equation is equivalent to the angular
part of Laplaces equation in spherical polar coordinates. Then the angular part of the
solution to the Schrdinger equation with a spherically symmetric potential will be
exactly the same as the angular part of the solution to Laplaces equation in spherical
polar coordinates: the spherical harmonics we shall discover below.
Of course the radial part of the solution will be different because here the potential will
have an effect.
So although we are here examining solutions to Laplaces equation, the solutions we shall
find will have relevance to other equations which involve the laplacian.
Rectangular Cartesian Coordinates
In rectangular cartesian coordinates Laplaces equation takes the form

=
2
2
2
2
2
2
0

x y z
.
The solution by the separation of variables method is accomplished in a number of steps.
Step 1: Write the field variable as a product of functions of the independent variables.
x y z X x Y y Z z , , = .
Step 2: Substitute the product solution into the partial differential equation. The
derivatives are now total derivatives.
d
d
d
d
d
d
2
2
2
2
2
2
0
X x
x
Y y Z z X x
Y y
y
Z z X x Y y
Z z
z





+ + = .
Step 3: Divide through by the product expression for the solution.
Royal Holloway University of London Department of Physics
PH2130 Mathematical Methods 2
1 1 1
0
2
2
2
2
2
2
X x
X x
x Y y
Y y
y Z z
Z z
z




d
d
d
d
d
d
+ + = .
Now the first term is a function of the independent variable x only, the second term a
function of the independent variable y only and the last term a function only of the
independent variable z. Since x, y, and z are independent of each other, each of the three
terms in the equation must be constant and their sum equal to zero. We will set the first
term equal to a, the second term equal to b, and the third term equal to + a b . There
may well be restrictions on the allowed values of the separation constants from the
boundary conditions on the system.
We have the three ordinary differential equations
d
d
d
d
d
d
2
2
2
2
2
2
0
0
0
X x
x
aX x
Y y
y
bY y
Z z
z
a b Z z






=
=
+ + = .
These are three Simple Harmonic Oscillator equations and their solutions are
X x e
Y y e
Z z e
ax
ay
i a bz



=
=
=

+
const
const
const .
The solution, corresponding to particular values of a and b is then

ab
ax ay i a bz
x y z e e e , , =
+
const .
But since the original equation is linear, any linear combination of possible solutions is
also a solution. So we may write the general form of the solution as
x y z C e e e
ab
ax ay i a bz
ab
, , . =
+

The constants C
ab
remain to be determined from the boundary conditions that the
particular solution must satisfy.
In two dimensions (let us say there is no dependence on the z coordinate), we may put
a b + = 0 and we obtain solutions of the form
x y e e
ax i ay
, . =

const
Royal Holloway University of London Department of Physics
PH2130 Mathematical Methods 3
Cylindrical Polar Coordinates
In cylindrical polar coordinates Laplaces equation takes the form

=
2
2 2
2
2
2
2
1 1
0

z
.
We proceed by the three standard steps for solution by the separation of variables
method.
Step 1: Write the field variable as a product of functions of the independent variables.
, , z R Z z = .
Step 2: Substitute the product solution into the partial differential equation. The
derivatives are now total derivatives.
d
d
d
d
d
d
d
d
2
2 2
2
2
2
2
1 1
0
R
Z z
R
Z z R Z z R
Z z
z

+ + + = .
Step 3: Divide through by the product expression for the solution.
1 1 1 1 1 1
0
2
2 2
2
2
2
2
R
R
R
R
Z z
Z z
z

d
d
d
d
d
d
d
d
+ + + =

.
Now the last term is a function of z only, and so it must be a constant. Lets set this
constant equal to a. Then we have an ordinary differential equation for Z:
d
d
2
2
0
Z z
z
aZ z

= ,
a familiar SHO equation.
We are left with an equation in and :
1 1 1 1 1
0
2
2 2
2
2
R
R
R
R
a

d
d
d
d
d
d
+ + + =

.
This may be separated by multiplying through by
2
, giving

2
2
2
2
2
2
1 1 1
0
R
R
R
R
a





d
d
d
d
d
d
+ + + =

.
The last term may now be separated. For convenience we shall use the separation
constant n
2
; we shall justify this choice later. Separating off the last term gives the
ordinary differential equation for as
d
d
2
2
2
0

+ = n ,
a straightforward SHM equation, together with the ODE for the radial function, which we
write in standard form as:
Royal Holloway University of London Department of Physics
PH2130 Mathematical Methods 4
d
d
d
d
2
2
2
2
1
0
R R
a
n
R

+ +

= .
This is Bessels equation of order n. We have previously encountered Bessels equation
of order zero when we studied the (circularly symmetric) vibrations of a circular drum.
Now we see how the equation arises in the general case.
Allowed values for n
The solution to the SHM equation for is, to within an arbitrary constant


= e
in
.
Now there is an important property of angular variables: as the angle increases the path
traced out returns again and again to the same point. But the field variable, which we are
trying to solve for, must have a given value at a given point; it must be single-valued.
The physical requirement of single-valuedness translates into the mathematical
requirement that
= + 2 .
So if

= e
in
is a solution to the equation, then so is


= e e
in in2
.
And for this to be the case, we must have
e
in2
1

= ,
which is only true when n is a positive or negative integer.
Previously we have found that restrictions on the allowed values of the separation
constants are imposed by boundary conditions. In this case restrictions on the allowed
values of the separation constant are imposed by the requirements of single-valuedness of
functions of angular variables. This is a second cause of quantisation.
It follows from this discussion that the radial solutions of cylindrical problems will
involve Bessel functions of integer order n.
Spherical Polar Coordinates
In spherical polar coordinates Laplaces equation takes the form

=
r
r
r
2
2
2
2
1 1
0
sin
sin
sin

.
We proceed by the three standard steps for solution by the separation of variables
method.
Step 1: Write the field variable as a product of functions of the independent variables.
r R r , , = .
Step 2: Substitute the product solution into the partial differential equation. The
derivatives are now total derivatives.
Royal Holloway University of London Department of Physics
PH2130 Mathematical Methods 5
d
d
d
d
d
d
d
d
d
d r
r
R r
r
R r R r
2
2
2
2
0

+ =

sin
sin
sin
.
Step 3: Divide through by the product expression for the solution.
1 1 1
0
2
2
2
2
R r r
r
R r
r




d
d
d
d
d
d
d
d
d
d

+ =

sin
sin
sin
Now the first term is dependent only on r, thus it must be constant and we choose
l l +1 as the separation constant. This choice must be justified later. This gives us the
ordinary differential equation for the R r function:
1
1
2
R r r
r
R r
r
l l



d
d
d
d

= +
or
d
d
d
d r
r
R r
r
l l R r
2
1 0

+ = .
This equation has solutions of the form
R r Ar Br
l l


= +
+1
.
Because the solutions of this equation are quite simple, particularly since the solutions
can be expressed in terms of the elementary functions, this equation has no special name.
However for completeness, we shall refer to it as the spherical R equation.
Now we have separated off the spherical R equation, the remainder of the Laplace
equation is
l l + +

+ = 1
1
0
2
2
2





sin
sin
sin

d
d
d
d
d
d
,
where we have tidied things up so that the last term is a function only of .
The last term is thus now separable, and similar to the cylindrical case, we shall use m
2
.
The justification for this is the same as in the cylindrical case: the single-valuedness of
the solutions requires that m be a positive or negative integer. The equation for is
then simply an SHM equation
d
d
2
2
2
0

+ = m .
When the part is separated off we are left with the equation for :
l l m + +

= 1 0
2 2


sin sin sin

d
d
d
d
.
This equation is conventionally transformed into one of the standard equations through
change of the independent variable to u by:
Royal Holloway University of London Department of Physics
PH2130 Mathematical Methods 6
u = cos .
Now
d
d
u

= sin
so that
d
d
d
d
d
d
d
d
= =
u
u
u
sin .
Substituting this into the equation gives
l l m
u u
+ +

= 1 0
2 2 2 2

sin sin sin
d
d
d
d
and eliminating in favour of u:
l l u m u u
u
u
u
u
+ +

= 1 1 1 1 0
2 2 2 2

d
d
d
d
where now is regarded as a function of u. Conventionally this is written (not quite in
standard form) as
1 2 1
1
0
2
2
2
2
+ +

= u
u
u
u
u
u
l l
m
u
u



d
d
d
d
2

.
This is the Associated Legendre equation. In the particular case that m= 0we get the
simpler equation
1 2 1 0
2
2
+ + = u
u
u
u
u
u
l l u



d
d
d
d
2

known as Legendres equation. You should recognise that the m= 0 case is when we are
considering cylindrically symmetric solutions.
Summary of special equations
We are now in a position to list the various ordinary differential equations we have
discovered in looking at solutions of Laplaces equation in different coordinate systems.
1 The SHM equation.
d
d
2
2
2
0
Z z
z
n Z z

+ =
Here the independent variable z can be a linear or an angular variable. In the linear case
boundary conditions will usually restrict the allowed values of the separation constant n.
In the angular case the requirement that the solution be single-valued restricts the allowed
values of n to integers.
The solutions of the SHM equation are sines and cosines (equivalently complex
exponentials).
Royal Holloway University of London Department of Physics
PH2130 Mathematical Methods 7
2 Bessels equation
d
d
d
d
2
2
2
2
1
0
R R
a
n
R

+ +

=
Recall that n is restricted to integer values, following from the SHM equation for .
The solutions to Bessels equation are the Bessel functions J r
n
and Y r
n
. These are
known to Mathematica as BesselJ[n, r] and BesselY[n, r].
3 The spherical R equation
d
d
d
d r
r
R r
r
l l R r
2
1 0

+ = .
This equation has simple solutions:
R r Ar Br
l l


= +
+1
;
we dont need any special functions.
Although we have not shown this, the separation constant l is restricted to integer
variables by the requirement that solutions to the equation are a single-valued function
of its angular argument .
4 Legendres equation
1 2 1 0
2
2
+ + = u
u
u
u
u
u
l l u



d
d
d
d
2

.
Expressed as functions of u = cos , the solutions of Legendres equation are
polynomials, known as Legendre polynomials, P u
l
. Since u = cos , it follows that the
variable u ranges between < < 1 1 u and this explains the interval for Legendre
polynomials.
Mathematica knows the Legendre polynomials as LegendreP[l, u].
5 The associated Legendre equation
1 2 1
1
0
2
2
2
2
+ +

= u
u
u
u
u
u
l l
m
u
u



d
d
d
d
2

.
When expressed as functions of u = cos , the solutions of the associated Legendre
equation are polynomials, known as the associated Legendre polynomials, P u
l
m
.
Mathematica knows the associated Legendre polynomials as LegendreP[l, m, u].
Spherical Harmonics
The angular part of the solutions of Laplaces equation (and any other equation involving

2
which has spherical symmetry) is contained in the product of the azimuthal function
and the polar function . The azimuthal function will comprise a
complex exponential e
im
and the polar function will be a solution of the associated
Royal Holloway University of London Department of Physics
PH2130 Mathematical Methods 8
Legendre equation P
l
m
cos . Thus a solution corresponding to given values of the
separation constants l and m will be
Y P e
l
m
l
m im


, cos
to within an arbitrary factor.
Since these parts always go together in this way for spherical problems the Y
l
m
, are
given their own name; they are called spherical harmonics. By convention the
assocoated Legendre polynomials and the spherical harmonic have their own (different)
normalisation conventions.
Mathematica knows the spherical harmonics as SphericalHarmonicY[l, m, , ]