Chebyshev Functions and their use for two-dimensional electrostatic

problems in elliptic coordinates

RAUL PEREZ-ENRIQUEZ1, IVAN MARIN-ENRIQUEZ2, J. L. MARIN2 AND R. RIERA2
1
Departamento de Fisica
Universidad de Sonora
Apdo. Postal 1620
83000 Hermosillo, Son. MXICO
2

Departamento de Investigacin en Fisica

Universidad de Sonora Apdo. Postal 5-088
83190 Hermosillo, Son. MXICO
Abstract: Separation of the two-dimensional Laplace equation in elliptic coordinates leads to a Chebyshev-like
differential equation for both the angular and radial variables, namely ( , ) ; in the case of the well
behaved solution in [-1,1] (its range of definition), are the usual first class Chebyshev polynomials. However,
since is defined in [1, ) , there is a need to construct another solution by, for example, a Frobenius series
representation. Using these functions, the complete solution of two-dimensional Laplaces equation in this
coordinate system can then be constructed accordingly, and could be used to study a variety of boundary-value
electrostatic problems involving infinite conductors and lines of charge. Moreover, the corresponding Greens
function for the Laplace operator can also be readily obtained using this procedure, a matter that can be useful
in the study of many problems in solid state physics which involve energy levels and/or optical properties of
hydrogenic impurities within nanostructures of elliptic shape. These subjects are afforded and discussed in the
present communication, some useful trends regarding applications of the result are also given.
Keywords: Chebyshev functions, Chebyshev differential equation, Laplace operator, Elliptic coordinates,
nanostructures with elliptic shape, Frobenius method

Introduction

Laplace equation plays a fundamental role in

potential theory since many two-dimensional
boundary-value problems are of crucial importance
for both, physics and mathematics. This is the case,
for instance, in electrostatics, fluid flow through
obstacles, conformal mapping and so on.
The solution of this equation for a specific
boundary-value problem in electrostatics, can give
information that a priori is unknown; namely, when
an initially isolated conductor (charged or raised to
a given potential) is perturbed by a charge
distribution, the charge on the conductors surface
after the perturbation redistributes to an unknown
distribution, then the conventional solution for the
potential as an integral involving the surface charge
cannot be used. In those cases, the general solution
of Laplace equation becomes an important tool to
obtain the new potential.
In most electrostatic problems, a given charge
distribution is usually involved and one must solve
instead the Poisson equation, but either in this case

the general solution of Laplace equation is still

important since it can be used to construct an
auxiliary function, the Greens Function, which
allows one to find the particular solution of Poisson
equation that satisfies all the boundary conditions.
The construction of the general solution of
two-dimensional Laplace equation involves its
separability in a given coordinate system, its is
separable, for instance, in rectangular, polar,
parabolic, elliptic and other less common
coordinate systems (See for Ref.[1]). In the specific
case of elliptic coordinates, its separation leads to a
Chebyshev-type ordinary differential equation for
both angular () and radial () coordinates.
The solution associated to the angular variable are
the well known first-class Chebyshev polynomials
but in the case of the radial one, they are not longer
useful because this coordinate is defined in [1,)
and clearly the polynomials diverge at infinity.
This fact implies that we need to find a
different solution which must be properly behaved
in this interval; once such a solution is known, the

construction of the Greens Function associated to

the Laplace operator in this coordinate system can
be readily done.
The knowledge of both, the general solution
and the Greens Function for the Laplace operator
can used to solve a variety of electrostatic
boundary-value problems which involve infinite
conductors and infinite charged lines in elliptic
coordinates.
In addition, the two-dimensional Greens
function representation allows one to find the twodimensional Coulomb potential, a matter that can
be useful to study many properties of solid state
physics systems which involves hydrogenic
impurities in nanostructured materials of elliptic
shape in a similar way as it has been done for other
shapes [2,3,4].
The same potential in the moment space was
studied by Ditrich [5]. Other authors, as Furman
[6], have treated elliptical charge configurations.
The aim of this work is to stress at both,
academic and research levels, the importance
inherent to the knowledge of the general solution of
the Laplace equation and the wide possibilities of
applications.
For the sake of clarity, this communication has
been structured as follows: In section 2, we discuss
the form of the two-dimensional Coulomb potential
and its relation with the well known form of the
Green Function, as reported in [7,8]; the general
solution of Laplace equation in elliptic coordinates
is discussed in section 3; a representation of the
Green function in this coordinates is constructed in
section 4 and finally, in section 5, we made some
conclusion and discussion on the possible
applications of the results presented here-in.

2 Two-dimensional Coulomb
potential
The two-dimensional Green function, as known in
textbooks (see for instance, Ref [7]) is of the form:
1
(1)
G (, ') =
ln | ' | ,
2
which can be interpreted as the electrostatic
potential at due to an infinite line of charge, with
unit charge per length, located at ' and it is
solution of the equation:
2 G (, ') = ( ')
(2)
where ( ') is the Dirac delta function.
You most notice that this function is also
solution of the Laplace equation in charge-free twodimensional space, i.e., when ' .

In two dimensions, the electrostatic potential

does not coincide with the usual Coulomb
potential, since the former is associated with an
infinite line of charge (or an infinite charged
conducting cylinder) while the later is associated
with the interaction of two point charges, a matter
which sometimes causes confusion. Anyway, the
Coulomb potential for a unit point charge in two
dimensions can be readily related with the twodimensional Greens function from Eq. (1), as:
1
= exp [ 2 G (, ') ]
(3)
| '|
In this way, a two-dimensional problem which
involves Coulomb potential can be afforded by
firstly constructing the general solution and the
Greens function associated with the Laplace
operator in the system of coordinates adequately
selected for the specific geometry or shape of the
object or system under study.
In the next sections, we shall describe how this
can be done in the specific case of elliptic
coordinates, but it can be readily adapted to other
orthogonal coordinate system in which Laplace
equation is separable or, at least, partially
separable.

3 General Solution for Laplaces

Equation in elliptic coordinates
In this section, we will develop Laplaces operator
in two dimensions using an elliptic coordinate
system. This analysis becomes apparent when we
are analyzing the problem of a line of charge
parallel to an elliptic-cylindrical conductor because
we can look for a solution by considering the
bidimensional problem of obtaining Greens
function, which is solution to Poissons equation
2 G (, ') = (, ') .
(4)
We already know that it can be developed by a
harmonic expansion of functions which are
solutions to Laplaces equation
2 () = 0 .
(5)
The points considered are those points different
from those of the source ( ' ). Our first step
towards the solution will be to consider the
transformation of Laplaces operator from
Cartesian coordinates to an elliptic system. Using
the transformation equations
x = a ; [1, );
(6)
y = a( 2 1)1 / 2 (1 2 )1 / 2 ; [1,1) .

The scale factors related to this transformation are

1/ 2

x 2 y 2
h = +

1/ 2

2 2
=a 2

(7)

and

(1 2 )1 / 2 d
dH ( )
(1 2 )1 / 2

H ( ) d
d
( 2 1)1 / 2 d 2
dS ( )
( 1)1 / 2

S ( ) d
d
= const.

=
x y
h =
+

2 1/ 2

1/ 2

=a
2
1
2

. (8)

3.1 Laplacian operator

The scale factors that has been previously
calculated will allow us to construct the Laplacian
operator, which will have the form
1 h h
2 =
+

(9)

h h h h
we can obtain
( 2 1)1 / 2 (1 2 )1 / 2 ( 2 2 )1 / 2
2 =
a
2 1/ 2
a 2 ( 2 2 )
(1 )
1 ( 2 1)1 / 2 ( 2 2 )1 / 2
+
(10)
a
a ( 2 2 )1 / 2 ( 2 1)1 / 2
1 (1 2 )1 / 2

a ( 2 2 )1 / 2

or
2 =

2
2

1
( 1)1/ 2
( 1)1/ 2

2


a ( )
2

(11)

2 1/ 2
(1 )
+ (1 )
.


This is the expression for the Laplace operator in
elliptic coordinates.
Now, we can consider the Laplaces equation
given in Eq. (5). This can be rewritten using Eq.
(11) as
2

1
2 ( , ) = 2 2
( 1)1 / 2
2

a ( )
2 1/ 2

( 2 1)1 / 2
+ (1 2 )1 / 2
(12)

(1 2 )1 / 2
( , ),

and then solved by the variable separation method.

So, we introduce
( , ) = S ( ) H ( ) ,
(13)
and rearrange to obtain

(14)

3.2 Angular equation: Chebyshev

polynomials
The effect of variable separation allows us to define
a constant. Assuming that const. = m 2 , from the
left term of Eq. (14), we obtain the well known
Chebyshev equation
d
dH
(1 2 )1 / 2
(1 2 )1 / 2
+ m 2 H = 0 (15)

d
d
which has solutions for m = 0,1, 2, " , valid for
[1,1] :
H ( ) = Tm ( ) ,
(16)
known simply as the Chebyshev polynomials.
These polynomials are orthogonal and satisfy the
following condition
1 T ( )T ( )
m
m'
(17)
1 (1 2 )1 / 2 d = am m,m ' .
Their general expression is
Tm ( ) = cos(m cos 1 ) .
(18)

3.3 Radial equation: second solution to

Chebyshevs equation
From the right hand term of Eq. (14), and with the
constant of separation already defined, one can
obtain
d 2
dS
( 2 1)1 / 2
( 1)1 / 2
m 2 S = 0 , (19)
d
d
that must be solved in the [1, ) interval. We can
recognize it as the Chebyshev equation of second
class. In this range, the Chebyshev polynomials
remain to be a valid solution, but they are irregular
at infinity, so we will need another linearly
independent solution for this equation, in order to
completely describe an electrostatic problem such
as the one we are interested in.
We have to decide how to obtain the second
solution to Chebyshevs equation, as there are
several ways to obtain it; one of them will be
discussed here.
3.3.1 Second solution by the Frobenius method
In order to obtain the solution using the Frobenius
method, we have used the traditional way, as it is
shown in the Appendix. The functions obtained
have the following form and properties:

a ln + 2 1
m=0
0
(20)
S m ( ) =

m
2 l
a0 1 + m al , m 1

l =1

where a0 0 , [1, ) and

(m + 2l )4 l
.
(m + l + 1)(l + 1)
They decay very fast as grows and have a finite
value in = 1 . Those functions, in conjunction
with Chebyshevs polynomials are a complete set
of functions that will allow us to develop the
Greens function corresponding to a singular point
in the elliptic coordinate system.
al =

With the aid of the functions Tm ( ) for the angular

function, and the functions Tm ( ) and S m ( ) for
the radial variable, we can construct the Greens
function in the two-dimensional space.
As mentioned before, we need to solve
Poissons equation for a linear distribution of
charge, located at ' = ( ', ') , which is
2G (, ') = ( ')
(21)

where 2 is given by Eq. (11).

Because the {Tm ( )} polynomials constitute a
complete set of basis functions in [-1,1], i.e. in the
angular coordinate, Diracs delta function can be
expressed as a linear combination of them, in the
form

T ( )Tm ( ')
.
(22)
( ') = m
2 1/ 2
m = 0 (1 )
Furthermore, the delta function can be used to
define a function by the means of
1

( ')Tm ' ( )d = Tm ' ( ')

then, using Eq. (22) we have

Tm ' ( ') = Am Tm ( ')

m

(23)

Tm ' ( )Tm ( )
d
(1 2 )1 / 2

= Am am m ,m 'Tm ( ').
m

That is,
Tm ' ( ') = Am ' am 'Tm ' ( ')

Thus, we have Eq. (22) as

then
2
1
d
d
( 1)1 / 2
( 2 1)1 / 2
2
d
d
a ( )
2

d
d
(1 2 )1 / 2

d
d
Bm g m ( )Tm ( )
+ (1 2 )1 / 2

(28)

1
( ') ( ') ,
=
h h

( 2 1)1 / 2
1
( ') Tm ( )Tm ( ')
2
2
2
a ( )
m am
or, assuming that Greens function can be separated
as a product of functions of the base space, i.e., if
G (, ') = G ( , ; ', ')
(27)
= B ( ', ') g ( )T ( )
=

Greens Function

1 Tm ( )Tm ( ')
.
(25)
2 1/ 2
m = 0 am (1 )
Using this result and the definition of the scale
factors for elliptical coordinates, we can return to
our differential equation for Greens function
( 2 1)1 / 2 (1 2 )1 / 2
2 G ( , ; ', ') =
a 2 ( 2 2 )

1 Tm ( )Tm ( ')
( ')
(26)
2 1/ 2
m = 0 am (1 )

( ') =

Am ' =

1
. (24)
am '

( 2 1)1 / 2 ( ') Tm ( )Tm ( ')

.
m
am
a 2 ( 2 2 )
Regrouping and using the results of Eq. (15), we
have

dg

d
m Bm ( 2 1)1 / 2 d ( 2 1)1 / 2 d m m2 g m

(29)

(
')
T

+ ( 2 1)1 / 2 ( ') m
Tm ( ) = 0
am
=

But we already know that the {Tm ( )} set is

linearly independent, thus we have
dg
d
( 2 1)1 / 2
( 2 1)1 / 2 m m 2 g m =
d
d
(30)
( 2 1)1 / 2 ( ')Tm ( ')

am Bm ( ', ')
We are going to analyze this equation in the region
of the singularity, i.e. the region where ' . In
this case, we have Eq. (28) and it will have two
different solutions: one regular at infinity, and other
regular and finite at = 1 . Those functions will be
used to construct Greens function, which has to be
continuous at = ' . They are:

BTm ( )
g m ( ) =
ASm ( )

1< < '

1< ' < <

(31)

where A and B are coefficients that need to be

calculated. Integrating Eq. (30) around = ' , we
have
' +
' + d

dg
lim
( 2 1)1 / 2 m d m 2 g m d

d
'
' d
(32)
'+
( ')Tm ( ')d
= lim

am Bm ( ', ')
'
The second integral in the left-hand side of Eq. (32)
vanishes, while the first one and that of the right
hand side simplify to the functions evaluated at
= ' ; then we have

( 1)
2

1/ 2

dg

'+

=
'

Tm ( ')
am Bm ( ', ')

(33)

Assuming that
Bm ( ', ') = B 'm ( ')Tm ( ') ,
(34)
and considering Eq. (31), we can rewrite Eq. (33)
as
( '2 1)1 / 2 AS 'm ( ') ( '2 1)1 / 2 BT 'm ( ') =
(35)
1

am B 'm ( ')
But the continuity of Greens function puts the
condition
A Tm ( ')
=
; (36)
AS m ( ') = BTm ( ')
B S m ( ')
then, from Eq. (35),

S ' ( ')
( '2 1)1 / 2 B Tm ( ') m
T 'm ( ') =
S m ( ')

1
,
am B 'm ( ')

or

Sm ( ')
(37)
am B 'm ( ')
where W {Sm , Tm } is the Wronskian between these
pair of functions, and is precisely
1
W {S m , Tm } = 2
(38)
;
( ' 1)1 / 2
then, the coefficient of the B 'm is
S ( ')
T ( ')
AB 'm = m
BB 'm ( ') = m
, (39)
am
am
when we use Eq. (36). With these factors we can
construct the functions for the regions above and
below ' , i.e.
( '2 1)1 / 2 BW {Sm , Tm } =

g m+ ( ') = AB 'm ( ') =

Tm ( ')
am

(40)
Sm ( ')
.
g ( ') = BB 'm ( ') =
am
Finally, we arrive to the expression for Greens
function using Eqs. (27), (34) and (40), to have

T ( )Tm ( ') f m ( , ')

(41)
G (, ') = m
,
am
m=0
where
Sm ( ')Tm ( ), 1 < < '
f m ( , ') =
(42)
Tm ( ') Sm ( ), 1 < ' <
and
, m = 0

am =
(43)
2 , m 0
This function has the required properties and the
condition of being symmetrical respect to the
exchange between variables.
Using Eq. (3) and (41), we obtain an
expression for the Coulomb potential:

2 Tm ( )Tm ( ') f m ( , ')

1
= exp
(44)
| ' | m=0
am

This form of the Coulomb potential can be

implemented in a numerical approach.

Conclusion

In this communication we have obtained the

general solution of Laplace equation and its
corresponding Greens Function in elliptic
coordinates. In addition, a representation of the
two-dimensional Coulomb potential was given. The
reported expressions for these functions can be
used to study an interesting class of twodimensional problems which range from purely
electrostatic to actual solid state physics problems.
In the former case, we can mention, for instance,
the boundary-value problem of an elliptic
conducting cylinder, with a given potential or
surface charge, and a charged line; in the latter, the
study of hydrogenic impurities in bidimensional
nanostructured quantum dots of elliptic shape.
As a collateral result, the solution of the
Chebyshev differential equation in [1,) was
constructed using the Frobenius method, which
allowed us to define the Second Chebyshev
Functions, and to construct both the Greens
function and the Coulomb potential in this
coordinate system.
The formalism followed here-in to obtain
these results can be extended to any two-

dimensional coordinate system in which the

Laplace equation is separable or, at least, partially
separable.
Work is in progress to apply some of the
results to specific systems and will be published
elsewhere.
References:
[1] Ley-Koo E. and Gngora A. Desarrollos
armnicos del potencial Coulombiano en
coordenadas cilndricas, parablicas y
esferoidales. Rev. Mex. Fis. Vol. 39, No. 5,
1993, pp. 785-793
[2] Marin J. L. and Cruz S. A. Enclosed quantum
systems: use of the Direct Variational Method.
J. of Phys. B: At. Mol. Opt. Phys. Vol. 24, pp.
2899, 1991
[3] Cruz S. A., Ley-Koo E., Marin J. L. and
Taylor-Armitage A. Variational Calculation
for the hydrogen atom confined in space with
paraboloidal and spheroidal boundaries. Int. J.
of Quantum Chem. Vol. 54, pp. 3, 1995
[4] Marin J. L., Riera R. Confined systems and
nanostructured materials, in Handbook of
advanced electronic and photonic materials
and devices V. 6, Ed. Nalwa H. S., Academic
Press, 2001
[5] Ditrich W. The Coulomb Greens function in
two dimensions. Am. J. Phys. Vol. 67, No. 9
(1999) pp. 768-775
[6] Furman M. A. Compact complex expressions
for the electric field of two-dimensional
elliptical charge distributions. Am. J. Phys.
Vol. 62, No. 12, 1999, pp. 1134-1140
[7] Arfken G. and Weber H. J. Mathematical
Methods for Physicists. Academic Press,
2001
[8] Kellogg O. D. Foundations of Potential
Theory. Dover Publications, 1953
[9] Marin J. L. and Cruz S. A. On the use of
Direct Variational methods to study confined
quantum systems. Am. J. Phys. Vol 59, pp.
931, 1991
[10] Boyd John P. Chebyshev and Fourier Spectral
Methods. Dover Publications, 2001
Aknowledgements: Ral Prez-Enrquez and Ivan
Marn-Enrquez are PhD students at Departamento
de Inv. En Fisica (Universidad de Sonora) and are
supported by CONACyT and UNISON. This work
has been done under project CONACyT-40629 and
47682.

Appendix
The Chebyshevs differential equation outside the
[-1, 1] interval can be expressed as follows:
d2 H
dH
( 2 1)
+
n 2 H ( ) = 0
(45)
2
d

d
where, H has a removable singularity in = 1
and is regular at = . In order to find a solution,
we will use the Frobenius method; in doing so, we
suggest the following form for the function

H ( ) = al k l

(46)

l =0

Then the first and second derivatives will be

H '( ) = (k l ) al k l 1

(47)

l =0

and

H ''( ) = (k l )( k l 1) al k l 2

(48)

l =0

Making the substitution of those expressions in Eq.

(45), we will have

( 2 1) (k l )(k l 1) al k l 2
l =0

+ (k l ) al
l =0

k l 1

n al
2

(49)
k l

= 0,

l =0

i.e.,

l =0

(k l )(k l 1) + (k l ) n 2 k l

(k l )(k l 1) a l

k l 2

(50)

=0

l =0

After replacing the index in the second summation

( l l 2 ), while we simplify the first one, we
will obtain

a
l =0

(k l ) 2 n 2 k l

(k l + 2)(k l + 1) a l + 2

(51)
k l

= 0,

l =2

and from it, the recurrence relations for the

coefficients will appear:
a0 (k 2 n 2 ) = 0
(52)
a1 [(k 1) 2 n 2 ] = 0
a l [(k l ) n ] = (k l + 2)( k l + 1) a l + 2
2

(53)

(54)
for l 2
From the secular equation (52) we will find the
allowed values of k; if we assume that
a0 0, a1 = 0 then k 2 n 2 = 0
k = n
(55)

But the requirement for the function H ( ) to

vanish as , makes the powers to be
considered strictly negative; thus,
k = n, with n > 0 .
(56)
The recurrence relation (54) for the case l 2
(even) and condition (56) will give
(n + l 2)(n + l 1)
a l =
a l + 2 ,
(57)
l (l + 2n)
a recurrence relation between coefficients that will
allow us to find all coefficients as a function of a0
that we assumed different from zero. In
consequence,
n(n + 1)
n(n + 1)
a2 =
a0 =
a0
2(2n + 2)
2 2(n + 1)
(58)
n
=
a0 ,
22
(n + 2)(n + 3)
n(n + 2)(n + 3)
a4 =
a2 =
a0
4(4 + 2n)
2 2 4 2(n + 2)
(59)
n(n + 3)
= 2
a0 ,
2 42
(n + 4)(n + 5)
a6 =
a4
6(2n + 6)
n(n + 3)(n + 4)(n + 5)
=
a0
(60)
23 6 4 2(n + 3)
n(n + 4)(n + 5)
=
a0
23 6!!
Continuing this way, we will derive a compact
expression for the coefficients,

We call then this set of functions the Chebyshev

functions of 2nd Class that are solution to Eq. (45),
and are defined by
a ln + 2 1 , for n = 0
0

( n + 2l ) (2 ) 2l
n
S n ( ) = a0 1 + n
(65)
l =1 ( n + l + 1) (l + 1)

for n 1

In Figs. 1 through 4 we show graphs of those

functions for values of the index n = 0,1, 2, " , 6

Fig. 1. Graph for the Chebyshev function of 2nd

class, S0 ( ) .

2 l 1

a2l =

n (n + s)
s = l +1
2l

a0 , with l = 1, 2, 3," (61)

2 l!
Functions of well-defined parity will then be built
with the aid of these coefficients:
2 l 1

(n + s)

H n ( ) = a0 n + n s =l +1 2l
n 2l , (62)

2 l!
l =1

where + stands for n even and for n odd. In the

special case where n = 0, the function H 0 ( ) is
solution to the differential equation
dH 0 ( )
d
= 0, (63)
( 2 1)1 / 2
( 2 1)1 / 2
d
d
where we have put Eq. (45) in self-adjoint form;
this can be solved by direct integration, and yields
the function

H 0 ( ) = C ln + 2 1

(64)

Fig. 2. Graphs for the Chebyshev functions of 2nd

class for n = 1, 2 .

S0 ( ) = A0 ln + 2 1

2
(67)
S1 ( ) = A1 1

S 2 ( ) = A2 (2 2 1) + 2 2 1

Both representations are compatible when

calculated for > 1 + , but the series form of
functions S n ( ) is easiest to implement in a
numerical calculation as that of Greens function on
elliptic coordinates.

Fig. 3. Graphs for the Chebyshev functions of 2nd

class for n = 3, 4 .

Fig. 4. Graphs for the Chebyshev functions of 2nd

class for n = 5, 6 .
Finally, we consider necessary to point out
that this method to obtain Chebyshev Functions of
2nd class is not unique; an alternative way to build
these functions would involve the direct evaluation
form the Wronskian and the Chebyshev
polynomials of 1st class as discussed by Arfken and
Weber for the Legendre polynomials [7].
The closed form of those Chebyshev
Functions of 2nd class would be expressed as

dx
(66)
S n ( ) = Tn ( ) An + Bn 2
1/ 2
2
[Tn ( x)]

1 ( x 1)
with An a constant to be determined and the

Tn (x) , Chebyshev polynomials, evaluated in the

interval of interest. The problem with this method
is that we have to calculate them one by one. As an
illustration, the three first ones are the following: