Professional Documents
Culture Documents
Polinomios de Legendre en Electrostatica PDF
Polinomios de Legendre en Electrostatica PDF
Contents
1 Laplace Equation in Spherical Coordinates 2
1.1 Legendre Equation and Polynomials . . . . . . . . . . . . . . . . . . . 4
1.2 Solution of Boundary Value Problems with Azimuthal Symmetry . . 12
1.2.1 Example: A Sphere With a Specified Potential . . . . . . . . . 13
1.2.2 Example: Hemispheres of Opposite Potential . . . . . . . . . . 14
1.2.3 Example: Potential of an Isolated Charge . . . . . . . . . . . . 17
1.3 Behavior of Fields in Conical Holes and Near Sharp Points . . . . . . 19
1.4 Associated Legendre Polynomials; Spherical Harmonics . . . . . . . . 23
1.5 The Addition Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.6 Expansion of the Greens Function in Spherical Harmonics . . . . . . 30
1
Although this is a new chapter, we continue to do things begun in
the previous chapter. In particular, the first topic is the separation of
variable method in spherical polar coordinates.
2 1 2 1 1 2
= r+ 2 sin + 2 2 . (1)
r r2 r sin r sin 2
This is also a coordinate system in which it is possible to find a solution
in the form of a product of three functions of a single variable each:
2
C
from which it follows that Q e .
Now, a change in by 2 corresponds to no change whatsoever in
spatial position; therefore, we must have Q( + 2) = Q() because
a function describing a measurable quantity must be a single-valued
function of position. Hence we can conclude that C = im where m
is an integer so that eim2 = 1. Thus C = m2 , and Q() Qm () =
eim , with m = 0, 1, 2,... . We recognize that the functions Qm can
be used to construct a Fourier series and are a complete orthogonal set
on the interval 0 + 2.
Returning now to Laplaces equation, Eq. (4), and using m2 for
1 d2 Q
Q d2 , we find
!
r2 sin2 d2 U sin d dP
2
+ sin m2 = 0 (6)
U dr P d d
or !
r 2 d2 U 1 d dP 1 2
2
+ sin 2 m = 0. (7)
U dr P sin d d sin
In this expression we recognize that the first term depends only on r
and next two, only on , so we as usual conclude that each term must
be separately a constant and that the constants must add to zero. The
first equation extracted by this device is
d2 U A
= U (8)
dr2 r2
where A is the constant. It is a standard convention to write A as
l(l + 1) which is still quite general if l is allowed to be complex. Thus
3
the preceding equation becomes
d2 U l(l + 1)
= U. (9)
dr2 r2
The solutions of this ordinary, second-order, linear, differential equation
are two in number and are U r l+1 and U 1/r l . Before commenting
further on that, let us go on to the equation for P ().
4
call such a potential azimuthally invariant; there are many interesting
systems which are more or less of this type. The equation for P is
!
d dP
(1 u2 ) + l(l + 1)P = 0, (14)
du du
which is called the Legendre equation.
A standard procedure for solving this equation (and other similar
second-order differential equations) is to assume that the solution can
be written as a power series. Then there must be a smallest power
in the series, so we can write
X j X
P (u) = u aj u = aj uj+ (15)
j=0 j=0
d2 P X
2
= (j + )(j + 1)aj uj+2 , (17)
du j=0
and
d 2 dP X
u = (j + )(j + + 1)aj uj+ . (18)
du du j=0
Substitution into the Legendre equation gives
h
X i
( + j)( + j 1)u+j2 aj ( + j)( + j + 1)u+j aj + l(l + 1)u+j aj = 0,
j=0
(19)
or, if we shift the zero of j in each term so as to isolate individual
powers of u,
( 1)u2 a0 + ( + 1)a1 u1 +
5
X
[( + j + 2)( + j + 1)aj+2 + (l(l + 1) ( + j)( + j + 1)) aj ] u+j = 0.(20)
j=0
The only way that this power series can vanish for all u on the
interval is to have the coefficient of each power of u vanish separately.
Thus I will list the coefficients:
j Coefficient of u+j
2 a0 ( 1)
1 a1 ( + 1)
j 0 [( + j + 2)( + j + 1)aj+2 + (l(l + 1) ( + j)( + j + 1)) aj ]
= 0, 1 (21)
6
Consider this relation when j is very large, much larger than 1. In
this limit it simplifies to the statement that aj+2 = aj (1+O(1/j)) which
will produce a power series (for large powers) in the form of a sum of
terms proportional to u2j , all with the same coefficient. At u 1, this
sum will not converge, and so P is singular at u = 1. This is not an
allowed behavior for a solution to the Laplace equation, so we cannot
have such a function representing the potential. What must therefore
happen is that the series terminates which means that there must be
some j such that aj 6= 0 while aj+2 = 0. Examining Eq. (22), we see
that this j is such that
lZ l (24)
7
= 1 and l odd
= 1 and l even
Since l cannot be both odd and even, we can only have an even or
an odd series (they are actually equivalent). The other must be zero.
Since by convention, we choose a0 6= 0, it must be that the odd series
vanishes, and thus a1 = 0. Remembering that l is odd when = 1 and
is even when = 0, we see that the solutions are polynomials of degree
l. They are known as Legendre polynomials. It is easy to generate a few
of them, aside from normalization, starting from l = 0 and using the
recurrence relation. As for normalization, they are traditionally chosen
to be such that P (1) = 1. Let us add a subscript l to P to designate
the particular Legendre polynomial. The first few are
l Pl (u)
0 P0 (u) = 1
1 P1 (u) = u
2 P2 (u) = 32 u2 1
2
5 3 3
3 P3 (u) = 2u 2u
35 4 15 2 3
4 P4 (u) = 8u 4u + 8
8
There are other ways to generate the Legendre polynomials. For
example, one has Rodrigues formula which is
1 dl 2
Pl (u) = l l
(u 1)l ; (27)
2 l! du
it is easy to see that this generates a polynomial of degree l; one may
show that it is a solution to the Legendre equation by direct substitution
into that equation. Thus, it must be the Legendre polynomial (one
should also check normalization). If one expands the factor (u2 1)l
in Rodrigues formula using the binomial expansion and then takes the
derivatives, she/he will find that
(1)l X
l (1)n (2n)!
Pl (u) = l
u2nl (28)
2 nl/2 n!(l n)!(2n l)!
Another way of generating Pl (u) is via the generating function
X
T (u, x) = (1 2ux + x2 )1/2 = xl Pl (u). (29)
l=0
If one takes l derivatives of this function with respect to x and then
sets x = 0, the result is l!Pl (u).
The Legendre functions have many properties that we will need to
make use of from time to time. For summaries of these, see e.g., ,
the section on Legendre functions in Abramowitz and Stegun starting
on p. 332 and also the section on orthogonal polynomials starting on
p. 771. Here we summarize some of the most significant properties.
First, orthogonality and normalization. Consider the integral
0
Z 1 1 Z 1 dl dl 0
duPl (u)Pl0 (u) = l+l0 0 du l (u2 1)l l0 (u2 1)l . (30)
1 2 l!l ! 1 du du
9
Suppose, without loss of generality, that l 0 l and start by integrating
by parts,
Z 1
duPl (u)Pl0 (u) =
1
1
l l0 1 Z 1 l+1 l0 1
1 d d 0 d d 0
(u2 1)l l0 1 (u2 l
1) du (u2 1)l l0 1 (u2 1)l (.31)
2l+l0 l!l0 ! du l du
1
1 du l+1 du
The first term in brackets vanishes because (u2 1) is zero at the end
points of the interval. Continuing from the right-hand side, integrate
in like fashion l0 1 more times. The result is
0
Z 1 1 Z 1 dl+l (u2 1)l 2
l0 l0
duPl (u)Pl0 (u) = (1) l+l0 0 du 0 (u 1) . (32)
1 2 l!l ! 1 dul+l
0 0
Now, dl+l (u2 1)l /dul+l = 0 if l0 > l, and so the integral is zero in this
case. If l0 = l, we have
Z 1 (2l)! Z 1 l
duPl (u)Pl (u) = (1) 2l 2 du(u2 1)l
1 2 (l!) 1
l 2
(2l)! (2l)!! (2 l!) (2l 1)!! 2
= (1)l 2l 2 2(1)l = 2 2l 2 = . (33)
2 (l!) (2l + 1)!! 2 (l!) (2l + 1)!! 2l + 1
Thus we have derived the relation
Z 1 2
duPl (u)Pl0 (u) = ll0 (34)
1 2l + 1
which expresses the orthogonality and normalization of the Legendre
polynomials.
Consider next recurrence relations. These provide, among other
things, a good way to generate values of Legendre polynomials on
10
computers. A number of recurrence relations can be derived using
Rodrigues formula and the Legendre equation. Consider, for example,
dPl+1 1 dl+2 2
= l+1 l+2
(u 1)l+1
du 2 (l + 1)! du
l + 1 dl+1 2 l
= l (u 1) u
2 (l + 1)! dul+1
1 dl h 2 l 2 2 l1
i
= l (u 1) + 2lu (u 1)
2 l! dul
1 dl h 2 l 2 l 2 l1
i
= l (u 1) + 2l(u 1) + 2l(u 1)
2 l! dul
dPl1 (u)
= (2l + 1)Pl (u) + , (35)
du
or
dPl+1 (u) dPl1 (u)
(2l + 1)Pl (u) = 0. (36)
du du
From this relation and the Legendre equation
!
d dPl
(1 u2 ) + l(l + 1)Pl = 0 (37)
du du
one may derive additional standard recurrence relations for the Legen-
dre polynomials. Several of these are
11
1.2 Solution of Boundary Value Problems with Azimuthal
Symmetry
Using what we have learned in the previous two sections, we are now
in a position to construct a general solution to the Laplace equation in
spherical coordinates under conditions of azimuthal invariance, that is,
when (x) is independent of . The most general form that a solution
can have is
X
(r, ) = Al rl + Bl r(l+1) Pl (cos ). (39)
l=0
12
1.2.1 Example: A Sphere With a Specified Potential
(a, , ) = Vo cos3 ()
Since our volume contains all r > a, physics demands that Al = 0 for
all l. The constants Bl are then determined by matching the terms in
the series the boundary condition on the surface of the sphere. Recall
that
P0 (x) 1
P1 (x) x
1 2
P2 (x) 2 (3x 1)
1 3
P3 (x) 2 (5x 3x)
1 4 2
P4 (x) 8 (35x 30x + 3)
1 5 3
P5 (x) 8 (63x 70x + 15x)
So that,
!
3 2 3
(a, , ) = Vo cos () = Vo P3 (cos()) + P1 (cos())
5 5
13
Thus
!4 !2
2 a 3 a
(r, , ) = Vo P3 (cos()) + P1 (cos()) .
5 r 5 r
For the first, suppose that we need to solve the Laplace equation in-
side of a sphere of radius a given that on the surface, the potential is
specified as follows:
V, 0 /2
(a, ) = (40)
V, /2 .
=V
= -V
14
term. The powers of a are included for the same reasons; r is of order
a and has the same dimensions so that leading coefficients Al are of
order unity and have dimension unity.
On the spherical surface, we have
X
(a, ) = V Al Pl (cos ). (42)
l=0
(45)
To complete the integral for the case of odd n we use a recurrence
15
relation
dPn+1 dPn1
= (2n + 1)Pn + (46)
du du
so that:
Z 1 Z 1 " #
dPn+1 dPn1
An = (2n+1) duPn (u) = du = Pn1 (0)Pn+1 (0)
0 0 du du
(47)
where we make use of the fact that Pn (1) = 1, independent of n. Fur-
ther, for even l,
(l 1)!!
Pl (0) = (1)l/2 (48)
l!!
where the double factorial sign means l!! = l(l 2)(l 4)...(2 or 1).
Hence
(n 2)!! n!!
Pn1 (0) Pn+1 (0) = (1)(n1)/2 +
(n 1)!! (n + 1)!!
(n 2)!! (n 2)!!
= (1)(n1)/2 (n + 1 n) = (1)(n1)/2 (2n + 1). (49)
(n + 1)!! (n + 1)!!
Now set n = 2m + 1, m = 0,1,2,..., and have
!2m+1
X r
(r, ) = V Bm P2m+1 (cos ) (50)
m=0 a
where
(2m 1)!!
Bm = (1)m (4m + 3). (51)
(2m + 2)!!
The first few terms in the expansion are
!3 !5 !7
3 r 7 r 11 r 25 r
(r, ) = V P (cos )
a 1
P3 + P5 P7 + ...
2 12 a 24 a 64 a
(52)
16
1.2.3 Example: Potential of an Isolated Charge
then we can infer the form of the expansion by expanding g(r) in powers
of r and recognizing that the coefficient of r l must be Pl (cos 0 ) times
a coefficient Al while the coefficient of (1/r)(l+1) must be Bl Pl (cos 0 ).
The most convenient value of 0 to use is certainly 0 or since we know
immediately the value of Pl (cos ) in these instances.
Consider the following specific example: Suppose that there is a
charge q at a position x = az
(0,0,a)
17
this function has a simple power series expansion,
q q 1 q r r2 r3
(r, 0) = = = 1 + + 2 + 3 + ... . r<a
ar a 1 r/a a a a a
(55)
Hence, associating Pl with (r/a)l , we have
!l
r
qX
(r, ) = Pl (cos ); r<a (56)
a l=0 a
the point is, the uniqueness of the expansion in terms of Legendre
polynomials tells us that this must be the solution. A similar expansion
done for r > a yields
!l
a
qX
(r, ) = Pl (cos ) r > a. (57)
r l=0 r
There are two points that are worth making in connection with
these expansions. First, as stated earlier, there is a generating function
T (u, x) for the Legendre polynomials; see Eq. (29). We have just de-
rived it; that is, it is , Eq. (54), equal to the sum in Eq. (56). Also,
we have obtained a convenient and useful expansion for the potential
of a point charge; in more general notation, we have derived
1 rl
X <
0
= l+1 Pl (cos ) (58)
|x x | l=0 r>
where is the angle between x and x0 while r< (r> ) is the smaller
(larger) of |x| and |x0 |.
18
1.3 Behavior of Fields in Conical Holes and Near Sharp Points
19
the variable y 21 (1 u) or u = 1 2y; then Eq. (60) becomes
!
1 d 1 dP
1 (1 2y)2 + ( + 1)P = 0 (61)
2 dy 2 dy
or !
d dP
y(1 y) + ( + 1)P = 0. (62)
dy dy
Let us look once again for a solution in the form of a power series
expansion,
X
P = y aj y j , (63)
j=0
with 0 y y0 1. Then
dP X
= ( + j)aj y +j1 , (64)
dy j=0
dP X
y(1 y) = ( + j)aj (y +j y +j+1 ), (65)
dy j=0
and
!
d dP X h i
y(1 y) = aj ( + j)y +j1 ( + j + 1)y +j ( + j).
dy dy j=0
(66)
Now combine these equations to find
h
X i
aj ( + j)2 y +j1 + aj (( + j)( + j + 1) + ( + 1)) y +j = 0
j=0
(67)
or, isolating individual powers of y,
a0 2 y 1 = 0 (68)
20
which implies that = 0, and
j(j + 1) ( + 1)
aj+1 = aj . (69)
(j + 1)2
If one lets = l, a non-negative integer, the result is just the Legendre
polynomials (no surprise), viewed as functions of y. More generally, for
any real > 0, one finds that the solutions are Legendre functions of
the first kind of order .
X
P = aj ()y j , (70)
j=0
These are well-behaved (that is, they are not singular) functions of y
for y < 1 corresponding to u > 1 and are singular at y = 1.
For 1 > > 0, P (y) has a single zero; for 2 > > 1, P (y)
has two zeroes, etc. This is important because if we have a cone of
half-angle with equipotential surfaces, we need to be such that
P ((1 cos )/2) = 0. There will thus be a sequence of allowed values
of , which we designate by k , k = 1, 2, 3,..., which are such that
y 12 (1 cos ) k th zero of P .
The general solution at finite values of r, and including the point
r = 0 , is
X
(r, ) = Ak rk Pk (cos ). (71)
k=1
For small r, the leading term is the one with the smallest power of r,
that is the k = 1 term. Hence we may approximate the sum sufficiently
close to the origin by its leading term
21
The dominant contribution to the electric field in this region comes
from this term; we have, by the usual E(x) = (x),
d
Er = = 1 Ar1 1 P1 (cos ) (73)
dr
and
1 d 1 1 dP1 (u)
E = = A sin r (74)
r d du cos
The behavior of 1 as a function of is shown below. For less than
2.405
about 0.8, one has1 1 21 , while for larger than about
1
2
the same number, 1 2 ln . As , 1 0 and so
Er E 1/r in this limit. The enhancement of a field near e.g., a
lightning rod is thus (R/) if R is the size of the system and is the
radius of curvature of the tip of the rod. Recall that in two dimensions
1
This relation comes from study of the properties of the Legendre functions.
22
we found an enhancement of order (R/)1/2 The enhancement is much
more pronounced in three dimensions; a three dimensional tip is a much
sharper thing than an edge.
23
we see that
(1 u2 )m/2 dl+m
Plm (u) = (1)m l l+m
[(u2 1)l ]. (77)
2 l! du
This last formula is also valid for negative m 2 ; comparing the two
cases, one may see that
(l m)! m
Plm (u) = (1)m P (u). (78)
(l + m)! l
As for the Legendre polynomials, there is a generating function for
the associated Legendre functions as well as a variety of recurrence
relations. For example, a recurrence relation in degree is given by
24
but orthogonality is quite straightforward using the formulae we have
already written down. Consider the integral
Z 2 Z Z 1
0 0
im
I= d d sin e Plm (cos )eim Plm
0 (cos ) = 2 mm0 duPlm (u)Plm
0 (u)
0 0 1
(82)
m
Assume l0 l, m 0, and write Plm
0 in terms of Pl0 :
(l0 + m)! Z 1
I = 2mm0 (1)m
0
duPlm 0 (u)Plm (u)
(l m)! 1
0
0
m (l + m)! 1 Z 1 dl m 2 l0 d
l+m
= 2mm0 (1) l+l 0 0
du l 0 m (u 1) l+m
(u2 1)l
(l m)! 2 l!l ! 1 du du
0 Z 1 l+l0
l0 (l + m)! 1 2 l0 d 2 l
= 2mm0 (1) 0 l+l 0 0
du(u 1) l+l 0 (u 1)
(l m)! 2 l!l ! 1 du
(l + m)!(2l)!(2l)!! 4 (l + m)!
= 2ll0 mm0 2 = ll 0 mm0 (.83)
(l m)!22l (l!)2 (2l + 1)!! 2l + 1 (l m)!
Thus we may construct an orthonormal set of functions on the surface
of the unit sphere; these are called spherical harmonics and are defined
as v
u
u 2l + 1 (l m)! m
u
Yl,m (, ) t Pl (cos )eim , (84)
4 (l + m)!
with m = l, l + 1, ..., l 1, m and l = 0, 1, 2, .... These functions have
the property that
Yl,m (, ) = (1)m Yl,m (, ). (85)
25
The completeness relation (not derived as usual) is
X
X m
Yl,m (0 , 0 )Yl,m (, ) = ( 0 )(cos cos 0 ). (87)
l=0 m=l
26
is
4 X l
Pl (cos ) = Yl,m (0 , 0 )Yl,m (, ). (90)
2l + 1 m=l
We shall derive this expression as an example of the use of spherical
harmonics and their properties. First, let us set up a second coordinate
system rotated relative to the original one in such a way that its polar
axis lies along the direction of x0 . In this system, the vector x has
components (rR , R , R ).
zR
x
x
= R
yR
R
xR
27
Now, the spherical harmonics satisfy the differential equation
l(l + 1)
2 Yl,m (, ) + Yl,m (, ) = 0 (93)
r2
and they also satisfy this equation with variables R , R . But the
Laplacian operator 2 = is a scalar object which is invariant
under coordinate rotations which means we can write it in the unrotated
frame while writing the spherical harmonic in the rotated frame:
l(l + 1)
2 Yl,m (R , R ) + Yl,m (R , R ) = 0. (94)
r2
Now recall that
v v
u u l 0
u 2l +1 u 2l +1 X X
t t
Yl,0 (R , R ) = Pl (cos ) = Al0 m0 (0 , 0 )Yl0 m0 , (, ).
4 4 l0 =0 m0 =l0
(95)
If we plug this into the differential equation above, we obtain:
X l0
X l(l + 1)
Al0 m0 (0 , 0 )[2 Yl0 m0 , (, ) + Yl0 m0 , (, )] = 0 (96)
l0 =0 m0 =l0 r2
or
X l0
X l 0
(l0 + 1) l(l + 1)
0
Al0 m0 ( , 0 ) + Yl0 m0 , (, ) = 0. (97)
l0 =0 m0 =l0 r2 r2
This equation can be true only if l = l 0 or if Al0 m0 = 0. Thus we have
demonstrated that Al0 m0 = 0 for l0 6= l, and the expansion of Pl reduces
to
l
X
Pl (cos ) = Alm (0 , 0 )Yl,m (, ). (98)
m=l
28
The coefficients in this expansion are found in the usual way for an
orthogonal function expansion,
Z Z
Alm = dPl (cos )Yl,m (, ) = dR Pl (cos R )Yl,m (, ). (99)
q
Following the same line of reasoning, we may express 4/(2l + 1)Yl,m (, )
as a sum of the form
v
u l
u
t 4 X
Yl,m (, ) = Blm0 (m)Ylm0 (R , R ) (100)
2l + 1 m0 =l
However, from Eq. (76), it is clear that when u = 1 Plm (u) = 0 when
m 6= 0,and Plm (u) = Pl (u) when m = 0, thus
v v v
u u u
u 4 u 2l + 1 u 2l + 1
t
Yl,m (, ) =0 = Bl0 (m)t Pl (1) = Bl0 (m)t ,
2l + 1 R 4 4
(102)
so
4 4
Alm = Bl0 (m) =
Yl,m (, ) =0 = Yl,m
(0 , 0 ) , (103)
2l + 1 R 2l + 1
where the last step follow since when R = 0, x = x0 . Thus,
4 X l
Pl (cos ) = Y (0 , 0 )Yl,m (, ). (104)
2l + 1 m=l l,m
Thus ends our demonstration of the spherical harmonic addition theo-
rem.
29
1
An application if this theorem is that we can write |xx0 | as an ex-
pansion in spherical harmonics:
1 rl
X X 4 r l l h
X i
< <
0
= l+1 Pl (cos ) = l+1 Yl,m (0 , 0 )Yl,m (, ) .
|x x | l=0 r> l=0 2l + 1 r> m=l
(105)
This expansion is often useful when faced with common integrals in
R
electrostatics such as as d3 x0 (x0 )/|x x0 |.
b
a
30
X
X l
A
rl al
+ lm l+1 + Blm l+1 Yl,m (, ) (107)
l=0 m=l b r
where Alm and Blm can be functions of x0 . The first term on the
1
right-hand side is |xx0 | and the second is a general solution of the
Laplace equation. The coefficients are determined by requiring that
the boundary conditions on G(x, x0 ) are satisfied (G(x, x0 ) = 0 for x
on one of the two bounding spherical surfaces). At r = a (r< = r = a
r> = r0 ), we have
4 al 0 0
X al 1
0= Y
0l+1 l,m
( , ) + A lm l+1 + B lm Yl,m (, ) = 0,
l,m 2l + 1 r b a
(108)
from which it follows, using the orthogonality of the spherical harmon-
ics, that
4 al 0 0 al 1
Y ( , ) + A lm + B lm = 0. (109)
2l + 1 r0l+1 l,m bl+1 a
By similar means applied at r = b (r< = r0 r> = r = b), one may show
that
4 r0l 0 0 1 al
Y ( , ) + Alm + Blm l+1 = 0. (110)
2l + 1 bl+1 l,m b b
These present us with two linear equations that may be solved for A lm
and Blm ; the solutions are
, !2l+1
4 0 0 a
2l+1
r0l a
Alm = Y ( , ) l 0l+1 l 1 (111)
(2l + 1) l,m br b b
and
, !2l+1
4 al+1 r0l al+1 a
Blm = Yl,m (0 , 0 ) 2l+1 0l+1 1 . (112)
2l + 1 b r b
31
From these and the expansion Eq. (107), we find that the Greens
function is
X 4
G(x, x0 ) = 0 0
2l+1 Yl,m ( , )Yl,m (, )
a
l,m (2l + 1) 1 b
!2l+1 !2l+1 !2l+1
a rl
< +
a rl r0l rl a r 0l
a 2l+1
1 l+1 + (113)
.
b r> b r0l+1 b2l+1 b rl+1 rl+1 r0l+1
This result, if we can call it that, can be written in a somewhat more
compact form by factoring the quantity {...}. Suppose that r> = r0
and r< = r; then
l a2l+1 1 r0l
{...} = r l+1
r r0l+1 b2l+1
2l+1 l
l a 1 r >
r< l+1 l+1 2l+1 . (114)
r< r> b
If, on the other hand, r> = r and r< = r0 , then
0l a2l+1 1 rl l a2l+1 1 l
r>
{...} = r 0l+1
l+1
2l+1 r< l+1
l+1 2l+1 .
r r b r< r> b
(115)
Comparing these results, we see that we may in general write
X 4/(2l + 1) a2l+1 1 l
r>
G(x, x0 ) = 0 0 l
2l+1 Yl,m ( , )Yl,m (, ) r< l+1 l+1 2l+1 .
l,m
a
1 b r< r> b
(116)
Notice that the Greens functions for the interior of a sphere of radius
b and for the exterior of a sphere of radius a are easily obtained by
taking the limits a 0 and b , respectively. In the former case,
32
for example, one finds
l l l
0 X 4 0 0 r< r< r
G(x, x ) = Yl,m ( , )Yl,m (, ) l+1 2l+1>
l,m 2l + 1 r> b
1 X 4 0 0 b rl
= Y ( , )Y l,m (, )
|x x0 | l,m 2l + 1 l,m r0 (b2 /r0 )l+1
1 b/r0
= (117)
|x x0 | |x x0R |
where x0R = (b2 /r0 , 0 , 0 ) in spherical coordinates.
34
set.
k Z(z) R()
real incomplete (ekz ) complete (oscillatory)
imaginary complete (ei|k|z ) incomplete
35
choose them is as follows:
!
x X (1)j (x/2)2j
J (x) = (125)
2 j=0 j!(j + + 1)
and !
x X (1)j (x/2)2j
J (x) = (126)
2 j=0 j!(j + 1)
where
Z
(z) dt tz1 et (127)
0
36
For = n, a non-negative integer, the Neumann function has a series
representation which is
! !
1 x n n1
X (n j 1)! x 2j 2
Nn (x) + ln(x/2)Jn (x)
2 j=0 j! 2
!n
1 x X (x/2)2j (1)j
[(j + 1) + (n + j + 1)] (129)
2 j=0 j!(n + j)!
and
H(2) = J (x) iN (x). (131)
It is not easy to see what are the properties of the various kinds of
Bessel functions from the expansions we have written down so far. As it
turns out, their behavior is really quite simple; many of the important
features are laid bare by their behavior at small and large arguments.
For x << 1 and real, non-negative , one finds
!
1 x
J (x) = [1 O(x2 )] (132)
( + 1) 2
and
2 [ln(x/2) + 0.5772
+ ...] = 0
N (x) = (133)
()
2
x 6= 0
37
For x >>> 1, ,
v
u !
u
t 2
J (x) cos x (134)
x 2 4
and v
!
u
u
t 2
N (x) sin x . (135)
x 2 4
1
Ordinary Bessels functions
0
J0(x)
-1 J1(x)
N0(x)
-2 N1(x)
-3
0 5 10 15
x=k
38
Bessel functions of all kinds satisfy certain recurrence relations. It
is a straightforward if tedious matter to show by direct substitution of
the series expansions that they obey the following:
2
+1 + 1 = 0 (136)
x
and
d
+1 + 2 1 = 0. (137)
dx
By taking the sum and difference of these relations, we find also
d
1 = . (138)
x dx
These are valid for all three kinds of Bessel functions.
The Bessel function J (x) can form a complete orthogonal set on
an interval 0 x x0 in much the same way as the sine function
sin(nx/x0 ) does (Note that x0 is a zero of the sine function.) Similarly,
let us denote the nth zero of J (x) by xn and then form the functions
J (xn y), with n = 1,2,... . Then it turns out that for fixed , these
functions provide a complete orthogonal set on the interval 0 y 1.
As usual, we shall not demonstrate completeness. Orthogonality can
be demonstrated by making use of the Bessel equation and recurrence
relations. It is a useful exercise to do so. Start from the Bessel equation
for J (xy),
1 d dJ (xy) 2 2
y + x 2 J (xy) = 0 (139)
y dy dy y
39
Multiply this equation by yJ (x0 y) and integrate from 0 to 1:
Z 1
0 d dJ (xy) 0 2 2
dy J (x y) y + yJ (x y) x 2 J (xy) = 0
0 dy dy y
(140)
or if we integrate the first term by parts:
1
0 dJ (xy) Z 1 dJ (x0 y) dJ (xy)
J (x y)y dy y +
dy 0 0 dy dy
Z 1 2
0 2
dyyJ (x y)J (xy) x 2 = 0 (141)
0 y
Similarly, if we start from the differential equation for J (x0 y) and per-
form the same manipulations, we find the same equation with x and x0
interchanged. Subtract the second equation from the first to find
0 dJ (u)
0 dJ (u) 02
Z 1
J (x )x J (x)x
2
+(x x ) ydyJ (xy)J (x0 y) = 0
du x du x0 0
(142)
If we let x = xn and x0 = xn0 , two distinct zeros of the Bessel function,
then the integrated terms vanish and we may conclude that the Bessel
functions J (xn y) and J (xn0 y) are orthogonal when integrated over
y from 0 to one, provided a factor of y is included in the integrand.
We still have to determine normalization in the case n = n0 . In the
preceding equation, let x0 = xn and rearrange the terms to have
dJ (u) Z 1
J (x)xn = dy y(x2 x2n )J (xn y)J (xy), (143)
du xn 0
or
Z 1 J (x)xn (dJ (u)/du)|xn
dy yJ (xn y)J (xy) = . (144)
0 x2 x2n
40
Use LHopitals Rule to evaluate the limit of this expression as x x n :
Z 1 1 dJ (x) dJ (u) 1
2
dy y[J (xn y)] = = [J0 (xn )]2 , (145)
0 2 dx xn du xn 2
41
form Z(z) eiz , and the functions of will be Bessel functions of
imaginary argument, e.g., J (i). For given 2 , there are two linearly
independent solutions which are conventionally chosen to be J and
H(1) , the reason being that they have particularly simple behaviors
at large and small arguments. Let us introduce the modified Bessel
functions I (x) and K (x),
42
From these equations we see that I (x) is well-behaved for x < ,
corresponding to < , while K (x) is well-behaved for x > 0 or
> 0, from which we can decide which function(s) to use in expanding
any given potential problem.
Lets look at some examples of expansions in cylindrical coordinates.
2.1 Example I
43
= V ( , ) z
a
c
=0
x
44
form
X
X
(, , z) = Amn eim J|m| (xmn /a) sinh(xmn z/a) (158)
m= n=1
where xmn is the nth zero of J|m| (x). Each term in the sum is itself
a solution of the Laplace equation; each one satisfies the boundary
conditions on z = 0 and = a, and, for given n, we have a complete
set of functions of while for given m, we have a complete set of
functions of .
The coefficients in the expansion are determined from the condition
that reduce to the given potential V on the top face of the cylinder.
Making use of the orthogonality of the basis functions of both and ,
we have
Z a Z 2
d dV (, )J|m| (xmn /a)eim (159)
0 0
or R 2 R
d 0a deim J|m| (xmn /a)V (, )
0
Amn = . (160)
a2 sinh(xmn c/a)[J|m|+1 (xmn )]2
For any given function V , one may now attempt to complete the inte-
grals.
2.2 Example II
Consider the same geometry as in the first example but now with
boundary condition = 0 on the constant-z faces and some given
value V (, z) on the surface at = a.
45
= 0 z
a
c
= V( ,z)
=0
x
46
Thus the expansion is
X
X
(, , z) = Amn eim sin(nz/c)I|m| (n/c) (162)
m= n=1
47
asymptotic form of the Bessel function, valid at large argument, is
v
u
u
t
2
Jm+1 (xmn ) cos[n + (m 1/2)/2 (m + 1)/2 /4]
xmn
v v
u u
u 2 nu 2
= t cos[(n 1)] = (1) t (167)
.
xmn xmn
Using this substitution in the closure relation for the finite interval
and taking the limit of large a, one finds that the sum becomes the
integral, Eq. (165). Of course, this is not a rigorous derivation because
the asymptotic expression is not arbitrarily accurate for all roots.
0 1 Z ik(zz 0 ) 1Z
(z z ) = dke = dk cos[k(z z 0 )] (169)
2 0
48
and
1 X 0
( 0 ) = eim( ) . (170)
2 m=
Similarly, expand the -dependence and z-dependence of G using the
same basis functions,
X Z dk
0 0 im(0 )
G(x, x ) = 2
g m (k, , )e cos[k(z z 0 )]. (171)
m= 0 2
Now operate on this expansion with the Laplacian using cylindrical
coordinates:
X Z dk d2 1 d m2 im(0 )
2 G(x, x0 ) = 2 2
+
2
+ k 2
g m e cos[k(z z 0 )]
m= 0 2 d d
4 Z dk
X 0
0
= ( ) 2
eim( ) cos[k(z z 0 )].
(172)
m= 0 2
0 0
Multiply by members of the basis sets, i.e., eim ( ) and cos[k 0 (z z 0 )]
and integrate over the appropriate intervals of 0 and z z 0 to find
a differential equation for gm ,
d2 gm 1 dgm 2 m2 4
2
+ k + 2
g m = ( 0 ). (173)
d d
For 6= 0 , this is Bessels equation with solutions (viewed as functions
or ) which are Bessel functions of imaginary argument, or, as we have
described, modified Bessel functions of argument k. Because of the
delta function inhomogeneous term, the solution for < 0 is different
from the solution for > 0 . Hence we may write that, for < 0 ,
49
and, for > 0 ,
50
The remaining constant in the determination of gm can be found
from the required normalization of G. Let us integrate Eq. (173) across
the point = 0 :
Z 0 + d2 1 d 2 m2 4
d + k + g m = (178)
0 d2 d 2 0
If we take the limit of 0 and realize that gm is continuous while its
first derivative is not, we find that this equation gives
0
dgm + dKm (x) dIm (x) 4
lim = A Im (k0 )k Km (k0 )k
= ,
0 d 0 dx dx 0
k0 k0
(179)
or
0 0 4
A[Im (x)Km (x) Km (x)Im (x)] = ; (180)
x
the primes denote derivatives with respect to the argument x. The
quantity in [...] here is the Wronskian of Im and Km . One may learn
by consulting, e.g., the section on Bessel functions in Abramowitz and
Stegun, that Bessel functions have simple Wronskians:
0 0 1
Im (x)Km (x) Km (x)Im (x) W [Im (x), Km (x)] = . (181)
x
Comparison of the two preceding equations leads one to conclude that
A = 4. Hence our expansion of G(x, x0 ), which is just 1/|x x0 |, is
1 Z
2 X 0
0
G(x, x ) = 0
= dkeim( ) cos[k(zz 0 )]Im (k< )Km (k> )
|x x | m= 0
(182)
51
which may also be written entirely in terms of real functions as
1 4Z
0
= dk cos[k(z z 0 )]
|x x | 0
1
X
I0 (k< )K0 (k> ) + cos[m( 0 )]Im (k< )Km (k> ) . (183)
2 m=1
52