Chapter 3: Boundary-Value Problems

i Electrostatics:
in El t t ti II
We begin this chapter with 3 sections (Secs. 3.2, 3.5, & 3.6) on
mathematics.
3.2 Legendre Equation and Legendre Polynomials
Legendre Equation :

d 
dx  
d  1  x 2 du     1 u  0,  1  x  1
ddx 
(3.9)
The solutions are: u (x)  AP ( x)  BQ ( x)
 P ( x) : Legendre
d function
f i off theh first
fi kind
ki d
Q ( x) : Legendre function of the second kind
 
Ref. 1: Gradshteyn & Ryzhik, "Table of Integrals, Series, and
Products," Chs. 7 & 8.
R f 2: Abramowitz
Ref. Ab it & Stegun,
St "Handbook
"H db k off Mathematical
M th ti l
Functions," Ch. 8. 1
 3.2 Legendre Equation and Legendre Polynomials (continued)

Rewrite
R it the
th solution:
l ti u (x)  AP ( x)  BQ ( x)
Q ( x) diverges as x  1. Hence, Q ( x) is not commonly used
i physics.
in h i
P ( x) is finite for x  1 and x  1, but P ( 1) diverges unless 
i an integer
is i t (
(see p.105.)
105 )
In many physics problems, boundary conditions require  to be
an integer.
i t Since
Si th
the fform off th
the Legendre
L d equation
ti isi unchanged
h d if
    1, we have P 1 ( x)  P ( x). Hence, when  is an integer
(denoted by l ),) negative l is redundant. Thus l  00, 11, 2 and Pl ( x)
redundant Thus,
becomes a polynomial (properties on following pages).
Note: The range (  1  x  1) considered here is often encountered
in physics problems. Mathmatically, the range of P ( x) and Q ( x)
can be extended to the entire complex x  iy plane. Furthermore,  can
plane Furthermore
also be a complex number (See Gradshteyn & Ryzhik). 2
 3.2 Legendre Equation and Legendre Polynomials (continued)

1 l 2
Legendre Polynomial : Pl ( x )  l d
l ( x  1) l
, l  0, 1, 2... (3.16)
2 l ! dx
Pl (  x )  ( 1)l Pl ( x)
Ql (1)  
Pl (1)  1

Pl (1)  (1)l
Lengendre polynomials P2(x) - P5(x) Second Lengendre functions
[P0(x) = 1, P1(x) = x] Q0(x), Q1(x), and Q2(x) 3
 1
The set Pl ( x) is orthogonal: 1 Pl ( x) Pl ( x)dx  2l21 l l (3.21)
It is also complete in index l. Hence, any function f ( x) can be

expanded
d d as f ( x)   Al Pl ( x) (3 23)
(3.23)
l0

3 5 Associated Legendre Functions and the

3.5
Spherical Harmonics
Associated Legendre Equation :

dx 
d 1  x 2 du     1  m2  u  0 , for  1  x  1
dx  1 x2 
The solutions are: u ( x)  APm ( x)  BQm ( x)
 Pm : associated
i d Legendre
L d function
f i off theh first
fi kind
ki d
 m
Q : associated Legendre function of the second kind
(Refs.: Gradshteyn & Ryzhik; Abramowitz & Stegun)
4
 3.5 Associated Legendre Functions and the Spherical Harmonics (continued)

Rewrite the solution: u( x )  APm ( x )  BQm ( x )

Qm ( x ) diverges as x  1, hence is not commonly used in physics.
Pm ( x ) iis finite
fi it on the l  1  x  1 only
th interval
i t l when
h
 is zero or a positive integer (  l  0, 1, 2...) and
 [p. 107.]]
[p
m  l ,   l  1 ,,  1,1 0, 1 ,  l  1 , l
0 1,
Under these conditions, we have (for positive or negative m )
2 m2 d l  m 2
m ( 1)m
Pl ( x )  l (1  x ) dx
2 l!
  (x  1)l (3.50)
 Pl0 ( x )  Pl ( x )
 m l m m
P
 l (  x )  (  1) Pl ( x )
 m (l  m )! m
with the properties:  Pl ( x )  ( 1)m Pl ( x ) (3.51)
 (l  m)!
 1 P m ( x ) P m ( x )dx  2 (l  m )!
 1 l  l ll  (3.52)
2l  1 (l  m )! 5

 3.5 Associated Legendre Functions and the Spherical Harmonics (continued)

The set Plm ( x ) is complete in index l in the sense any function f ( x )

  m : a fixed integer 
can be expanded as f ( x )   Cl Plm ( x ) 
l m See (A.3) A 
(A 3) in Appendix A.
Spherical Harmonics Ylm ( , ):
2l  1 (l  m )! m
Ylm ( , )  Pl (cos  )eim , (3.53)
4 (l  m )!
where l  0 or a positive integer; m  l ,  (l  1),, 0,, (l  1), l
Y0,0 ( , )  1 z x
 4
Y ( , )  3 sin  e i 
 1,
1 1 8 r
Examples : 
Y1,0 ( , )  43 cos  y

Y1,1 ( , )   3 sin  ei 
 8 x 6
 3.5 Associated Legendre Functions and the Spherical Harmonics (continued)

Properties of spherical harmonics:

(i) Using the orthogonality relation,
1 2 (l  m )!
P m
1 l  ( x ) P m

l ( x ) dx  ll  (3.52)
2l  1 (l  m )!
we can show that the spherical harmonics are orthonormal, i.e
*
 d  Yl m ( ,  )Ylm ( ,  )   ll  mm , (3.55)
where
2  2 1 z
 d  0 d 0 sin  d  0 d 1 d cos  x


r
y

x
7
 3.5 Associated Legendre Functions and the Spherical Harmonics (continued)

l ( ,  ) is complete,
(ii) The set Ylm i e an arbitrary function g ( ,  )
complete i.e.
can be expanded as
 l
g ( ,  )    AlmYlm ( ,  ) (3.58)
l 0 m l
*
Multiplying both sides by Ylm l ( ,  )), integrating over  ,  , and
making use of (3.55), we obtain
*
Alml   d  Yl ( ,  ) g ( ,  )
lm
Substitution of Alm into (3.58) gives the following expression
for g ( ,  )),
  l
g ( ,  )   d    Ylm *
( ,  )Ylm ( ,  )  g ( ,  )
l 0 ml 
 * l
   Ylm ( ,  )Ylm ( ,  )   (   ) (cos   cos  ) (3.56)
l 0 m  l
This is the completeness relation of Ylm ( ,  ) [cf. (2.34) & (2.35).]
8
 3.5 Associated Legendre Functions and the Spherical Harmonics (continued)

(iii) Other ti off Ylm ( , ):

Oth properties )
Yl , m ( , )  ( 1)) m Ylm
*
( , )

 2l  1
Yl ,0 ( , )  Pl (cos  )
 4
This can be seen from the definition of Ylm ( , ) :

2l  1 (l  m )! m
Ylm ( , )  Pl (cos  )eim
4 (l  m )!
and the relations:
(l  m )! m
Pl m ( x )  ( 1) m Pl ( x ) (3 51)
(3.51)
(l  m )!
Pl0 ( x )  Pl ( x )

9
 3.6 Addition Theorem for Spherical Harmonics
The addition theorem for spherical harmonics is derived on pp. pp
110-111. Here we write the theorem without derivation:
4 l *
( )
Pl (cos  Ylm ( ,  )Ylm ( ,  )), (3 62)
(3.62
2l  1 ml z x
w e e  iss the
where t e angle
a g e between
betwee x and a d x. 
Setting l  1 in (3.62) gives r x
4 *  
P1 (cos  )  [Y1,1 ( ,  )Y1,1 ( ,  ) r
3
* y
 Y1,0 ( ,  )Y1,0 ( ,  ) 
* x
 Y1,1 ( ,  )Y1,1 ( ,  )] 
(  )  cos  , Y1,1 
U i P1 (cos
Using 3
8
i  e i , Y1,0 
sin 3
4
cos  ,

11  
and Y1,1 3
8
sin  ei , we obtain a useful expression:
p
cos   cos  cos    sin  sin   cos(   ). (1) 10
 3.1 Laplace Equation in Spherical Coordinates
 2  ( x)  0
 
2 2
1 
 r 2  r   2 1  sin    1   0
r r sin    r sin   2
2 2
x
U (r )
Let  (x)  r P( )Q( ) 
r
 
2 2
UQ d d Q
 PQ 2  2
d U sin  d  2 2
dP UP 0
dr r sin  d r sin  d 2
2 2 
i 
r sin
Multiply by
UPQ
Dividing all terms by sin2, we see that the
r-dependence is isolated within this term. So
 ( 1) this term must be a constant. Let it be  (  1).
 
2 1 2 d 2U 1 d 2Q
 sin  [U r 2  P sin  d (sin  d )]  Q
d dP 1
2 0 (3.3)
dr d

 2
The -dependence is isolated within this term, m
so this term must be a constant. Let it be -m2. 11
 3.1 Laplace Equation in Spherical Coordinates (continued)
 ( 1))  m

2


Rewrite (3.3): sin 2 1
[U r 2 d 2U
dr
1
 
2  P sin  d sin  d ]  Q
d d P 1 d 2Q
d 2 0

d 2Q 2
The equation for Q( ) is: 2  m Q0 (3.4)
d
 Q  eim , e im m is
i to
t be
b determined
d t i d
The equation for P ( ) is from the b.c.


sin  d d 
1 d sin  dP   (  1)  m2  P  0.
sin 2  
(3.6)

Let x  cos  , then the equation takes the form of the associated
Legendre equation:

dx  
d 1  x 2 dP     1  m2  P  0
dx  1 x 2 
 Pm ( x)   Pm (cos )   is to be determined
P m  m  (2)
Q ( x)  Q (cos  )  from the b.c.
12
 3.1 Laplace Eq. in Spherical Coordinates (continued)

 ( 1) 2
 m




2
Rewrite (3.3): sin  1
[U r 2 d 2U 1
 d 2Q
 P sin d sin  d ]  Q 2  0
dr 2
d d P 1
d

2U  ( 1)
The equation for U (r ) is: d
2  2 U 0 (3.7)
dr r
U Since  is
Si i determined
d t i d
 U  r 1 , r    r , r  1
r from the b.c. for (3.6),
Thus
Thus, this is not an eigenvalue
r   Pm (cos  )  eim  problem.
    1   m   im  ,
r (  )  e
 Q (cos 
where each bracket represents a linear combination of the two
functions inside. Because the differential equation is linear, the linear
combination of any number of solutions is also a solution.
Note that  and m are arbitrary constants until we apply boundary
conditions.
13
 3.3 Boundary-Value Problems with
Azimuthal Symmetry
y y
Problem 1: Find  inside 2 hemispheres held at opposite
potentials as shown in the figure.
 V , 0    
 2  0,  (a,  )     
2
 V , 2 a
r   Pm (cos  )  eim 
    1   m   im 
r  Q (cos  )  e 
(i)  is independent of  .  m  0
(ii)  is fi i at   0 andd  (i.e.
i finite (i at Note :
1. P (1) diverges unless  is
cos   1 and  1).
an integer
g (p.105.)
(p )
   l  0, 1, 2, and drop Q m
2. We have set l  0, 1, 2,
 1
(iii)  is finite at r  0.  drop r because Pl 1( x)  Pl ( x).
 3. Q ( x)   as x  1.
  (r ,  )   A r l P (cos  )
l l
l 0 14
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)

 V , 0    
The b.c. at r  a is :  (a, )   Al al Pl (cos  )     
2
l  V , 2

Use 11 Pl ( x) Pl ( x)dx  2l21  ll (3.21)

 
 11 Pl (cos  ) ( a,  ) d cos   Al al 11 Pl2 (cos  )d cos   Al al 2l21
 Al  Vl 2l21 [ 01 Pl (cos  )d cos   01 Pl (cos  )d cos  ]
a
l 1

 V   2   2l  1 l  2  !! , for odd l
1 2
(2n+1)!!=(2n+1)(2n1)(2n3)…5×3×1

  al 2  l 21  !

pp. 99 100 
99-100 0 , for even l
  (r ,  )  V 2 a P1 (cos  )  8  a  P3 (cos  )   , r  a
 3 r 7 r 3
(3.36)
 
To find  for r  a, replace  a  in (3.36) by  r  [see (2.27)]
r l a l 1
15
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)

Problem 2: A conducting sphere of radius a with net charge

Q on its surface is placed in a uniform electric field E0ez . Use the
method of expansion to find  outside the sphere and  on the
spher
h e.
E0
r   P (cos  )  e
m im
 r
    1   m   im  a 
 Q (cos  )  e
E0 z
r  metal
sphere
(i)  is independent of  .  m  0
(ii)  is finite at   0 and  (i.e. at c os   1 and  1).
   l  0,, 1,, 2,, and drop
p Qm Question:
Q

]Pl (cos  )   (l  1) as
 (l 1) l
Hence,  (r ,  )   [ Al r  Bl r
l r
l 0 r  . Why y keep
Q
b.c. at r   :    E0 r cos   r the Al r l terms?
 
external field due to net chrage Q
 A1   E0 , Al  0 (for l  1), and B0  Q 16
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)

  (r ,  )   E0 r cos   r   Bl r  (l 1) Pl (cos
Q
( )
l 1
b.c. at r  a:  (r  a )  const.
B1 Q   (l 1)
  (r  a )  ( E0 a  2 ) cos
 a     B

a Pl (cos  )
a
 not a l 2
l  
0 not a const.
0 const.
 B1  E0 a3 and Bl  0 for l  2
Q 3
  (r ,  )   E0 r cos   r  E0 2 cos 
a
r
due to induced surface charge
density  on the sphere
3
As will become clear in Ch. 4 [Eq. (4.56)], the E0 2 cos  term in 
a
r
is due to an electric dipole of dipole moment p  4 0 a3 E0 . (see p.64)
The induced surface charge g densityy  is
 Q
   0 r  3 0 E0 cos   0 2 17
r a a
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)
Problem 3:  due to a unit point source at x in infinite space
First, let's assume the point source is on the z -axis (at a distance
r  from the origin) and divide the space into two regions: r  r  and
r  r . In each region, we have  2  0 with the solution z
x  x
r   Pm ((cos  )  eim  r
    1   m   im  r

r  Q (cos  )  e 
y
indep of  .  m  0
(i)  is indep. x
(ii)  is finite at   0 and  .    l  0,1, 2,  and drop Qm
at r  0.  drop r l 1 in region r  r 
(iii)  is finite 
 as r   .  drop r l
in region r  r 
 
Al r l Pl (cos  ), r  r
l  0

  Bl r l 1Pl (cos  ), r  r 
l  0 18
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)

The formal method to solve for Al and Bl is to match the b.c. at

r = r’ (as will be done in Sec. 3.9). Here we obtain Al and Bl by
g the fact that we alreadyy know Φ = 1 xx’|| ((for a unit
exploiting
p
point source, q  . So, by the uniqueness theorem, we have

 
 Al Pl (cos  )
r l
, r  r
1 l  0
 
x  x   l 1
 Bl r Pl (cos  ) , r  r 
l 0
F   00, we hhave Pl (1)  1 and
For d x  x  r  r  . H
Hence,
 
 Al r l
, r  r
1 l  0

r  r   l 1
 Bl r , r  r
l 0
19
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)

n(n  1)) n2 2 n(n  1)(

)(n  2)) n3 3
( x  y ) n  x n  nx n1 y  x y  x y 
2! 3!
 1 1 1 1  r l  rl
     ( r )   l 1 , r  r

 r  r r  1  r  r  l 0
r l 0 r 
1
But 
r  r  1 1 1 1  r l  r l
   ( )   l 1 , r  r
 r  r  r 1  r  r l 0 r l 0 r
 r
Equating the RHS of this equation to the RHS of the equation on
the previous page, we obtain   rl
  l 1 Pl (cos  ), r  r 
1 1 l  0 r 
Al  l 1 , Bl  r l

r x  x   r l
l  P (cos  ) , r  r 
l 1 l
0 r
 r l
1
or   l1 Pl (cos  ), [two equations in one] (3.38)
x  x l 0 r
where r (r ) is the smaller (larger) of r and r . 20
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)

Rewrite (3.38):
z
1  rl
  l1 Pl (cos  ) point x x
x  x l 0 r source
r
r
This equation was derived with the 
unit point source located on the z -axis
axis y
(upper figure). However, it depends only
on the magnitudes (r , r ) of x and x and x
the angle ( ) between x and x. So we z x point
expect
p the expression
p in (3.38)
( ) can be cast  source
into a general form which holds true when r x
the unit point charge is at an arbitrary point  
r
(lower figure). We may obtain the general y
form by way of the addition theorem. 
x

21
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)

S b th
Sub. the RHS off the
th addition
dditi theorem
th x  point
z
4 l *  source
Pl (cos  )   Ylm ( ,  )Ylm ( ,  )
r
(3.62)
2l  1 ml x
 
 rl
r
1
for Pl (cos  ) in   l1 Pl (cos  ), y
x  x l 0 r 
x

1  l 1 rl *
we gett  4   
Ylm ( ,  )Ylm ( ,  ) (3.70)
(3 70)
x  x l 0 m l 2l  1 r
l 1

So, we started with a physics problem (the potential of a point

So
charge in infinite space), but end up with a mathematical relation
in (3
(3.70).
70)
Question: Why write a simple function   1/ | x  x | in such
a complicated
li d form?
f ? (See
(S next problem.)
bl )
22
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)

Problem 4: Potential due to charge

g q uniformly
y distributed on a
circular ring of radius a.
Let  ( x )  K (   ) ( r  c ) in spherical
p coordinates
q    ( x )d 3 x 3x
d

 K   (   ) ( r  c ) r 2 sin  drd d
 2 Kc 2 sin  a
q
 K
2 c2 sin 
q c 
  (x)  2
 (   ) ( r  c )
2 c sin 
q
 2
 (cos   cos  ) ( r  c )
2 c
 ( x a )
  f ( x)  
f ( a )
23
 3.3 Boundary-Value Problems with Azimuthal Symmetry (continued)

q
 ( x )   (cos    cos  ) ( r   c )
2 c 2
1  1 rl *
 (x) 3
l
1  4   Ylm ( ,  )Ylm ( ,  )
 ( x)   d x x  x  l 2l  1 r
l 1
4 0 x  x
v l  0 m 

q  l 1  r
l
Y *
( ,  ) Y ( ,  ) 
2
    r  d
dr d cos  d    r l 1 lm

lm 
2 0c l 0 ml 2l  1
2 v
 (cos    cos  ) (r   c) 
(l m)! m
Ylm ( ,  )  24l1 (  )eim 
Pl (cos
(l  m)!
Apparently, only the m  0 terms survive the   integration.
  2  r
l
P (cos  ) P (cos  ) 
q   1
  
  ( x)    r dr  d cos  r l 1 l

l
4 0c l 0
2 0  1
 
 (cos    cos  ) (r   c) 
q  rl
  P (cos  ) Pl (cos  )
4 0 l 0 rl 1 l
Jackson uses a slightly different method to derive this. See p.103. 24
 3.4 Behavior of Fields in a Conical Hole or
Near a Sharp Point
Consider the source-free configurations shown
P
g
in the figures. r
2 r   Pm (cos  )  eim  
   0      1   m   im 
r  Q ((cos  )  e  

(i) The geometry is indep. of  (We also assume

 P
indep. of  .)  m  0
that the b.c. is indep. r
(ii) Qm (cos ) diverges at   0 or cos  1. 

 drop Qm (cos )

r 
Hence,,     1  P ((cos  )
r 
Note: P ( x) diverges at x  1 unless   integer. However, in this
problem,
bl we have
h       cos   1 in i the
th region
i off
interest. Hence,  is not required to be an integer. 25
 3.4 Behavior of Fields in a Conical Hole or Near a Sharp Point (continued)

r 
Rewrite:     1  P (cos  )
r 
(iii)  is finite at r  0.
(a) demand   0 and drop r  1    r P (cos  )

d   1  0 and
(b) ddemand d drop
d r    r  1P ((cos  )
But P (cos  )  P 1 (cos  ), hence   r  1P 1 (cos  )
 Either option (a) or option (b) gives   r P (cos  ),   0
(iv)   0 at     P (cos  )  0    1 ,  2 ,  3 , (  0)
Note: In the boundary condition: P (cos  )  0,  is fixed and
 is the eigenvalue to be solved.
solved r  0

    Ak r k P k (cos  )  A1r1 P1 (cos  ), (3.44)
k 1
where 1 is the smallest eigenvalue [the first root of P (cos  )  0].26
 3.4 Behavior of Fields in a Conical Hole or Near a Sharp Point (continued)

r 0 P
 Er     1 A1r1 1P (cos  )  r1 1 r
 r 1 

 1  1 1 1 1 
  E   r   A1r sin  P1 (cos  )  r

   0 E (   )   A1 0 r1 1 sin  P (cos  )  r1 1
 1
 P
Behavior of 1 as a function of  is shown r

in the figure below. Note that
1
1  1,
1 if   90
 2
1  1, if   90
 1
1  1, if   90
0 180

90
When  < 90o (conical hole), both E and   0 as r  0.
27
 3.4 Behavior of Fields in a Conical Hole or Near a Sharp Point (continued)

However, when  > 90o (sharp point),

However point) both E and    as
r  0. Large electric field (E > 2.5  104 V/cm) can cause the air to
breakdown and form a conducting path in the air for the sharp point
to discharge. This is the principle of the lightning rod (pp. 77-78.)
If the region
g of interest is bounded byy the surface at r  a, the
coefficients Ak in (3.44) can be determined by the b.c.  (r  a )   ( )
g
through

 ( )   Ak a k P k (cos  )
k 1
If  (r  a )   ( )  0, then all Ak  0.    0 everywhere
In reality, the lightning rod is not perfectly sharp. Hence,  is
finite at the tip, and on a clear day when there is a small potential
difference between the g ground the clouds,, the lightning
g g rod will
not discharge.
28
3.4 Behavior of Fields in a Conical Hole or Near a Sharp Point (continued)

A physical picture of the lightning rod

29
 3.7 Laplace Equation in Cylindrical Coordinates;
Bessel Functions
2  2 1  1  2
 2
  (x)  0  2     2 2  2  0 z 
   z
Let  (x)  R(  )Q( ) Z ( z ) P
 2Z z
2  kz
k x
 z 2  k Z  0  Z  e
 2

   Q2   2Q  0  Q  ei
y
  
 x
  2 R 1 R  2  2 
  2      k   2  R  0  R  J (k  ),
) N (k  )

where J and N are Bessel functions of the first and second kind,,
respectively (see following pages).
 J (k  )  ei  ekz 
   i    kz  (3)
 N (k  )  e  e  30
 3.7 Laplace Equation in Cylindrical Coordinates; Bessel Functions (continued)

Bessel Functions : If we let x  k  , the equation for R takes

the standard form of the Bessel equation,

dx 2 x d
dx 
d 2 R  1 dR  1   2 R  0
x 2  (3 77)
(3.77)

with solutions J ( x) and N ( x), from which we define the Hankel

f ti
functions:
 H(1) ( x)  J ( x)  iN ( x)
 (2) (3 86)
(3.86)
 H ( x)  J ( x)  iN ( x)
and the modified Bessel functions (Be (Bessel
ssel functions of imaginary
argument)
 I ( x)  i  J (ix) (3 100)
(3.100)
  i 1H (1) (ix )
 
K ( x )  2  (3.101)
S JJackson
See k pp. 112
112-116,
116 G Gradshteyn
dh & Ryzhik,
R hik andd Abramowitz
Ab i
& Stegun for properties of these special functions. 31
 3.7 Laplace Equation in Cylindrical Coordinates; Bessel Functions (continued)

Bessel functions J0(x), J1(x), and J2(x)

Modified
difi d Bessell
functions I0(x), I1(x),
Neumann functions N0(x),
) N1(x),
) and K0(x),
) and K1(x)
N2(x)
32
3.8 Bounday-Value Problems in Cylindrical Coordinates
Example
E l 1: Potential
P t ti l inside
i id a charge-free
h f cylinder
li d (see
( figure)
fi )
with the b.c.  ( z  L)  V ( , ) and   0 on other surfaces.
 J ( k  )   i   kz 
2  e e z
  ( x)  0       i    kz  VV(( ,, ) )
 
N ( k  )  e  e 
(i) Z ( z )  Aekz
k
 Be kz
k

  0 at z  0  Z (0)  0  B   A   0 a L
 
 Z ( z )  A ekz  e kz  A sinh kz
y
(ii)  ( )   (  2 )), ii. ee.  is single-valued
single valued. 00
   m  integer x
 im 
 Q( )   Cm e    Am sin
i m  Bm cos m 
m  m 0
((iii))  is finite at   0.  drop
p N m (k  )  R  J m (k  )

33
 3.8 Bounday-Value Problems in Cylindrical Coordinates (continued)

Rewrite: R  J m (k  )
x
(iv)   0 at   a  J m (ka )  0  k  kmn  mn a , n  1, 2, 3
where xmn is the n-th root of J m ( x)  0. Thus,,
 
   ,  , z     J m  kmn   sinh  kmn z  Amn sin m  Bmn cos m 
m 0 n 1
With k fixed by the boundary condition to a set of dicrete values
(kmn ), we may introduce two properties of J m (kmn  ) :
The set J m ( kmn  ) is orthogonal in index n : [m : a fixed number.]
 a
     a 2 2
 nn ((3.95))
 0 m mn
J ( k ) J ( k
m mn ) d [ J
2 m1 mn( k a )]
The set J ( k x) is complete in index n. Hence, any function
 m mn

 f ( x) can be expanded as f ( x)   Cn J m (kmn x)
 n 1
Questions: (See last page of Appendix A.)
1 Why
1. Wh is i J m (kmn x) orhtog
h onall andd complete l in
i index
i d n instead
i d off m?
2. Why is there a factor  in the integrand of (3.95), but not in (3.52)? 34
 3.8 Bounday-Value Problems in Cylindrical Coordinates (continued)
Rewrite:
 
   ,  , z     J m  kmn   sinh  kmn z  Amn sin m  Bmn cos m 
m 0 n 1

(v)    ,  , z  L   V   ,  

 V   ,     sinh  kmn L  J m  kmn   Amn sin m  Bmn cos m 

m,n
2 a
Operating both sides with 0 d 0  d  J m (kmn  )
sin m
cos m
and  
making use of the orthogonal properties of sin m and cos m , and
a 2
the relation: 0 J m ( kmn  ) J m ( kmn  )  d   2 [ J m1 ( kmn a)]2  nn (3.95)
a

 Amn  2cosech  kmn L  2 a sin m 

  2 2  d 0  d V (  ,  ) J m (kmn  )  
 mn   a J m1  kmn a 
B 0
 cos m 
(for m  0, use 1B )
2 0n 35
 3.8 Bounday-Value Problems in Cylindrical Coordinates (continued)

Example 2: Potential in the charge-free semi-infinite space z  0

   ,  , z  0   V   ,  
subject to the b.c. 
     ,  , z   0
2  J (k  )  ei  ekz 
  ( x)  0       i    kz 
 N (k  )  e  e 
(i)  remains fini te as z  .  drop ekz  Z ( z )  Ae  kz
(ii)  ( )   (  2 )    m  integer

 Q( )    Am sin m  Bm cos m 
m 0
(iii)  is finite at   0.  drop N m (k  )  R  J m (k  )
(iv)   0 at     J m (k  )  0  continuous eigenvalue k

    ,  , z    0 dke kz J m (k  )  Am (k ) sin m  Bm (k ) cos m 

m 0
(3.106)36
 3.8 Bounday-Value Problems in Cylindrical Coordinates (continued)

g to k :
Rewrite ((3.106)) with variable k changed

   ,  , z    0 dk e k z J m (k  )  Am (k ) sin m  Bm ( k ) cos m 

m 0
(v)    ,  , z  0   V   ,  

 V   ,     0 dk J m  k    Am (k ) sin
i m  Bm (k ) cos m 
m 0

Operating both sides with 02 d 0  d  J m (k  )

 sin m
cos m and 
making use of the orthogonal properties of sin m and cos m , and
 1
the relation: 0 xJ m ( kx) J m ( k x) dx   ( k  k ) (3.108)
k

 
 Am (k )  k 2
 Bm (k )  

  0 d 0  d V (  ,  ) J m (k  ) cos m
sin m

(3.109)

1
For m  0, use B0 (k ) in series (3.106).
2 37
 3.9 Expansion of Green Functions in
S h i l Coordinates
Spherical C di t
The Green function for an electrostatic potential problem satisfies
 2G  x, x   4  x  x 
with G  x, x   0 for x on the boundary
y surface.
Question: Jackson p.120 states the b.c. as "G (x, x)  0 for either
x or x on the boundary surface
surface." Why?
Case 1: Green function in infinite space
Th simplest
The i l formf i G  x, x   1 (Sec.
is (S 1.10). point
1 10) source
xx
x x
p
It can be expressed as an expansion
p in r
spherical coordinates as (Sec. 3.7) r
1  l 1 rl *
G  x, x    4   
Ylm ( ,  )Ylm ( ,  ) (3.70)
x  x l 0 ml 2l  1 r
l 1
38
 3.9 Expansion of Green Functions in Spherical Coordinates (continued)
Case 2 : Green function outside a conducting sphere
point
i t
By the method of images, we have obtained image x  source
the Green function in Sec. 2.6, charge
r x
1 a aa22 x r
G  x, x    xx22
(2.16)
x  x x x  a x
2

x2 a G0
The first term in (2.16) is expanded in (3.70).
2
Th secondd term
The t can be
b expandedd d using
i (3.70). Since x  2 x ,
(3 70) Si a
x
2
a2
we substitute r = x  r and r  a 2 x  r into (3.70) to obtain
x

  2 l
a
a  l 1 a r *
 4  

Yl ( ,  )Ylm
lm l ( ,  )
l 0 m l 2l  1 r r
2 l 1
x x  a 2 x
x
r
1  rl 1 a 2 l 1  *
 G  x, x   4   l 1  ( )  Ylm ( ,  )Ylm ( ,  ), (3.114)
l ,m 2l  1  r a rr  39
 3.9 Expansion of Green Functions in Spherical Coordinates (continued)

Case 3: Green function inside a spherical shell bounded by

grounded conductors (see figure)
 2G  x, x   4  x  x 
x
with the b.c. G (r  a )  G (r  b)  0
r x
Thiss pproble
ob em iss ddifficult
cu to o solve
so ve by thee r
method of images. We will solve it by a a
systematic method: method of expansion.
b
Write  ( x  x) in spherical coordinates,
 (x  x)  12  (r  r ) (   ) ((cos   cos  )
r
Use the completeness relation (3.56) for  (   ) (cos   cos  )
 l *
  (x  x)  1 
2  (r  r )   Ylm ( ,  )Ylm ( ,  ) (3.117)
r l 0 m l
Note that, in (3.117), we have decomposed a point charge into an
infinite number of spherical "charge layers", all of radius r .
40
 3.9 Expansion of Green Functions in Spherical Coordinates (continued)
 l
  G  x, x   4
2 1  ( r  r )   Y * ( ,  )Y ( ,  )
lm lm (4)
r2 l 0 m l

variable co
constant
sa constants variables
The RHS of this equation is an expansion in spherical harmonics,
which suggests that we expand G  x, x  similarly. This is possible
since Ylm ( ,  ) form a complete set.
 l
G  x, x     Alm  r | r , ,    Ylm ( ,  ), (3.118)
l 0 m l
variable constants variables
where Alm is a function of r to be solved from (4).
Expressing
i Alm as

Alm  r | r , ,     gl (r , r )Ylm ( ,  ) (5)
and sub. (5) into (4), we get the equation for gl ( r , r ) (see Sec. 3.1),
41
 3.9 Expansion of Green Functions in Spherical Coordinates (continued)

1 d2 l (l  1) 4
2  rgl (r , r )  2 gl (r , r )   2  (r  r ) (3.120)
r dr r r
Divide the space into r  r  and r  r . In each region, (3.120)
reduces to
1 d2 l (l  1)
2  rggl (r , r )  2 gl (r , r )  0
r dr r
 Ar l  Br l 1 , r  r 
 gl (r , r )   (6)
 Ar l  Br l 1 , r  r  r
The remainingg job j is to find 4 r
a
boundary conditions to solve for the
b
4 constants A, B, A, and B in (6). ( )

42
 3.9 Expansion of Green Functions in Spherical Coordinates (continued)

(i) gl (r  a, r )  0  gl (r , r )  A 2 l 1
r  l 1 ,
l a
r
 r  r

(ii) g (r  b, r )  0  g (r , r )  B  1  rl
 , r  r r
r
l l r l 1 b 2l 1 a
(iii) gl (r , r ) is continuous at r  r . b
Physical reason:  is continuous across the charge
layer at r  r . (E is finite at r  r .    lim E r  0). Thus,
r 0
1  r l

 A  C 1  rl


A r l a
r
2 l 1
 l 1   B  1
r l 1
 r l
b 2l 1  A
 
B
r l 1 b 2l 1
l a 2l 1
r   l 1

 r l 1 b 2l 1


l a 2l 1
 B  C r   rl 1 
r 
C 1  rl

 rl 1 b2l 1
 gl (r , r )  
l a 2l 1
r  l 1 ,
r   r  r

C r  

l a

2l 1 1  rl ,
rl 1 r l 1 b2l 1   r  r

 l a 2l 1   1 rl 
 C  r  l 1   l 1  2l 1  (3.122)
 r   r b  43
 3.9 Expansion of Green Functions in Spherical Coordinates (continued)

1 d2 l (l  1) 4
Rewrite (3.120): 2  rgl (r , r )  2 gl (r , r )   2  (r  r )
r dr r r
((iv)) We need one more condition to get g the remainingg constant C in
(3.122). Physically, this condition is related to the discontinuity
of Er ( drd g ) across the charge
d l
g layer y at r  r . Mathematicaly, y, we
integrate the delta funtion in (3.120) to bring out the discontinuity.
p y ((3.120)) by
Multiply y r and integrateg from r    to r    (  0))
d d 4
  l
rg ( r , r ) r   l
rg ( r , r ) r  
dr dr r
rgl (solid line)

  r 1  ( r) 2l 1    2l 1 
 l  (l  1)( b )
C a r
 as a function of r

C (l  1)  l ( a )2l 1  1  ( r)2l 1    4

r  r  b  r
4
 C
(2l  1) 1  ( ba )2l 1  r r
44
 3.9 Expansion of Green Functions in Spherical Coordinates (continued)

Sub. C into (3.122), we get

G  x, x  
 l * ( , )Y ( , )
Ylm  l a 2l 1  1 rl 
 4   
lm
2 1   r  r l 1  r l 1  b2l 1  (3.125)
l 0 m  l (2l 1) 1( )
a l

    
 b 

Limitingg case 1: a  0 & b  , (3.125)

( )  (3.70)
( )
1  l 1 rl *
G  x, x     4   
Ylm ( , )Ylm ( , ) (3.70)
x  x  l 2l  1 r
l 0 m 
l 1

Limiting case 2: b  , (3.125)  (3.114)

1  rl 1 a 2 l 1  *
G  x, x  4   l 1  ( )  Ylm ( , )Ylm ( , ), (3.114)
l ,m 2l  1  r a rr 
45
 3.10 Solution of Potential Problems with the
Spherical Green Function Expansion
Example 1: Potential inside a charge-free
n
sphere
p of radius b subject
j to V ( , )
the b.c.  ( r  b)  V ( ,  ) b
Since we already have the Green function for volume of
i
interest
this problem, it is convenient to use the formal
solution derived in Sec. 1.10:
1 3 1 
 ( x)   (x ) G (x, x )d x   s  (x ) G (x, x)da (1.44)
v     
4 0 4 n
0
1 
There is no charge inside.   (x)    s  (x)  G (x, x)da
4 n
Note: The unit vector n is normal to the surface and pointing
outward from volume of interest. n is a differentiation
along n ( n  r for this example).
46
 3.10 Solution of Potential Problems with the Spherical Green Function Expansion (continued)
Rewrite (3.125):
 l* ( , )Y ( , )  2l 1   1 rl 
G  x, x   4  
Ylm

lm
a
r
2 l 1   
l
 l 1   l 1  2l 1 
a

 
l 0 ml (2 l 1) 1
 b
( ) r   r b 
For this example, a  0, r  r , and r  r , hence
1 1 l
G  x, x   4 
 *
Ylm  
l ( ,  )Ylm l ( ,  ) r ( l 1  2l 1 )
l r
l ,m 2l  1 r b
 1 * 1 l 1
 G  4  Ylm ( ,  )Ylm ( ,  )r ( l  2  2l 1 )
l l l r
r l ,m 2l  1 r b
  4 *
 G  G   2  ( r )l Ylm ( ,  )Ylm ( ,  ) (7)
n rb r rb b l ,m b
da  r 2 sin  d d   b 2 d  (8)
  x  s    r   b   V  ,    (9)
Sub. (7) - (9) into  ( x)   41  s  ( x)  G ( x, x) da, we get
n
  x      V ( ,  )Ylm *
( ,  )d  ( r )l Ylm ( ,  ) (3.128)
l ,m b 47
 3.10 Solution of Potential Problems with the Spherical Green Function Expansion (continued)

Example 2: Potential due to a uniformly charged ring of radius a and

total charge Q located on the x-y plane inside a grounded
conducting sphere of radius b
In spherical coordinates, the x-y plane
is at    2.The charge density  (x) can
be written as
b Q
Q
 ( x)  2
 (r  a ) (cos  ) a (3 129)
(3.129)
2 a
The potental is given by
1 3 1 
 ( x)  v  ( x )G ( x , x ) d x    s   G (x, x)da (1.44)
 ( x )
4 0 4 n
0
There is no inner conductor in this problem.  (3.125) reduces to
 l l 1 rl 
G  x, x   4   2l 1 Ylm
1 *
l ( ,  )Ylm
l ( ,  ) r  l 1  2l 1  (10)
l 0 m l  r b 
48
 3.10 Solution of Potential Problems with the Spherical Green Function Expansion (continued)

Sy e y in   m  0. Hence,
Symmetry e ce,
Ylm ( ,  )  Yl 0 ( ,  )  2l 1P (cos  )
4 l

 G  x , x 

  Pl (cos  ) Pl (cos  )rl l11
l 0 r  
b
rl
2 l 1  (11)
Q
S b (11) and
Sub. d  ( x)  2  (r  a ) ((cos  )
2 a b Q
into (1.44), we obtain
a
1
 x (x)G  x, x 
3
 ( x)  d
4 0
 (r   a ) (cos
(  ) 

 
Q
 2 2  r 2 dr d cos  d    rl 
8  0 a   P (cos  ) P (cos  ) r
l 1  
 l 0
l l rl 1 b 2 l 1 

Q 
4 0 l 0
l 1
r b
rl
 Pl (0) Pl (cos  ) r l 1  2 l 1  (3.130)

where r (r ) is the smaller (larger) of r and a. 49

 3.10 Solution of Potential Problems with the Spherical Green Function Expansion (continued)

Example 3: Potential due to a uniformly charged line of length 2b and

total charge Q located on the z-axis inside a grounded
conductingg sphere
p of radius b ((see figure)
g ) z
linear
li
The charge density  ( x) can be written in density 2Qb
b
spherical
p coordinates as (see ( pproblem below.)) y
Q 1
 ( x)  2  (cos   1)   (cos   1) x (3.132)
2b 2 r
The potential is given by
1 3 1 
 ( x)  v  ( x )G ( x , x ) d x    s   G (x, x)da
 ( x ) d  (1
(1.44)
44)
4 0 4 n
0
Rewrite (11)
(11), which is applicable to this problem:
G  x, x  


  Pl (cos  ) Pl (cos  )rl l11
l 0 r

rl
b 2 l 1  (11)

Sub. (11) into (1.44), we obtain

50
 3.10 Solution of Potential Problems with the Spherical Green Function Expansion (continued)

  (cos 1) (cos 1) 

Q 2  2 r2 
 ( x)   r  dr d cos  d   
  (  )rl  l11  2l 1  
l
8 0b Pl ((cos  ) Pl (cos
r
 l 0  r b  
Q  b l 1 rl 
   Pl ((1))  Pl ( 1))  Pl ((cos  ) 0 r  l 1  2l 1  dr  ((3.133))
8 0b l 0   r b 

  r
1 rl
b
 r l
  l b 1
l 1  2 l 1 0 r dr  r r 
r l 1

b
r l
2 l 1  dr
 l ( l 1) 
2 l 1 [1  r l ]
b

Pl (1))  (1))l and Pl ((1))  1  Odd l terms cancel.

 b
Q  4 j 1  r )2 j  P (cos  )  (3.136)
  ( x)  ln
 r ( )   1  ( 
4 0b  j 1 2 j (2 j  1)  b  2j

Note that the l  0 term in (3.133) is given by ln( br ). See p.124. 51
 3.10 Solution of Potential Problems with the Spherical Green Function Expansion (continued)

oblem: Show
Problem S ow tthee charge
c a ge density
de s ty in (3.132):
(3. 3 ): z
li
linear
Q 1 density 2Qb
 ( x) 
2b 2 r 2  (cos   1)   (cos   1) b
y
represents a unifom charge distribution along z.
x
Solution: The total charge is
3 Q b 2 1 2  (cos 1) (cos 1)
  (x)d x  2b 0 r dr 1 d cos  0 d 2 r 2
Q b 1
 0 dr 1 d cos  [  (cos   1)   (cos   1) ]
2b  
 0, +z -axis   ,  z -axis
Q b
 b dz  uniform distribution from z  b to z  b.
2b
Note: The above integration over cos starts from cos  1 and
and ends at cos  1. It does not cross 1 or  1. This issue can be
resolved byy a limitingg pprocedure;; namely, y, we write
 (cos   1)   (cos   1)  lim[ (cos   1   )   (cos   1   )]
 0 52
optional 3.11 Expansion of Green Functions in
Cylindrical
y Coordinates
Consider the Green equation:
 2G  x, x   4  x  x  , with G  x, x   0 as x  
An obvious solution is 1/ x  x . We have also z
solved this equation by the method of expansion in
spherical coordinates [(3.70)]. Here, by the same P
z
method, we solve it again g in cylindrical
y coordinates. x
Write   x  x  as
y
  x  x   1                z  z    

x
      1  eim(  )
   2
m
with 
  z  z   1   dkeik  z  z  1   dk cos  k  z  z  
 2   0
       
2
  G  x, x    
2
  0 dke im (  )
cos  k  z  z  (12)
m 53
optional 3.11 Expansion of Green Functions in Cylindrical Coordinates (continued)

Since eim and eikz are complete sets,

sets we may expand G  x, x  in
variables  and z
  im (  )
G  x, x   1 1
2   0 dkg m ( k ,  ,  ) e cos[k  z  z ] (3
(3.140)
140)
m
where the coefficient g m (k ,  ,  ) is a function of m, k ,  and  .
Sub. (3.140) into (12) we get
 2 2 2

 1   
1
2  0 dk ( 2     2 2  2)
1
2 m    z
im (  )
 g m (k ,  ,  )e cos[k ( z  z)]
       im (  )
2
    0 dke cos[k ( z  z)] (13)
m

In (13), 2 2
 m , 2 2
 k , 2  1   1    . Hence,
 2 z 2  2
 1     (k 2  m2 )  g (k ,  ,  )   4        (3.141)
     2  m  54
optional 3.11 Expansion of Green Functions in Cylindrical Coordinates (continued)

See (3.98)-(3.101)
(3 98) (3 101) in Jackson
Jackson.
 AI m (k  )  BK m (k  ),   
 g m (k ,  ,  )  
 AI m (k  )  BK m (k  ),   
(i) g m is finite at   0.  B  0
(ii) g m remains finite as   .  A  0
((iii)) g m is continuous at    .
 AI m (k  )  BK m (k  )
A K m (k  )  A  CK m (k  )
  
B I m (k  )  B  CI m (k  )
CK m (k  ) I m (k  ),   
 g m (k ,  ,  )  
CI m (k  ) K m (k  ),   
 CI m (k  ) K m (k   ) (14) 55
optional 3.11 Expansion of Green Functions in Cylindrical Coordinates (continued)

((iv)
v) Too obta
obtain the coe c e t C in g m (k ,  ,  )  C
t e coefficient CI m (k  ) K m (k   ),
mutiply (3.141) by  and integrate form     to     (  0)
  4
dgm dg
 m
d    d   
 Ck[ I m (k  ) K m (k  )  K m (k  ) I m (k  )]   4
Use the relation: I m ( x) K m ( x)  I m ( x) K m ( x)  1/ x (3.147)
 Ck ( k1 )   4  C  4  g m (k ,  ,  )  4 I m (k  ) K m (k   )
Sub. the above expression for g m (k ,  ,  ) into (3.140)

 G  x, x     0 dkeim(  ) cos  k  z  z   I m (k  ) K m (k   )
2
m 
Since G (x, x)  1 , we have by the uniqueness theorem
xx
1 
dk im(  ) cos  k  z  z I m (k  ) K m (k   ) (3.148)
   0 dke
2
x  x m 
56
 3.12 Eigenfunction Expansion for Green Functions
Eigenfunction Expansion of Green Function in 3 Dimensions :
We have obtained the Green function for the Poisson equation by
the method of eigenfunction expansion in 2 dimensions [e [e.g.
g (3
(3.118),
118)
in  ,  ]. Here, we develop a general technique to obtain the Green
function by y eigenfunction
g expansion
p in 3 dimensions. Consider the
Green function for a more general equation (with homogeneous b.c.):
 2G  x, x    f (x)    G  x, x   4  x  x  ((3.156))
a given real function a given constant
We shall solve (3 156) by expanding G  x, x  and   x  x  in
(3.156)
eigenfunctions of a related problem formulated as follows.
an eigenvalue to be determined by the
same f (x) as in (3.156) b.c., not the same λ as in (3.156)
 2  x    f (x)     x   0 (3 153)
(3.153)
with the same boundary surface and homogeneous b.c. as for (3.156).
57
 3.12 Eigenfunction Expansion for Green Functions (continued)
Assume the ((3-dimensional)) eigenfunctions
g for
 2  x    f (x)     x   0
are  n (x)). Since the operator
p [ 2  f (x)] is Hermitian, we have
 3

v m ( x ) n ( x ) d x   mn
and  n form a complete set with real eigenvalue n [see Apendix A].
A]
Write G (x, x)   an (x) n (x) (3.157)
n
S b (3
Sub. (3.157) d  (x  x)   n (x) n (x) [[see (2
157) and (2.35)]
35)] iinto
n
 G  x, x    f (x)    G  x, x   4  x  x  , we obtain
2

 an ( x){  n ( x)   f (x)    n (x)}  4  n ( x) n (x)

2 
n n
Since  n satisfies  2 n (x)  [ f (x)  n ] n (x)  0,
0 we have

 an ( x)(  n ) n ( x)  4  n ( x) n ( x)
n n
  ( x)   ( x) ( x)
 an (x)  4 n  G (x, x)  4  n n (3.160)
n  n n  58
 3.12 Eigenfunction Expansion for Green Functions (continued)

We now specialize
p to the Green function for the Poisson equation
q
i.e. (3.156) with f (x)    0. z
Example 1: Green function for a rectangular box c
 2G  x, x   4  x  x 
 x  0 and a
 y
with G  x, x   0 at  y  0 and b b
 z  0 and c x a
C id the
Consider th corresponding di eigenvalue
i l problem
bl [(3.153)
[(3 153) with
ith f (x)
 0 and   k 2 ]:  2 (x)  k 2 (x)  0 with the same b.c.
1 2 X 1 d 2Y 1 d 2Z 2
Let  (x)  X ( x)Y ( y ) Z ( z )  X d
2  Y 2  Z 2  k 0
 dx dy dz
 
 kl2  2  k 2
 X ( x)  Aeikl x  Be ikl x km n

 Y ( x)  Beikl y  Ce ikl y with k 2  kl2  km2  kn2
 Z ( x)  Deikl z  Ee ikl z
 59
 3.12 Eigenfunction Expansion for Green Functions (continued)

 X ( x)  0 at x  0 and a kl  la  X  sin lax ,

   m y

b.c. Y ( x)  0 at y  0 and b  km  b and Y  sin b ,
m
  n  Z  sin n z ,
 Z ( x )  0 at z  0 and c k 
 n c  c

k  2 2
klmn  2 l2
 a 2  2
2
 2
m
b
n
c
2

  ( x)  8 sin l x sin m y sin n z
abc
b a b c
 j (x) j (x)
Sub.  (x) into (3.160): G (x, x)  4  , we obtain
j j 
2
  ;  j  klmn ;  0
G (x, x) j l ,m ,n

32  sin lax sin lax sin mb y sin mb y sin nc z sin nc z
  ((3.167))
 abc l ,m,n1 l2  m2  n2
a2 b2 c2 60
 3.12 Eigenfunction Expansion for Green Functions (continued)

p 2: Green function for infinite space

Example p
 2G  x, x   4  x  x  with G  x, x   0 as x  
Consider the corresponding eigenvalue problem
 2 (x)  k 2 (x)  0 in infinite space
Th solution
The l ti isi
1 ik x 1 ik x x ik y y ik z z
 k ( x)  32
e  32
e (3.162)
 2   2 
where k  k xe x  k y e y  k z e z
Since the region of interest is infinite space, we have continuous
32
engenvalue k 2 and the factor 1/  2  gives the normalization
condition for  k (x) :


 k ( x ) k ( x ) d 3
x    k  k  (3.163)
[see p.69 for a one dimensional example.]
61
 3.12 Eigenfunction Expansion for Green Functions (continued)

 n (x) n (x)
S the
So h series
i expansion:
i G (x, x)  4  [(3.160)]
[(3 160)]
n n  
becomes
 k (x) k (x) 3
G (x, x)  4  d k
k  
2 1 ik x
With k  k ,   0, and  k  32
e , we have
(2 )
ik ( x  x)
1 e
G (x, x)  2  d 3k
2 k2
Since G (x, x)  1 , we get another mathematical expression
xx
for 1 by the uniquess theorem
xx
ik ( x  x)
1 1 e
 2  d 3k (3 164)
(3.164)
x  x 2 k2
62
 3.12 Eigenfunction Expansion for Green Functions (continued)
Solution of Inhomogeneous Differential Equation by the Green
Function Method :
To show the usefulness of the 3-dimensional Green function just
obtained,
b i d we consider id an inhomogeneo
i h us differential
diff i l equation:
i
 2u (x)   f (x)    u (x)  4 S (x) (15)
wth homogeneous b.c. In (15), S (x) is a distributed source. We have
shown that the solution for the same equation with a point source:
 2G  x, x    f (x)    G  x, x   4  x  x  (3
(3.156)
156)
is G (x, x)  4  n (x) n (x) /(n   ), (3.160)
n
where  n (x) is the eigenfunction of  2 n (x)   f (x)  n  n (x)  0.
( ) is u (x)  v G (x, x) S (x)d 3 x, (16)
Then,, the solution of (15) ( )
which can be verified if we operate both sides with  2  f (x)   and
apply (3.156)
(3 156) to the RHS.
RHS
Note: If   n , there is no solution unless v un (x) S (x)d 3 x  0. 63
 Homework of Chap. 3

Problems: 1,, 2,, 3,, 6,, 7,,

9, 17, 20, 22

64