8.9 Problem 8.

8.9.1

R h i
V E
r  r  E~ d x
~ 3

k = 2
R
V E
Ed x
~ ~ 3

R R h  i R h i R  
V
~
E  ~

Ed
3


x V
~
E 
rr ~
E +  ~
E ~
V E r  r  E~ d x E~ 
E~ + Ed
3 ~ x 3

k2 = R 
2 R 
~ 
Ed
~
2
~ 
V E Ed x
~ 3

V E x 3

R h i
V
~
E 
rr E~ + rr
 ~
E
= R 
V E~  Ed

~ 3
x
R h i R  3  R h R i
V
~
E  rr
~
E d 3
x E
~  ~
Ed x V
E~  r  r  E~ d3x E~ 
Ed
~ 3x
R 
2 R 2
V E
Ed x
~ ~ 3
V E
Ed x
~ ~ 3
R h i R h i
V E~ 

r  r  E~ ~
V E r  r  E~ d x RV E~ 
Ed
3
~ x 3
= R R R
V E
Ed x
~ ~ 3 ~ 
Ed
V E {z
~ x 3
E~ 
Ed
~ x 3
| } V
k2
R h i R
V E~ 
r  r  E~ d x k E~ 
Ed
3 ~ x 2 3

= R (1)
V E
Ed x
~ ~ 3

R h i
V E
r  r  E~ d x:
We now examine the integral ~ 3

Z h i Z
@ @
E~ 
r  r  E~ = Ei E " " d3 x
V V @x` @xj k |mjk{z mi`}
ij k` j` ik
Z Z Z Z 
@ @ @ @
= Ei E d3 x Ei Ei dxj dxk dxi
V @xk @xi k i k i | {z } | {z } @xj @xj
u dv
Z Z Z  >
0 Z
@ @ @  @ @  
= Ei E d3 x 
Ei  Ei 
Ei E dx dx dx
V @xi @xk k i k @xj
 V i |@x{z
j @xj i j k i
} | {z }
dv u

1
 Z Z Z  >
0 Z
@ @ @   @ @  
= Ei Ek d3 x Ei  Ei  + Ei E dx dx dx
V @x i @x k i k @xj
 V i @xj @xj i j k i
| {z } | {z }
dv u
Z Z
@@ @ @  3
= E
i @x @x Ek d x
3
Ei Ei dx
V
Z h
i k V @x j @xj
  i
= E~ 
r r
 E~  E~
r2 E~  d3 x
V
 
*0
  
*0


~
r
Using the fact that E 
r

~
 E = ~
r
0 = E r
E~ :
 
Z h i Z h   i
E~ 
r  r  E~ =  E~
r r
E~  r 2
E~  d3 x
V V | {z }
Z
rrE~ 
h i
=  E~
r  r  E~  d x 3

V
Plugging this into equation (1) yields:
R h i R
r  r  E~  d x k V E~ 
Ed
V  E~
~ x 3 2 3

k =2
R
~ 
Ed
~ x
V E i
3

R h 
V  ~
E
r  r  E~  k ~
E  dx 2 3

= R
V E
Ed x
~ ~ 3

=0
where, in the last step, we've used the vector Helmholtz equation, r  r  E ~ = k2 E~ .
Thus, there are no rst-order changes in k2 , indicating that there are only second-order and
higher changes in k2 .

8.9.2

~ = E0 cos(=2R)^z ,
Given that E
   
r  r  E~ =r r
E~ r E~ 2

| {z }
0

! 0 0
!
1 @ @ E~ ~7
1 @ 2E


@ 2 E~7

=  + 2  2 + 2
 @ @  @' @z
      
 2
= E0 sin + 2 cos z^
2R 2R 4R 2R

2
Plugging this into the denition of k2 gives:
RRh  i

sin 
+ 2 cos2 
d
0 2R
cos 2R 2R 4R 2 2R
k2 = RR

0
cos2 2R d
(4 +  2 )
1

= 1 2 16

4 2
R ( 4 + 2)
2 2 + 4
= 2 2
4R  4
r
 2 + 4
=) Rk = = 2:4146
2 2 4
Note that we can ignore the integrals in ' and z . Since neither of the integrands depend on
either of these variables, the values of the ' and z integrals will cancel.
The rst root of J0 (x) is 2.4048, so this approximate solution is about 0.4% away from the
exact answer.

8.9.3

~ = E0 [1 + (=R)2
Given that E (1 + )(=R)4 ],
! 0 0
!
  1 @ @ E~ ~7
1 @ 2E
@ 2 E~7

r  r  E~ = r E~ =
2

 @ @

+ 2  2 + 2
 @' @z
4

= E0 R2 4r 2 4r2  z^
R4
Plugging this into the denition of k2 yields:
RR
(1 + (=R)2 (1 + )(=R)4 ) R44 (R2 4r2 4r2 ) d
k = 2 0
RR
0
(1 + (=R)2 (1 + )(=R)4 )2 d
2 + 43  + 13 2
= R2
60
(16 + 7 + 2 )
20 6 + 4 + 2
= 2 (2)
R 16 + 7 + 2
We want to nd the value of  which minimizes the above expression:
dk2 20 (16 + 7 + 2 )(4 + 2) (6 + 4 + 2 )(7 + 2)
= =0
d R2 (16 + 7 + 2 )2
1 p
=)  = (10  34)
3
3
+ is a maximum while a is a minimum. Plugging  into equation (2) yields:
s p
17 2 34
kR = 80 p = 2:4050
68 + 34
This approximate solution is about 0.009% away from the exact answer.

8.14 Problem 8.14

8.14.1

We start with equation 8.117 in Jackson:

 2
dx
n 2
= n2 (x) n 2
dz
Letting n(x) = n(0) sech(x), n = n(0) sech(xmax ), and sinh(x) = sinh(xmax ) sin(z ):
 2
d
n (0) sech (xmax )
2 2
sinh 1 [sinh(xmax ) sin(z )] = n2 (0) sech2 (x) n2 (0) sech2 (xmax )
dz
 2
d 1
sech (xmax )
2
sinh 1 [sinh(xmax ) sin(z )] = sech2 (xmax )
dz 1 + sinh (xmax ) sin (z )
2 2

(3)
I then used following Maple commands:
x:=x->arcsinh(sinh(a*xmax)*sin(a*z)/a);
simplify(sech(a*xmax)^2*diff(x(z),z)^2\
- 1/(1+sinh(a*xmax)^2*sin(a*z)^2) + sech(a*xmax)^2);

The output was zero. Hence, the two sides of equation (3) are equal and
x = sinh 1 [sinh(xmax sin(x))] is a solution to Jackson's equation 8.117.
Because n = n(0) sech(xmax ) = n(0) cos((0)), sech(xmax ) = cos((0)). Hence,
1
xmax = cosh 1 [cos((0))]

To plot the rays, we note:
sech(x) = cos((0))
q
1
1 + sinh2 (x) =
cos (0)
s s
1 1 cos2 (0)
sinh(x) = 1= = jtan (0)j
cos2 (0) cos2 (0)

4
  
=) x = sinh 1
jtan (0)j sin(z)

These paths are plotted in gure 1 for three dierent launch angles (0).

(0) = =6 (0) = =4 (0) = =3

Figure 1: Plots of x(z ) for three values of (0) where  = 1.

8.14.2

To nd the half-period of the ray Z = 2z (xmax ), we rst solve our given equation for x(z )
for z in terms of x:
 
1 sinh(x)
z= sin 1
 sinh(xmax )

Solving for Z yields:

2 2 
Z = z (xmax ) = sin 1 (1) =
  

5
8.14.3

Starting with the paragraph above equation 8.119 in Jackson:

Z xmax
Lopt = 2 n(x)ds
0
Z Z=2
n(x)
=2 n(x) dz
0 n
Z Z=2
sech(x)
=2 n(0) sech(x) dz
0 sech(xmax )
Z Z=2
cosh(xmax )
= 2n(0) dz
0 cosh2 (x)
Z =2
cosh(xmax )
= 2n(0) dz
0 1 + sinh2 (x)
Z =2
1
= 2n(0) cosh(xmax ) dz
0 1 + sinh (xmax ) sin2 (z )
2

where we've usedR the identity Jackson gives, sinh(x) = sinh(xmax ) sin(z ). Using the
integral identity 0=2 1+C 2dzsin2 (z) = 2p1+C 2 yields:
" #

Lopt = 2n(0) cosh(xmax ) q
2 1 + sinh2 (xmax )
| {z }
cosh( xmax )

= n(0) = n(0)Z

The optical path length is independent of xmax , and hence it is also independent of (0).
Thus, optical path length is the same for all launch angles.