1 Problem 11.

1.1 Part a

We begin by dierentiating the velocity addition formula (we will assume the space ship is
traveling parallel to the Earth):
u0 + v
u=
1 + uc2v0

 0  0
1+ u0 v
du0 7

+ dt
dv (u0 + v) 1 7
du0 v + u0 dv
du
= c2 dt c2 dt dt
dt 1+ u v
2
0

c2
1 vc22 du0 dt0
=

dt0 |{z}
1 + uc2v 0 2
dt
1=

Letting u0 = 0 because the ship is, by denition, at rest in its own instantaneous reference
frame:  =
du0
3 2
du v 2

dt
= 1 c dt0 2

Letting u = v because the space ship's velocity (according to Earth) is equal to the velocity
of the space ship's reference frame relative to Earth's:
 3=2
dv v2 dv 0
dt
= 1 c2 dt0
Z Z
dv dv 0
v2
3=2
= dt0
dt
1 c2
0 0
Given that dv =dt = g = constant,
v
q = gt
1 v2
v2
Solving for v yields:
= q gt 2 2v (1)
1 + gc2t
Now, we integrate both sides of the equation for time dilation:
Z Z
dt
dt0 =

Z r
v2
t0 = 1 dt
c2

1
and substitute in equation (1):
Z v
u
g 2 t2
t0
u
= t1   dt
c2 1 + gc22t2
Z   1=2
g 2 t2
= 1+ c2
dt

Looking up this integral in a table, we nd that it is equal to:

 
c gt
t0 = arcsinh
g c
Solving for t in terms of t0 yields:
 
c gt0
t= sinh
g c
For the rst leg of the journey, t0 = 5 years. In addition, g = 9:86 m/s , and c = 3  10 2 8

m/s:
 
t=
3  10 8
sinh (9:86)(5  3:16  10 )
 1 7

9:86 3  10 3:16  10
8 7

= 86 years
The total journey is 4  70 years = 344 years. Hence, it is the year 2100 + 344 = 2444 .

1.2 Part b

In the rst two legs, the rocket ship traveled:

Z T
d=2 vdt
0
Z : 
5 3 16 107
gt
=2 q dt
0
1 + gc22t2
= 7:83  10 m 16

2 Problem 11.11

Letting A = eL and A = e L

1 2
( + L) ,

A() = A2 A1 1 = e L ( + L) e L (2)

2
We want to prove that to rst order in L,
1
X 
A() =
n(L; L) (3)
n=0
n!

where
= I ,
(L; L) = L and
n(L; L) = [L;
n (L; L)] for n  2. Replacing the
0 1 1
left hand side of equation (3) with the Taylor series of A() yields:
1 n
X 1
X
 
n!
An( )
(0) = n!

n(L; L)
n=0 n=0

Hence, it is sucient to prove:

A(n) (0) =
n (L; L)
We will prove this by induction.
We begin by showing that this holds for n = 0 and n = 1:
A(0) (0) = I X
A (1)
(0) = (L + L) L = L X
Now, we assume that to rst order, A n (0) =
n(L; L) in order to prove that to rst
( )

order, A n (0) =
n (L; L). Now, we assume f n (0) =
n(A; B ) in order to prove that
( +1)
+1
( )

f n (0) =
n (A; B ) (induction).
( +1)
+1

A(n) (0) =
n (A; B ) = [L;
n (L; L)]
1
 
=) A(n) (0) = L; A(n 1)
(0) = LA n (0) ( 1)
A(n 1)
L

Taking the derivative of both sides,

 
A(n+1) (0) = LA(n) (0) A(n) L = L; A(n) (0)
= [L;
n(L; L)]
=
n+1(L; L)
Hence, we have proved equation (3) by induction.
A(n) (0) =
n (L; L)

Letting  = 1 in equation (3) and expanding the summation yields:

1 1
A = I + L + [L + L] + [L + [L + L]] + : : :
2! 3!

3
3 Problem 11.14

3.1 Part a

3.1.1 F  F

We solve for the Lorentz scalar F  F :

= F  F
F  F
= F  F  
We know that F  F is index notation for matrix multiplication and   takes the trace of
the resulting multiplication. Hence:


F  F = trace F  F (4)
where F  is dened in Jackson's equation 11.137:
2 3
0 Ex Ey E z
6 x
F  = 64E 0 Bz By 7 7
Ey Bz 0 Bx 5
Ez By Bx 0
To nd F , we invert the signs on only the rst row and rst column (or, just using Jackson's
equation 11.138 also works):
2 3
0 Ex E y Ez
6
F = 64 E x 0 Bz By 7 7
Ey Bz 0 Bx 5
Ez By Bx 0
Plugging these matrices into equation (4) yields:
02 32 31
0 Ex E y Ez 0 Ex Ey E z
F  F = trace B
B6Ex
6 0 Bz By 7 6
7 6 Ex 0 Bz By 7 7C
C
@4Ey Bz 0 Bx 5 4 Ey Bz 0 Bx 5A
Ez By Bx 0 Ez By Bx 0
= 2Bx + 2By + 2Bz 2Ex 2Ey 2Ez
2 2 2 2 2 2

= 2 jBj jEj
2 2

3.1.2 F  F
Next, we solve for the Lorentz scalar F  F :
F  F = F  F
= F  F  

4


F  F = trace F  F (5)
where F  is dened in equation 11.140:
2 3
0 Bx By Bz
6Bx
F  = 64B 0 Ez Ey 7
7
y Ez 0 Ex 5
Bz 0 Ey Ex
To nd F , we invert the signs on only the rst row and rst column:
2 3
0 Bx By Bz
6 Bx
F = 6
0 Ez Ey 77
4 By Ez 0 Ex 5
Bz Ey Ex 0
Plugging the matrices F  and F into equation (5) yields:
02 32 31
0 Bx By Bz 0 Ex Ey Ez
F  F = trace B
B 6
6Bx 0 E z E y 7 6 Ex 0
7 6 Bz By 7 C
7C
@4By Ez 0 Ex 5 4 Ey Bz 0 Bx 5A
Bz Ey Ex 0 Ez By Bx 0
= 4BxEx 4By Ey 4Bz Ez
= 4E
B

3.1.3 F  F
Finally, we solve for the Lorentz scalar F  F :
F  F = F  F
= F  F  


F  F = trace F  F (6)
Plugging the matrices F  and F into equation (6) (dened above) yields:
02 32 31
0 Ex Ey Ez 0 Bx By Bz
F  F = trace B
B6Ex
6 0 Bz By 7 6
7 6 Bx 0 Ez Ey 77C C
@4Ey Bz 0 Bx 5 4 By Ez 0 Ex 5A
Ez By Bx 0 Bz Ey Ex 0
= 2Ex + 2Ey + 2Ez 2Bx 2By 2Bz
2 2 2 2 2 2

= 2 jEj jBj 2 2

Because jBj jEj , jEj jBj , and E
B are the only three possible ways we can combine
2 2 2 2

vectors E and B to produce scalars which are quadratic in E and B, there are not any other
invariants quadratic in the eld strengths E and B.
5
3.2 Part b

No. Proof by contradiction:

Assume that there exists one reference frame in which we see a purely electric eld (i.e.,
jBj = 0) and another frame where we see a purely magnetic eld (i.e., jEj = 0), disregarding
the trivial case where jEj = jBj = 0.
We have just shown that jEj jBj is an invariant{ that is, it must remain constant across
2 2

all inertial frames. This invariant is equal to jEj in the frame where jBj = 0 and it's equal
2

to jBj in the frame where jEj = 0. Hence:

2

jEj = jBj
2 2

which is impossible since the square of a magnitude of a vector cannot be negative. So, no,
it is not possible to have an electromagnetic eld which appears as a purely electric eld in
one inertial frame and appears as a purely magnetic eld in another inertial frame.
Let S be a reference frame where there exists a nonzero electric eld and let S 0 be a reference
frame where the electric eld vanishes. Equating the invariants between these two elds
yields:
jBj2
jEj = jB0j =) jEj = jBj
2 2 2 2
jB0j
2

E
B = 0

(
=) jEj < jBj
2 2

E
B = 0