Assignment 4

Goldstein 3.35 Another version of the truncated repulsive Coulomb potential has the form

k k
(
V (r) = − r<a
r a
0 r>a

Obtain closed-form expressions for the scattering angle and the differential scattering cross section. These
are most conveniently expressed in terms of a parameter measuring the distance of closest approach in units
of a. What is the total cross section?

As the potential is zero outside of r = a, the particle is free. The impact parameter is then s < a. (If
s > a, then there is no scattering.) Much of the problem is getting the geometry of the system right. We
will set a coordinate system with origin at the center of the sphere. The particle enters the sphere of nonzero
potential at a height s above the horizontal plane. The particle will experience (due to the 1/r potential) a
hyperbolic orbit. It’s point of closest approach to the origin will occur at an angle (as in Goldstein Fig 3.20)
of Ψ. It will then, by symmetry, move through another angle of Ψ to exit the sphere at a scattering angle
of Θ. With respect to a line drawn from the origin at an angle of Θ to the horizontal, the particle, again by
symmetry, will exit the sphere at a perpendicular distance of s from that line.
It is useful to split up Ψ into two other angles. One is the angle that a vector from the origin to the
point of entry would make with the horizontal. Call this α. It is such that sin α = s/a. The remainder of Ψ
we will call φ. Its significance can be seen by considering the equation for the path inside the sphere, namely
a hyperbola:
1 mk
= u(θ) = 2 1 + e cos(θ − θ0 )


r(θ) l
If we take θ0 = π (see Goldstein p. 109), periapsis (the point of closest approach to the origin) corresponds
to θ = 0 and
mk
u(θ) = 2 (e cos θ − 1)
l
and finally θ = ±φ is thus the angles of entry and exit of our sphere in terms of our Kepleriean coordinate.
The relation for our scattering angle is thus

π = Θ + 2(α + φ)

So to make sense of the scattering problem, we need relation between s and Θ. In other words we need to
eliminate α and φ in terms of s and Θ if we can. From the angular momentum, we know
√
l = mv0 s = s 2mE

We also know that inside the nonzero potential we have

1 1
E= mṙ2 + mr2 θ̇2 + V (r)
2 2
1  dr 2 l2 k k
= m θ̇ + + −
2 dθ 2mr2 r a
l2  du 2 l2 2 k
= + u + ku −
2m dθ 2m a
l2  mke 2 l2  mk 2 mk 2
k
= sin θ + (e cos θ − 1) + 2 e cos θ − 1 −
2m l2 2m l2 l a
mk 2 2 k
= (e − 1) −
2l2 a

1
So we can solve for the eccentricity
2l2  Ea 
e2 = 1 + +1
mka k
Now note that
Θ
sin = cos α cos φ − sin α sin φ
2
where we have
s
sin α =
a
1p 2
cos α = a − s2
a
1  l2 
cos φ = +1
e mka
At this point, it’s just horrendous algebra. Using dimensionless variables
s
x=
a
Ea
=
k

we plug in these relations to the equation (squared) for sin(Θ/2)

Θ
sin2 = cos2 α cos2 φ + sin2 α(1 − cos2 φ) − 2 sin α sin φ cos α cos φ
2

and do all sorts of things, we find at the end (miraculously enough) that

Θ 1 − x2
sin2 =
2 1 + 4x2 (1 + )

Remembering that x is related to the scaled (by the radius) impact parameter, we can solve for s2 to get

cos2 (Θ/2)
s2 = a2
1 + 4(1 + ) sin2 (Θ/2)

The differential cross section is then

s  ds 
σ(Θ) =
sin Θ dΘ
 
a2 1+η
=
4 (1 + η sin2 Θ)2

with η = 4(1 + ).

2
Goldstein 4.4 (a) By examining the eigenvalues of an antisymmetric 3 × 3 real matrix A, show that 1 ± A
is nonsingular.
(b) Show then that under the same conditions the matrix


−1
B= 1+A 1−A

is orthogonal.

(a) First consider an antisymmetric 3 × 3 real matrix

 
0 a b
A =  −a 0 c 
−b −c 0

The corresponding characteristic equation for this matrix is

0 = det A − λ1
= (−λ)3 − abc + abc − λ a2 + b2 + c2

Letting s2 = a2 + b2 + c2 > 0, we can see that the eigenvalues of A are λ = 0, ±is and the A is a singular
matrix.
Now consider A ± 1. The characteristic equation for A can be made a statement for the nonsingular
nature of this new matrix. In particular, we simply set λ = ±1 in the above relation (not an eigenvalue any
more) and do not set it to zero. We have

det A ± 1 = ± 1 + r2

which, by virtue of r2 > 0 is always nonzero. Hence 1 ± A is nonsingular.

(b) With the same conditions, construct the new matrix

−1
B= 1+A 1−A .

To establish orthogonality, we will assume that the inverse to this matrix exists, B−1 , and show that it is
the same as BT . To this end, recall that by definition, AT = −A. Hence,

T
1±A = 1 ∓ A.

We will need a couple of standard results from the algebra of matrices. Recall that for two nonsingular
matrices, C and D, it is true that (CD)−1 = D−1 C−1 as well as (CD)T = DT CT . In addition, the order
in which we take the transpose and the inverse of a matrix does not matter. To show this, note that taking
the transpose of both sides of 1 = CC−1 yields
 T
1T = CC−1

T
1 = C−1 CT

−1
Multiplying this from the right by CT , we get

−1
T
−1
T
CT = C−1 CT CT = C−1

and the order of taking the transpose and the inverse doesn’t matter.

3
 With these results, we now look at
h
−1 iT
−1
BT B = 1 + A 1 − A



1+A 1−A
h
−1 iT
T

−1
= 1−A 1+A 1+A 1−A
h
T i−1


−1
= 1−A 1−A 1+A 1−A

−1

−1
= 1+A 1 − A + A − AA 1 − A

−1


−1
= 1+A 1+A 1−A 1−A
=1

which shows that BT = B−1 and B is orthogonal.

Technically, we should also establish that BBT = 1 as well to demonstrate that both the left and right
inverses are the same:

−1 h
−1 iT
BBT = 1 + A 1 − A



1+A 1−A


−1 h
−1 iT
T
= 1+A 1−A 1−A 1+A


−1 h
T i−1

= 1+A 1−A 1−A 1−A


−1
−1

= 1+A 1−A 1+A 1−A

h

i−1

= 1+A 1+A 1−A 1−A

h

i−1

= 1+A 1−A 1+A 1−A


−1
−1

= 1+A 1+A 1−A 1−A
=1
which should do it.

4
Goldstein 4.15 Show that the components of the angular velocity along the space set of axes are given in
terms of the Euler angles by
ωx = θ̇ cos φ + ψ̇ sin θ sin φ
ωy = θ̇ sin φ − ψ̇ sin θ cos φ
ωz = ψ̇ cos θ + φ̇

The angular velocity is given by ω ~ where, as the discussion in Section 4.9 indicates, we can simply
combine the infinitesimal rotations with angular velocities ωφ = φ̇, ωθ = θ̇, and ωψ = ψ̇. In addition, we can
add up the infinitesimal rotations to provide the finite angular velocities along each of these Euler angles:

ω
~ =ω~φ + ω ~θ + ω ~ψ
= φ̇ ẑ + θ̇ ξˆ + ψ̇ ẑ 0

where we have identified the different directions (unit vectors) around which each rotation was made. Of
course, these unit vectors do not belong to the same system (refer to Figure 4.7) so we will need to refer
each to the space set of axes. ẑ is the unit vector along the z-direction in the space set of axes and hence is
fine and we can leave it. ξˆ is the unit vector along the line of nodes and can be thought of as the original
x-axis rotated by φ around the z-axis. This is described (with an abuse of notation) by
    
cos φ sin φ 0 1 cos φ
ξˆ =  − sin φ cos φ 0   0  =  − sin φ 
0 0 1 0 0

or, in terms of the unit vectors:

ξˆ = cos φ x̂ − sin φ ŷ
ẑ 0 is the unit vector along the direction that arises by rotating the the original z-axis about the line of nodes
(the φ rotated x-axis) by an angle θ. This can be represented by
    
1 0 0 0 0
ẑ 0 =  0 cos θ sin θ   0  =  sin θ 
0 − sin θ cos θ 1 cos θ

where the rotation is with respect to the (in Goldstein’s notation) (ξ, η, ζ) system. Hence we can write this
as
ẑ 0 = sin θ η̂ + cos θ ζ̂
Note that ζ̂ = ẑ and that η̂ can be written akin to ξˆ as
    
cos φ sin φ 0 0 sin φ
η̂ =  − sin φ cos φ 0   1  =  cos φ  = sin φ x̂ + cos φ ŷ
0 0 1 0 0

We now have, on putting it all together, the following for the angular velocity



ω = φ̇ ẑ + θ̇ cos φ x̂ − sin φ ŷ + ψ̇ sin θ sin φ x̂ + sin θ cos φ ŷ + cos θ ẑ

except that this doesn’t look right. The only remaining thing to do is to note that we have described this
in terms of a passive transformation in which our transformation has worked on the coordinate systems. In
fact, we should have done an active transformation that works on the vector itself. The only difference is
that we allow the rotations to take place in the opposite directions. In other words, φ → −φ and so on for
each angle. On doing this, we can peel off the components, and we get exactly the result we have been asked
to get.

5
Goldstein 4.21 A particle is thrown up vertically with initial speed v0 , reaches a maximum height and
falls back to the ground. Show that the Coriolis deflection when it again reaches the ground is opposite in
direction, and four times greater in magnitude, than the Coriolis deflection when it is dropped at rest from
the same maximum height.

This is a fairly standard problem. The most difficult aspect, perhaps, is picking a coordinate system.
Because we are looking at a small region on the Earth (a sphere), it makes sense to put a coordinate origin
there. However, there is a temptation to set an origin at the center of the Earth with the z-axis running
through the North Pole. Resist this temptation! Instead, let us take our origin on the surface of the Earth
where we are with the z direction pointed vertically, the y-axis pointed north and the x-axis pointed east.
The Coriolis force (or “effect” if we concede terminology to Goldstein) is given by

F~c = −2m ω


~ × ~v

where ω~ is the angular velocity of the Earth at our location and ~v is the velocity of our particle. The angular
velocity, of course, will vary with latitude. We will measure latitude from the equator with the angle α.
Of course, ω~ will not, in this coordinate system, lie along any one of its axes (but would have if we had
succumbed to temptation and put the z-axis along the spin axis of the Earth). It does lie in the y-z plane
so we can write
ω
~ = ω (cos α ŷ + sin α ẑ)
Crossing this with the velocity of the particle gives


ω̂ × ~v = x̂ vz cos α − vy sin α + ŷ vx sin α − ẑ vx cos α

Note that as long as the velocity has no x component there will be no Coriolis effect in the z direction (or
the y direction). So in the case that the particle is thrown directly upwards (in the z direction), the vertical
motion will be determined entirely by the influence of gravity. Hence, we can immediately write
1
z(t) = − g t2 + vz0 t + z0
2
where vz0 and z0 are the initial velocity and z position, respectively. Similarly, the motion in the y direction
(north-south) will be trivial in this case.
The remaining component of the motion is that in the x direction and it will be dictated by

mẍ = −2mωvz (t) cos α = 2mω cos α(gt − vz0 )

which is easily integrated to yield

g 1 
x(t) = x0 + vx0 t + 2ω cos α t3 − vz0 t2 .
6 2
We will take the initial position of the particle to be at the origin of coordinates with initial speed of
v0 in the z direction. Of course, the maximum height is attained at t = v0 /g and is zmax = v02 /(2g). The
particle hits the ground at t = 2v0 /g and the total deflection in the x (east-west) direction is
 v03  8 
x(t) = 2ω cos α − 2

2v0 /g g2 6

Note that if the particle were dropped from the maximum height of zmax = v02 /(2g) then it would take
t = v0 /g to reach the ground. In that case there is no initial velocity in the z direction (vz0 = 0) and the
deflection becomes  v3 1
x(t) = 2ω cos α 02

v0 /g g 6
which is 4 times smaller and in the opposite direction of the first.

6
Goldstein 4.23 The Foucault pendulum experiment consists in setting a long pendulum in motion at a
point on the surface of the rotating Earth with its momentum originally in the vertical plane containing the
pendulum bob and the point of suspension. Show that the pendulum’s subsequent motion may be described
by saying that the plane of oscillation rotates uniformly 2π cos θ radians per day, where θ is the co-latitude.
What is the direction of rotation? The approximation of small oscillations may be used, if desired.

The Foucault pendulum is effectively a spherical pendulum. In a local coordinate system, the pendulum
is described by three coordinates, (x, y, z). If the Earth did not rotate, the force of gravity and the tension in
the pendulum’s string would be the only forces we would need to consider. Taking the origin of coordinates
to be the lowest point of the pendulum in its swing, the force of gravity to be F~g = −mgẑ, and the tension
to be T~ = x̂ T cos α + ŷ T cos β + ẑ T cos γ (where α, β and γ are direction cosines the pendulum makes with
the coordinate axes), Newton’s equations become
x
m ẍ = −T
l
y
m ÿ = −T
l
l−z
m z̈ = T − mg
l
where we have used cos α = x/l, cos β = y/l and cos γ = (l − z)/l. In the limit of small angle, we can say that
x and y are small in comparison to l and that z is even smaller (second order) and thus the z equation can
be solved to first order as T = mg. The other two equations for x and y, of course, are harmonic oscillator
equations and indicate that if, for instance, y(0) = ẏ(0) = 0, then y = 0 for all time and the pendulum would
oscillate in only a single plane (the x-z plane). Naturally if we had started with a different initial condition,
the result would be the same, but the plane would be different.
If we now incorporate the rotation of the Earth with a Coriolis term, we would have to add, as in the
last problem, a term such as

F~c = −2m ω


~ × ~v
h
i
= −2mω x̂ vz cos ψ − vy sin ψ + ŷ vx sin ψ − ẑ vx cos ψ

where we have used ψ as the latitude. Including these terms in the above, we have
T

m ẍ = − x − 2mω ż cos ψ − ẏ sin ψ
l
T
m ÿ = − y − 2mω ẋ sin ψ
l
z−l
m z̈ = −T − mg + 2mω ẋ cos ψ
l
Again, for small oscillations, the smallness of z allows us to solve the z equation as T = mg. Likewise
dropping the ż term from the x equation, we have
T
m ẍ = − x + 2mω sin ψ ẏ
l
T
m ÿ = − y − 2mω sin ψ ẋ
l
and we have a coupled set of equations. To solve these, it is helpful to find variables that will uncouple
them. One way to do this is to diagonalize the matrix that defines the coupling. More specifically, let
X = (mẍ + T x/l)/(2mω sin ψ) and similarly for Y . Then we have the relation
    
X 0 1 ẋ
=
Y −1 0 ẏ

7
The eigenvalues of the coupling matrix satisfy the characteristic equation (−λ)2 + 1 = 0 or λ = ±i. Finding
the corresponding eigenvectors we get  
1 1
vi = √
2 i
 
1 1
v−i = √
2 −i
Constructing the diagonalizing matrix, we have
 
1 1 1
A= √
2 i −i

T
The adjoint of this multiplying the original vector ( ẋ ẏ ) gives us our uncoupled variables, namely ẋ ± iẏ.
Using these variables in the form z = x + iy, we can write our equations as

T
mz̈ = − z − i 2mω sin ψ ż
l
This is just a damped harmonic oscillator with a complex damping. It will therefore have solutions of the
form eiΩt where on substituting in, we get the algebraic relation for Ω

T
−Ω2 + − 2ω sin ψ Ω = 0
ml
which has solutions r
T
Ω± = −ω sin ψ ± + ω 2 sin2 ψ
ml
s
ω2
= −ω sin ψ ± ω0 1 + 2 sin2 ψ
ω0

where we have defined ω02 = T /ml = g/l, the frequency of the pendulum. Provided the frequency of the
pendulum is much larger than the frequency of the earth’s rotation (a pretty good assumption), then the
solutions become
Ω± ≈ ±ω0 − ω sin ψ
Our original coordinates are thus

z = x + iy = Aeiω0 t e−iωt sin ψ + Be−iωt e−iω sin ψt

where A and B are complex coefficients. If we choose initial conditions such that the pendulum is pulled
back and released from rest, we have z(0) = x0 and ż(0) = 0 where we’ll take the initial displacement in the
x direction. As a result, we get

x0 = A + B
0 = iA(ω0 − ω sin ψ) + iB(−ω0 − ω sin ψ)

It should be clear that the rate of precession will be ω sin ψ.