Physics 401 Intermediate Mechanics Fall 2024

Homework Assignment #10 — Solutions

1. [Based on Taylor, Exercise 8.14] Consider a particle of reduced mass µ orbiting in a
central force with U = krn where k and n are constants with kn > 0. (This condition
ensures that the force is attractive.)
(a) Find the radius at which the particle (with given angular momentum ℓ) can orbit
at a fixed radius (ie in a circular orbit).
The effective potential is
ℓ2
Ueff = + krn (1)
2µr2
and the effective force is
dUeff ℓ2
Feff = − = 3 − knrn−1 (2)
dr µr
This force has to vanish for a circular orbit
ℓ2 ℓ2
Feff = 0 ⇒ = knrn−1 ⇒ rn+2 =
µr3 µkn
Thus circular orbits can occur at
1/(n+2)
ℓ2

r0 = (3)
µkn

where r0 denotes the fixed radius.

(b) For what values of n is this circular orbit stable?
We can check for stability by examining the second derivative of the effective
potential (1)
d2 Ueff 3ℓ2 ℓ2
 
n−2 µkn(n − 1) n+2
= 4 + kn(n − 1)r = 4 3+ r
dr2 µr µr ℓ2
Substituting in r0 from (3) gives
d2 Ueff ℓ2
= (n + 2) (4)
dr2 r=r0 µr04

The potential should be concave-up (positive second derivative) for stability. This
gives the condition n > −2 for a stable circular orbit.
(c) For the stable case,√show that the period of small oscillations about the circular
orbit is τosc = τorb / n + 2 where τorb is the orbital period.
The radial equation is µr̈ = Feff where the effective force is given in (2). Hence
ℓ2
µr̈ = − knrn−1
µr3

1
 Substituting in r = r0 + ϵ and expanding to first order in ϵ gives
ℓ2
   
ϵ n−1 ϵ
µϵ̈ = 1 − 3 + · · · − knr0 1 + (n − 1) + · · ·
µr03 r0 r0
2 2
   
ℓ µkn n+2 ℓ µkn(n − 1) n+2
= 1 − 2 r0 − 4 3+ r0 ϵ + ···
µr03 ℓ µr0 ℓ2
ℓ2
= −(n + 2) 4 ϵ + · · ·
µr0
where in the last line we substituted in the value of r0 from (3). Incidentally, the
factor multiplying ϵ is just minus the second derivative of the effective potential
given in (4). In any case, comparing with the harmonic oscillator equation ẍ +
Ω2 x = 0 gives the frequency for small oscillations as
ℓ √
Ω= n+2
µr02
Recalling that the angular momentum
√ is ℓ = µr02 ω where ω is the angular fre-
quency of the orbit gives Ω = ω n + 2. Since the frequencies are inversely related
to the periods, we finally obtain
1 1 τorb
= √ ⇒ τosc = √
Ω ω n+2 n+2

2. A particle of mass m and angular momentum ℓ moves about a fixed force center. The
path of the particle is given by a logarithmic spiral, namely r(ϕ) = r0 eaϕ for constants
r0 and a. What is the force F (r) that gives rise to this path?
We rewrite the transformed radial equation
m
u′′ + u = − F
ℓ2 u2
as
ℓ2 u2
F =− (u + u′′ )
m
We now substitute in
1 1 a2 −aϕ
u= = e−aϕ ⇒ u′′ = e
r r0 r0
to obtain
ℓ2 u2 1 + a2 −aϕ ℓ2 u2 (1 + a2 )ℓ2 3
 
F =− e =− (1 + a2 )u = − u
m r0 m m
Since u = 1/r, we finally have

(1 + a2 )ℓ2
F (r) = −
mr3
This is an inverse cube force.

2
3. [Based on Taylor, Exercise 8.23] A particle of mass m moves with angular momentum
ℓ in the field of a fixed force center with
k λ
F (r) = − 2
+ 3
r r
where k and λ are positive.
(a) Show that the orbit of the particle has the form
c
r(ϕ) =
1 + ϵ cos(βϕ)
where c, β and ϵ are positive constants. Find c and β in terms of the given
parameters.
The transformed radial equation is
 
′′ m m 2 3

λm km
u = −u − 2 2 F = −u − 2 2 −ku + λu = − 1 + 2 u + 2
ℓu ℓu ℓ ℓ
or
km
u′′ + β 2 u =
ℓ2
where β 2 = 1+λm/ℓ2 . The homogeneous equation is that of a harmonic oscillator,
uh = A cos(βϕ − δ) where δ is a phase. It is also easy to see that a constant value
of up = km/β 2 ℓ2 can solve the inhomogeneous equation. Hence the complete
solution is
km km
u= 2 2
+ A cos(βϕ − δ) = 2 2 (1 + ϵ cos(βϕ − δ))
β ℓ β ℓ
where ϵ = Aβ 2 ℓ2 /km. This gives the orbit
c
r= (5)
1 + ϵ cos(βϕ − δ)

where r
β 2 ℓ2 ℓ2 λ λm
c= = + , β= 1+
km km k ℓ2

(b) Describe the orbit for the case that 0 < ϵ < 1. For what values of β is the orbit
closed?
The orbital equation (5) is very similar to that for Kepler orbits, except that
there is precession when β ̸= 1. The radius of closest approach is rp = c/(1 + ϵ),
while the radius of furthest approach is ra = c/(1 − ϵ). The choice 0 < ϵ < 1
ensures that the orbit is bound, and lies between the closest and furthest approach
distances. (This is the corresponding equivalent of elliptical Kepler orbits.) To
have a closed orbit, the orbit must come back to itself after the angle ϕ advances
by some integer multiple of 2π. This can only happen if β is a rational number ,
which can be written as a fraction β = p/q.

3
 (c) What happens when λ is negative with λ < −ℓ2 /m?
The effective force is
ℓ2 k ℓ2 /m + λ
Feff = +F =− 2 +
mr3 r r3
For λ < −ℓ2 /m, the coefficient of the 1/r3 term becomes negative and the at-
tractive 1/r3 force overcomes the centrifugal force. The effective force is always
attractive, and in this case bound orbits do not exist — the particle will either
hit the force center or escape to infinity.
4. [Based on Taylor, Exercise 8.17] Prove the virial theorem for any periodic orbit of a
particle.
(a) Find the time derivative of the quantity G = ⃗r · p⃗ and, by integrating from time
0 to t, show that
G(t) − G(0)
= 2⟨T ⟩ + ⟨F⃗ · ⃗r ⟩
t
where F⃗ is the net foce on the particle and
Z t
1
⟨f ⟩ ≡ f (t′ )dt′
t 0

is the time average of any quantity f .

For G = ⃗r · p⃗, we have
dG
= ⃗r˙ · p⃗ + ⃗r · p⃗˙
dt
For the first term, we use the standard definition of momentum p⃗ = m⃗r˙ , while for
the second we use Newton’s second law, p⃗˙ = F⃗ . This gives
dG
= m|⃗r˙ |2 + F⃗ · ⃗r
dt

Since the kinetic energy is T = 12 m|⃗r˙ |2 , we have

dG
= 2T + F⃗ · ⃗r
dt
We now integrate both sides from 0 to t
Z t Z t
G(t) − G(0) = 2 ′
T dt + F⃗ · ⃗r dt′
0 0

Dividing by t and using the definition of time average then gives

G(t) − G(0)
= 2⟨T ⟩ + ⟨F⃗ · ⃗r ⟩ (6)
t

4
 (b) Show that the left side approaches zero as t → ∞, and hence deduce that
⟨T ⟩ = − 12 ⟨F⃗ · ⃗r ⟩
when the averages are taken over a very long time.
For a periodic orbit, the displacement ⃗r and momentum p⃗ remain bounded. As
a result, the quantity G is bounded. What this means in particular is that there
is a maximum as well as a minimum value of G along the orbit. This gives the
time-independent bound
|G(t) − G(0)| ≤ Gmax − Gmin
So as t → ∞, the left hand side of (6) goes to zero since the numerator is bounded
while the denominator goes to infinity. It immediately follows that as t → ∞ we
must have
0 = 2⟨T ⟩ + ⟨F⃗ · ⃗r ⟩
which implies that
⟨T ⟩ = − 21 ⟨F⃗ · ⃗r ⟩ (7)
Note that, if we integrated over any number of complete periods, then G(t) = G(0)
(since the motion has ended up at the same displacement and velocity as the
staring point). In this case, the left hand side vanishes without having to take
time to infinity.
(c) Show that ⟨T ⟩ = n⟨U ⟩/2 for motion with the potential U = krn .
For U = krn the force is
F⃗ = −∇U
⃗ = −∇(kr
⃗ n
) = −knrn−1 r̂
Inserting this in (7) then gives
1 n n
⟨T ⟩ = − ⟨−knrn−1 r̂ · ⃗r ⟩ = ⟨krn ⟩ = ⟨U ⟩
2 2 2

5. [Based on Taylor, Exercise 8.31] Consider the motion of two particles subject to a
repulsive inverse-square force F = k/r2 with positive k. Show that this system has no
states with E < 0 (as measured in the CM frame), and that in all states with E > 0,
the relative motion follows a hyperbola. Assume ℓ ̸= 0, and note that we should have
r > 0 for the orbit.
The effective potential is
ℓ2 k
Ueff = 2
+
2µr r
(This is just like for the Kepler problem, except that here the sign of the k/r
potential is positive.) This potential is a sum of two positive terms, so it follows
that Ueff > 0. The total energy E cannot be less than Ueff (otherwise the kinetic
energy would be negative). But since Ueff > 0, it immediately follows that we
must have E > 0. This is obvious if we plot the effective potential

5
 U_eff

For the orbit, we may solve the modified radial equation

µ µk
u′′ + u = − F (u) = −
ℓ2 u2 ℓ2
The solution is just as in the standard (attractive) case, except that we change
the sign of the force constant k. As a result, we have

ℓ2 /µk
r=−
1 + ϵ cos ϕ

Since r should be positive, and since ℓ2 /µk > 0, the motion must be such that
the denominator is negative. This can only occur when ϵ > 1 and cos θ < −1/ϵ.
We can show that this is a hyperbola by following the same manipulations as for
the Kepler problem. We start with

ℓ2 ℓ2
r(1 + ϵ cos ϕ) = − ⇒ r=− − ϵx
µk µk
Squaring both sides gives
2
ℓ2 ℓ2

2 2
x +y = + 2ϵ x + ϵ2 x2
µk µk

We now complete the square to get

2 2
ϵ ℓ2 ℓ2
 
2 2 1
(ϵ − 1) x + 2 −y =
ϵ − 1 µk µk ϵ2 −1
or 2 2
ϵ ℓ2 ℓ2
 
2 2 2 2
(ϵ − 1) x + − (ϵ − 1)y =
ϵ2 − 1 µk µk
This matches the hyperbolic Keplerian case, except for the sign of the x-shift.
This shift ensures that the body comes in from the left, and always remains to
the left of the origin (center of mass)

6
 y y

x x

Attractive Repulsive
6. [Computer] In general relativity, a particle of mass m moving with angular momentum
ℓ about a fixed force center experiences a force
k 3kℓ2
F (r) = − −
r2 m2 c2 r4
where k = GM m and c is the speed of light. Numerically solve for the orbit r(ϕ) of the
particle for the parameters k = 1, m = 1, ℓ = 1 and c = 100. (These numbers are not
necessarily realistic, but are chosen to be simple.) Choose as your initial conditions
r(0) = 5 and r′ (0) = 0. Note that it may be easier to solve the transformed radial
equation u′′ = −u − mF (1/u)/ℓ2 u2 where u = 1/r(ϕ). Make a plot of the orbit (plot
y = r sin ϕ versus x = r cos ϕ) for ϕ from 0 to 8π. Now do the same for c = 10 and
c = 5. What can you conclude about the c → ∞ limit of general relativistic orbits?
We numerically solve the transformed radial equation
3kℓ2 4
 
′′ m m 2 km 3k 2
u = −u − 2 2 F (u) = −u + 2 2 ku + 2 2 u = −u + 2 + u
ℓu ℓu mc ℓ mc2
Here is a python script that solves the equation and plots the orbit.

Homework #10 problem 6

Orbits in a Schwarzschild geometry
[1]: # We use numpy and matplotlib
import numpy as np
import matplotlib.pyplot as plt

Simulation parameters
[2]: # These parameters are not necessarily realistic
k = 1 # Gravitational constant
m = 1 # mass of the particle

7
 l = 1 # angular momentum

# We will run the simulation with different speeds of light

# c = 100

# initial radius
r0 = 5

# phi increment (can adjust)

dphi = .01
# total simulation angle
phimax = 4*(2*np.pi) # Simulation for 4 orbits

We define a python procedure orbit() that takes one argument, c the (fake)
speed of light
[3]: def orbit(c=100):

# starting angle
phi = 0
# starting radius
u = 1/r0
up = 0

# initialize the phi and r arrays for the orbital data

phivals = [phi]
rvals = [1/u]

# loop until we reach the maximum number of orbits

# (This implements the Euler-Cromer method)
while phi < phimax:
# here is u''
upp = -u+k*m/l**2+3*k*u**2/m/c**2
# update u', and then u
up = up + upp*dphi
u = u + up*dphi
# increment the angle
phi = phi + dphi
# save the datapoint
phivals.append(phi)
rvals.append(1/u)

# return the data

return phivals, rvals

8
 Run the simulation for c = 100
[4]: phivals, rvals = orbit(100)

# make the x y plot of the orbit

fig, axes = plt.subplots()
axes.set(title='Orbital motion for␣
,→$c=100$',xlabel="$x$",ylabel="$y$")

axes.plot(rvals*np.cos(phivals),rvals*np.sin(phivals))
plt.savefig('p401_hw10_orbit100.pdf')
plt.show()

Orbital motion for c = 100

1.5

1.0

0.5

0.0
y

0.5

1.0

1.5

0 1 2 3 4 5
x

Run the simulation for c = 10

[5]: phivals, rvals = orbit(10)

# make the x y plot of the orbit

fig, axes = plt.subplots()
axes.set(title='Orbital motion for␣
,→$c=10$',xlabel="$x$",ylabel="$y$")

axes.plot(rvals*np.cos(phivals),rvals*np.sin(phivals))
plt.savefig('p401_hw10_orbit10.pdf')

9
 plt.show()

Orbital motion for c = 10

1
y

0 1 2 3 4 5
x

Run the simulation for c = 5

[6]: phivals, rvals = orbit(5)

# make the x y plot of the orbit

fig, axes = plt.subplots()
axes.set(title='Orbital motion for␣
,→$c=5$',xlabel="$x$",ylabel="$y$")

axes.plot(rvals*np.cos(phivals),rvals*np.sin(phivals))
plt.savefig('p401_hw10_orbit5.pdf')
plt.show()

10
 Orbital motion for c = 5
5
4
3
2
1
y

0
1
2
3
4 2 0 2 4
x

The orbit for c = 100 is elliptical and almost closed, with only a very small amount
of precession. On the other hand, the precession of the orbit is more pronounced
for c = 10. When c = 5 the precession is very large. Of course, the actual speed of
light is fixed, so changing c is a bit fake. But what this demonstrates is that as c
gets larger, the general relativistic effect of precession gets smaller. In the c → ∞
limit, the general relativity effect disappears, and we are left with a conventional
Kepler orbit.

11