Question 1: Spherical Pendulum

Consider a two-dimensional pendulum of length l with mass M at its end. It is easiest to p use spherical
coordinates
p centered at the pivot since the magnitude of the position vector is constant: |~
r | = (lr̂) · (lr̂) =
2
l (r̂ · r̂) = l. In other words, the mass is restricted to move along the surface of sphere of radius r = l that
is centered on the pivot. The motion of the pendulum can therefore be described by the polar angle θ, the
azimuthal angle φ, and their rates of change.

(a) The Lagrangian for a spherical pendulum

Let’s assume that the mass is on “bottom half” of the sphere, so that the mass has a Cartesian coordinate
z = −l cos θ. Since gravity is the only external, non-constraint force acting on the mass, with potential
energy U = M gz = −M gl cos θ, the Lagrangian (L) can be first written as:
1
L=T −U = M |~v |2 + M gl cos θ
2
In spherical coordinates, ~v = ṙr̂ + rθ̇θ̂ + rφ̇ sin θφ̂, and for the problem under consideration we note that
ṙ = l˙ = 0. So we can write L as an explicit function of the spherical coordinates,

1
M l2 θ̇2 + l2 φ̇2 sin2 θ + M gl cos θ


L= (1)
2
There are two equations of motion for the spherical pendulum, since Lin Equation 1 is a function of both θ
and φ; we therefore use the Euler-Lagrange equation for both coordinates to obtain them. For θ and φ,

 
d ∂L ∂L
− = 0 =⇒ M l2 θ̈ − M l2 θ̇2 sin θ cos θ + M gl sin θ = 0 (2)
dt ∂ θ̇ ∂θ
 
d ∂L ∂L d
M l2 φ̇ sin2 θ = 0


− = 0 =⇒ (3)
dt ∂ φ̇ ∂φ dt

Equation 3 suggests that the quantity M l2 φ̇ sin2 θ is a constant in time. This quantity has the same form
as the z-component of the angular momentum vector for the mass (Lz ), so Equation 3 states that angular
momentum is conserved in the z direction for our spherical pendulum.

(b) Energy in terms of θ and Lz

For Lagrangians that don’t explicitly depend on time, and for potentials that only depend on the coordinates
(and not explicitly in their rates of change), the total mechanical energy E = T + U is another constant of
the motion. Equation 1 satisfies both of these conditions, so E is constant for the spherical pendulum. We
can eliminate the φ̇ term in T since, according to Equation 3, Lz = M l2 φ̇ sin2 θ is constant in time. So the
expression for E becomes

1
M l2 θ̇2 + l2 φ̇2 sin2 θ − M gl cos θ


E=
2
  2 
1 2 2 2 Lz 2
= M l θ̇ + l sin θ − M gl cos θ
2 M l2 sin2 θ
1 L2z
= M l2 θ̇2 + − M gl cos θ = E (4)
2 M l2 sin2 θ

1
Notice that E was re-written as a function of only one coordinate (θ); one can therefore talk about an
equivalent one-body problem, where E = Teff + Ueff , Teff = M l2 θ̇2 /2, and

L2z
Ueff = − M gl cos θ (5)
M l2 sin2 θ
Given the one-body form of Equation 4, we can technically solve for the solution for θ as a function of time
(or vice versa). Noting that θ̇ = dθ/dt and solving for it in Equation 4, we get
r
dθ 2[E − Ueff (θ)] √ dθ
= =⇒ dt = M l p
dt M l2 2[E − Ueff (θ)]
Z t √ Z θ
dθ
t − t0 = dt = l M p (6)
t0 θ0 2[E − Ueff (θ)]

For the full solution of the spherical pendulum, we also need to find the solution for φ. This can be done
using the integrated form of Equation 3, Lz = M l2 φ̇ sin2 θ, and using the chain rule of advanced calculus to
put the eventual integral in terms of θ. In the form of equations, this gives

Lz dφ
=
M l2 sin2 θ dt
dφ dθ
=
dθ r
dt
dφ 2[E − Ueff (θ)]
=
dθ M l2
and, solving for dφ/dθ, we finally get
Z φ Z θ
dφ Lz 1 Lz dθ
= √ =⇒ φ − φ0 = dφ = √ (7)
l M sin2 θ 2[E − Ueff (θ)] sin2 θ
p p
dθ φ0 l 2M θ0 E − Ueff (θ)

(c) Max and min values of θ

At the maximum and minimum values of θ, the mass has no (effective) kinetic energy (i.e. θ̇ = 0) and so
the total mechanical energy at those points is equal to the effective potential energy:

E = Ueff
L2z
= − mgl cos θ
2M l2 sin2 θ
L2z
= − mgl cos θ (8)
2M l2 (1 − cos2 θ)

We can then find an algebraic (cubic) equation for the maximum and minimum values of cos θ by solving
for it in Equation 8:

L2z
M gl(cos3 θ − cos θ) + E(cos2 θ − 1) + =0 (9)
2M l2

2
 Question 2: Radial Oscillations
In the classic two-body problem, the total mechanical energy for a pair of point-like particles with an
interaction potential V (r) can be re-written to reflect an effective one-body problem when using constants of
the motion (i.e. conservation laws), as well as a suitable reference frame and coordinate system. Therefore,
instead of describing the motion of two particles with masses m1 and m2 undergoing motion dictated by the
potential V (r), one can instead talk about a particle of reduced mass µ = m1 m2 /(m1 + m2 ) moving in an
“effective potential” Veff , which has the form

L2
Veff = V (r) + (10)
2µ2 r2
and r is the radial coordinate of the particle with mass µ measured relative to the center of field. Note
that Equation 10 yields the potential, which is the potential energy per unit (reduced) mass, and that the
potential energy can be computed to be Ueff = µVeff .

For circular orbits, r = r0 = constant in time, and for general central-force problems in Newtonian mechanics
the orbital angular momentum (L) is a conserved quantity1 of the motion: L = µr02 φ̇2 . The latter equation
can be solved to find the period of orbital motion,
Z 2π Z T
dφ L L L
φ̇ = = 2 =⇒ dφ = 2π = 2 dt = 2 T
dt µr0 0 0 µr0 µr0
and finally can be put in terms of r0 and L,

µr02
T = 2π (11)
L
(a) Newtonian potential for inverse-square-law force
Consider the case when the interaction potential V (r) = −V0 /r. The effective potential can be explicitly
written as a function of r,

V0 L2
Veff = − + 2 2. (12)
r 2µ r
A circular orbit corresponds to a system with total mechanical energy E that is equal to Ueff at all times. In
other words, a circular orbit has a value of r0 that corresponds to the minimum value of Ueff (or, equivalently,
Veff ), which can be found to be

L2 L2

dVeff  V0
= − = 0 =⇒ r0 = (13)
dr r=r0 r02 µ2 r03 µ2 V0
We can find the orbital period for this potential by combining the general formula for T (Equation 11) with
Equation 13 to eliminate L; this is most easily done by finding T 2 :

µ2 r04 µ2 r04 r3
T 2 = 4π 2 2
= 4π 2 2
= 4π 2 0 (14)
L (r0 µ V0 ) V0

If V0 = GM , where G is Newton’s gravitational constant and M = m1 + m2 is the total (gravitational) mass

of the system, then Equation 14 is Kepler’s third law of planetary motion.

1 These equations are written for an orbit with its plane embedded in the x − y Cartesian plane, so that θ = π/2 is constant

in time.

3
For small perturbations in r relative to r0 , we can express Veff as a Taylor expansion in r:

1 d2 Veff 
 
dVeff 
Veff (r) ≈ Veff (r0 ) + (r − r0 ) + (r − r0 )2 + . . . (15)
dr r=r0 2 dr2 r=r0
where the first derivative of Veff is equal to 0 by definition. We are therefore left with a constant term
and a quadratic term in r for the Taylor-expanded Veff . This has an equivalent form to the potential for a
simple harmonic oscillator in one dimension, since the constant term will vanish when we compute the Euler-
Lagrange equation of motion for r. We can therefore find the frequency of oscillations in r by computing
the second derivative in Veff ,

d2 Veff V0 L2 V0 (r0 µ2 V0 ) V0 V0
= −2 3 + 3 2 4 = −2 3 + 3 = −2 3 + 3 3
dr 2 r0 µ r0 r0 µ2 r04 r0 r0
V0
= 3 (16)
r0

where the frequency (ω) can be finally found to be:

s
V0
ω= (17)
r03
We can compare this to the frequency of orbital motion (Ω) by noting that Ω = 2π/T , and from Equation
14 this can be found to be:
s
V0
Ω= =ω (18)
r03

So, in short, the orbital and oscillation frequencies are equal for the case where the interaction potential
corresponds to an inverse-square-law force.

(b) Two-dimensional harmonic potential

Now consider the case when the interaction potential V (r) = kr2 /2, where k is positive constant. The
effective potential can be explicitly written as a function of r,

1 2 L2
Veff = kr + 2 2 . (19)
2 2µ r
We can now apply the same procedures as done for part (a) of this problem to find the various quantities of
interest. For starters, we can compute the radius of a circular orbit for this effective potential by setting the
first derivative of Equation 19 to 0:
 2 1/4
L2

dVeff  L
= kr 0 − 2 3 = 0 =⇒ r0 = (20)
dr r=r0
 µ r0 µ2 k
We can combine the above result with the general relation for the orbital period T and orbital frequency Ω
to find them in terms of r0 and k:

4π 2 µ2 r04 4π 2 µ2 r04 4π 2 2π √
T2 = 2
= 2 4 = =⇒ Ω = = k (21)
L kµ r0 k T
Using Equation 15, we can find the solution for small oscillations in r about r0 by computing the second
derivative of Veff with respect to r:

4
 d2 Veff  L2 kµ2 r04

= k + 3 = k + 3 = 4k (22)
dr2 r=r0 µ2 r04 µ2 r04
and so the frequency of oscillations is found to be:
√
ω = 2 k = 2Ω (23)