Classical Mechanics Spring 2021
Daniel K. Mark
August 16, 2021
Classical Mechanics 1: A relativistic pendulum
[Note: In this problem we do not assume that angles are small]
To take a break from their usual work, a tenacious phD student has decided they need to calculate relativistic corrections
to the motion of a simple pendulum, and verify they are indeed negligible. The pendulum is consistituted by a bob of mass
m at the end of an inextensible wire of length l and zero mass. The other end of the wire is fixed to a fulcrum. Unless
otherwise states, all quantities are defined in a frame where the fulcrum of the pendulum is at rest (“Lab” frame).
A uniform gravitational field exists, characterized by a magnitude g. In order to keep mathematics simpler, we assume
that gravity can be treated as a force proportional to the “relativistic mass” of the particle and we take the force on the
particle to be:
mg
|F | = q , (1)
2
1 − vc2
where v is the norm of the velocity vector.
a) [0.5 pt] Write an expression for the differential of the particle’s potential — dV — as its position varies by dθ. Use
polar coordinates and set the origin at the fulcrum (See Fig. 1).
dV = γmgl sin θdθ = −γmgl d(cos θ) (2)
b) [1 pt] Remembering that the Lagrangian of a free relativistic particle of mass m and velocity ~v is
r
2 v2
L(v) = −mc 1 − 2 , (3)
c
write down the Lagrangian and the Hamiltonian of the relativistic pendulum.
[Note: for the potential, you might use the appropriate integral of what you found in part a), without solving it]
θ cos θ
mc2 mc2
Z Z
L(θ, v) = − − mgl γ(θ0 ) sin θ0 dθ0 = − + mgl γ(cos θ0 )d(cos θ0 ) (4)
γ θ0 γ cos θ0
To find the Hamiltonian,
∂L
p= = γmv (5)
∂v
Z cos θ
H = pv − L = γmc2 + V = γmc2 − mgl γ(cos θ0 )d(cos θ0 ) (6)
cos θ0
Note: In Eq. (5), we do NOT differentiate the γ in the integral expression of V . The justification is not quite clear to
me, but the alternative is a road filled with pain.
c)[1 pt] Assume the bob is released from an initial angle θ0 with zero velocity. Write an expression for the gamma factor
of the bob at a generic angle θ as function of the potential, the value of the potential at θ0 , and constants.
Using the fact that Ḣ = 0 (conservation of energy),
γmc2 + V (θ) = mc2 + V (θ0 ) (7)
d) [2 pt] Write an expression for the potential that does not contain integrals or derivatives. Your answer must have the
form:
V (θ) = h(θ) + K (8)
where h(θ) and K are a function and constant, respectively.
1
From Eq. (7), mc2 γ(θ1 ) + V (θ1 ) = mc2 γ(θ2 ) + V (θ2 ) for any θ1 , θ2 . Therefore mc2 γ 0 (θ) = −V 0 (θ), or more conveniently:
dγ dV
mc2 =− = mglγ(cos θ) (9)
d cos θ dcos θ
gl
⇒ γ(cos θ) = exp 2 (cos θ − cos θ0 ) (10)
c
Using Eq. (7) again,
2 gl
V (θ) = V (θ0 ) − mc exp 2 (cos θ − cos θ0 ) − 1 (11)
c
e) [1 pt] Write the velocity of the bob at position θ. You answer must not contain the potential.
From Eq. 10,
gl
γ(cos θ) = exp 2 (cos θ − cos θ0 ) (12)
c
s
gl
⇒ v(θ) = c 1 − exp 2 2 (cos θ0 − cos θ) (13)
c
f ) [1.5 pt] Write the equation of motion for θ. You answer must not contain integrals or derivatives.
[Note: You do not need to solve the equation]
Our answer does contain one derivative, θ̇. I will assume this is acceptable.
s
gl
v(θ) = lθ̇ = c 1 − exp 2 2 (cos θ0 − cos θ) (14)
c
Note: I am not sure if this is the intended answer: it is extremely short relative to the allocated points.
g) [1 pt] Using only Newtonian mechanics, find an integral expression for the Newtonian period of the pendulum.
Using conservation of energy:
1 2 2
ml θ̇ = mgl(cos θ − cos θ0 ) (15)
2
s Z Z
l dθ
∓ √ = dt (16)
2g cos θ − cos θ0
s Z
θ0
l dθ
⇒T =4 √ (17)
2g 0 cos θ − cos θ0
The sign depends on the quadrant of the oscillation, just choose it such that T > 0.
h) [2 pt] Calculate the period of the relativistic pendulum and show that you recover the Newtonian result in the appropriate
limit. Explicitly write down the leading order relativistic correction. Is the relativistic period smaller or larger than the
Newtonian one?
[Note: Your answer will contain an integral, you do not need to solve it]
Analogously, from Eq. (14):
θ0
c θ0
Z Z
c dθ dθ
T =4
l
r h i ≈ 4l r 2 (18)
0 0
1 − exp 2 cgl2 (cos θ0 − cos θ) 2 cgl2 (cos θ − cos θ0 ) − 21 2 cgl2 (cos θ − cos θ0 )
s Z
θ0
l 1 gl p
≈4 dθ √ + 2 cos θ − cos θ0 (19)
2g 0 cos θ − cos θ0 2c
The second term in Eq. (19) is always positive, therefore the relativistic period will be longer than the Newtonian one (as
expected).
Note: After preparation of this note I found the reference which this question was based on [1], with consistent answers.
Note 2: The most important steps in this problem are Eqs. (7) and (10), in particular turning (7) into a differential
equation to simplify the integral expression for V .
2
Classical Mechanics 2: The precession of the perihelion
A particle moves in a region described by the central potential
k
V (r) = − e−r/a (20)
r
with k > 0 and a > 0. We will work in the center-of-mass frame, and call µ the reduced mass of the system.
a) [0.5 pt] Write the Lagrangian of the system. You can assume planar motion and call φ your angular variable;
1 2
L= µṙ + r2 φ̇2 − V (r) (21)
2
b) [1 pt] Write the equations of motion;
d ∂L ∂L L2
= ur̈ = = −V 0 (r) + µrθ̇2 = −V 0 (r) + 3 (22)
dt ∂ ṙ ∂r µr
d ∂L d ∂L
= (µr2 φ̇) = = 0 ⇒ µr2 φ̇ = L (23)
dt ∂ φ̇ dt ∂φ
c) [1 pt] Find an equation for r of the form
2
dr
= τ [E − Veff (r)] (24)
dt
Where E is the total energy of the system, Veff (r) an effective potential and τ a real constant that you must express in
terms of known variables;
From conservation of energy:
1 2 L2
µṙ + + V (r) = E (25)
2 2µr2
2
L2
dr 2
= E− − V (r) (26)
dt µ 2µr2
So τ = 2/µ and Veff (r) = V (r) + L2 /(2µr2 ).
d) [2.5 pt] Show that there exist a maximum angular momentum `max above which no bound orbits are possible. Give
an expression for `max in terms of µ, k, a and numerical constants. Your answer must be explicit and must not contain
exponential functions.
Note: I originally provided an incorrect solution, assuming that an orbit with E > 0 is equivalent to an unbound orbit.
This is incorrect for the Yukawa potential, see e.g. the Fall 2002 Mechanics exam [2].
From Eq. (26), ṙ2 = µ2 (E − Veff (r)). A bound orbit exists in minima of Veff (see illustration below). Therefore, no bound
0
orbits can exist if Veff has no local minima. That is, Veff (r) 6= 0 everywhere.
L2
0 k −r/a 1 1
Veff (r) = − 3 + e + (27)
µr r r a
0
This has no solution Veff (r) = 0 if:
L2
k −r/a 1 1
> e + everywhere, i.e. (28)
µr3 r r a
hr r −r/a i
L2 > µka 1+ e everywhere (29)
a a
L2 0 0 L2
In the first line, we do not need to consider the other case µr 3 < −Veff (r) because Veff (r) decays exponentially, while µr 3
2
L 0
decays as a power law, so it is impossible to have µr 3 < −Veff (r)
everywhere.
r r
e−r/a . This quantity is
2
To satisfy Eq. (29), L /(µka) has to be greater than the largest possible value of 1+
√ a a
maximized at r/a = (1 + 5)/2:
d hr r −r/a i
0= 1+ e (30)
dr a a
−r/a r r 2
=e 1+ − (31)
a a
√
r 1+ 5
⇒ = , (32)
a 2
3
√ √
and this quantity has maximum value (2 + 5) exp −(1 + 5)/2 ≈ 4.2 × e−1.6 ≈ 4.2 × 0.2 = 0.84, where we have used
the “potentially useful data” below. Therefore,
√ p p
`max ≈ 0.84 µka ≈ 0.92 µka (33)
e) [2 pt] Find a differential equation for u(φ), where u ≡ 1/r;
From Eq. (26),
2 2 2
L2
dr L dr 2
= = E − − V (r) (34)
dt µr2 dφ µ 2µr2
2
L2 du L2 2
= E− u − V (u) (35)
2µ dφ 2µ
dr dr/dt dφ
where I have used dφ = dφ/dt , dt = µrL2 , and dφ
du
= − r12 dφ
dr
.
Consider now a situation where ` < `max so that a stable circular orbit is possible at r = r0 . We will focus on an orbit
which is very close to this one, i.e. r(t) = r0 + δr(t) with |δr(t)| r0 .
f ) [3 pt] Assuming r0 a, find an approximate expression for the advance of the perihelion in each revolution. Your
solution must be of the form
A B
∆θ = 2 + + Cr0 + Dr02 (36)
r0 r0
Where A and B, C and D are real constants you must calculate (some or all of them might be zero) [Note: your answer
cannot depend on k, µ or `]
A crucial observation is the following: The intersection of the Veff (r) curve with E describes the turning points rmin , rmax
of the orbit. Therefore, for a circular orbit, E intersects Veff (r) at its minimum. (This is equivalent to the fact that a
circular orbit minimizes E for a given L, and can also be seen from Eq. (26), demanding that ṙ = 0 and r̈ = 0.)
0.0
0.2
0.4
Generic orbit
Veff
0.6
rmin rmax E
0.8
1.0
1.2 Circular orbit
0.0 0.5 1.0 1.5 2.0 2.5 3.0
r
After much thought, we realize that we are being asked to calculate the frequency (with respect to φ) of small oscillations
about the minimum of Veff . Calling δu = u − u0 ≈ −u20 δr, the angle φ required for δu to make one oscillation is equal to
2π + ∆θ, where ∆θ is the advance of the perihelion.
Expanding Eq. (35) about u0 :
2
L2
d(δu) 0 :0 1 00 2
≈ E − Veff (u0 ) +
Veff (u0 ) δu − Veff (u0 ) (δu) (37)
2µ dφ 2
1 3 0 :0 4 00 2
= δE − 2r0 Veff (r0 ) + r0 Veff (r0 ) (δu) (38)
2
Eq. (38) has solution δu = A cos(Ω(φ − φ0 )), with
µr04 00
Ω2 = V (r0 ) (39)
L2 eff
To see this, differentiate Eq. (38) by φ and divide by 2(δu)0 to obtain our friend (δu)00 = −Ω2 (δu).
4
00
Lastly we compute Veff (r0 ):
L2 k
Veff (r0 ) = 2 − e−r0 /a = E (40)
2µr0 r0
2
0 L k 1 1
Veff (r0 ) = − 3 + e−r0 /a + =0 (41)
µr0 r0 r0 a
" 2 #
2
00 L k 1 1 1
Veff (r) = 3 4 − e−r/a + + 2 (42)
µr r r a r
Using Eq. (41), Eq. (42) becomes:
L2
00 r0 1
Veff = 4 3− 1+
(r0 ) + (43)
µr0 a 1 + r0 /a
2 r0 1
Ω =3− 1+ + (44)
a 1 + r0 /a
r 2
0
≈1− (45)
a
Finally we obtain: !
2π 1 r 2
0
∆θ = − 2π ≈ 2π p −1 ≈π (46)
Ω 1 − (r0 /a)2 a
So A = B = C = 0 and D = π/a2 . Treat yourself to a pie! To verify these answers, scroll down to the last line of Ref. [3].
— Potentially useful data:
√ √ √ √
2 = 1.4, 3 = 1.7, 5 = 2.2, 7 = 2.6 (47)
e−3.4 = 0.03, e−1.6 = 0.2, e−0.8 = 0.45, e−0.1 = 0.90 (48)
This turns out to be important (but not sufficient?) for part d). What I initially thought was a red herring turned out
to be a regular herring.
Note: The reasoning behind this question is likely to measure, through orbits, whether the gravitational potential is
really a Yukawa potential (20) with a large a, by measuring the precession of a perihelion. By dimensional analysis and
the hint, ∆θ must be a function of r0 /a. Since a large a is undetectable, in hindsight A and B must be zero (not that it
would have helped to solve this problem). I cannot think of a simple argument for why C should be 0.
References
[1] Cahit Erkal. The simple pendulum: a relativistic revisit. European Journal of Physics, 21(5):377–384, jul 2000. doi:
10.1088/0143-0807/21/5/302. URL https://doi.org/10.1088/0143-0807/21/5/302.
[2] URL http://physrefs.mit.edu/sites/default/files/documents/P2Fall02.pdf.
[3] URL http://www.physics.usu.edu/Wheeler/ClassicalMechanics/CMYukawaPotential.pdf.
