Central-force motion

Topics

1. Central-force motion: central force from potential, 2 7→ 1, conservation of angular

momentum, effective potential energy

2. Gravitation: equation of orbits, ellipses

3. Kepler’s laws: elliptical orbits, equal-time-equal-area, period-squared-length-cubed

Goals

1. Apply the Lagrange formulation to study the central-force motion, an important problem
in classical mechanics

2. Understand how the various types of orbits, bounded or unbounded, arise in the gravita-
tional problem of two masses

3. Explain Kepler’s empirical laws concerning the motion of planets in the solar system.

Remarks

1. Finally, the last chapter!

1
1. GENERAL CENTRAL-FORCE MOTION

As the very last topic of our course (yay!), let us study the “central-force” problem and under-
stand how that relates to the planetary or even stellar motion. This problem has, quite obviously,
played a crucial role in spurring the development of physics over the past millennia. After all,
physics is about understanding how things work, and nothing is nearly as amazing as realizing the
Earth is a spaceship cruising in a nearly empty space! In the following, we study the central-force
problem using the Lagrangian formalism.
Let us consider two objects, with masses m1 and m2 , which are located respectively at ~r1
and ~r2 . In this context, we model each of the objects as a point mass, and so the only degrees
of freedom involved are ~r1 and ~r2 (the story is different if, e.g., we also consider the individual
rotation of the objects etc.). For simplicity, let us restrict the motion of the masses to a two-
dimensional plane. The motion in three-dimensional space is essentially the same since we can
always describe our problem in the plane orthogonal to the (conserved) angular momentum of the
system [1]. We suppose the masses interact with each other through a conservative force described
by the potential energy

U = U (|~r1 − ~r2 |), (1)

i.e., it depends only on the magnitude of the separation between the two masses.
# central force from potential

We will first show that the potential energy U defined above lead to a central force. Let
r = |~r1 − ~r2 |, then in Cartesian coordinates one has the explicit form

p
r= (x1 − x2 )2 + (y1 − y2 )2 . (2)

From Chapter 3, we know the force acting on either of the masses is given by the corresponding
gradient of the potential energy. For instance, the force acting on the first mass is
   
∂U ∂U ∂U ∂r ∂r
F~1 = −∇
~ ~r1 U (r) = − êx + êy =− êx + êy . (3)
∂x1 ∂y1 ∂r ∂x1 ∂y1

The derivatives can be evaluated explicitly as

∂r (x1 − x2 ) ∂r (y1 − y2 )
=p ; =p , (4)
∂x1 (x1 − x2 )2 + (y1 − y2 )2 ∂y1 (x1 − x2 )2 + (y1 − y2 )2

2
which gives
∂r ∂r (x1 − x2 )êx + (x1 − x2 )êy ~r1 − ~r2
êx + êy = p = , (5)
∂x1 ∂y1 2
(x1 − x2 ) + (y1 − y2 ) 2 |~r1 − ~r2 |

the unit vector pointing from mass 2 to mass 1. Let us denote this unit vector by r̂, then we have

∂U ∂U
F~1 = − r̂; F~2 = −F~1 = r̂. (6)
∂r ∂r
Since the direction of the force points along the line joining masses 1 and 2, we call this a “central
force”.
# reduction to a single-particle problem

To analyze the motion of the masses, we consider the Lagrangian

1 1
L = m1 |~r˙1 |2 + m2 |~r˙2 |2 − U (r). (7)
2 2
However, as r depends implicitly on ~r1 and ~r2 , and we know that the center of mass momentum
~ CM
is conserved for this isolated two-mass system, it is natural to rewrite the Lagrangian using R
and ~r = rr̂ as the generalized coordinates. Recall

~ CM = m1~r1 + m2~r2 ,
R (8)
m1 + m2
and so one can write
m1~r1 + m2~r2 ~r1 − ~r2 ~ CM + m2
~r1 = + m2 =R ~r;
m1 + m2 m1 + m2 m1 + m2
(9)
m1~r1 + m2~r2 ~r1 − ~r2 ~ CM − m1
~r2 = − m1 =R ~r.
m1 + m2 m1 + m2 m1 + m2
So, we may rewrite the kinetic energy as
2
 
1 ˙~ ˙~ 2m2 ~˙ m
T = m1 RCM · RCM + RCM · ~r˙ + 2
~r˙ · ~r˙
2 m1 + m2 (m1 + m2 )2
2
 
1 ˙~ ˙~ 2m1 ~˙ m
+ m2 RCM · RCM − RCM · ~r˙ + 1
~r˙ · ~r˙ (10)
2 m1 + m2 (m1 + m2 )2
1  ~˙ 2 1 m1 m2 ˙ 2
 
= (m1 + m2 ) RCM  + |~r| .
2 2 m1 + m2
In other words, the kinetic energy can be separated into two pieces, as if there is a particle of the
mass M = m1 + m2 sitting at R ~ CM , and another independent particle of mass m1 m2 sitting at ~r
m1 +m2

(although ~r is physically the difference vector pointing from mass 2 to 1). Let us define
 −1
m1 m2 1 1
µ= = + , (11)
m1 + m2 m1 m2

3
and we call this the reduced mass (note: in the primer we referred to the same expression as “a
kind of effective mass.” The terminology reduced mass is more cryptic but at the same time more
standard.).
In terms of these variables, we can rewrite the Lagrangian as

1  ~˙ 2 1 ˙ 2
L = M R r| − U (r).
CM  + µ|~ (12)
2 2

As we have anticipated, the center-of-mass coordinates do not appear in the Lagrangian, and so
their conjugate momenta

∂L ∂L
Px = = M ẊCM ; Py = = M ẎCM (13)
∂ ẊCM ∂ ẎCM

are independently conserved. Since R ~ CM can only move in an inertial manner, let us pick the
~˙ CM = ~0. This
center-of-mass frame as the inertial frame and so, in this frame, we simply have R
leaves us with

1
LCM = µ|~r˙ |2 − U (r), (14)
2

which is, formally, the Lagrangian of a single particle of mass µ, position vector ~r, and experi-
encing a potential energy given by U (r)! This shows that having two masses is really the same as
having one, and so an isolated two-body problem is fundamentally the same as a single-particle
system.

Figure 1. Two masses m1 and m2 undergoing central force motion is mathematically equivalent to a single
mass µ moving in the central force potential U (r) around the center-of-mass. In the limit when m2  m1 ,
the center of mass of the system practically coincides with the position of m2 and the reduced mass µ → m1 .

4
# angular momentum conservation in central-force problems

As the potential energy depends on r = |~r|, it will be natural for us to adopt polar coordinates
represent ~r by (r, θ). Rewriting the kinetic energy in terms of these coordinates, we have

# Lagrangian for the relative motion in the center-of-mass frame

1  2 2 2

LCM = µ ṙ + r θ̇ − U (r), (15)
2

and now we can crank through the same procedures for deriving equations of motion. First, com-
pute the conjugate momenta

∂LCM ∂LCM
pr = = µṙ; pθ = = µr2 θ̇. (16)
∂ ṙ ∂ θ̇
As ∂LCM /∂θ = 0, pθ is a conserved momentum and we denote its value by

# conservation of angular momentum in central-force problems

` = µr2 θ̇. (17)

# radial equation of motion

At this step, we can choose to proceed with the Lagrangian formulation and compute the Euler-
Lagrange equations, or to pass to the Hamiltonian formulation and obtain the canonical equations
of motion. Of course, the two formulation are equivalent, and for this problem there is little
difference when we solve for the analytical solution using either approaches. In the following, let
us find the radial equation of motion from the Euler-Lagrange equation:
d ∂L
pr − =0
dt ∂r
∂U
µr̈ − µrθ̇2 + =0 (18)
∂r
`2 ∂U
µr̈ − 3 + =0.
µr ∂r
Compared to the expectation from the Newton’s law, we can identify

`2 ∂U
Fr = 3
− (19)
µr ∂r

5
as an effective force acting on the reduced mass µ in the radial direction.
As is typical for the equation of motion obtained either from Newton’s second law or the Euler-
Lagrange equation, Eq. (18) is second order in r. One can, however, use the conservation of energy
to obtain an “integrated” version of the equation of motion which involves only first derivative [2]:
 2 
1  2 2 2
 1 2 `
E = µ ṙ + r θ̇ + U (r) = µṙ + + U (r) . (20)
2 2 2µr2

It is natural to define an effective potential energy

# effective potential energy for the corresponding one-dimensional problem

`2
V` (r) = + U (r), (21)
2µr2

such that the total energy, E = 12 µṙ2 + V` (r), can be interpreted as the sum of the kinetic energy
of a particle moving in one dimension and the potential energy V` (r). Notice also that Eq. (19)
can be understood simply as Fr = −∂r V` (r).
Warning: this notion of effective potential energy could be confusing. Note that, in reality, the
only potential energy is U (r) and the term proportional to `2 is ultimately a contribution to the
kinetic energy. Consistently, the value of V` (r) depends on the angular momentum `, which is
conserved and could be treated as a parameter specifying the system. However, if we follow the
derivation above and forget about the physical system, then it appears like we have a single point
mass moving in one dimension and subjected to the force corresponding to the effective potential
energy V` (r).
A new feature of the present problem, however, is that the angular momentum ` enters as a
parameter in the effective potential energy V` . In any case, since E is also a conserved quantity,
one can obtain a first-order differential equation
r
2
ṙ = (E − V` (r)), E ≥ V` (r) (22)
µ

which has the formal solution

Z r1
dr
q = t1 − t0 , (23)
2
r0
µ
(E − V` (r))

where we let r0 = r(t0 ) and r1 = r(t1 ).

6
# orbits

When a concrete form of the central force potential U (r) is given, one could attempt to solve
for r(t) (subjected to a given value of `). Once r(t) is known (either analytically or numerically),
one can solve for θ(t) by integrating θ̇ = `/(µr2 ). In the study of planetary motion, however, one
is often more interested in the shape of the orbit rather than the solutions r(t) and θ(t). To this
end, we note that we can also think of r as a function of angle θ, such that

dr dr
ṙ = = θ̇. (24)
dt dθ

As θ̇ can be expressed in terms of r through Eq. (17), we may rewrite Eq. (22) as
r
dr ` dr 2
θ̇ = 2 = (E − V` (r))
dθ µr dθ µ
Z Z (25)
d(1/r)
⇒ p = − dθ,
2µ (E − V` (r)) /`2

which, when solved, provides us with the trajectory r(θ).

7
2. GRAVITATION

Let us now specialize the general solutions above to the problem of gravitation, i.e., we take
the central-force potential to be

Gm1 m2 GM µ
U (r) = − =− , (26)
r r

where we have set the reference energy to be U (r → ∞) = 0. Curiously, as the second equality
shows, the potential energy could equivalently be interpreted by that between the total mass M
(sitting at the origin) and the reduced mass µ (sitting at ~r). Correspondingly, the effective potential
energy is now

# effective potential energy for two masses interacting through gravity

`2 GM µ
V` (r) = 2
− . (27)
2µr r

A sketch of the generic shape of V` (r) is shown in Fig. 2. It attains a global minimum at

`2
rc = . (28)
GM µ2

When E = V` (rc ), the system has the minimum possible energy and it can only perform circular
motion (c.f., Eq. (22)). That’s why we label the value as rc with “c” stands for “circular.” This
minimum energy is given by

`2 GM µ µ
Ec = 2
− = − 2 (GM µ)2 . (29)
2µrc rc 2`

Since r = rc is a constant in this case, we also have θ̇ = `/µrc2 to be a constant. If we denote this
constant by ωc = θ̇, we have

µ2 rc4 ωc2
rc = ⇒ GM = rc3 ωc2 , (30)
GM µ2

which one may recognize as the more familiar relation between the radius and the angular velocity
when one consider a circular orbit of e.g., a satellite around the Earth.
Another important feature in the effective potential energy here is that V` (r) → +∞ as r → 0+
but V` (r) → 0 as r → +∞. Here, the 0 energy reference is taken with respect to the gravitational

8
Figure 2. Effective potential V` (r) for gravitation, assuming ` > 0. The purple line indicates closed/
bounded orbits with total energy E < 0.

potential energy when the two masses are infinitely far apart. So long as the system has E < 0, the
value of r will be bounded by the two solutions of E = V` (r), and so r remains finite at all time.
We call these are closed/ bounded orbits. Alternatively, if E > 0 then r can grow unbounded, and
we call these open/ unbounded orbit. The case of E = 0 represents the borderline case, and it
could be characterized by the distance

`2 1
r∗ = = rc . (31)
2GM µ2 2

For any trajectory with angular momentum `, we may consider the minimum distance rmin attained
between the two masses. The orbits/ trajectories are (i) closed/ bounded if rmin > r∗ , (ii) borderline
if rmin = r∗ , and (iii) open/ unbounded if rmin < r∗ .
In fact, one can see graphically from Fig. 2 that the minimum distance rmin encodes the energy
of the system: draw any horizontal line, which we interpret as the energy E of the system. If the
line is below the curve, then E < V` (r) for all r and that is impossible (as E = 12 µṙ2 + V` (r) ≥
V` (r)). If E = V` (rc ), then rmin = rc . For V` (rc ) < E < 0, the horizontal line intersects the
curve at exactly two values of r, which we may label as rmin and rmax . Finally, for E ≥ 0, the line
intersect the curve at exactly one value which we interpret as rmin . Visually, we can see that so
long as E ≥ V` (rc ), E is in one-to-one correspondence with the value of rmin . More quantitatively,

9
rmin can be determined by solving
`2 GM µ
E = V` (rmin ) = 2
− , (32)
2µrmin rmin
which can be understood as a quadratic equation in the variable 1/rmin . Its solution is
p
1 GM µ + (GM µ)2 + 2`2 E/µ
= , (33)
rmin `2 /µ
where we have taken the root with the positive sign since 1/rmin > 1/rmax in general. We can also
rearrange it to find
p s !
2
1 GM µ + (GM µ)2 + 2`2 E/µ GM µ 2`2 E
= = 1+ 1+ . (34)
rmin `2 /µ `2 (GM µ)2 µ

That’s alright, but ugly. By this point of the course, you might have acquired a taste of how to do
“prettier” physics: we can instead try to measure the quantities based on some inherent scales in
the problem. The natural scales here would be the radius of the circular orbit, rc from Eq. (28),
and its corresponding energy Ec from Eq. (29). Using them, we can rewrite
r ! r !−1
1 1 E E
= 1+ 1− ⇒ rmin = rc 1 + 1 − . (35)
rmin rc Ec Ec

The qualitative description of the nature of the orbits above can be analyzed more quantitatively
by carrying out the integration in Eq. (25). Using the concrete form of V` (r) for the gravitational
problem, and using a substitution of variable u = 1/r, we can evaluate the integral
Z Z
du
r   = − dθ, (36)
2µ `2 2
`2
E − 2µ u + GM µ u

where suitable integration bounds are assumed on both sides. This is the “standard” way to solve
this problem.
But integrations are hard! Instead, let us introduce a trick and transform the problem to an
easier one. Consider the radial equation of motion in Eq. (18), which contains r̈. So, we first use
the same change of variable as above and evaluate
dr ` dr ` du
ṙ =θ̇ = 2 =− , (37)
dθ µr dθ µ dθ
and with that the second derivative can be evaluated as
` d2 u `2 d2 u `2 2 d2 u
 
d ` du
r̈ = − =− θ̇ = − = − u . (38)
dt µ dθ µ dθ2 µ2 r2 dθ2 µ2 dθ2

10
We can now rewrite Eq. (18) as

`2 2 d2 u `2 3
− u − u + GM µ u2 =0
µ dθ2 µ
2
 2  (39)
` 2 du 1
− u +u− =0,
µ dθ2 rc

where we have used rc−1 = GM µ2 /`2 from Eq. (28). So long as u does not vanish identically (i.e.,
unless we keep the two masses infinitely far apart), the variable u = u(θ) satisfies

d2 u 1
2
+u= . (40)
dθ rc

Hopefully this equation looks familiar: when the dust settles we are down to a linear, second-order,
inhomogeneous ordinary differential equation!
Let us solve this, one last time, following the steps outlined in Chapter 2:

1. Find any solution to the inhomogeneous equation: here, we simply take the constant

1
uP (θ) = (41)
rc

which one can verify is a solution.

2. Find the general solution to the homogeneous counterpart: the homogeneous version of the
equation is

d2 u
+ u = 0, (42)
dθ2

which is hopefully familiar: it is the same differential equation we first solved in this course
(c.f. Eq. (6) in the primer)! Let’s just write down the general solution

u(θ) = α cos(θ + θ0 ), (43)

where α and θ0 are the two undetermined coefficients which are fixed by the initial condi-
tions (one could also parameterize the general solution as α cos θ + β sin θ, which is equiv-
alent up to some trigonometric identities and a relabeling of the undetermined coefficients).

3. Fix the undetermined coefficients by the initial conditions: our overall solution reads

1
u(θ) = α cos(θ + θ0 ) + . (44)
rc

11
 Since it is up to us to orientate the coordinate system, let us pick θ0 such that the distance
1
r attains a minimum when θ = 0, i.e., we set θ0 = 0 and demand u(θ = 0) = rmin
. This
corresponds to
1 1 1 1
=α+ ⇒ α= − . (45)
rmin rc rmin rc
Note that we always have rmin ≤ rc and so α ≥ 0.

All-in-all, we find the solution

 
1 1 1 1
− = − cos θ, (46)
r(θ) rc rmin rc
which can be rearranged into
rc
r(θ) = . (47)
1 + (rc /rmin − 1) cos θ
Let us define
rc
= − 1, (48)
rmin
and with that we can write

# equation for the trajectory/ orbit

rc
r(θ) = . (49)
1 +  cos θ

Notice that the value of  depends on the relative size of rc vs rmin . From Eq. (28), we can
think of rc as encoding the angular momentum ` of the trajectory, whereas from Eq. (35) and its
surrounding discussion we can think of rmin as representing the total energy E of the trajectory.
To complete the analysis, we consider how our solution r(θ) changes as we sweep the energy
E (i.e., vary rmin ). The four cases labeled by (a) to (d) below are summarized pictorially in Fig. 3.

(a) E = Ec : circular (closed/ bounded)

Starting with the minimum possible value of E = Ec , we have rmin = rc . In this case, we
see from Eq. (48) that  = 0, and so

r(θ) = rc for  = 0, (50)

which is independent of θ, as one would expect for a circular orbit.

12
Figure 3. Types of orbits in an ideal two-body gravitational problem. Shown for the same angular momen-
tum ` and so orbits which get closer to M correspond to higher total energy E. Geometrically, the curves
can be identified as (a) a circle, (b) an ellipse, (c) a parabola, and (d) a hyperbola. This family of curves
are mathematically called the “conic sections,” for they are the curves one obtains by sectioning a cone at
different angles.

(b) Ec < E < 0: elliptical (closed/ bounded) We will spend more time on this case, for this
is the most relevant case for planetary motion around the Sun (as well as satellite motion
around the Earth etc).

As we increase the energy, rmin decreases (Fig. 2). Our energy conditions imply

rc
Ec <E ⇒ rmin < rc ⇒  = − 1 > 0; &
rmin
rc rc (51)
E <0 ⇒ rmin > r∗ = ⇒ = −1<1
2 rmin

13
i.e., we have

0 <  < 1. (52)

Given a fixed energy E (and hence ), the minimum and maximum values of r(θ) is attained
respectively at θ = 0 and π, for which
rc rc
rmin = ; & rmax = , for 0 <  < 1. (53)
1+ 1−
Notice rmax is finite as we have  < 1. As such, when θ goes from 0 to π one finds r to go
smoothly from rmin to rmax , and it goes back to rmin when we further increase θ to 2π. This
is, intuitively, a deformed circle. Indeed, Eq. (49) is the defining relation for an ellipse in
polar coordinates.  is known as the eccentricity of the ellipse.

# geometry of an ellipse

Let us show that Eq. (49) defines an ellipse. Geometrically, an ellipse can be constructed
as follows: first, take an inextensible string and pin its end points. Using a pen, straighten
out the string such that it is kept taut. The shape obtained by moving the pen around while
keeping the string taut is defined as an ellipse (Fig. 4). If the two end points of the string
coincide, then what we described would simply give a circle with its radius being half the
length of the string, and the fixed end points sit at the center of the circle. In this sense, an
ellipse is a natural generalization of a circle when we allow the two end points to be pinned
at different locations. We call the two end points of the string the “foci.” Focusing on the
triangle f M µ in Fig. 4, the geometric condition we imposed is

f µ + µM = constant, (54)

where AB denotes the length of the line segment connecting A and B.

Now, let’s go back to the gravitational problem. We will show that the orbit described by
Eq. (49) with 0 <  < 1 can be understood as an ellipse with the total mass M sitting at one
of the foci and the reduced mass µ moving around it. By considering the length of the string
(purple line in Fig. 4) at θ = 0, we see that

f µ + µM = rmin + rmax . (55)

Similarly, one also sees that the distance between the two foci is

f M = rmin − rmax . (56)

14
For convenience, let us define

r± = rmin ± rmax . (57)

Then, our geometrical constraint can be expressed in terms of r± , µM = r, and θ as

q
2 (58)
r2 + r− − 2rr− cos(π − θ) + r = r+ ,

where we have used the “law of cosines” to write down f µ. Since cos(π − θ) = − cos θ, we
obtain

r 2 + r−
2
+ 2rr− cos θ =(r+ − r)2
2 2
r− + 2rr− cos θ =r+ − 2rr+
2 2 (59)
2r(r+ + r− cos θ) =r+ − r−
2 2
r+ − r−
r= .
2(r+ + r− cos θ)
Re-expressing using r± = rmax ± rmin , we can simplify the expression by noticing

1 2 2 1
(r+ − r− ) = (r+ + r− )(r+ − r− ) = 2rmax rmin , (60)
2 2
and so Eq. (59) becomes
2rmax rmin
r=
((rmax + rmin ) + (rmax − rmin ) cos θ)
 −1
rmax + rmin rmax − rmin
= + cos θ (61)
2rmax rmin 2rmax rmin
     −1
1 1 1 1 1 1
= + + − cos θ .
2 rmin rmax 2 rmin rmax
−1 −1
We massage it into this form since the sum and differences between rmin and rmax are simple
in view of Eq. (53):

1 1 1+ 1−
± = ± , (62)
rmin rmax rc rc

which equal to 2/rc and 2/rc respectively. So, we finally find

 −1
1  rc
r= + cos θ = , (63)
rc rc 1 +  cos θ
which connects our (geometrical) description of an ellipse to the orbit trajectory found in
Eq. (49).

15
Figure 4. In the context of a bounded/ closed orbit in gravitation, we can think of the total mass M as sitting
on one of the foci of the ellipse, chosen as the origin, and the reduced mass µ is moving around it along an
ellipse. Physically, nothing sits at the other focus (which we label by f ). The shortest and longest distances
between M and µ are respectively denoted by rmin and rmax . We have chosen the (polar) coordinate system
such that r(θ = 0) = rmin .

Note that we could have absorbed the case of circular orbit ( = 0) into the elliptical orbit
as a special case. The closed/ bounded orbits with 0 ≤  < 1 describe the orbital motion of
bounded objects, like planets, in our solar system.

(c) E = 0: parabolic (borderline open/ unbounded)

When E = 0, we have rmin = r∗ = rc /2, which gives  = 1. This limit could be approached
from an ellipse by sending rmax = 1/(1 − ) → ∞. In terms of Eq. (49), we see that µ
goes around M as we increase θ in the range −π < θ < π. Geometrically, this defines a
parabola, but in the interest of time we would not dwell into the details of that.

(d) E > 0: hyperbolic (open/ unbounded)

Going beyond the “borderline” case, we get unbounded orbits (which is a kind of a mis-
nomer, as we don’t get an “orbit” when the object flies away and does not return. Since
rmin < r∗ = rc /2, we get  > 1 from Eq. (48). As the distance r = |~r| ≥ 0, Eq. (49) is
well-defined only when the denominator is positive, i.e.,

1 +  cos θ ≥ 0 ⇒ cos θ ≥ −1/. (64)

16
This implies there is a critical angle θ∗ > 0 such that r(θ) is well-defined when −θ∗ <
θ < θ∗ . As t → ±∞, the trajectory approaches the straight lines with slopes ∓ tan θ∗
asymptotically.

17
3. KEPLER’S LAWS

We probably cannot end a course on classical mechanics without mentioning the Kepler’s laws
of planetary motion, so, let’s do that!
Consider a planet revolving around the Sun. Since the mass of the Sun is way bigger than that
of any of the planets, we can take M ≈ mSun and µ ≈ mplanet (Fig. 1). Kepler made the following
observations (which were explained by Newton a century later)

1. The orbit of the planet is an ellipse with the Sun sitting at one of the two foci.

# elliptical orbits

We have established this already when we discussed the closed/ bounded orbits, i.e., case
(b), in the interpretation of the equation of trajectory/ orbit r(θ).

2. Given the same interval of time, the area swept out by a line connecting the planet to the
Sun is the same.

# equal time, equal area

This claim is a bit awkward to state in words, but much better illustrated by a figure. In
Fig. 5, the blue and red areas are the same provided the same time interval ∆t is involved.
Heuristically, this is because when the planet is closer to the Sun, it is also moving faster
due to the conservation of energy. So, given the same time interval ∆t, the planet moves a
longer distance along the arc of the ellipse. This compensates the closer distance to the Sun
exactly such that the area swept will be the same.

Now, more quantitatively, the Kepler’s second law corresponds to the claim that

dA
= constant, (65)
dt

with the area A as defined by Fig. 5. This implies

Z t2
dA
A= = constant × (t2 − t1 ), (66)
t1 dt

18
 i.e., A depends only on the time interval ∆t = t2 − t1 . To see why Eq. (65) holds, it suffices
to consider an infinitesimal version, for which the area is that of a small triangle (purple in
Fig. 5):

1 1
δA ≈ (rδθ)r = r2 δθ, (67)
2 2

and so we have

dA δA 1 `
= lim = r2 θ̇ = , (68)
dt δt→0 δt 2 2µ

where in the last equality we used the conservation of angular momentum in central force
motion. This shows that dA/dt is indeed a constant, and hence the second law follows.

Figure 5. The areas A1 = A2 when the same time interval δt is spent for the planet to “sweep” them out.
This can be understood by consider an infinitesimal version when the angle swept out is δθ.

3. Let τ be the period of the orbit and a be the semi-major axis. Then τ 2 ∝ a3 .

# period-squared, length-cubed

To understand this claim, we need to first introduce another vocabulary: the “semi-major
axis” of an ellipse is defined to be the distance from the center (note: not a focus unless we
have a circle!) to the furthest point on the ellipse (Fig. 6). We denote the semi-major axis
by a. One can similarly define the semi-minor axis, denoted by b, as the shortest distance

19
between any point on the ellipse and the center. In a sense, a and b are the appropriate
(anisotropic) generalization of the radius of a circle to the ellipse. We note that the area of
the ellipse is πab (see Appendix A if you want more details).

To establish Kepler’s third law, let us first connect the semi-major/ minor axes a and b to
the quantities we had defined when we first discussed the geometry of an ellipse. First, we
notice
 
r+ rmax + rmin rc 1 1 rc
a= = = + = . (69)
2 2 2 1+ 1− 1 − 2
Note that a is the arithmetic mean of rmax and rmin (i.e., the “ordinary” mean). The expres-
sion for b requires a little more work. In particular, we recall the distance between the two
foci is r− , and the total length of the purple line in Fig. 6 is r+ = 2a, so
r
2 2
 r 2  r 2
− + r+ − r− (70)
b2 + = ⇒ b= ,
2 2 4
but we have already evaluated this combination in Eq. (60), so
√ rc
b = rmax rmin = p . (71)
(1 − )2
Notice that, curiously, b is the geometric mean of rmax and rmin .

With these preparations, we can now consider the period τ . When we set the time interval
δt = τ , the area defined under the Kepler’s second law becomes the same as the area of an
ellipse. Using Eq. (68), we can write
Z τ
dA `
πab = dt = τ. (72)
0 dt 2µ
If you still remember, the parameter rc encodes the angular momentum ` in our problem
through its definition in Eq. (28). So, expressing everything in terms of rc , we can write this
“area equation” as
p
rc rc GM µ2 rc
π p = τ
1 − 2 (1 − )2 2µ
 3/2 √ (73)
rc GM
⇒ π = τ.
1 − 2 2
Since the bracketed expression above is simply the semi-major axis a, we arrive at
√
3/2 GM GM 2 (74)
a = τ ⇒ a3 = τ ,
2π 4π 2
in other words, a3 ∝ τ 2 with the proportionality constant depending only on (fundamental)
constants and M ≈ MSun .

20
Figure 6. The distance a is called the semi-major axis of the ellipse and b is called the semi-minor axis. The
area of the ellipse is simply πab. (As a sanity check, for a circle one has a = b = r and we recover the
familiar expression of πr2 .)

21
...OUTRO

At long last, we have reached the end of our course! We have promised that after finishing
this course you will “see equations dancing around when you gaze upon the starry sky.” That
would be the case if you truly appreciate the power of a mathematics-based model of our physical
universe, or if you were so overwhelmed by our course that you couldn’t shake it off your mind
even when you stared at something as magnificent as a starry sky. Either way, you have made the
right connection: theoretical physics is about how to develop (mathematical) models to understand
and predict phenomena, big or small, in our physical world. If you think about how insignificant
we are in the vast universe, it is truly spectacular that we managed to understand anything at all
about our world!
We started this course by warning you that it will be hard (c.f., course overview). I’m not sure
if it’s good or bad that we lived up to that promise (or not). Anyway, the materials we need to cover
correspond to the accumulation of over a thousand years of human knowledge. That said, the most
modern idea we have seriously introduced in this course was only the Hamiltonian’s formulation
of classical mechanics, which emerged in around 1830s [3]. In other words, you have around 200
years’ worth of brilliant physics insights to learn in the remaining of your university life [4]!
But, as we also emphasized in the course overview, in 3032 we reward you for trying. If you
care to read up to this line, I’m confident you will not need to take this course again :) Thank you
for taking 3032; good luck with your finals.

22
 Appendix A: Ellipses in Cartesian coordinates

For completeness, we provide a brief overview on how to discuss an ellipse using Cartesian
coordinates in this optional appendix.

Our problem is setup in Fig. 7. We let a and b respectively be the semi-major and semi-minor
axes. We further let the separation between the two foci be ξ. The parameters a, b, and ξ can be
related to the minimum distance to one of the focus rmin (nerdy name = perigee), the maximum
distance rmax (nerdy name = apogee), and eccentricity  as discussed in the main text. (Note: these
are not independent parameters. Geometrically, an ellipse is only characterized by two parameters.
For instance, one has relations like rmax /rmin = (1 + )(1 − )−1 .) Unlike our discussion in the
polar form, which placed the origin at the focus on the right, this time we place the origin at the
center for convenience.

Figure 7. Studying an ellipse in Cartesian coordinate systems.

Our goal here is to derive an equation between x and y for points moving on the ellipse defined
by the “taut pinned string” condition we discussed in the main text. To this end, we first note that
the total length of the string (purple line in Fig. 7) is 2a. Our geometrical condition translates into

p p
(x − ξ)2 + y 2 + (x + ξ)2 + y 2 = 2a, (A1)

23
and we can now massage this to simplify
 p 2
(x + ξ)2 + y 2 = 2a − (x − ξ)2 + y 2
p
x2 + 2xξ +  ξ2 + y2 =4a2 − 4a (x − ξ)2 + y 2 + x2 − 2xξ + 
ξ2 + y2
p ξ
(x − ξ)2 + y 2 =a − x
a
ξ2 2 (A2)
x2 −   + ξ 2 + y 2 =a2 − 
2xξ 2ξx
+ x
a2
ξ2
 
x2 1 − 2 + y 2 =a2 − ξ 2
a
2
x y2
+ =1
a2 a2 − ξ 2
To bring it into a more familiar form, let us get rid of ξ in favor of b. If we bring the point (x, y) to
where the ellipse intersects the y axis (blue line in Fig. 7), we see that

b 2 + ξ 2 = a2 ⇒ a2 − ξ 2 = b 2 , (A3)

and so we may rewrite Eq. (A4) as

x2 y 2
+ 2 =1, (A4)
a2 b
which is a “standard form” of the equation for an ellipse in Cartesian coordinates.
This form also allows one to see that the area of the ellipse is πab. of course, one can integrate
p
y = b 1 − x2 /a2 directly (and multiply by 2). A simpler “calculation” is as follows: let R define
the region bounded by the ellipse. The area of the ellipse can be defined as
Z
A= dxdy. (A5)
R

Now, let us define the scaled (dimensionless) variables

x y
x̃ = ; ỹ = , (A6)
a b
and in these scaled variables the region R̃ is a circle of unit radius, so the area of R̃ is π. We can
now conclude
Z x y  Z
A = ab d d = ab dx̃dỹ = πab. (A7)
R a b R̃

[1] What if the angular momentum vanishes? In that case, we actually have motion along a single line, i.e.,
motion in one dimension!

24
[2] Here, the total energy E is the energy of the system in the inertial frame in which the center of mass is
stationary.
[3] This roughly marks the transition from what one usually calls the “early modern era” to the “modern
era.” Both special and general relativity are within classical (defined as non-quantum) mechanics, but
they are from the modern era instead. It might have been more accurate if we named this course as
“early modern mechanics” :p.
[4] Don’t be mistaken to think that 200 years is short compared to the thousands of years we have spent
on accumulating the knowledge discussed in this course: knowledge accumulated, like the human
population size, grows exponentially with time!

25