Central forces

AK

A. Equation of motion such that

F⃗ = −∇V
⃗ (r) ≡ −∂r V (r)r̂. (5)
• Consider a particle of mass m is subjected to a
force which is only a function of the distance of the Thus, these forces are conservative.
particle from a origin and is along the line joining
• Lets formulate the whole problem in polar coordi-
the particle and the origin. This means, in polar
nates. From the earlier classes, we know
coordinate, the force is    
⃗r¨ = r̈ − rϕ̇2 r̂ + rϕ̈ + 2ṙϕ̇ . (6)
m⃗r¨ = F⃗ = F (r)r̂.
Thus, we can write the second law of Newton as
Let is consider the angular momentum of the par-
⃗ = ⃗r × (m⃗r˙ ).
   
ticle: L m r̈ − rϕ̇2 r̂ + m rϕ̈ + 2ṙϕ̇ = −∂r V (r)r̂. (7)
⃗
dL
= m⃗r˙ × ⃗r˙ + ⃗r × (m⃗r¨) • Lets take a dot product of ϕ̂ of the above equation
dt on both sides. This imples,
= ⃗r × F⃗ = F (r)⃗r × r̂ = 0. (1)
rϕ̈ + 2ṙϕ̇ = 0,
Thus, the angular momentum is dynamically con- 1 2 
⃗
stant, i.e, dL/dt = 0. ⇒ r ϕ̈ + 2rṙϕ̇ = 0,
r
• We also know that, d  2 
⇒ r ϕ̇ = 0. (8)
dt
⃗ = m⃗r · (⃗r × ⃗r˙ ) = 0.
⃗r · L (2) The last equation is infact the same as the angular
momentum conservation, Eq. (1). To see this, lets
Eq. (1)also means that, L̂ direction remains con- write the angular momentum in the polar coordi-
stant throughout the motion and Eq. (2) implies nate,
that the motion of the particle at any point of time
is in a plane perpendicular to the direction L̂. Thus, ⃗
L
= ⃗r × ⃗r˙ = rr̂ × (ṙr̂ + rϕ̇ϕ̂) = (r2 ϕ̇) ẑ. (9)
in many cases it is useful to describe such motion m
using 2D polar coordinates.
Here we used ⃗r˙ = ṙr̂ + rϕ̇ϕ̂, and the fact that r̂ ×
• An interlude: gradient in 2D polar coordinate. ϕ̂ = ẑ. Thus, the angular momentum L ⃗ is indeed
Consider a function in two dimension, f (x, y). We constant and along the perpendicular direction ẑ,
can equivalently write f as a function of r, ϕ, where with an amplitude
x = r cos ϕ and y = r sin ϕ. Further,
L = mr2 ϕ̇.
df = ∂x f dx + ∂y f dy.
• Now, lets take a dot product of r̂ with both sides
This implies df = ∇f⃗ ·d⃗r, with ∇f ˆ
⃗ = ∂x f x̂+∂y f dy of Eq. (7), to arrive at
ˆ Equivalently, we want to write,
and d⃗r = xx̂ + y dy. mr̈ − mrϕ̇2 = −∂r V.
2
⃗ · d⃗r,

df = ∂r f dr + ∂ϕ f dϕ = ∇f L
⇒ mr̈ − mr = −∂r V.
mr2
where (from last class) d⃗r = drr̂ + rdϕϕ̂, this im-
L2
 2 
L
plies, in the polar coordinates, ⇒ mr̈ = − ∂r V = −∂r + V (r) .
mr3 2mr2
⃗ = ∂r f r̂ + 1 ∂ϕ f ϕ̂.
∇f (3) Thus, the equation of motion simply looks like
r
mr̈ = −∂r Veff (r), L = mr2 ϕ̇ = const., (10)
• The above means that, for any central force, which
L2
a function of only r, we can always define a poten- with Veff (r) = + V (r). (11)
tial, 2mr2
Z r This reduces the whole problem (originally in three
V (r) = − F (r)dr, (4) dimensions), in an one-dimensional problem of an
effective potential Veff !
 2

• If the energy E is such that V (0) > E as well as

V (r → ∞) > E, then the particle can be orbit-
ing the origin forever, and we call such trajectories
bounded orbits.
• Further, if the Veff has a single minimum and E is
such that E = Veff (r1 ) = Veff (r2 ), then the motion
is bounded between r = r1 and r = r2 , see Fig. 1
for example. An example would be an ellipse. The
angles ϕ between successive r = r1 and r = r2 will
appear are called apsidal angles. An illustration is
in Fig. 2.

1. An example
FIG. 1. Energy E is such that the orbit is bounded between
r = r1 and r = r2 .
Lets take an example of inverse-square law, that is
when V (r) = −α/r. Thus
α L2
Veff = − + .
r 2mr2
′
From Veff (r) = 0, easy to see that, the minimum of the
Veff is at r0 = L2 /km, and Veff (r0 ) = −α2 m2 /2L2 . Now,
few things can happen.
• When E = Veff (r0 ) = −α2 m2 /2L2 , we simply have
ṙ = 0. That is r = r0 ! This is an equation of circle!
So, the orbit will be circular.
• When E > Veff (r0 ) but E < 0. In this case the
particle orbits is between two values of the radius
r1 and r2 (at which Veff (r) = E, and ṙ = 0). We
shall see later the motion traes an ellipse.
The smallest value of r = r1 is called the periapsis
(for motion around the Sun, we call them perihe-
FIG. 2. For this orbit, the apsidal angle is π/4. lion). The largest r = r2 is called the apoapsis (for
motion around the Sun, they are called aphelion).
Together they are called apsides.
• What about energy? We expect that, from the
above 1D problem E should be equal to 12 mṙ2 + • When E > 0, then the particle will be unbounded.
Veff (r). Lets see this explicitly. From earlier class That is, the particle comes from infinity, reaches a
minimum distance r, when Veff (r) = E, and then
1 it again goes back to infinity. This is what is called
E = mv 2 + V (r), a scattering process.
2
1 What about the stability of the circular orbit, when
= m(ṙ2 + r2 ϕ̇2 ) + V (r),
2 E = Veff (r0 ), i.e, is the particle is going to remain mostly
2
around the circular orbit if a small perturbation is given.

1 2 1 2 L
= mṙ + mr + V (r) This is the same question as asking whether at r = r0 ,
2 2 mr2
Veff has a minimum or maximum. Consider, V (r) =
1 L2 −α/r2 . Then,
= mṙ2 + + V (r) ≡ Veff (r). (12)
2 2mr2
′ nα L2 nα L2
Thus, if the energy is known, then another way to Veff (r) = − = 0 ⇒ n−2 = . (15)
rn+1 mr3 r0 m
solve for the motion of the particle is by using the
following two equations Then
′′ n(n + 1)α 3L2
1 2 Veff (r0 ) = − + ,
mṙ + Veff (r) = E, L = mr2 ϕ̇ = const., (13) n+2
r0 mr04
2
L2 n + 1 L2 3L2 L2
with Veff (r) = + V (r). (14) =− 4 + 4 = (2 − n). (16)
2mr2 r0 m mr0 mr04
 3

′′
Thus, Veff (r0 ) > 0 as long as n < 2. This imples
(1 + a2 )L2
V (r) = − + const.
2. Orbit equation 2mr2

• We also can do little more maths as followin: 3. Bertrand’s theorem (1873)

 2  2
1 2 1 dr 1 L dr
mṙ = m ϕ̇ = m = E − Veff ,
2 2 dϕ 2 mr2 dϕ Only two kinds of central forces, the attractive lin-
2
ear and the inverse-square, can result in closed, if

1 dr 2m
⇒ = 2 (E − Veff (r)) . (17) bounded, orbits.
r2 dϕ L
This equation, in principle, allows us to obtain how Closed orbits are orbits in which the particle eventually
r depends on ϕ, thus allowing us to know the full retraces its own footsteps. One example of closed orbit
trajectory of a particle. Using a new variable u = is a circle. This happens if the energy of the particle
1/r, the equation can also be written as E is such that E = Veff (r0 ), where r0 is a minima of
′
 2    the potential, that is Veff (r0 ) = 0. Then the motion is
du 2m 1 possible only with r = r0 and with ṙ = 0.
= 2 E − Veff . (18)
dϕ L u To have the orbit such that particle retraces its own
path, it is important to understand that the apsidal an-
• The same equation can be also cast in the following gle ϕA need to be constants. If an orbit is closed after
way: some M revolutions so that r(ϕ + 2πM ) = r(ϕ), then an
 2
1 ∗ du integral number N periods of the radial oscillation must
m + Ṽ (u) = E, (19) fit at the same point, thus 2ϕA N = 2πM , or
2 dϕ
π
ϕA = , (20)
where
m∗ = L2 /m and Ṽ (u) = Veff u1 = 21 m∗ u2 +


R
V u1 . where R must be a rational number.
Let us explore this for an orbit, which is nearly circu-
lar. Below, we follow Ref. [1] and [2]. Let us remember
Eq. (19):
 2
1 ∗ du 1
m + m∗ u2 + V (u−1 ) = E, (21)
2 dϕ 2
where the first two terms are like a simple-harmonic
oscillator’s kinetic and potential energy, with the time
replaced by u, and V (u−1 ) is like a perturbation. If
-15 -10 -5 0 5 10 15 V (u−1 ) is also of quadratic in u, then the net poten-
tial is 12 m∗ u2 + V (u−1 ) = 12 γu2 , and the solution would
be of the form u ∝ cos(ωϕ + ϕ0 ), with ω 2 = γ/m∗ . In
this case the condition for closed orbit is that ω needs to
be rational.
Now, the key point is, for almost any kind of potential,
the potential can be approximated to be ∝ u2 , near its
extreme (minimum) point. Let us perturb the potential
Ṽ (u) near its extremum, in Taylor expansion:
FIG. 3. The orbit is r = eϕ/10 .
dṼ (u)
An example of orbit is a spiral, given by r = r0 eaϕ . Ṽ (u) = Ṽ (u0 ) + (u − u0 )
du
What is the form of the potential? u0
2
1 d Ṽ (u)
We have + (u − u0 )2 + ··· . (22)
2 du2
u0
L aL
ṙ = a(r0 eaϕ )ϕ̇ = ar = .
mr 2 mr Now, Ṽ (u) has a minimum at u = u0 . Let us write
d2 Ṽ (u)
From Eq. (14), we have then Ṽ (u0 ) = E0 , du2 = α and u − u0 = δ. Then,
u0
2
L2 1

m aL
+ + V (r) = E = cont. Ṽ (u) = E0 + αδ 2 + · · · . (23)
2 mr 2mr2 2
 4

Then the orbit equation is

 2
1 ∗ dδ 1
m + αδ 2 = E − E0 . (24)
2 dϕ 2

Now, we can do the derivatives in r by writing d/du =

(dr/du)d/dr. Then a little exercise gives,

3V ′ (r0 ) + r0 V ′′ (r0 )
α = m∗ × (25)
V ′ (r0 )

Then, the solution of the orbit equation (which is again

a Harmonic oscillator problem), is u = u0 + A cos(Ωϕ),
with
s
3V ′ (r0 ) + r0 V ′′ (r0 )
Ω= , (26)
V ′ (r0 ) FIG. 4. The effective potential for an attractive inverse-
square-law force.
with minimum and maximum of u are u ± A. If umax
reaches at ϕ = 0, then the umin reaches at the apsidal
angle ϕA = π/Ω. But, to have a stable orbit, this must From the analysis of the effective potential, you should be
be a positive, real number, independent of E. For any able to show that, only attractive, inverse law potential
potential V (r), let us write, can have bounded orbit. For example, in case of circular
orbit, dVeff /dr = 0 gives,
3V ′ (r) + rV ′′ (r)
f (r) = . (27)
V ′ (r) dV l2 L2
− 3
= 0 ⇒ f (r) = − 3 . (30)
dr mr mr
Near r = r0 , f (r) = C, a real, positive number. This
gives, So, the force must be attractive (essentially to cancel the
centrifugal force, which is fictitious).
k n Also, we notice,
3V ′ + rV ′′ = CV ′ ⇒ V (r) = r , (28)
n
dV l2 α L2 L2
− = 0 ⇒ = ⇒ r0 = . (31)
where n = C − 2. So, only for such potentials, the f (r) dr mr3 r2 mr3 mα
is a constant! Further,
Then, Veff (r0 ) = −mα2 /2L2 .
π π π For such a potential, we need to solve
ϕA = =√ =√ . (29)
Ω C n+2  2   
du 2m 1
We immediately see that for n = −1 and 2, ϕ is a rational = 2 E − Veff , (32)
dϕ L u
multiple of π. But what about all other values of n, for
which this is also possible? We’ll discuss that no other with Veff ( u1 ) = L2 u2
+ V ( u1 ) = L2 u2
− αu, giving
2m 2m
value of n is possible (using again Newtonian dynamics),
in the next assignment. 
du
2
2mE 2mα
= 2
− u2 + 2 u
dϕ L L
B. Orbits for inverse-square-law force
 mα 2 2mE  mα 2
=− u− 2 + +
L L2 L2
 2
In astronomical problems, the potential V (r) ∝ r2 is dz
⇒ = −z 2 + B 2 , (33)
unphysical and we are left with only V (r) ∝ −1/r as dϕ
the possible potential for closed orbits. Thus, if we see
closed orbits, the gravitational force must follow an in- where
verse square-law! This neat agreement is what led physi- r
cists to astonish even today about the powerful theorem mα  mα  2EL2
z =u− 2 ; B = 1+ . (34)
we just described. L L2 mα2
Now the question is, what kind of orbits are there for Then we write,
inverse square-law potential? We’ll take Z Z
dz
α √ = dϕ (35)
V (r) = − . −z 2 + B 2
r
 5

This is a small exercise to solve the equation in obtaining,

1 mα
= 2 (1 + ϵ cos(ϕ − ϕ0 )). (36)
r L
In this we can also drop integration constant ϕ0 by setting
the motion at ϕ = 0 accordingly. The above equation is
very important one.
r
2EL2
ϵ= 1+ (37)
mα2
is called the eccentricity of the particle’s motion. ϵ is
always real (that is E > −mα2 /2L2 , why?). Defining

L2
k= ,
mα
the equation can also be written in Cartesian coordinates FIG. 5. The hyperbola solution when ϵ > 1. The two defini-
as (with ϕ0 = 0 for simplicity) tions of b (red arrows) are equivalent.
k = r + ϵx. (38)

Solving for r and squaring, we get to write the orbit equation (39) as (exercise)
2
x2 + y 2 = k 2 − 2kϵx + ϵ2 x2 . (39) (x − aϵ) y2
− = 1, (42)
a2 b2
Its another small exercise to conclude that the trajectory
which, of course, is a hyperbola with center at aϵ,
is a
in between the two straight lines:
• Circle, if ϵ = 0. x − aϵ y
± = 0. (43)
• Ellipse, if 1 < ϵ < 0. a b

• Parabola, if ϵ = 1, and, This hyperbola can have two branches. But, we

need to keep satisfying Eq. (38). For a positive
• Hyperbola, if ϵ > 1. value of k, as r increases, x needs to decrease such
that the sum of them remains constant. This allows
Other comments are following: only one solution.
• From Eq. (36), the minimum value of r can be • The impact parameter is defined as the perpendic-
found when ϕ + ϕ0 = 0, giving us ular distance of the straight lines, Eq. 43, from the
origin.
L2
rmin = . (40) Lets call the distance
mα(1 + ϵ)
p p
rmax is slightly tricky. When ϵ < 1, then the RHS d0 = x2 + y 2 = x2 + (x − aϵ)2 (b/a)2 .
of the equation is never 0 and rmax is
Then
2
L dd0
rmax = (if ϵ < 1). = 0 ⇒ x + (b/a)2 (x − aϵ) = 0, (44)
mα(1 − ϵ) dx
If ϵ > 1, then the RHS can be zero for some value with (b/a)2 = ϵ2 − 1. So, the minimum of d0 occurs
of ϕ, resulting in rmax = ∞. If it is a closed orbit, when (from the above equation),
that is, if ϵ < 1, then the apsidal angle is π.
x + (ϵ2 − 1)(x − aϵ) = 0,
• ϵ < 1 for E < 0 and ϵ ≥ 1 for E > 0. So, basically,
⇒ϵ2 x = aϵ(ϵ2 − 1) = kϵ,
for E > V (∞) = 0, the orbits are not bounded
(parabola or hyperbola). ⇒x ≡ x0 = k/ϵ.

• For ϵ > 1, let us define these positive numbers We also note that at the minimum of d0 , Eq. (44)
is also satisfied. Thus,
k k
a= , b= √ , (41) q
√
ϵ2 − 1 2
ϵ −1 d0min = x20 − x0 (x0 − aϵ) = x0 aϵ = b.
 6

So, this small exercise shows that this distance is

also nothing but b.

• What is the minimum distance of the incom-

ing particle to the origin? For y = 0, we get
x = rmin = aϵ − a. Check that this is same as
Eq. (40).

Also, check that, at this position, the y coordinate FIG. 6. The elliptical solution when ϵ < 1.
of the asymptotic lines is also nothing but b (i.e,
in Eq. (42), use x = aϵ − a, show y = b). This is 1. Elliptical orbit
another definition of impact parameter, also shown
in Fig. 5. Also see Problem 7.14 of Morin.
For 0 < ϵ < 1, we write the orbit equation, Eq. (39),
as
2
(x + aϵ) y2 k k
• Lets have a close look at Fig. 5. What is the angle + = 1, a = , b= √ . (49)
of deflection ϕd ? From the geometry: a2 b2 1 − ϵ2 1 − ϵ2
are the semi-major and the semi-minor axes. The center
  √ the ellipse is at x2 = −aϵ. The focal length is c =
of
b a2 − b2 = kϵ/(1 − ϵ ) = aϵ, which is at the origin.
ϕd = π − 2tan−1 . (45)
a Three laws of Kepler are the following:

1. The planets move in elliptical orbits with sun

But, at the focus.

2. The radius vector to a planet sweeps out an

r
b p 2EL2
= ϵ2 − 1 = . (46) area at a rate that is independent of the position
a mα2
of the planet.

If a particle is coming from infinity with the impact This is essentially stating that the angular momentum
parameter b with initial velocity v0 , then the angu- is conserved. In a small time dt, if an area dA is swept
lar momentum is L = mv0 b, kinetic energy E = (see above figure), then, dA = r(rdϕ)/2. So,
mv02 /2. For gravitational force, α = GmM , where
dA r2 dϕ L
M is the mass at the center, with M ≫ m. Then, = = . (50)
we get b/a = v02 b/GM . If we define, γ = v02 /GM , dt 2dt 2m
then,
3. The square of the period T of an orbit is propor-
tional to the cube of the semi-major-axis length.
ϕd = π − 2tan−1 (γb). (47)
From the second-law, the area of the ellipse:
Z T Z T
dA L LT
This allows us to determine the b as A = πab = dt = = .
0 dt 0 2m 2m
  Squaring this,
1 ϕd
b= cot . (48) 4m2 π 2 a2 b2
γ 2 T2 = .
L2
Now, L2 = kmα = mα(1 − ϵ2 )a. Also, b2 = a2 (1 − ϵ2 ).
So, from the angle of deflection, one can compute Putting these, we get
the impact parameter.
4mπ 2 a3
T2 = .
α
Again, for gravitational force on a planet, α = GmM,
• For a repulsive force, where α is now negative, the where M ≫ m is the mass of the star. Thus,
orbit equation becomes always a hyperbola! This
is part of your assignment! 4π 2 a3
T2 = .
GM
 7

[1] Jared Galbraith, and Jacob Williams, Journal of Under-

graduate Reports in Physics 29, 100005 (2019).
[2] S.A. Chin, Am. J. Phys. 83, 320-323 (2015).