v02

<< 1
2gl

g
Then the equation θ̈ + sin θ = 0 is approximated by
l
g
θ̈ + θ = 0
l
g1
2
⇔ θ̈ + ω 2 θ = 0, ω := ,
l

which gives rise to simple harmonic motion.

Example: Sheet 10 Qn 3.

9 Motion under Central Forces.

9.1 Central forces

A central force is a generalisation of the gravitational force (§§5.1, 5.3). Consider a particle of
−−→
mass m at P with nonzero position vector x = OP .
P
m

Definition 9.1 F(x) is a “central force” if it is always directed towards (or away from) O,
and its magnitude depends only on the distance r = kxk.

So,
f (r)
F(x) = f (r) x̂ = x,
r

75
 x
with x̂ := , and f (r) a scalar function. The central force is one of attraction towards O if
r
f (r) < 0, or repulsion away from O if f (r) > 0.
Examples: 1. For a particle of mass m in orbit about a planet of mass M , the gravitational
force is a central force with
µm
f (r) = − , µ = GM.
r2
2. The 3-dimensional analogue of a simple harmonic motion (a mass on a spring): mẍ = −kx
(k the stiffness of the spring), f (r) = −kr.
3. In electrostatics, the force between two charged particles P1 and P2 is

K q1 q2
F(x) = x̂,
r2

where q1 and q2 are charges of P1 and P2 , K > 0 a physical constant (i.e. like particles repel,
opposite attract). [The above is called Coulomb’s law of electrostatics, due to the same Coulomb
as for the friction law.] Hence, assuming particle P2 fixed, f (r) = Kq1 q2 /r 2 .

9.2 Motion under central forces

The equation of motion (N2) is of the form:

f (r)
mẍ = F(x) = x. (9.1)
r

Notation: In this section we assume that the particle has mass m = 1, and so from now on the
term m won’t explicitly enter into the analysis. (An alternative way of thinking of this is that
we assume the central force is a force “per unit mass”. There is no loss in generality, since in
our analysis f (r) is arbitrary and so can be replaced by m f (r) to this effect.) So

f (r)
ẍ = x. (9.2)
r

Properties of motion:
Property 1: Equation (9.2) describes motion in a plane.
Proof:

f (r) f (r)
x ∧ ẍ = x ∧ x= x ∧ x = 0.
r r
d
∴ (x ∧ ẋ) = ẋ ∧ ẋ} +x ∧ ẍ = 0.
dt | {z
=0

∴ x ∧ ẋ = h = constant . (9.3)

76
Take the dot product of (9.3) with x (cf. Sheet 7 Qn 1):

x.h = x.x ∧ ẋ = [x, x, ẋ] = 0

∴ x.h = 0.

If h 6= 0 this is the equation of plane, Π say, perpendicular to h and containing the origin. Then
x(t) ∈ Π, ∀t i.e. the motion is in the plane Π. [In the special case of h = 0, from (9.3) x ∧ ẋ = 0,
and so ẋ is parallel to x, and we have motion along a line (and hence in any plane containing
this line).]
Since motion is in a plane it is 2-Dimensional, and we can use polar coordinates in the plane
to describe the motion. With k, the unit vector in the h direction, choose the polar coordinates
on Π so that er , eθ , k form a right-handed system. Hence (9.3) can be written as

x ∧ ẋ = h k, h =: khk.

eθ er
h
x(t)
O
Π

Angular momentum: (cf. Sheet 11 Qn 1) What is the physical meaning of x ∧ ẋ = constant?

For a particle of mass m we define the angular momentum by l = x ∧ mẋ. N2 states that
F = mẍ, so taking the vector product with x gives,
d
x∧F= (x ∧ mẋ),
dt
(you need to fill in a step here!). The term x ∧ F is the torque, which measures the turning
d
effect produced by F on a particle, and the term (x ∧ mẋ) is the rate of change of angular
dt
momentum. If F = 0, or F || x, then x∧F = 0, i.e. the torque is zero, and so x∧mẋ = constant.
For motion under a central force, F || x and so angular momentum is conserved.
Property 2: For motion under any central force, r 2 θ̇ = constant.
First, recall (§4.3)

x = rer

ẋ = ṙer + r θ̇eθ
   
ẍ = r̈ − r θ̇ 2 er + r θ̈ + 2ṙ θ̇ eθ .

77
Then (with m = 1)

constant = h = h k = x ∧ ẋ
 
= (rer ) ∧ ṙer + r θ̇eθ

= r 2 θ̇ er ∧ eθ

∴ hk = r 2 θ̇ k

=⇒ h = r 2 θ̇ = constant (9.4)

This is precisely K2 (Kepler’s equal areas law, §5.1), which is thus seen to be true for any
central force, not only for gravitation. Finally, as expected from N2, the angular component of
acceleration is always zero since,

1 d  2  1 dh
r θ̈ + 2ṙ θ̇ = r θ̇ = = 0. (9.5)
r dt r dt

Property 3 : The orbit equation.

  f (r) f (r)
With ẍ = r̈ − r θ̇ 2 er and x = f (r)er , then ẍ = x, implies
r r

r̈ − r θ̇ 2 = f (r),

or, since r 2 θ̇ = h, we have,

h2
∴ r̈ − = f (r) . (9.6)
r3
This is a 2nd order nonlinear ODE for r(t).
Conservation of Energy Equation As in §7, we can integrate (9.6) once (write v = ṙ) to
obtain Z
1 2 h2
ṙ + 2 − f (r) dr = E,
2 2r
for some constant E. This can be regarded as an equation describing the conservation of energy.
Now return to the question of finding the orbit (= the trajectory of the particle). We seek to
eliminate t in (9.6) to obtain r = r(θ). To do this we make the following change of variable:

1
r= , u = u(θ),
u

which makes sense only if

θ̇ 6= 0,

78
since otherwise θ is constant. Now,

d −1
d
dθ
ṙ = u = u−1
dt dθ dt
1 du du 2 du
= − 2 θ̇ = − r θ̇ = − h .
u dθ dθ |{z} dθ
=h

h
So, again using θ̇ = = hu2 ,
r2
 
d dṙ d du d2 u
r̈ = ṙ = θ̇ = − θ̇ h = − h2 u2 .
dt dθ dθ dθ dθ 2

Hence (9.6) becomes

d2 u
−h2 u2 − h2 u3 = f (u−1 ),
dθ 2
or,
d2 u f (u−1 )
+ u = − , assuming (h 6= 0). (9.7)
dθ 2 h2 u2

Definition 9.2 Equation (9.7) is called the orbit equation. It holds provided θ̇ 6= 0 (or, equiv-
alently, h 6= 0).

Example: Usually, (9.7) will be nonlinear, but in the following example it reduces to a linear
µ
equation. Consider f (r) = − 2 , that is, f (u−1 ) = − µu2 , the case of gravitation. Equation
r
(9.7) becomes
d2 u µ
2
+ u = 2. (9.8)
dθ h
This is a 2nd order linear ODE with solution

µ
u = A cos θ + B sin θ + ,
h2

where A, B are arbitrary constants. Thus

1 µ
= u = C cos(θ − θ0 ) + 2 , (9.9)
r h

where C, θ0 are alternative arbitrary constants. (We can choose C ≥ 0, with A = C cos θ0 , B =
C sin θ0 .) We are also free to choose the line θ = 0 (that is, the x-axis), so without loss of
generality, we take θ0 = 0. Also, we choose

µe
C = , e > 0,
h2

79
 Ch2
so e := . Therefore,
µ

h2 /µ
= 1 + e cos θ
r
l h2
∴ r = , l := .
1 + e cos θ µ

Hence, we recover the usual polar form of a hyperbolic function. For example, if 0 < e < 1
we obtain an ellipse (see §4.4), for the motion of planets, comets, etc. [e = 1 parabola, e > 1
hyperbola, for asteroids etc.]
Another example: Sheet 11 Qn 3 (inverse cube law).

9.3 Escape velocity and black holes

Suppose a particle is projected radially from the surface of a planet of radius R with speed v0 .
(Remember that we assume m = 1.) Thus, ẋ(0) = v0 er , and from ẋ = ṙer + r θ̇eθ the initial
conditions are:
r = R, ṙ = v0 , θ = 0, θ̇ = 0, at t = 0.

v0

O
M

For the case of Newtonian gravitation,

f (r) µ
ẍ = x = − 3x
r r

with,
µ
f (r) = − , µ = GM.
r2
In polar coordinates r 2 θ̇ = h, and the orbit equation (9.6) becomes

h2 µ
r̈ − 3
= − 2. (9.10)
r r

The initial conditions give:

h = r 2 (0) θ̇(0) = R2 × 0 = 0,

80
and so r 2 (t)θ̇(t) = 0. Therefore, for r 6= 0,

θ(t) = constant = 0,

since we take θ(0) = 0. Hence motion is always in the radial direction. Since h = 0, (9.10) gives,
µ
r̈ = − .
r2
To solve this introduce v = ṙ (cf. §7), and so
dv µ
= − 2.
dt r
Assuming v = v(r) and using the chain rule,
dv dv dr dv dv
= = ṙ =v ,
dt dr dt dr dr
and so,
dv µ
v = − 2.
dr r
Integrating, we obtain
1 2 µ
v = + C,
2 r
for some constant C. From the initial conditions
1 µ
C = v02 − ,
2 R
so
2µ 2µ
v 2 (r) = + v02 − . (9.11)
r R
Note that v(r) decreases with r.
Consider the following two cases:
2µ
(i) v02 < .
R
Equation (9.11) is
2µ 2µ
0 ≤ v 2 (r) = + v02 − .
r
|{z} | {z R}
→0 as r→∞ <0
This implies
2µR
r ≤ =: rmax ,
2µ − v02 R
with v 2 = 0 at r = rmax and v 2 > 0 for r < rmax .

Thus r cannot exceed rmax , since then v 2 would be negative. Thus, the particle travels
out to r = rmax and then returns to the planet. It cannot therefore escape the planet’s
gravitational attraction.

81
 rmax

R
O

2µ
(ii) v02 ≥ .
R
In this case
2µ 2µ
v2 = + v02 − > 0, ∀r,
r
|{z} | {z R}
>0 ≥0

and v is never zero. Thus the particle never stops and r keeps increasing indefinitely, i.e.
the particle escapes the planet.

Conclusion: To escape the planet the initial speed v0 must be greater than (or equal to)
r r
2µ 2M G
vesc := = .
R R

Definition 9.3 The escape speed is the speed vesc , such that for v0 ≥ vesc a particle escapes
the gravitational attraction of the planet. Similarly, the escape velocity is vesc = vesc er .

For the earth vesc ≈ 11.2km s−1 .

The escape speed for a general f (r) is discussed in Question 4 Sheet 11.
Black Holes: Consider a star and assume its radius satisfies

2M G
R< ,
c2

where c is the speed of light (c ≃ 3 × 108 km s−1 ). Since G and c are universal constants, this is
a condition relating the radius of the star to its mass. Rearranging, we have
r
2M G
c< ,
R

so the escape speed is greater than the speed of light. Since not even light can escape from such
a star, it will appear invisible at a distance greater than rmax . Such a star is called a black
2M G
hole. The value is called the “Schwarzschild radius”. [For a planet with the mass of our
c2
sun to be a black hole R must be less than 1cm!]

82
A rough definition of a black hole is as follows: A black hole is an object in space that has
collapsed under its own gravitational forces to such an extent that its escape speed is equal to
the speed of light. Black holes are believed to be formed in the gravitational collapse of massive
stars at the end of their life. The existence of black holes was first postulated by Laplace in
1798! (The above simplified analysis assumes of course that light also satisfies the Newton’s
laws of mechanics: the actual physical mechanism is more subtle, referring to general relativity
and cosmology which we do not discuss here.)

9.4 Circular orbits

For motion under a central force is it possible to have circular orbits? For a particle P to be
in a circular orbit about a mass M , we must have r ≡ a, a constant, and this simplifies the
equations considerably.
P

11111111
00000000
00000000
11111111
00000000
11111111
00000000
11111111
00000000
11111111
00000000
11111111
00000000
11111111
00000000
11111111 M

The basic equations of motion under a central force are (9.4) and (9.6):

h2
r 2 θ̇ = h, r̈ − = f (r),
r3

with h and f (r) given. If r = a =constant, then r̈ = 0. Hence

h
θ̇ = = a constant, ω say,
a2

and,
h2 + a3 f (a) = 0, h = a2 ω. (9.12)

This implies
aω 2 + f (a) = 0, (9.13)

which can only have a solution if f (a) < 0 (i.e. in the case of attraction). In this case,
 1/2
− f (a)
ω = ,
a

83
i.e. a circular orbit does exist for any a with f (a) < 0 with the above given ω.
For example, for gravity f (r) = − µ/r 2 , µ > 0, and there exists a unique circular orbit for any
 µ 1/2
a with θ̇ ≡ ω = = constant.
a3

9.5 On Stability of Circular Orbits

Stability is one of the most overused words in Applied Mathematics and so it is important to
say carefully what we mean. By “stability”, in the present context, we mean: “If the particle
is in some way perturbed slightly from its orbit it will remain close to its original orbit for all
future time.” Otherwise, we say that the orbit is “unstable”.
Consider a circular orbit r ≡ a, θ̇ ≡ ω, hence f (a) < 0 and (9.13) holds.

perturbation

Seek the solution in the form

a r(t) = a + η(t)
θ̇(t) = ω + ǫ(t)

Let this circular orbit be perturbed slightly to

r(t) = a + η(t), θ̇(t) = ω + ǫ(t). (9.14)

Here the functions η(t) and ǫ(t) are “small”, with initial values,

η(0) = δ1 , η̇(0) = δ2 , ǫ(0) = δ3 , (9.15)

and with |δ1 |, |δ2 |, |δ3 | all “small”.

The following gives a rigorous mathematical definition for the stability in the present context.

Definition 9.4 The circular motion is stable (in the sense of Lyapunov) if ∀ǫ > 0 ∃δ > 0 such
that as long as the initial conditions satisfy |δ1 | + |δ2 | + |δ3 | < δ, for all positive times holds

|η(t)| < ǫ, |ǫ(t)| < ǫ.

Otherwise the motion is called unstable.

84
Theorem 9.5 (Stability Theorem) A circular motion due to a smooth central force f , f (a) <
0, is stable if and only if
3f (a)
f ′ (a) + < 0.
a

Before discussing the proof, let us first consider two consequences of the theorem. First, for
3f (a)
f (r) = − µ/r 2 , f ′ (a) + = −µ/a3 < 0. Therefore, according to the Theorem, circular
a
orbits under the inverse square law of Newtonian gravitation are stable (a good news for the
3f (a)
world we live in!). Next, for f (r) = −µ/r n , n > 0, f ′ (a) + = −(n − 3)µ/an+1 . Thus, if
a
n < 3 circular orbits are stable, whereas, for n ≥ 3 circular orbits are unstable.
Idea of proof: (the following material is non-examinable.
The following is a sketch of a typical stability perturbation analysis based on ideas of “lineari-
sation” of the related ODE about its unperturbed solution.
The basic equations for motion under a central force hold, namely, (9.4), (9.6), are

h2
r 2 θ̇ = h, r̈ − = f (r). (9.16)
r3

The unperturbed solution r ≡ a, θ̇ ≡ ω solves (9.16) with h = a2 ω and (9.13) held.

The further plan is the following:

(i) If h1 denotes the constant of the perturbed orbit, to show that the change in the constant,
δh = (h1 − h) is “small”.

(ii) To show that, to the main-order approximation, η(t) satisfies the linear 2nd order ODE
 
′ 3f (a)
η̈ − f (a) + η = α, (9.17)
a

where α is a “small” constant.

(iii) Hence to show that, if

3f (a)
f ′ (a) + < 0,
a
the perturbed orbit is stable, whereas if

3f (a)
f ′ (a) + ≥ 0,
a

the perturbed orbit is unstable.

85
Remark Since η and ǫ are “small”, in the following analysis we ignore any terms involving
η 2 , η ǫ, ǫ2 , and any terms containing higher powers. For convenience, we will write h.o.t. for
these “higher order terms” which are to be ignored.

(i) First, from (9.5) we have

r 2 (t)θ̇(t) = h = r 2 (0)θ̇(0),

so if we perturb r(0) and θ̇(0), the value of the constant h changes. Let h1 be the constant
for the perturbed orbit. Then from (9.16) and (9.14),

h21
η̈(t) − = f (a + η(t)), (9.18)
(a + η(t))3

and,
(a + η(t))2 (ω + ǫ(t)) = h1 . (9.19)

Now, (9.19) together with (9.15) shows that the constant h1 satisfies:

h1 = (a + η(0))2 (ω + ǫ(0)) = (a + δ1 )2 (ω + δ3 )

= (a2 + 2aδ1 + h.o.t)(ω + δ3 )

= a2 ω + 2a ω δ1 + a2 δ3 + h.o.t
| {z }
δh
∴ h1 = h + δh,

with all terms in δh “small”. So we are justified in saying that δh is a “small” constant.

(ii) Since |η| << a, and |ǫ| << ω,

1 1  η −3 1  η 
= 3 1+ = 3 1 − 3 + h.o.t. .
(a + η)3 a a a a

Also, Taylor’s Theorem gives,

f (a + η) = f (a) + ηf ′ (a) + h.o.t.

Hence, from (9.18)

 
(h + δh)2 3η
η̈ − 1− + h.o.t. = f (a) + f ′ (a)η + h.o.t.
a3 a

Thus, using (9.12) and neglecting the higher-order terms,

3h2 2hδh
η̈ + η − f ′ (a)η − 3 = 0,
a4 a

86
 or, using (9.12) again,  
′ 3f (a)
η̈ − f (a) + η = α, (9.20)
a
where
2hδh
α= .
a3
Thus we obtain the required constant coefficient second order ODE given by (9.17).

(iii) We need to consider the following cases:

f (a)
(a) f ′ (a) + 3 = − λ2 < 0.
a
Therefore (9.20) becomes:
η̈ + λ2 η = α,

which has solution

α
η(t) = A cos λt + B sin λt + ,
λ2
where A, B are integration constants. Using the initial conditions (9.15) then,

α δ2
A = δ1 − ; B= .
λ2 λ

Therefore
 α δ2 α
η(t) = δ1 − 2 cos λt + sin λt + 2 .
λ λ λ
Thus
α |δ2 | |α|
|η(t)| ≤ δ1 − 2
+ + 2 , ∀t,
λ λ λ
which is independent of t and remains small for all t. The particle oscillates about the
original circular orbit, and remains close to it for all time. Thus the orbit is stable.
3f (a)
(b) f ′ (a) + = γ 2 > 0.
a
Then (9.20) becomes
η̈ − γ 2 η = α,

which has solution,

α
η(t) = A cosh γt + B sinh γt − .
γ2
Using the initial conditions (9.15), we have

α δ2
A = δ1 + ; B= .
γ2 γ

87
 So  
α δ2 α
η(t) = 2
+ δ1 cosh γt + sinh γt − 2 .
γ γ γ
α δ2
Thus if 2
+ δ1 + 6= 0 then |η(t)| −→ ∞ as t −→ ∞. Therefore the orbit
γ γ
is unstable. (Notice that the original assumption of the smallness of η, η << a,
thereby ceases to remain valid for large t: a more accurate argument would therefore
have to be required for a completely rigorous proof.)
3f (a)
(c) Finally, f ′ (a) + = 0.
a
This case is also unstable. To see this, note that equation (9.20) becomes

η̈ = α,

t2
which has solution η(t) = α + δ2 t + δ1 . If either δ2 6= 0, or α 6= 0, then
2

|η(t)| −→ ∞, as t → ∞.

Thus the orbit is unstable.

88