THE 2-BODY PROBLEM

ROBERT J. VANDERBEI

A BSTRACT. In this short note, we show that a pair of ellipses with a common focus is a solution to
the 2-body problem

1. I NTRODUCTION .
Solving the 2-body problem from scratch is doable but difficult. But, what if we simply want to
verify that there are elliptical orbits where the two ellipses share a common focus, which is also the
center of mass of the system? With these suppositions, maybe this problem isn’t so bad. In other
words, let’s try to use guess-n-check method. It should be easier. (We all believe that P 6= NP,
right!) Here we go...
First, choose the coordinate system so that the foci lie on the horizontal x-axis and so that the
shared focus is at the origin (see Figure).
One body’s second focus is on the positive x-axis and the other body’s second focus is on the
negative x-axis. Call the right body “body one” and the left body will be “body two”. The orbit of

Key words and phrases. Celestial mechanics, n-body problem, ellipse.

(c,b)
↓

c c+a

F IGURE 1. A pair of ellipses sharing a common focus.

1
2 ROBERT J. VANDERBEI

body one can be given parametrically as

x1 = c + a cos θ
y1 = b sin θ.
Here, a, b, and c are constants whereas x1 , y1 , and θ are functions of time t. The constant c is the
x-coordinate of the center of the ellipse and the constants a and b are the semi-major and semi-
minor axes, respectively. Clearly all three constants are positive numbers and a > b. Furthermore,
an
√ important property of ellipses is that the distance from the center of the ellipse to a focus is
a2 − b2 . Since (c, 0) is the center of the ellipse and (0, 0) is a focus, it follows that
√
c = a2 − b 2 .
The assumption that the center of mass of the system coincides with the focus at the origin implies
that    
x2 x1
= − .
y2 y1
The distance r between the two bodies plays and important role in Newton’s law of gravity, so we
start by computing it:
p
r = (x2 − x1 )2 + (y2 − y1 )2
p
= (2x1 )2 + (2y1 )2
q
= 2 x21 + y12
p
= 2 (c + a cos θ)2 + (b sin θ)2
p
= 2 c2 + 2ac cos θ + a2 cos2 θ + b2 sin2 θ
q
= 2 c2 + 2ac cos θ + a2 cos2 θ + (a2 − c2 ) sin2 θ
√
= 2 c2 cos2 θ + 2ac cos θ + a2
= 2(a + c cos θ).
Newton’s laws involve accelerations and so we differentiate once
ẋ1 = −a sin θ θ̇
ẏ1 = b cos θ θ̇.
and then a second time
ẍ1 = −a sin θ θ̈ − a cos θ θ̇2
ÿ1 = b cos θ θ̈ − b sin θ θ̇2 .
For simplicity, and without loss of generality, suppose that units are chosen in such a way that the
gravitational constant G equals one. Also for simplicity, but with some loss of generality, suppose
 THE 2-BODY PROBLEM 3

that the masses of the two bodies are both equal to one. With these assumptions, Newton’s law of
gravity is
x2 − x1 −x1
ẍ1 = =
r3 4(a + c cos θ)3
y2 − y1 −y1
ÿ1 = 3
= .
r 4(a + c cos θ)3
Substituting the formulas for x1 , y1 , and their second derivatives derived above, we get
−(c + a cos θ)
−a sin θ θ̈ − a cos θ θ̇2 =
4(a + c cos θ)3
−(b sin θ)
b cos θ θ̈ − b sin θ θ̇2 = .
4(a + c cos θ)3
Negating both sides and writing in matrix form, we see that
 
   c/a + cos θ
sin θ cos θ θ̈  4(a + c cos θ)3 
= sin θ

− cos θ sin θ θ̇ 2  
4(a + c cos θ)3
Multiplying by the matrix inverse, we get
 
    c/a + cos θ
θ̈ sin θ − cos θ  4(a + c cos θ)3 
= 
sin θ

θ̇2 cos θ sin θ  
4(a + c cos θ)3
" #
1 sin θ(c/a + cos θ) − cos θ sin θ
=
4(a + c cos θ)3 cos θ(c/a + cos θ) + sin θ sin θ
" #
1 (c/a) sin θ
=
4(a + c cos θ)3 (c/a) cos θ + 1
 c sin θ 
3 
=  4a(a + c1 cos θ) 


4a(a + c cos θ)2

In other words, for a solution to exist, the function θ must simultaneously be a solution to both of
the following, possibly inconsistent, equations:
c sin θ
θ̈ =
4a(a + c cos θ)3
1
θ̇ = √
2 a(a + c cos θ)
4 ROBERT J. VANDERBEI

To verify consistency, we check that differentiating the equation for θ̇ brings us to the equation
derived above for θ̈:
1 c sin θ θ̇
θ̈ = √
2 a (a + c cos θ)2
1 c sin θ 1
= √ 2
√
2 a (a + c cos θ) 2 a(a + c cos θ)
c sin θ
= .
4a(a + c cos θ)3
Hence, we see that the formulas for the first and second derivative of θ are indeed consistent with
each other.
Picking θ(0) can be given any value as it merely determines where the bodies are in their orbit at
time t = 0. The semi-major and semi-minor axes a and b determine c (as already mentioned) and
θ̇(0):
c sin θ(0)
θ̇(0) = .
4a(a + c cos θ(0))3

Suppose, for simplicity, that θ(0) = 0. The differential equation for θ̇(t) can be rewritten as
dt
(a + c cos θ) dθ = √ .
2 a

Since ε = c/a is the well-known eccentricity of the ellipse, it is common to divide both sides of
this equation by a thereby replacing c with the eccentricity ε:
dt
(1 + ε cos θ) dθ = .
2a3/2
Integrating from 0 to t, we get
Z θ(t) Z t
dt
(1 + ε cos θ) dθ = .
0 0 2a3/2
The integrals can be computed explicitly. The result is
t
θ(t) + ε sin θ(t) = .
2a3/2
Unfortunately, this is a transcendental equation for θ(t) and so it does not have a simple “closed
form” solution. However, in the case where the eccentricity is zero, we have an exact solution:
t
θ(t) = .
2a3/2
 THE 2-BODY PROBLEM 5

And, when the eccentricity is not zero, the following recursion quickly converges on the correct
answer:
θ(0) (t) = 0
t
θ(1) (t) = − ε sin θ(0) (t)
2a3/2
..
.
t
θ(k+1) (t) = − ε sin θ(k) (t)
2a3/2
..
.

D EPARTMENT OF O PERATIONS R ESEARCH AND F INANCIAL E NGINEERING , P RINCETON U NIVERSITY, P RINCE -

TON , NJ 08544.
E-mail address: [email protected]