Chapter 5

Few-Body Problems: The Motion

of the Planets

©2005 by Harvey Gould, Jan Tobochnik, and Wolfgang Christian

27 May 2005

We apply Newton’s laws of motion to planetary motion and other systems of a few particles and
explore some of the counter-intuitive consequences of Newton’s laws.

5.1 Planetary Motion

Planetary motion is of special significance because it played an important role in the conceptual
history of the mechanical view of the universe. Few theories have affected Western civilization as
much as Newton’s laws of motion and the law of gravitation, which together relate the motion of
the heavens to the motion of terrestrial bodies.
Much of our knowledge of planetary motion is summarized by Kepler’s three laws, which can
be stated as:

1. Each planet moves in an elliptical orbit with the Sun located at one of the foci of the ellipse.
2. The speed of a planet increases as its distance from the Sun decreases such that the line from
the Sun to the planet sweeps out equal areas in equal times.
3. The ratio T 2 /a3 is the same for all planets that orbit the Sun, where T is the period of the
planet and a is the semimajor axis of the ellipse.

Kepler obtained these laws by a careful analysis of the observational data collected over many
years by Tycho Brahe.
Kepler’s first and third laws describe the shape of the orbit rather than the time dependence
of the position and velocity of a planet. Because it is not possible to obtain this time dependence

111
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 112

in terms of elementary functions, we will obtain the numerical solution of the equations of motion
of planets and satellites in orbit. In addition, we will consider the effects of perturbing forces on
the orbit and problems that challenge our intuitive understanding of Newton’s laws of motion.

5.2 The Equations of Motion

The motion of the Sun and Earth is an example of a two-body problem. We can reduce this
problem to a one-body problem in one of two ways. The easiest way is to use the fact that the
mass of the Sun is much greater than the mass of the Earth. Hence we can assume that, to a good
approximation, the Sun is stationary and is a convenient choice of the origin of our coordinate
system.
If you are familiar with the concept of a reduced mass, you know that the reduction to a
one-body problem is more general. That is, the motion of two objects of mass m and M whose
total potential energy is a function only of their relative separation can be reduced to an equivalent
one-body problem for the motion of an object of reduced mass µ given by
Mm
µ= . (5.1)
m+M
Because the mass of the Earth, m = 5.99 × 1024 kg is so much smaller than the mass of the Sun,
M = 1.99 × 1030 kg, we find that for most practical purposes, the reduced mass of the Sun and the
Earth is that of the Earth alone. In the following, we consider the problem of a single particle of
mass m moving about a fixed center of force, which we take as the origin of the coordinate system.
Newton’s universal law of gravitation states that a particle of mass M attracts another particle
of mass m with a force given by
GM m GM m
F=− r̂ = − r, (5.2)
r2 r3
where the vector r is directed from M to m (see Figure 5.1). The negative sign in (5.2) implies that
the gravitational force is attractive, that is, it tends to decrease the separation r. The gravitational
constant G is determined experimentally to be

m3
G = 6.67 × 10−11 . (5.3)
kg · s2

The force law (5.2) applies to the motion of the center of mass for objects of negligible spatial
extent. Newton delayed publication of his law of gravitation for twenty years while he invented
integral calculus and showed that (5.2) also applies to any uniform sphere or spherical shell of
matter if the distance r is measured from the center of each mass.
The gravitational force has two general properties: its magnitude depends only on the separa-
tion of the particles, and its direction is along the line joining the particles. Such a force is called
a central force. The assumption of a central force implies that the orbit of the Earth is restricted
to a plane (x-y), and the angular momentum L is conserved and lies in the third (z) direction. We
write Lz in the form
Lz = (r × mv)z = m(xvy − yvx ), (5.4)
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 113

y
Fx m

Fy

θ
x
M

Figure 5.1: An object of mass m moves under the influence of a central force F . Note that
cos θ = x/r and sin θ = y/r, which provide useful relations for writing the equations of motion in
component form suitable for numerical solutions.

where we have used the cross-product definition L = r × p and p = mv. An additional constraint
on the motion is that the total energy E is conserved and is given by
1 GM m
E= mv 2 − . (5.5)
2 r

If we fix the coordinate system at the mass M , the equation of motion of the particle of mass
m is
d2 r GM m
m =− r. (5.6)
dt2 r3
It is convenient to write the force in Cartesian coordinates (see Figure 5.1):

GM m GM m
Fx = − cos θ = − x (5.7a)
r2 r3
GM m GM m
Fy = − sin θ = − y. (5.7b)
r2 r3
Hence, the equations of motion in Cartesian coordinates are

d2 x GM
=− 3 x (5.8a)
dt2 r
d2 y GM
= − 3 y, (5.8b)
dt2 r
where r2 = x2 + y 2 . Equations (5.8a) and (5.8b) are examples of coupled differential equations
because each equation contains both x and y.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 114

5.3 Circular and Elliptical Orbits

Because many planetary orbits are nearly circular, it is useful to obtain the condition for a circular
orbit. The magnitude of the acceleration a is related to the radius r of the circular orbit by

v2
a= , (5.9)
r
where v is the speed of the object. The acceleration always is directed toward the center and is
due to the gravitational force. Hence we have

mv 2 GM m
= , (5.10)
r r2
and ( GM )1/2
v= . (5.11)
r
The relation (5.11) between the radius and the speed is the general condition for a circular orbit.
We also can find the dependence of the period T on the radius of a circular orbit using the
relation,
2πr
T = , (5.12)
v
in combination with (5.11) to obtain
4π 2 3
T2 = r . (5.13)
GM
The relation (5.13) is a special case of Kepler’s third law with the radius r corresponding to the
semimajor axis of an ellipse.
A simple geometrical characterization of an elliptical orbit is shown in Figure 5.2. The two
foci of an ellipse, F1 and F2 , have the property that for any point P , the distance F1 P + F2 P is
a constant. In general, an ellipse has two perpendicular axes of unequal length. The longer axis
is the major axis; half of this axis is the semimajor axis a. The shorter axis is the minor axis; the
semiminor axis b is half of this distance. It is common to specify an elliptical orbit by a and by
the eccentricity e, where e is the ratio of the distance between the foci to the length of the major
axis. Because F1 P + F2 P = 2a, it is easy to show that
√
b2
e= 1− 2 (5.14)
a
with 0 < e < 1. (Choose the point P at x = 0, y = b.) A special case is b = a for which the ellipse
reduces to a circle and e = 0.

5.4 Astronomical Units

It is convenient to choose a system of units in which the magnitude of the product GM is not too
large and not too small. To describe the Earth’s orbit, the convention is to choose the length of
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 115

B
P

F1 F2
A
x
O
ea
a

Figure 5.2: The characterization of an ellipse in terms of the semimajor axis a and the eccentricity
e. The semiminor axis b is the distance OB. The origin O in Cartesian coordinates is at the center
of the ellipse.

the Earth’s semimajor axis as the unit of length. This unit of length is called the astronomical
unit (AU) and is
1 AU = 1.496 × 1011 m. (5.15)
The unit of time is assumed to be one year or 3.15 × 107 s. In these units, the period of the Earth
is T = 1 years and its semimajor axis is a = 1 AU. Hence from (5.13)

4π 2 a3
GM = = 4π 2 AU3/years2 . (astronomical units) (5.16)
T2
As an example of the use of astronomical units, a program distance of 1.5 would correspond to
1.5 × (1.496 × 1011 ) = 2.244 × 1011 m.

5.5 Log-log and Semilog Plots

The values of T and a for our solar system are given in Table 5.1. We first analyze these values
and determine if T and a satisfy a simple mathematical relationship.
Suppose we wish to determine whether two variables y and x satisfy a functional relationship,
y = f (x). To simplify the analysis, we ignore possible errors in the measurements of y and x. The
simplest relation between y and x is linear, that is, y = mx+b. The existence of such a relation can
be seen by plotting y versus x and finding if the plot is linear. From Table 5.1 we see that T is not
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 116

a linear function of a. For example, an increase in T from 0.24 to 1, an increase of approximately

4, yields an increase in T of approximately 2.5.
For many problems, it is reasonable to assume an exponential relation

y = C erx , (5.17)

or a power law relation

y = C xn , (5.18)
where C, r, and n are unknown parameters.
If we assume the exponential form (5.17), we can take the natural logarithm of both sides to
find
ln y = ln C + rx. (5.19)
Hence if (5.17) is applicable, a plot of ln y versus x would yield a straight line with slope r and
intercept ln C.
The natural logarithm of both sides of the power law relation (5.18) yields

ln y = ln C + n ln x. (5.20)

If (5.18) applies, a plot of ln y versus ln x yields the exponent n (the slope), which is the usual
quantity of physical interest if a power law dependence holds.

planet T (Earth years) a (AU)

Mercury 0.241 0.387
Venus 0.615 0.723
Earth 1.0 1.0
Mars 1.88 1.523
Jupiter 11.86 5.202
Saturn 29.5 9.539
Uranus 84.0 19.18
Neptune 165 30.06
Pluto 248 39.44

Table 5.1: The period T and semimajor axis a of the planets. The unit of length is the astronomical
unit (AU). The unit of time is one (Earth) year.

We illustrate a simple analysis of the data in Table 5.1. Because we expect that the relation
between T and a has the power law form T = Can , we plot ln T versus ln a (see Figure 5.3). A
visual inspection of the plot indicates that a linear relationship between ln T and ln a is reasonable
and that the slope is approximately 1.50 in agreement with Kepler’s second law. In Chapter 8, we
will discuss the least squares method for fitting a straight line through a number of data points.
With a little practice you can do a visual analysis that is nearly as good.
The PlotFrame class contains the axes and titles needed to produce linear, log-log, and semi-
log plots. It also contains the methods needed to display data in a table format. This table can
be displayed programmatically or by right-clicking (control-clicking) at runtime. Listing 5.1 shows
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 117

ln T 2

-1
-2 -1 0 1 2 3 4 5 6
ln a

Figure 5.3: Plot of ln T versus ln a using the data in Table 5.1. Verify that the slope is 1.50.

a short program that produces the log-log plot of the semi-major axis of the planets versus the
orbital period. The arrays, a and T, contain the semimajor axis of the planets and their periods,
respectively. Setting the log scale option causes the PlotFrame to transform the data as it is being
plotted and causes the axis to change how labels are rendered. Note that the plot automatically
adjusts itself to fit the data because the autoscale option is true by default. Also the grid and the
tick-labels change as the window is resized.
Listing 5.1: A simple program that producs a log-log plot to demonstrate Kepler’s second law.
package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
import o r g . o p e n s o u r c e p h y s i c s . f r a m e s . PlotFrame ;

public c l a s s SecondLawPlotApp {
public s t a t i c void main ( S t r i n g [ ] a r g s ) {
PlotFrame frame = new PlotFrame ( " ln (a)" , " ln (T)" , " Kepler ' s second law " ) ;
frame . s e t L o g S c a l e ( true , true ) ;
frame . s e t C o n n e c t e d ( f a l s e ) ;
double [ ] p e r i o d = {
0 . 2 4 1 , 0 . 6 1 5 , 1 . 0 , 1 . 8 8 , 1 1 . 8 6 , 2 9 . 5 0 , 8 4 . 0 , 1 6 5 , 248
};
double [ ] a = {
0.387 , 0.723 , 1.0 , 1.523 , 5.202 , 9.539 , 19.18 , 30.06 , 39.44
};
frame . append ( 0 , a , p e r i o d ) ;
frame . s e t V i s i b l e ( true ) ;
// d e f i n e s t i t l e s o f t a b l e columns
frame . setXYColumnNames ( 0 , "T ( years )" , "a ( AU )" ) ;
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 118

x y1 (x) y2 (x) y3 (x)

0 0.00 0.00 2.00
0.5 0.75 1.59 5.44
1.0 3.00 2.00 14.78
1.5 6.75 2.29 40.17
2.0 12.00 2.52 109.20
2.5 18.75 2.71 296.83

Table 5.2: Determine the functional forms of y(x) for the three sets of data. There are no mea-
surement errors, but there are roundoff errors.

// shows d a t a t a b l e ; can a l s o be done from frame menu

frame . showDataTable ( true ) ;
frame . s e t D e f a u l t C l o s e O p e r a t i o n ( j a v a x . swing . JFrame . EXIT ON CLOSE ) ;
}
}

Exercise 5.1. Simple functional forms

a. Run SecondLawPlotApp and convince yourself that you understand the syntax.

b. Modify SecondLawPlotApp so that the three sets of data shown in Table 5.2 are plotted. Gen-
erate linear, semi-log, and log-log plots to determine the functional form of y(x) that best fits
each data set.

5.6 Simulation of the Orbit

We now develop a program to simulate the Earth’s orbit about the Sun. The PlanetApp class
shown in Listing 5.2 organizes the startup process and creates the visualization. Because this
class extends AbstractSimulation, it is sufficient to know that the superclass invokes the doStep
method periodically when the thread is running or once each time the Step button is clicked.
The preferred scale and the aspect ratio for the plot frame is set in the constructor. The state-
ment frame.setSquareAspect(true) ensures that a unit of distance will equal the same number
of pixels in both the horizontal and vertical directions; the statement planet.initialize(new
double[]{x, vx, y, vy, 0}) in the initialize method is used to create an array on the fly
as the argument to another method.

Listing 5.2: PlanetApp.

package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
import o r g . o p e n s o u r c e p h y s i c s . c o n t r o l s . * ;
import o r g . o p e n s o u r c e p h y s i c s . f r a m e s . * ;

public c l a s s PlanetApp extends A b s t r a c t S i m u l a t i o n {

PlotFrame frame = new PlotFrame ( "x ( AU )" , "y ( AU )" , " Planet Simulation " ) ;
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 119

P l a n e t p l a n e t = new P l a n e t ( ) ;

public PlanetApp ( ) {
frame . addDrawable ( p l a n e t ) ;
frame . setPreferredMinMax ( −5 , 5 , −5, 5 ) ;
frame . s e t S q u a r e A s p e c t ( true ) ;
}

public void doStep ( ) {

f o r ( i n t i = 0 ; i <5; i ++) { // do 5 s t e p s b e t w e e n s c r e e n draws
p l a n e t . doStep ( ) ; // a d v a n c e s time
}
frame . s e t M e s s a g e ( "t = "+decimalFormat . format ( p l a n e t . s t a t e [ 4 ] ) ) ;
}

public void i n i t i a l i z e ( ) {
p l a n e t . o d e S o l v e r . s e t S t e p S i z e ( c o n t r o l . getDouble ( " dt " ) ) ;
double x = c o n t r o l . g et Double ( "x" ) ;
double vx = c o n t r o l . g et Do uble ( " vx " ) ;
double y = c o n t r o l . g et Double ( "y" ) ;
double vy = c o n t r o l . g et Do uble ( " vy " ) ;
// c r e a t e an a r r a y on t h e f l y as t h e argument t o a n o t h e r method
p l a n e t . i n i t i a l i z e (new double [ ] {x , vx , y , vy , 0 } ) ;
frame . s e t M e s s a g e ( "t = 0" ) ;
}

public void r e s e t ( ) {
c o n t r o l . s e t V a l u e ( "x" , 1 ) ;
c o n t r o l . s e t V a l u e ( " vx " , 0 ) ;
c o n t r o l . s e t V a l u e ( "y" , 0 ) ;
c o n t r o l . s e t V a l u e ( " vy " , 6 . 2 8 ) ;
c o n t r o l . s e t V a l u e ( " dt " , 0 . 0 1 ) ;
initialize ();
}

public s t a t i c void main ( S t r i n g [ ] a r g s ) {

S i m u l a t i o n C o n t r o l . createApp (new PlanetApp ( ) ) ;
}
}

The Planet class in Listing 5.3 defines the physics and instantiates the numerical method. The
latter is the Euler algorithm, which will be replaced in Problem 5.2. Note how the argument to the
initialize method is used. The System.arraycopy(array1,index1,array2,index2,length
method in the core Java API copies blocks of memory such as arrays and is optimized for particular
operating systems. This method copies length elements of array1 staring at index1 into array2
starting at index2. In most applications index1 and index2 will be set equal to 0.
Listing 5.3: A class that models the rate equation for a planet acted on by an inverse square law
force.
package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 120

import j a v a . awt . * ;
import o r g . o p e n s o u r c e p h y s i c s . d i s p l a y . * ;
import o r g . o p e n s o u r c e p h y s i c s . n u m e r i c s . * ;

public c l a s s P l a n e t implements Drawable , ODE {

// GM i n u n i t s o f (AU) ˆ 3 / ( y r )ˆ2
f i n a l s t a t i c double GM = 4 * Math . PI * Math . PI ;
C i r c l e c i r c l e = new C i r c l e ( ) ;
T r a i l t r a i l = new T r a i l ( ) ;
double [ ] s t a t e = new double [ 5 ] ; // {x , vx , y , vy , t }
E u l e r o d e S o l v e r = new E u l e r ( t h i s ) ; // c r e a t e s n u m e r i c a l method

public void doStep ( ) {

odeSolver . step ( ) ; // a d v a n c e s time
t r a i l . addPoint ( s t a t e [ 0 ] , s t a t e [ 2 ] ) ; // x , y
}

void i n i t i a l i z e ( double [ ] i n i t S t a t e ) {
System . a r r a y c o p y ( i n i t S t a t e , 0 , s t a t e , 0 , i n i t S t a t e . l e n g t h ) ;
// r e i n i t i a l i z e s t h e s o l v e r i n c a s e t h e s o l v e r a c c e s s e s d a t a from p r e v i o u s s t e p s
odeSolver . i n i t i a l i z e ( odeSolver . getStepSize ( ) ) ;
trail . clear ();
}

public void g e t R a t e ( double [ ] s t a t e , double [ ] r a t e ) {

// s t a t e [ ] : x , vx , y , vy , t
double r 2 = ( s t a t e [ 0 ] * s t a t e [ 0 ] ) + ( s t a t e [ 2 ] * s t a t e [ 2 ] ) ; // r s q u a r e d
double r 3 = r 2 * Math . s q r t ( r 2 ) ; // r cubed
rate [ 0 ] = state [ 1 ] ; // x r a t e
r a t e [ 1 ] = (−GM* s t a t e [ 0 ] ) / r 3 ; // vx r a t e
rate [ 2 ] = state [ 3 ] ; // y r a t e
r a t e [ 3 ] = (−GM* s t a t e [ 2 ] ) / r 3 ; // vy r a t e
rate [ 4 ] = 1; // time r a t e
}

public double [ ] g e t S t a t e ( ) {
return s t a t e ;
}

public void draw ( DrawingPanel panel , G r a p h i c s g ) {

c i r c l e . setXY ( s t a t e [ 0 ] , s t a t e [ 2 ] ) ;
c i r c l e . draw ( panel , g ) ;
t r a i l . draw ( panel , g ) ;
}
}

The Planet class implements the Drawable interface and defines the draw method as described
in Section 3.3. In this case we did not use graphics primitives such as fillOval to perform the
drawing. Instead, the method calls the methods circle.draw and trail.draw to draw the planet
and its trajectory, respectively.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 121

Invoking a method in another object that has the desired functionality is known as forwarding
or delegating the method. One advantage of forwarding is that we can change the implementation
of the drawing within the Planet class at any time and still be assured that the planet object is
drawable. We could, for example, replace the circle by an image of the Earth. Note that we have
created a composite object by combining the properties of the simpler circle and trace objects.
These techniques of encapsulation and composition are common in object oriented programming.

Problem 5.2. Verification of Planet and PlanetApp for circular orbits

a. Verify Planet and PlanetApp by considering the special case of a circular orbit. For example,
choose (in astronomical units) x(t = 0) = 1, y(t = 0) = 0, and vx (t = 0) = 0. Use the relation
(5.11) to find the value of vy (t = 0) that yields a circular orbit. How small a value of ∆t is
needed so that a circular orbit is repeated over many periods? Your answer will depend on
your choice of differential equation solver. Find the largest value of ∆t that yields an orbit that
repeats for many revolutions using the Euler, Euler-Cromer, Verlet, and RK4 algorithms. Is
it possible to choose a smaller value of ∆t, or are some algorithms, such as the Euler method,
simply not stable for this dynamical system?
b. Write a method to compute the total energy (see (5.5)) and compute it at regular intervals as
the system evolves. (It is sufficient to calculate the energy per unit mass, E/m.) For a given
value of ∆t, which algorithm conserves the total energy best? Is it possible to choose a value
of ∆t that conserves the energy exactly? What is the significance of the negative sign for the
total energy?
c. Write a separate method to determine the numerical value of the period. (See Problem 3.9c for
a discussion of a similar condition.) Choose different sets of values of x(t = 0) and vy (t = 0),
consistent with the condition for a circular orbit. For each orbit, determine the radius and the
period and verify Kepler’s third law.

Problem 5.3. Verification of Kepler’s second and third law

a. Set y(t = 0) = 0 and vx (t = 0) = 0 and find by trial and error several values of x(t = 0) and
vy (t = 0) that yield elliptical orbits of a convenient size. Choose a suitable algorithm and plot
the speed of the planet as the orbit evolves. Where is the speed a maximum (minimum)?

b. Use the same initial conditions as in part (a) and compute the total energy, angular momentum,
semimajor and semiminor axes, eccentricity, and period for each orbit. Plot your data for the
dependence of the period T on the semimajor axis a and verify Kepler’s third law. Given the
ratio of T 2 /a3 that you found, determine the numerical value of this ratio in SI units for our
solar system.
c. The force center is at (x, y) = (0, 0) and is one focus. Find the second focus by symmetry.
Compute the sum of the distances from each point on the orbit to the two foci and verify that
the orbit is an ellipse.

d. According to Kepler’s second law, the orbiting object sweeps out equal areas in equal times. If
we use an algorithm with a fixed time step ∆t, it is sufficient to compute the area of the triangle
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 122

swept in each time step. This area equals one-half the base of the triangle times its height, or
2 ∆t (r × v) = 2 ∆t(xvy − yvx ). Is this area a constant? This constant corresponds to what
1 1

physical quantity?

e.∗ Show that algorithms with a fixed value of ∆t break down if the “planet” is too close to the
sun. What is the cause of the failure of the method? What advantage might there be to using
a variable time step? What are the possible disadvantages? (See Project 5.19 for an example
where a variable time step is very useful.)

Problem 5.4. Non-inverse square forces

a. Consider the dynamical effects of a small change in the attractive inverse-square force law, for
example, let the magnitude of the force equal Cm/r2+δ , where δ << 1. For simplicity, take
the numerical value of the constant C to be 4π 2 as before. Consider the initial conditions
x(t = 0) = 1, y(t = 0) = 0, vx (t = 0) = 0, and vy (t = 0) = 5. Choose δ = 0.05 and determine
the nature of the orbit. Does the orbit of the planet retrace itself? Verify that your result is
not due to your choice of ∆t. Does the planet spiral away from or toward the sun? The path
of the planet can be described as an elliptical orbit that slowly rotates or precesses in the same
sense as the motion of the planet. A convenient measure of the precession is the angle between
successive orientations of the semimajor axis of the ellipse. This angle is the rate of precession
per revolution. Estimate the magnitude of this angle for your choice of δ. What is the effect of
decreasing the semimajor axis for fixed δ? What is the effect of changing δ for fixed semimajor
axis?
b. Einstein’s theory of gravitation (the general theory of relativity) predicts a correction to the
force on a planet that varies as 1/r4 due to a weak gravitational field. The result is that the
equation of motion for the trajectory of a particle can be written as
[ ]
d2 r GM ( GM )2 1
=− 2 1+α r̂, (5.21)
dt2 r c2 r2

where the parameter α is dimensionless. Take GM = 4π 2 and assume α = 10−3 . Determine

the nature of the orbit for this potential. (For our solar system the constant α is a maximum
for the planet Mercury, but is much smaller than 10−3 .)

c. Suppose that the attractive gravitational force law depends on the inverse-cube of the distance,
Cm/r3 . What are the units of C? For simplicity, take the numerical value of C to be 4π 2 .
Consider the initial condition x(t = 0) = 1, y(t = 0) = 0, vx (t = 0) = 0, and determine
analytically the value of vy (t = 0) required for a circular orbit. How small a value of ∆t is
needed so that the simulation yields a circular orbit over several periods? How does this value
of ∆t compare with the value needed for the inverse-square force law?
d. Vary vy (t = 0) by approximately 2% from the circular orbit condition that you determined in
part (c). What is the nature of the new orbit? What is the sign of the total energy? Is the
orbit bound? Is it closed? Are all bound orbits closed?
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 123

(a) (b)

Figure 5.4: (a) An impulse applied in the tangential direction. (b) An impulse applied in the radial
direction.

Problem 5.5. Effect of drag resistance on a satellite orbit

Consider a satellite in orbit about the Earth. In this case it is convenient to measure distances in
terms of the radius of the Earth, R = 6.37 × 106 m, and the time in terms of hours. Because the
force on the satellite is proportional to Gm, where m = 5.99 × 1024 kg is the mass of the Earth, we
need to evaluate the product Gm in Earth units (EU). In these units the value of Gm is given by

m3 ( 1 EU )3 ( )2 ( )
Gm = 6.67 × 10−11 3.6 × 103
s/h 5.99 × 1024
kg
kg · s2 6.37 × 106 m
= 20.0 EU3/h2 . (Earth units) (5.22)

Modify the Planet class to incorporate the effects of drag resistance on the motion of an orbiting
Earth satellite. Assume that the drag force is proportional to the square of the speed of the satellite.
To be able to observe the effects of air resistance in a reasonable time, take the magnitude of the
drag force to be approximately one-tenth of the magnitude of the gravitational force. Choose initial
conditions such that a circular orbit would be obtained in the absence of drag resistance and allow
at least one revolution before “switching on” the drag resistance. Describe the qualitative change
of the orbit due to drag resistance. How does the total energy and the speed of the satellite change
with time?

5.7 Impulsive Forces

What happens to the orbit of an Earth satellite when it is hit by space debris? We now discuss
the modifications we need to make in Planet and PlanetApp so that we can apply an impulsive
force (a kick) by a mouse click. If we apply a vertical kick when the position of the satellite is as
shown in Figure 5.4a, the impulse would be tangential to the orbit. A radial kick can be applied
when the satellite is as shown in Figure 5.4b.
User actions, such as mouse clicks or keyboard entries, are passed from the operating system to
Java event listeners. Although this standard Java framework is straightforward, we have simplified
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 124

it to respond to mouse actions within the Open Source Physics panels and frames.1 In order
for an Open Source Physics program to respond to mouse actions, the program implements the
InteractiveMouseHandler interface and then registers its ability to process mouse actions with
the PlotFrame. This procedure is demonstrated in the following test program. You can copy the
handleMouseAction code into your program and replace the print statements with useful methods.
Other mouse actions, such as MOUSE CLICKED, MOUSE MOVED, and MOUSE ENTERED are defined in the
InteractivePanel class.

Listing 5.4: A test program that implements the InteractiveMouseHandler interface.

package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
import j a v a . awt . e v e n t . * ;
import j a v a x . swing . * ;
import o r g . o p e n s o u r c e p h y s i c s . d i s p l a y . * ;
import o r g . o p e n s o u r c e p h y s i c s . f r a m e s . * ;

public c l a s s MouseApp implements I n t e r a c t i v e M o u s e H a n d l e r {

PlotFrame frame = new PlotFrame ( "x" , "y" , " Interactive Handler " ) ;

public MouseApp ( ) {
frame . s e t I n t e r a c t i v e M o u s e H a n d l e r ( t h i s ) ;
frame . s e t V i s i b l e ( true ) ;
frame . s e t D e f a u l t C l o s e O p e r a t i o n ( JFrame . EXIT ON CLOSE ) ;
}

public void handleMouseAction ( I n t e r a c t i v e P a n e l panel , MouseEvent e v t ) {

switch ( p a n e l . getMouseAction ( ) ) {
case I n t e r a c t i v e P a n e l .MOUSE DRAGGED :
p a n e l . s e t M e s s a g e ( " Dragged " ) ;
break ;
case I n t e r a c t i v e P a n e l . MOUSE PRESSED :
p a n e l . s e t M e s s a g e ( " Pressed " ) ;
break ;
case I n t e r a c t i v e P a n e l .MOUSE RELEASED :
p a n e l . s e t M e s s a g e ( null ) ;
break ;
}
}

public s t a t i c void main ( S t r i n g [ ] a r g s ) {

new MouseApp ( ) ;
}
}

The switch statement is used in Listing 5.4 instead of a chain of if statements. The panel’s
getMouseAction method returns an integer. If this integer matches one of the named constants
following the case label, then the statements following that constant are executed until a break
statement is encountered. If a case does not include a break, then the execution continues with
1 See the Open Source Physics User’s Guide for an extensive discussion of interactive drawing panels.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 125

the next case. The equivalent of the else construct in an if statement is default followed by
statements that are executed if none of the explicit cases occur.
We now challenge your intuitive understanding of Newton’s laws of motion by considering
several perturbations of the motion of an orbiting object. Modify your planet program to simulate
the effects of the perturbations in Problem 5.6. In each case answer the questions before doing the
simulation.

Problem 5.6. Tangential and radial perturbations

a. Suppose that a small tangential “kick” or impulsive force is applied to a satellite in a circular
orbit about the Earth (see Figure 5.4a.) Choose Earth units so that the numerical value of
the product Gm is given by (5.22). Apply the impulsive force by stopping the program after
the satellite has made several revolutions and click the mouse to apply the force. Recall that
the impulse changes the momentum in the desired direction directly. In what direction does
the orbit change? Is the orbit stable, for example, does a small impulse lead to a small change
in the orbit? Does the orbit retrace itself indefinitely if no further perturbations are applied?
Describe the shape of the perturbed orbit.
b. How does the change in the orbit depend on the strength of the kick and its duration?

c. Determine if the angular momentum and the total energy are changed by the perturbation.
d. Apply a radial kick to the satellite as in Figure 5.4b and answer the same questions as in parts
(a)–(c).
e. Determine the stability of the inverse-cube force law (see Problem 5.4) to radial and tangential
perturbations.

Mouse actions are not the only possible way to affect the simulation. We also can add custom
buttons to the control. These buttons are added when the program is instantiated in the main
method.
public s t a t i c void main ( S t r i n g [ ] a r g s ) {
// OSPControl i s a s u p e r c l a s s o f S i m u l a t i o n C o n t r o l
OSPControl c o n t r o l = S i m u l a t i o n C o n t r o l . createApp (new PlanetApp ( ) ) ;
c o n t r o l . addButton ( " doRadialKick " , " Kick !" , " Perform a radial kick " ) ;
}

Note that SimulationControl (and CalculationControl) extend the OSPControl superclass and
therefore support the addButton method where this method is defined. We assign the variable
returned by the static createApp method to a variable of type OSPControl to highlight the object
oriented structure of the Open Source Physics library.
The first parameter in the addButton method specifies the method that will be invoked when
the button is clicked, the second parameter specifies the text label that will appear on the button,
and the third parameter specifies the tool tip that will appear when the mouse hovers over the
button. Custom buttons can be used for just about anything, but the corresponding method must
be defined.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 126

Exercise 5.7. Custom buttons

Use a custom button in Problem 5.6 rather than a mouse click to apply an impulsive force to the
planet.

5.8 Velocity Space

In Problem 5.6 your intuition might have been incorrect. For example, you might have thought
that the orbit would elongate in the direction of the kick. In fact the orbit does elongate, but in a
direction perpendicular to the kick. Do not worry, you are in good company! Few students have
a good qualitative understanding of Newton’s law of motion, even after taking an introductory
course in physics. A qualitative way of stating Newton’s second law is

Forces act on the trajectories of particles by changing velocity not position.

If we fail to take into account this property of Newton’s second law, we will encounter physical
situations that appear counterintuitive.
Because force acts to change velocity, it is reasonable to consider both velocity and position
on an equal basis. In fact position and momentum are treated in such a manner in advanced
formulations of classical mechanics and in quantum mechanics.
In Problem 5.8 we explore some of the properties of orbits in velocity space in the context of
the bound motion of a particle in an inverse-square force. Modify your program so that the path
in velocity space of the Earth is plotted. That is, plot the point (vx , vy ) the same way you plotted
the point (x, y). The path in velocity space is a series of successive values of the object’s velocity
vector. If the position space orbit is an ellipse, what is the shape of the orbit in velocity space?

Problem 5.8. Properties of velocity space orbits

a. Modify your program to display the orbit in position space and in velocity space at the same
time. Verify that the velocity space orbit is a circle, even if the orbit in position space is an
ellipse. Does the center of this circle coincide with the origin (vx , vy ) = (0, 0) in velocity space?
Choose the same initial conditions that you considered in Problems 5.2 and 5.3.

b.∗ Let u denote the radius vector of a point on the velocity circle, and w denote the vector from
the origin in velocity space to the center of the velocity circle (see Figure 5.5). Then the
velocity of the particle can be written as

v = u + w. (5.23)

Compute u and verify that its magnitude is given by

u = GM m/L, (5.24)

where L is the magnitude of the angular momentum. Note that L is proportional to m so that
it is not necessary to know the magnitude of m.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 127

vy

w
vx
O

Figure 5.5: The orbit of a particle in velocity space. The vector w points from the origin in velocity
space to the center of the circular orbit. The vector u points from the center of the orbit to the
point (vx , vy ).

c.∗ Verify that at each moment in time, the planet’s position vector r is perpendicular to u.
Explain why this relation holds.

Problem 5.9. Effect of impulses in velocity space

How does the velocity space orbit change when an impulsive kick is applied in the tangential or
in the radial direction? How does the magnitude and direction of w change? From the observed
change in the velocity orbit and the above considerations, explain the observed change of the orbit
in position space.

5.9 A Mini-Solar System

So far our study of planetary orbits has been restricted to two-body central forces. However, the
solar system is not a two-body system, because the planets exert gravitational forces on one another.
Although the interplanetary forces are small in magnitude in comparison to the gravitational force
of the Sun, they can produce measurable effects. For example, the existence of Neptune was
conjectured on the basis of a discrepancy between the experimentally measured orbit of Uranus
and the predicted orbit calculated from the known forces.
The presence of other planets implies that the total force on a given planet is not a central force.
Furthermore, because the orbits of the planets are not exactly in the same plane, an analysis of the
solar system must be extended to three dimensions if accurate calculations are required. However
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 128

r21 m2
m1

r1
r2

x
M

Figure 5.6: The coordinate system used in (5.25). Planets of mass m1 and m2 orbit a sun of mass
M.

for simplicity, we will consider a model of a two-dimensional solar system with two planets in orbit
about a fixed sun.
The equations of motion of two planets of mass m1 and mass m2 can be written in vector
form as (see Figure 5.6)

d2 r1 GM m1 Gm1 m2
m1 2
=− 3
r1 + r21 (5.25a)
dt r1 r21 3
d2 r2 GM m2 Gm1 m2
m2 2 = − r2 − r21 , (5.25b)
dt r2 3 r21 3
where r1 and r2 are directed from the sun to planets 1 and 2 respectively, and r21 = r2 − r1 is the
vector from planet 1 to planet 2. It is convenient to divide (5.25a) by m1 and (5.25b) by m2 and
to write the equations of motion as

d2 r1 GM Gm2
= − 3 r1 + r21 (5.26a)
dt2 r1 r21 3
d2 r2 GM Gm1
= − 3 r2 − r21 . (5.26b)
dt2 r2 r21 3

A numerical solution of (5.26) can be obtained by the straightforward extension of the Planet
class as shown in Listing 5.5. To simplify the drawing of the particle trajectories, the Planet2
class defines an inner class, Mass, which extends Circle and contains a Trail. Whenever a planet
moves, a point is added to the trail so that its location and path are shown on the plot. Inner
classes are an organizational convenience that save us the trouble of having to create another file,
which in this case would be named Mass.java. When we compile the Planet2 class, we will
produce a bytecode file named Planet2$Mass.class in addition to the file Planet2.class. Inner
classes are most effective as short helper classes which work in conjuction with the containing class
because they have access to all the data (including private variables) in the containing class.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 129

Listing 5.5: A class that implements the rate equation for two interacting planets acted on by an
inverse square law force.
package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
import j a v a . awt . * ;
import o r g . o p e n s o u r c e p h y s i c s . d i s p l a y . * ;
import o r g . o p e n s o u r c e p h y s i c s . n u m e r i c s . * ;

public c l a s s P l a n e t 2 implements Drawable , ODE {

// GM i n u n i t s o f (AU) ˆ 3 / ( y r )ˆ2
f i n a l s t a t i c double GM = 4 * Math . PI * Math . PI ;
f i n a l s t a t i c double GM1 = 0 . 0 4 *GM;
f i n a l s t a t i c double GM2 = 0 . 0 0 1 *GM;
double [ ] s t a t e = new double [ 9 ] ;
ODESolver o d e S o l v e r = new RK45MultiStep ( t h i s ) ;
Mass mass1 = new Mass ( ) , mass2 = new Mass ( ) ;

public void doStep ( ) {

odeSolver . step ( ) ;
mass1 . setXY ( s t a t e [ 0 ] , s t a t e [ 2 ] ) ;
mass2 . setXY ( s t a t e [ 4 ] , s t a t e [ 6 ] ) ;
}

public void draw ( DrawingPanel panel , G r a p h i c s g ) {

mass1 . draw ( panel , g ) ;
mass2 . draw ( panel , g ) ;
}

void i n i t i a l i z e ( double [ ] i n i t S t a t e ) {
System . a r r a y c o p y ( i n i t S t a t e , 0 , s t a t e , 0 , i n i t S t a t e . l e n g t h ) ;
mass1 . c l e a r ( ) ; // c l e a r s d a t a from t h e o l d t r a i l
mass2 . c l e a r ( ) ;
mass1 . setXY ( s t a t e [ 0 ] , s t a t e [ 2 ] ) ;
mass2 . setXY ( s t a t e [ 4 ] , s t a t e [ 6 ] ) ;
}

public void g e t R a t e ( double [ ] s t a t e , double [ ] r a t e ) {

// s t a t e [ ] : x1 , vx1 , y1 , vy1 , x2 , vx2 , y2 , vy2 , t
double r 1 S q u a r e d = ( s t a t e [ 0 ] * s t a t e [ 0 ] ) + ( s t a t e [ 2 ] * s t a t e [ 2 ] ) ; // r1 s q u a r e d
double r1Cubed = r 1 S q ua r e d * Math . s q r t ( r 1S qu ar e d ) ; // r1 cubed
double r 2 S q u a r e d = ( s t a t e [ 4 ] * s t a t e [ 4 ] ) + ( s t a t e [ 6 ] * s t a t e [ 6 ] ) ; // r2 s q u a r e d
double r2Cubed = r 2 S q ua r e d * Math . s q r t ( r 2S qu ar e d ) ; // r2 cubed
double dx = s t a t e [4] − s t a t e [ 0 ] ; // x12 s e p a r a t i o n
double dy = s t a t e [6] − s t a t e [ 2 ] ; // y12 s e p a r a t i o n
double dr2 = ( dx * dx )+( dy * dy ) ; // r12 s q u a r e d
double dr3 = Math . s q r t ( dr2 ) * dr2 ; // r12 cubed
rate [ 0 ] = state [ 1 ] ; // x1 r a t e
rate [ 2 ] = state [ 3 ] ; // y1 r a t e
rate [ 4 ] = state [ 5 ] ; // x2 r a t e
rate [ 6 ] = state [ 7 ] ; // y2 r a t e
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 130

rate [1] = (( −GM* s t a t e [ 0 ] ) / r1Cubed )+((GM1* dx ) / dr3 ) ; // vx1 r a t e

rate [3] = (( −GM* s t a t e [ 2 ] ) / r1Cubed )+((GM1* dy ) / dr3 ) ; // vy1 r a t e
rate [5] = (( −GM* s t a t e [ 4 ] ) / r2Cubed ) −((GM2* dx ) / dr3 ) ; // vx2 r a t e
rate [7] = (( −GM* s t a t e [ 6 ] ) / r2Cubed ) −((GM2* dy ) / dr3 ) ; // vy2 r a t e
rate [8] = 1; // time r a t e
}

public double [ ] g e t S t a t e ( ) {
return s t a t e ;
}

c l a s s Mass extends C i r c l e {
T r a i l t r a i l = new T r a i l ( ) ;

public void draw ( DrawingPanel panel , G r a p h i c s g ) {

t r a i l . draw ( panel , g ) ;
super . draw ( panel , g ) ;
}

void c l e a r ( ) {
trail . clear ();
}

public void setXY ( double x , double y ) {

super . setXY ( x , y ) ;
t r a i l . addPoint ( x , y ) ;
}
}
}

The target application, Planet2App, extends AbstractSimulation in the usual way. Because
it is almost identical to Listing 5.2, it is not shown here. The complete program is available in the
ch05 package.

Problem 5.10. Planetary perturbations

Use Planet2App with the initial conditions given in the program. For illustrative purposes, we have
adopted the numerial values m1 /M = 10−3 and m2 /M = 4×10−2 and hence GM1 = (m2 /M )GM =
0.04GM and GM2 = (m1 /M )GM = 0.001GM. What would be the shape of the orbits and the periods
of the two planets if they did not mutually interact? What is the qualitative effect of their mutual
interaction? Describe the shape of the two orbits. Why is one planet affected more by their
mutual interaction than the other? Is the angular momentum and the total energy of planet one
conserved? Is the total energy and total angular momentum of the two planets conserved? A
related, but more time consuming problem is given in Project 5.18.

Problem 5.11. Double stars

Another interesting dynamical system consists of one planet orbiting about two fixed stars of equal
mass. In this case there are no closed orbits, but the orbits can be classified as either stable or
unstable. Stable orbits may be open loops that encircle both stars, figure eights, or orbits that
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 131

b θ
target
2πb db

∝ 2π sinθ |dθ|

Figure 5.7: The coordinate system used to define the differential scattering cross section. Particles
passing through the beam area 2πb db are scattered into the solid angle dΩ.

encircle only one star. Unstable orbits will eventually collide with one of the stars. Modify Planet2
to simulate the double star system, with the first star located at (−1, 0) and the second star of
equal mass located at (1, 0). Place the planet at (0.1, 1) and systematically vary the x and y
components of the velocity to obtain different types of orbits. Then try other initial positions.

5.10 Two-Body Scattering

Much of our understanding of the structure of matter comes from scattering experiments. In this
section we explore one of the more difficult concepts in the theory of scattering, the differential
cross section.
A typical scattering experiment involves a beam with many incident particles all with the
same kinetic energy. The coordinate system is shown in Figure 5.7. The incident particles come
from the left with an initial velocity v in the +x direction. We take the center of the beam and
the center of the target to be on the x axis. The impact parameter b is the perpendicular distance
from the initial trajectory to a parallel line through the center of the target (see Figure 5.7). We
assume that the width of the beam is larger than the size of the target. The target contains many
scattering centers, but for calculational purposes we may consider scattering off only one particle
if the target is sufficiently thin.
When an incident particle comes close to the target, it is deflected. In a typical experiment,
the scattered particles are counted in a detector that is far from the target. The final velocity of
the scattered particles is v′ , and the angle between v and v′ is the scattering angle θ.
Let us assume that the scattering is elastic and that the target is much more massive than
the beam particles so that the target can be considered to be fixed. (The latter condition can be
relaxed by using center of mass coordinates.) We also assume that no incident particle is scattered
more than once. These considerations imply that the initial speed and final speed of the incident
particles are equal. The functional dependence of θ on b depends on the force on the beam particles
due to the target. In a typical experiment the number of particles in an angular region between
θ and θ + dθ is detected for many values of θ. These detectors measure the number of particles
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 132

scattered into the solid angle dΩ = sin θdθdϕ centered about θ. The differential cross section σ(θ)
is defined by the relation
dN
= nσ(θ)dΩ, (5.27)
N
where dN is the number of particles scattered into the solid angle dΩ centered about θ and the
azimuthal angle ϕ, N is the total number of particles in the beam, and n is the target density
defined as the number of targets per unit area.
The interpretation of (5.27) is that the fraction of particles scattered into the solid angle dΩ is
proportional to dΩ and the density of the target. From (5.27) we see that σ(θ) can be interpreted
as the effective area of a target particle for the scattering of an incident particle into the element
of solid angle dΩ. Particles that are not scattered are ignored. Another way of thinking about
σ(θ) is that it is the ratio of the area b db dϕ to the solid angle dΩ = sin θdθ dϕ, where b db dϕ is
the infinitesimal cross sectional area of the beam that scatters into the solid angle defined by θ to
θ + dθ and ϕ to ϕ + dϕ. The alternative notation for the differential cross section, dσ/dΩ, comes
from this interpretation.
To do an analytical calculation of σ(θ), we write
 
dσ b  db 
σ(θ) = = . (5.28)
dΩ sin θ  dθ 

We see from (5.28) that the analytical calculation of σ(θ) involves b as a function of θ, or more
precisely, how b changes to give scattering through an infinitesimally larger angle θ + dθ.
In a scattering experiment particles enter from the left (see Figure 5.7) with random values
of the impact parameter b and azimuthal angle ϕ and the number of particles scattered into the
various detectors is measured. In our simulation we know the value of b, and we can integrate
Newton’s equations of motion to find the angle at which the incident particle is scattered. Hence,
in contrast to the analytical calculation, a simulation naturally yields θ as a function of b.
Because the differential cross section is usually independent of ϕ, we need to consider only
beam particles at ϕ = 0. We have to take into account the fact that in a real beam, there are more
particles at some values of b than at others. That is, the number of particles in a real beam is
proportional to 2πb∆b, the area of the ring between b and b + ∆b, where we have integrated over
the values of ϕ to obtain the factor of 2π. Here ∆b is the interval between the values of b used in
the program. Because there is only one target in the beam, the target density is n = 1/(πR2 ).
The scattering program requires the Scatter, ScatterAnalysis, and ScatterApp classes.
The ScatterApp class in Listing 5.6 organizes the startup process and creates the visualizations.
As usual, it extends AbstractSimulation by overriding the doStep method. However, in this case
a single step is not a time step. A step calculates a trajectory and scattering angle for the given
impact parameter. After a trajectory is calculated, the impact parameter is incremented and the
panel is repainted. If necessary, you can eliminate this visualization to increase the computational
speed. If the new impact parameter exceeds the beam radius bmax, the animation is stopped and
the accumulated data is analyzed. Note that the calculateTrajectory method returns true if
the calculation succeeded and that an error message is printed if the calculation fails. Including a
failsafe mechanism to stop a computation is good programming practice.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 133

Listing 5.6: A program that calculates the scattering trajectories and computes the differential
cross section.
package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
import o r g . o p e n s o u r c e p h y s i c s . c o n t r o l s . * ;
import o r g . o p e n s o u r c e p h y s i c s . f r a m e s . * ;

public c l a s s ScatterApp extends A b s t r a c t S i m u l a t i o n {

PlotFrame frame = new PlotFrame ( "x" , "y" , " Trajectories " ) ;
S c a t t e r A n a l y s i s a n a l y s i s = new S c a t t e r A n a l y s i s ( ) ;
S c a t t e r t r a j e c t o r y = new S c a t t e r ( ) ;
double vx ; // s p e e d o f t h e i n c i d e n t p a r t i c l e
double b , db ; // impact parameter and i n c r e m e n t
double bmax ; // maximum impact parameter

public ScatterApp ( ) {
frame . setPreferredMinMax ( −5 , 5 , −5, 5 ) ;
frame . s e t S q u a r e A s p e c t ( true ) ;
}

public void doStep ( ) {

i f ( t r a j e c t o r y . c a l c u l a t e T r a j e c t o r y ( frame , b , vx ) ) {
a n a l y s i s . d e t e c t P a r t i c l e (b , t r a j e c t o r y . getAngle ( ) ) ;
} else {
c o n t r o l . p r i n t l n ( " Trajectory did not converge at b = "+b ) ;
}
frame . s e t M e s s a g e ( "b = "+decimalFormat . format ( b ) ) ;
b += db ; // i n c r e a s e s t h e impact parameter
frame . r e p a i n t ( ) ;
i f ( b>bmax ) {
c o n t r o l . c a l c u l a t i o n D o n e ( " Maximum impact parameter reached " ) ;
analysis . plotCrossSection (b ) ;
}
}

public void i n i t i a l i z e ( ) {
vx = c o n t r o l . g e tD ou bl e ( " vx " ) ;
bmax = c o n t r o l . g e tD ou ble ( " bmax " ) ;
db = c o n t r o l . g et D ou bl e ( " db " ) ;
b = db / 2 ; // s t a r t s b a t a v e r a g e v a l u e o f f i r s t i n t e r v a l 0−>db
// b w i l l i n c r e m e n t t o 3 * db /2 , 5 * db /2 , 7 * db /2 , . . .
frame . s e t M e s s a g e ( "b = 0" ) ;
frame . c l e a r D r a w a b l e s ( ) ; // removes o l d t r a j e c t o r i e s
analysis . clear ();
}

public void r e s e t ( ) {
c o n t r o l . s e t V a l u e ( " vx " , 3 ) ;
c o n t r o l . s e t V a l u e ( " bmax " , 0 . 2 5 ) ;
c o n t r o l . s e t V a l u e ( " db " , 0 . 0 1 ) ;
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 134

initialize ();
}

public s t a t i c void main ( S t r i n g [ ] a r g s ) {

S i m u l a t i o n C o n t r o l . createApp (new ScatterApp ( ) ) ;
}
}

The Scatter class shown in Listing 5.7 calculates the trajectories by expressing the equation of
motion as a rate equation. The most important method is calculateTrajectory, which calculates
a trajectory by stepping the differential equation solver and adding the resulting data to a trail to
display the path. Because the beam source is far away, we stop the calculation when the distance
of the scattered particle from the target exceeds the initial distance. Note the use of the ternary
?: operator. This very efficient and compact, operator uses three expressions. The first expression
evaluates to a boolean. If this expression is true, then the statement after the ? is executed. If this
expression is false, then the statement after the : is executed. However, because some potentials
may trap particles for long periods of time, we also stop the calculation after a predetermined
number of time steps.

Listing 5.7: A class that models particle scattering using a central force law.
package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
import j a v a . awt . * ;
import o r g . o p e n s o u r c e p h y s i c s . d i s p l a y . * ;
import o r g . o p e n s o u r c e p h y s i c s . f r a m e s . * ;
import o r g . o p e n s o u r c e p h y s i c s . n u m e r i c s . * ;

public c l a s s S c a t t e r implements ODE {

double [ ] s t a t e = new double [ 5 ] ;
RK4 o d e S o l v e r = new RK4( t h i s ) ;

public S c a t t e r ( ) {
odeSolver . setStepSize ( 0 . 0 5 ) ;
}

boolean c a l c u l a t e T r a j e c t o r y ( PlotFrame frame , double b , double vx ) {

s t a t e [ 0 ] = −5.0; // x
s t a t e [ 1 ] = vx ; // vx
state [2] = b; // y
state [ 3 ] = 0; // vy
state [ 4 ] = 0; // ti m e
T r a i l t r a i l = new T r a i l ( ) ;
t r a i l . c o l o r = Color . red ;
frame . addDrawable ( t r a i l ) ;
double r 2 = ( s t a t e [ 0 ] * s t a t e [ 0 ] ) + ( s t a t e [ 2 ] * s t a t e [ 2 ] ) ;
double count = 0 ;
while ( ( count <=1000)&&((2 * r 2 ) >(( s t a t e [ 0 ] * s t a t e [ 0 ] ) + ( s t a t e [ 2 ] * s t a t e [ 2 ] ) ) ) ) {
t r a i l . addPoint ( s t a t e [ 0 ] , s t a t e [ 2 ] ) ;
odeSolver . step ( ) ;
count++;
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 135

}
return count <1000;
}

private double f o r c e ( double r ) {

// Coulomb f o r c e law
return ( r==0) ? 0 : ( 1 / r / r ) ; // r e t u r n s 0 i f r = 0
}

public void g e t R a t e ( double [ ] s t a t e , double [ ] r a t e ) {

double r = Math . s q r t ( ( s t a t e [ 0 ] * s t a t e [ 0 ] ) + ( s t a t e [ 2 ] * s t a t e [ 2 ] ) ) ;
double f = f o r c e ( r ) ;
rate [ 0 ] = state [ 1 ] ;
rate [ 1 ] = ( f * state [ 0 ] ) / r ;
rate [ 2 ] = state [ 3 ] ;
rate [ 3 ] = ( f * state [ 2 ] ) / r ;
rate [ 4 ] = 1;
}

public double [ ] g e t S t a t e ( ) {
return s t a t e ;
}

double g e t A n g l e ( ) {
return Math . atan2 ( s t a t e [ 3 ] , s t a t e [ 1 ] ) / Math . PI ;
}
}

The ScatterAnalysis class performs the data analysis. This class creates an array of bins to
sort and accumulate the trajectories according to the scattering angle. The values of the scattering
angle between 0 and 180◦ are divided into bins of width dtheta. To compute the number of particles
coming from a ring of radius b, we accumulate the value of b associated with each bin or “detector”
and write bins[index] += b (see the detectParticle method), because the number of particles
in a ring of radius b is proportional to b. The total number of scattered particles is computed in
the same way
t o t a l N += b ;

You might want to increase the number of bins and the range of angles for better resolution.

Listing 5.8: The ScatterAnalysis class accumulates the scattering data and plots the differential
cross section.
package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
import o r g . o p e n s o u r c e p h y s i c s . f r a m e s . PlotFrame ;

public c l a s s S c a t t e r A n a l y s i s {
in t numBins = 2 0 ;
PlotFrame frame = new PlotFrame ( " angle " , " sigma " , " differential cross section " ) ;
double [ ] b i n s = new double [ numBins ] ;
double d t h e t a = Math . PI / ( numBins −1);
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 136

double t o t a l N = 0 ; // t o t a l number o f s c a t t e r e d p a r t i c l e s

void c l e a r ( ) {
f o r ( i n t i = 0 ; i <numBins ; i ++) {
bins [ i ] = 0;
}
totalN = 0;
frame . c l e a r D a t a ( ) ;
frame . r e p a i n t ( ) ;
}

void d e t e c t P a r t i c l e ( double b , double t h e t a ) {

t h e t a = Math . abs ( t h e t a ) ; // t r e a t s p o s i t i v e and n e g a t i v e a n g l e s e q u a l l y t o g e t b e t t e r s
int index = ( int ) ( theta / dtheta ) ;
b i n s [ i n d e x ] += b ;
t o t a l N += b ;
}

void p l o t C r o s s S e c t i o n ( double r a d i u s ) {
double t a r g e t D e n s i t y = 1/Math . PI / r a d i u s / r a d i u s ;
double d e l t a = ( d t h e t a * 1 8 0 ) / Math . PI ; // u s e s d e g r e e s f o r p l o t
frame . c l e a r D a t a ( ) ;
f o r ( i n t i = 0 ; i <numBins ; i ++) {
double domega = 2 * Math . PI * Math . s i n ( ( i +0.5) * d t h e t a ) * d t h e t a ;
double sigma = b i n s [ i ] / t o t a l N / t a r g e t D e n s i t y /domega ;
frame . append ( 0 , ( i +0.5) * d e l t a , sigma ) ;
}
frame . s e t V i s i b l e ( true ) ;
}
}

Problem 5.12. Total cross section

The total cross section σT is defined as
∫
σT = σ(θ) dΩ. (5.29)

Add code to calculate and display the total cross section in the plotCrossSection method. Design
a test to verify that the ODE solver in the Scatter class has sufficient accuracy.

In Problem 5.13, we consider a model of the hydrogen atom for which the force on a beam
particle is zero for r > a. Because we do not count the beam particles that are not scattered,
we set the beam radius equal to a. For forces that are not identically zero, we need to choose a
minimum angle for θ such that particles whose scattering angle is less than this minimum are not
counted as scattered (see Problem 5.14).
Problem 5.13. Scattering from a model hydrogen atom

a. Consider a model of the hydrogen atom for which a positively charged nucleus of charge +e
is surrounded by a uniformly distributed negative charge of equal magnitude. The spherically
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 137

symmetric negative charge distribution is contained within a sphere of radius a. It is straight-

forward to show that the force between a positron of charge +e and this model hydrogen atom
is given by {
1/r2 − r/a3 , r ≤ a
f (r) = (5.30)
0. r>a

We have chosen units such that e2 /(4πϵ0 ) = 1, and the mass of the positron is unity. What
is the ionization energy in these units? Modify the Scatter class to incorporate this force. Is
the force on the positron from the model hydrogen atom purely repulsive? Choose a = 1 and
set the beam radius bmax = 1. Use E = 0.125 and ∆t = 0.01. Compute the trajectories for
b = 0.25, 0.5, and 0.75 and describe the qualitative nature of the trajectories.

b. Determine the cross section for E = 0.125. Choose nine bins so that the angular width of a
detector is delta = 20◦ and let db = 0.1, 0.01, and 0.002. How does the accuracy of your results
depend on the number of bins? Determine the differential cross section for different energies
and explain its qualitative energy dependence.

c. What is the value of σT for E = 0.125? Does σT depend on E? The total cross section has
units of area, but a point charge does not have an area. To what area does it refer? What
would you expect the total cross section to be for scattering from a hard sphere?
d. Change the sign of the force so that it corresponds to electron scattering. How do the trajectories
change? Discuss the change in σ(θ).

Problem 5.14. Rutherford scattering

a. One of the most famous scattering experiments was performed by Geiger and Marsden who
scattered a beam of alpha particles on a thin gold foil. Based on these experiments, Rutherford
deduced that the positive charge of the atom is concentrated in a small region at the center
of the atom rather than distributed uniformly over the entire atom. Use a 1/r2 force in class
Scatter and compute the trajectories for b = 0.25, 0.5, and 0.75 and describe the trajectories.
Choose E = 5 and ∆t = 0.01. The default value of x0 , the initial x-coordinate of the beam, is
x0 = −5. Is this value reasonable?

b. For E = 5 determine the cross section with numberOfBins = 18. Choose the beam width
bmax = 2. Then vary db (or numberOfBins) and compare the accuracy of your results to the
analytical result for which σ(θ) varies as [sin(θ/2)]−4 . How do your computed results compare
with this dependence on θ? If necessary, decrease db. Are your results better or worse at small
angles, intermediate angles, or large angles near 180◦ ? Explain.
c. Because the Coulomb force is long range, there is scattering at all impact parameters. Increase
the beam radius and determine if your results for σ(θ) change. What happens to the total cross
section as you increase the beam width?

d. Compute σ(θ) for different values of E and estimate the dependence of σ(θ) on E.

Problem 5.15. Scattering by other potentials

CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 138

a. A simple phenomenological form for the effective interaction between electrons in metals is the
screened Coulomb (or Thomas-Fermi) potential given by

e2 −r/a
V (r) = e . (5.31)
4πϵ0 r
The range of the interaction a depends on the density and temperature of the electrons. The
form (5.31) is known as the Yukawa potential in the context of the interaction between nuclear
particles and as the Debye potential in the context of classical plasmas. Choose units such that
a = 1 and e2 /(4πϵ0 ) = 1. Recall that the force is given by f (r) = −dV /dr. Incorporate this
force law into class Scatter and compute the dependence of σ(θ) on the energy of the incident
particle. Choose the beam width equal to 3. Compare your results for σ(θ) with your results
from the Coulomb potential.
b. Modify the force law in Scatter so that f (r) = 24(2/r13 − 1/r7 ). This form for f (r) is
used to describe the interactions between simple molecules (see Chapter 9). Describe some
typical trajectories and compute the differential cross section for several different energies. Let
bmax = 2. What is the total cross section? How do your results change if you vary bmax?
Choose a small angle as the minimum scattering angle. How sensitive is the total cross section
to this minimum angle? Does the differential cross section vary for any other angles beside the
smallest scattering angle?

5.11 Three-body problems

Poincaré showed that it is impossible to obtain an analytical solution for the unrestricted motion
of three or more objects interacting under the influence of gravity. However solutions are known
for a few special cases, and it is instructive to study the properties of these solutions.
The ThreeBody class computes the trajectories of three particles of equal mass moving in a
plane and interacting under the influence of gravity. Both the physics and the drawing are imple-
mented in the ThreeBody class shown in Listing 5.9. Note that the getRate and computeForce
methods compute trajectories for an arbitrary number of masses. Note how the computeForce
method uses the arraycopy method to quickly zero the arrays. To simplify the drawing of the
particle trajectories, the ThreeBody class uses an inner class that extends a Circle and contains
a Trail.

Listing 5.9: A class that models the dynamics of the three-body problem.
package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
import j a v a . awt . * ;
import o r g . o p e n s o u r c e p h y s i c s . d i s p l a y . * ;
import o r g . o p e n s o u r c e p h y s i c s . n u m e r i c s . * ;

public c l a s s ThreeBody implements Drawable , ODE {

in t n = 3 ; // number o f i n t e r a c t i n g b o d i e s
// s t a t e= { x1 , vx1 , y1 , vy1 , x2 , vx2 , y2 , vy2 , x3 , vx3 , y3 , vy3 , t }

double [ ] s t a t e = new double [ 4 * n + 1 ] ;

CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 139

double [ ]
f o r c e = new double [ 2 * n ] , z e r o s = new double [ 2 * n ] ;
ODESolver o d e S o l v e r = new RK45MultiStep ( t h i s ) ;
Mass mass1 = new Mass ( ) , mass2 = new Mass ( ) , mass3 = new Mass ( ) ;

public void draw ( DrawingPanel panel , G r a p h i c s g ) {

mass1 . draw ( panel , g ) ;
mass2 . draw ( panel , g ) ;
mass3 . draw ( panel , g ) ;
}

public void doStep ( ) {

odeSolver . step ( ) ;
mass1 . setXY ( s t a t e [0] , state [ 2 ] ) ;
mass2 . setXY ( s t a t e [4] , state [ 6 ] ) ;
mass3 . setXY ( s t a t e [8] , state [ 1 0 ] ) ;
}

void i n i t i a l i z e ( double [ ] i n i t S t a t e ) {
System . a r r a y c o p y ( i n i t S t a t e , 0 , s t a t e , 0 , 1 3 ) ; // c o p i e s i n i t s t a t e t o s t a t e
mass1 . c l e a r ( ) ; // c l e a r s d a t a from o l d t r a i l
mass2 . c l e a r ( ) ;
mass3 . c l e a r ( ) ;
mass1 . setXY ( s t a t e [ 0 ] , s t a t e [ 2 ] ) ;
mass2 . setXY ( s t a t e [ 4 ] , s t a t e [ 6 ] ) ;
mass3 . setXY ( s t a t e [ 8 ] , s t a t e [ 1 0 ] ) ;
}

void computeForce ( double [ ] s t a t e ) {

System . a r r a y c o p y ( z e r o s , 0 , f o r c e , 0 , f o r c e . l e n g t h ) ; // s e t s f o r c e a r r a y e l e m e n t s t o 0
f o r ( i n t i = 0 ; i <n ; i ++) {
f o r ( i n t j = i +1; j <n ; j ++) {
double dx = s t a t e [ 4 * i ]− s t a t e [ 4 * j ] ;
double dy = s t a t e [ 4 * i +2]− s t a t e [ 4 * j + 2 ] ;
double r 2 = dx * dx+dy * dy ;
double r 3 = r 2 * Math . s q r t ( r 2 ) ;
double f x = dx/ r 3 ;
double f y = dy/ r 3 ;
f o r c e [ 2 * i ] −= f x ;
f o r c e [ 2 * i +1] −= f y ;
f o r c e [ 2 * j ] += f x ;
f o r c e [ 2 * j +1] += f y ;
}
}
}

public void g e t R a t e ( double [ ] s t a t e , double [ ] r a t e ) {

computeForce ( s t a t e ) ; // f o r c e a r r a y a l t e r n a t e s f x and f y
f o r ( i n t i = 0 ; i <n ; i ++) {
in t i 4 = 4 * i ;
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 140

rate [ i4 ] = s t a t e [ i4 +1]; // x rate i s vx

rate [ i 4 +1] = force [2* i ] ; // vx r a t e is fx
rate [ i 4 +2] = s t a t e [ i4 +3]; // y rate i s vy
rate [ i 4 +3] = f o r c e [2* i +1]; // vy r a t e is fy
}
r a t e [ s t a t e . l e n g t h −1] = 1 ; // t ime r a t e i s l a s t
}

public double [ ] g e t S t a t e ( ) {
return s t a t e ;
}

c l a s s Mass extends C i r c l e {
T r a i l t r a i l = new T r a i l ( ) ;
// Draws t h e mass .

public void draw ( DrawingPanel panel , G r a p h i c s g ) {

t r a i l . draw ( panel , g ) ;
super . draw ( panel , g ) ;
}

// C l e a r s t r a i l
void c l e a r ( ) {
trail . clear ();
}

// S e t s p o s t i o n and adds t o t r a i l
public void setXY ( double x , double y ) {
super . setXY ( x , y ) ;
t r a i l . addPoint ( x , y ) ;
}
}
}

The initial conditions for our examples are contained in the ThreeBodyInitialConditions
class. This file is available in the ch05 package but is not listed here because it contains mostly
numeric data.
In 1765 Euler discovered an analytical solution in which three masses start on a line and rotate
so that the central mass stays fixed. The EULER array in ThreeBodyInitialConditions initializes
the model to produce this type of solution. The first mass is placed at the center and the other
two masses are placed on opposite sides with velocities that are equal but opposite. Because of
the symmetry, the trajectories are ellipses with a common focus at the center.
A second analytic solution to the unrestricted three-body problem was found by Lagrange in
1772. This solution starts with three masses at the corners of an equilateral triangle. Each mass
moves in an ellipse in such a way that the triangle formed by the messes remains equilateral. The
LAGRANGE array initializes this solution.
A spectacular new solution that adds to the sparse list of analytic three-body solutions was
first discovered numerically by Moore and proven to be stable by Chenciner and Montgomery. The
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 141

MONTGOMERY array contains the initial conditions for this solution.

The ThreeBodyApp class in Listing 5.10 is the target class for the three-body program. The
doStep method merely increments the model’s differential equations solver and repaints the view.

Listing 5.10: A program that displays the trajectories of three bodies interacting via gravitational
forces.
package o r g . o p e n s o u r c e p h y s i c s . s i p . ch05 ;
import o r g . o p e n s o u r c e p h y s i c s . c o n t r o l s . * ;
import o r g . o p e n s o u r c e p h y s i c s . f r a m e s . * ;

public c l a s s ThreeBodyApp extends A b s t r a c t S i m u l a t i o n {

PlotFrame frame = new PlotFrame ( "x" , "y" , " Three - Body Orbits " ) ;
ThreeBody t r a j e c t o r y = new ThreeBody ( ) ;

public ThreeBodyApp ( ) {
frame . addDrawable ( t r a j e c t o r y ) ;
frame . s e t S q u a r e A s p e c t ( true ) ;
frame . s e t S i z e ( 4 5 0 , 4 5 0 ) ;
}

public void i n i t i a l i z e ( ) {
t r a j e c t o r y . o d e S o l v e r . s e t S t e p S i z e ( c o n t r o l . getDouble ( " dt " ) ) ;
t r a j e c t o r y . i n i t i a l i z e ( T h r e e B o d y I n i t i a l C o n d i t i o n s .MONTGOMERY) ;
frame . setPreferredMinMax ( −1.5 , 1 . 5 , −1.5 , 1 . 5 ) ;
}

public void r e s e t ( ) {
c o n t r o l . s e t V a l u e ( " dt " , 0 . 1 ) ;
e n a b l e S t e p s P e r D i s p l a y ( true ) ;
initialize ();
}

protected void doStep ( ) {

t r a j e c t o r y . doStep ( ) ;
frame . s e t M e s s a g e ( "t="+decimalFormat . format ( t r a j e c t o r y . s t a t e [ 4 ] ) ) ;
}

public s t a t i c void main ( S t r i n g [ ] a r g s ) {

S i m u l a t i o n C o n t r o l . createApp (new ThreeBodyApp ( ) ) ;
}
}

Problem 5.16. Stability of solutions to the three-body problem

Examine the stability of the three solutions to the three-body problem by slightly varying the initial
velocity of one of the masses. Before passing your new initial state to trajectory.initialize,
calculate the center of mass velocity and subtract this velocity from every object. Show that any
instability is due to the physics and not to the numerical differential equation solver. Which of the
three analytic solutions is stable? Check conservation of the total energy and angular momentum.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 142

5.12 Projects
Project 5.17. Effect of a “solar wind”

a. Assume that a satellite is affected not only by the Earth’s gravitational force, but also by a
weak uniform “solar wind” of magnitude W acting in the horizontal direction. The equations
of motion can be written as
d2 x GM x
=− 3 +W (5.32a)
dt2 r
d2 y GM y
=− 3 . (5.32b)
dt2 r
Choose initial conditions so that a circular orbit would be obtained for W = 0. Then choose a
value of W whose magnitude is about 3% of the acceleration due to the gravitational field and
compute the orbit. How does the orbit change?
b. Determine the change in the velocity space orbit when the solar wind (5.32) is applied. How
does the total angular momentum and energy change? Explain in simple terms the previously
observed change in the position space orbit. See Luehrmann for further discussion of this
problem.

Project 5.18. Resonances and the asteroid belt

a. A histogram of the number of asteroids versus their distance from the Sun shows some distinct
gaps. These gaps, called the Kirkwood gaps, are due to resonance effects. That is, if asteroids
were in these gaps, their periods would be simple fractions of the period of Jupiter. Modify class
Planet2 so that planet two has the mass of Jupiter by setting GM1 = 0.001*GM. Because the
asteroid masses are very small compared to that of Jupiter, the gravitational force on Jupiter
due to the asteroids can be neglected. The initial conditions listed in Planet2 are approximately
correct for Jupiter. The initial conditions for the asteroid (planet one in Planet2) correspond
to the 1/3 resonance (the period of the asteroid is one third that of Jupiter). Run the program
with these changes and describe the orbit of the asteroid.
b. Use Kepler’s third law, T 2/a3 = constant, to determine the values of a, the asteroid’s semimajor
axis, such that the ratio of its period of revolution about the Sun to that of Jupiter is 1/2, 3/7,
2/5, and 2/3. Set the initial value of x(1) equal to a for each of these ratios and choose the
initial value of vy(1) so that the asteroid would have a circular orbit if Jupiter were not present.
Describe the orbits that you obtain.
c. It is instructive to plot a as a function of time. However, because it is not straightforward to
measure a directly in the simulation, it is more convenient to plot the quantity −2GM m/E,
where E is the total energy of the asteroid and m is the mass of the asteroid. Because E is
proportional to m, the quantity −2GM m/E is independent of m. If the interaction of the
asteroid with Jupiter is ignored, it can be shown that a = −2GM m/E, where E is the asteroid
kinetic energy plus the asteroid-Sun potential energy. Derive this result for circular orbits. Plot
the quantity −2GM m/E versus time for about thirty revolutions for the initial conditions in
Problem 5.18b.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 143

4.0

2.0

0.0

-2.0

-4.0
-4.0 -2.0 0.0 2.0 4.0

Figure 5.8: Orbits of the two electrons in the classical helium atom with the initial condition
r1 = (3, 0), r2 = (1, 0), v1 = (0, 0.4), and v2 = (0, −1) (see Project 5.19c).

d. Compute the time dependence of −2GM m/E for asteroid orbits whose initial position x(1)
ranges from 2.0 to 5.0 in steps of 0.2. Choose the initial values of vy(1) so that circular orbits
would be obtained in the absence of Jupiter. Are there any values of x(1) for which the time
dependence of a is unusual?
e. Make a histogram of the number of asteroids versus the value of −2GM m/E at t = 2000. (You
can use the HistogramFrame class described on page 227 if you wish.) Assume that the initial
value of x(1) ranges from 2.0 to 5.0 in steps of 0.02 and choose the initial values of vy(1) as
before. Use a histogram bin width of 0.1. If you have time, repeat for t = 5000, and compare the
histogram with your previous results. Is there any evidence for Kirkwood gaps? A resonance
occurs when the periods of the asteroid and Jupiter are related by simple fractions. We expect
the number of asteroids with values of a corresponding to resonances to be small.

f. Repeat part (e) with initial velocities that vary from their values for a circular orbit by 1, 3,
and 5%.

Project 5.19. The classical helium atom

The classical helium atom is a relatively simple example of a three-body problem and is similar to
the gravitational three-body problem of a heavy sun and two light planets. The important differ-
ence is that the two electrons repel one another, unlike the planetary case where the intraplanetary
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 144

interaction is attractive. If we ignore the small motion of the heavy nucleus, the equations of motion
for the two electrons can be written as
r1 r1 − r2
a1 = −2 3 + 3 (5.33a)
r1 r12
r2 r2 − r1
a2 = −2 3 + 3 , (5.33b)
r2 r12

where r1 and r2 are measured from the fixed nucleus at the origin, and r12 is the distance between
the two electrons. We have chosen units such that the mass and charge of the electron are both
unity. The charge of the helium nucleus is two in these units. Because the electrons are sometimes
very close to the nucleus, their acceleration can become very large, and a very small time step
∆t is required. It is not efficient to use the same small time step throughout the simulation and
instead a variable time step or an adaptive step size algorithm is suggested. An adaptive step size
algorithm can be used with any standard numerical algorithm for solving differential equations.
The RK45 algorithm described in Appendix 3A is adaptive and is a good all-around choice for
these types of problems.

a. For simplicity, we restrict our atom to two dimensions. Modify Planet2 to simulate the classical
helium atom. Choose units such that the electron mass is one and the other constants are
absorbed into the unit of charge so that the force between two electrons is
1
|F | = . (5.34)
r2
Choose the initial value of the time step to be ∆t = 0.001. Some of the possible orbits are
similar to those we have seen in our mini-solar system. For example, try the initial condition
r1 = (2, 0), r2 = (−1, 0), v1 = (0, 0.95), and v2 = (0, −1).

b. Most initial conditions result in unstable orbits in which one electron eventually leaves the
atom (autoionization). The initial condition r1 = (1.4, 0), r2 = (−1, 0), v1 = (0, 0.86), and
v2 = (0, −1) gives “braiding” orbits. Make small changes in this initial condition to observe
autoionization.

c. The classical helium atom is capable of very complex orbits (see Figure 5.8). Investigate the
motion for the initial condition r1 = (3, 0), r2 = (1, 0), v1 = (0, 0.4), and v2 = (0, −1). Does the
motion conserve the total angular momentum? Also try r1 = (2.5, 0), r2 = (1, 0), v1 = (0, 0.4),
and v2 = (0, −1).

d. Choose the initial condition r1 = (2, 0), r2 = (−1, 0), and v2 = (0, −1). Then vary the initial
value of v1 from (0.6, 0) to (1.3, 0) in steps of ∆v = 0.02. For each set of initial conditions
calculate the time it takes for autoionization. Assume that ionization occurs when either electron
exceeds a distance of six from the nucleus. Run each simulation for a maximum time of 2000.
Plot the ionization time versus v1x . Repeat for a smaller interval of ∆v centered about one of
the longer ionization times. These calculations require much computer resources. Do the two
plots look similar? If so, such behavior is called “self-similar” and is characteristic of chaotic
systems and the geometry of fractals (see Chapters 7 and 14). More discussion on the nature
of the orbits can be found in Yamamoto and Kaneko.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 145

References and Suggestions for Further Reading

Harold Abelson, Andrea diSessa, and Lee Rudolph, “Velocity space and the geometry of plan-
etary orbits,” Am. J. Phys. 43, 579–589 (1975). See also Andrea diSessa, “Orbit: a mini-
environment for exploring orbital mechanics,” in O. Lecarme and R. Lewis, editors, Comput-
ers in Education, 359, North-Holland (1975). Detailed geometrical rather than calculus-based
arguments on the origin of closed orbits for inverse-square forces are presented. Sections 5.7
and 5.8 are based on these papers.

Ralph Baierlein, Newtonian Dynamics, McGraw-Hill (1983). An intermediate level text on me-
chanics. Of particular interest are the discussions on the stability of circular orbits and the
effects of an oblate sun.
John J. Brehm and William J. Mullin, Introduction to the Structure of Matter, John Wiley &
Sons (1989). See Section 3-4 for a discussion of Rutherford scattering.
Alain Chenciner and Richard Montgomery, “A remarkable periodic solution of the three-body
problem in the case of equal masses,” Annals of Mathematics 152, 881–901 (2000).
J. M. A. Danby, Computer Modeling: From Sports To Spaceflight . . . From Order To Chaos,
William-Bell (1997). See Chapter 11 for a discussion of orbits including an excellent treatment
of the Lagrange points.

R. P. Feynman, R. B. Leighton, M. Sands, The Feynman Lectures in Physics, Vol. 1, Addison-

Wesley (1963). See Chapter 9.
A. P. French, Newtonian Mechanics, W. W. Norton & Company (1971). An introductory level
text with more than a cursory treatment of planetary motion.
Ian R. Gatland, “Numerical integration of Newton’s equations including velocity-dependent forces,”
Am J. Phys. 62, 259 (1994). The author chooses a variable time step based on the difference
in the calculation of the positions rather than the energy as we did in Project 5.19.

Herbert Goldstein, Charles P. Poole, and John L. Safko, Classical Mechanics, third edition,
Addison-Wesley (2002). Chapter 3 has an excellent discussion of the Kepler problem and
the conditions for a closed orbit.
Myron Lecar and Fred A. Franklin, “On the original distribution of the asteroids. I,” Icarus 20,
422–436 (1973). The authors use simulations of the motions of asteroids and discuss the
Kirkwood gaps.

Arthur W. Luehrmann, “Orbits in the solar wind – a mini-research problem,” Am. J. Phys. 42,
361 (1974). Luehrmann emphasizes the desirability of student problems requiring inductive
rather than deductive reasoning.
Jerry B. Marion and Stephen T. Thornton, Classical Dynamics, fifth edition, Harcourt (2004).
Chapter 8 discusses central force motion, the precession of the Mercury, and the stability of
circular orbits.
CHAPTER 5. FEW-BODY PROBLEMS: THE MOTION OF THE PLANETS 146

Michael McCloskey, “Intuitive physics,” Sci. Am. 248 (4), 122–130 (1983). A discussion of the
counterintuitive nature of Newton’s laws.

John R. Merrill and Richard A. Morrow, “An introductory scattering experiment by simulation,”
Am. J. Phys. 38, 1104–1107 (1970).

C. Moore, “Braids in classical gravity,” Phys. Rev. Lett. 70, 3675–3679 (1993).
Bernard Schutz, Gravity from the ground up, Cambridge University Press (2003). The associated
Web site, <http://www.gravityfromthegroundup.org/>, has many Java programs includ-
ing a simulation of the orbit of a planet around a black hole or neutron star, using the
equation of motion appropriate for general relativity.
Tomomyuki Yamamoto and Kunihiko Kaneko, “Helium atom as a classical three-body problem,”
Phys. Rev. Lett. 70, 1928–1931 (1993).