First integrals. Reduction. The 2-body problem.

Physics 6010, Fall 2010

First integrals. Reduction. The 2-body problem.
Relevant Sections in Text: 3.1 3.5
Explicitly Soluble Systems
We now turn our attention to some explicitly soluble dynamical systems which are
very important in mechanics. The systems we will study include one dimensional systems,
two body central force systems, and linear (oscillatory) systems. In each of these cases
the explicit solubility of the equations of motion arises from the existence of suitable
symmetries/conservation laws. The method by which we solve for the motion of these
systems involves an important mechanics technique we shall call reduction.
First integrals and motion in one dimension
We now turn our attention to a class of dynamical systems which can always be under-
stood in great detail: the autonomous system with one degree of freedom. For autonomous
systems with one degree of freedom the motion is determined by conservation of energy.
We shall explore this for Lagrangians of the form L = T V .
Often times the study of motion of a dynamical system can be reduced to that of
a one-dimensional system. For example, the motion of a plane pendulum reduces to
dynamics on a circle. Another example that we shall look at in detail is the 2-body central
force problem. Finding the motion of this system amounts to solving for the motion of
the relative separation of the two bodies, which is ultimately a one-dimensional problem.
No matter how the one-dimensional system arises, in very general circumstances we can
completely solve the equations of motion.
Although we could proceed more generally, we will restrict our attention to the very
common situation where the Lagrangian for a one-dimensional system is of the form
L =
1
2
a(q, t) q
2
V (q, t),
where q is the generalized coordinate for the system and a, V are some given functions.
For example, a plane pendulum of length l and mass m has (exercise)
a(q, t) = ml
2
,
and
V (q, t) = mgl cos q.
Here q is the angular position of the pendulum from the vertical.
1
First integrals. Reduction. The 2-body problem.
As another example, consider a plane pendulum whose length oscillates according to
l = l
0
cos t. In this case (exercise)
a(q, t) = ml
2
0
cos
2
(t),
and
V (q, t) = mgl
0
cos(t) cos(q).
Here we see how the kinetic and potential energies can get an explicit time dependence.
Yet another example arises when we consider a particle moving on a curve y = 0,
z = f(x) under the inuence of gravity (a roller coaster!). This system has (exercise)
a(q, t) = m(1 + f
2
)
and
V (q, t) = mgf(q),
where we have chosen the position on the x-axis to be q. Here we see how the kinetic
energy can become a function of the generalized coordinate.
Our main result is the following. For Lagrangians of the form
L =
1
2
a(q) q
2
V (q),
i.e., with no explicit time dependence, the equations of motion can be solved up to
quadrature. This means that the solution can be found using only integration and algebra.
The proof uses conservation of energy. Since
L
t
= 0, we have that
E =
1
2
a(q) q
2
+ V (q)
is constant (time independent) when evaluated on a solution to the equations of motion
(EL equations). Thus we have (exercise)
1
2
a(q(t))
_
dq
dt
_
2
+ V (q(t)) = constant E.
For a given value of E, we thus get a rst-order (albeit non-linear) ODE for q(t). Since the
EL equations are second-order (exercise) we have, in eect, used conservation of energy to
integrate the equations of motion once. For this reason one often calls E a rst integral of
the equations of motion.
We can solve this rst-order ODE as follows. Fix a numerical value for E; this is equiv-
alent to partially xing the initial conditions.* We then rearrange the energy conservation
* The EL equations are a single second order ODE. Solutions are uniquely determined by
2 numbers, corresponding, e.g., to the initial values of q and q. By xing a value for the
energy we are putting one condition on the initial q and q values.
2
First integrals. Reduction. The 2-body problem.
equation given above into the form
dt =

a(q)
2(E V (q))
dq,
so that, upon integrating both sides,
t t
0
=
_
q
q
0

a(x)
2(E V (x))
dx.
Here q
0
is the other integration constant (besides E) which arises when solving a second
order ODE. t
0
is free parameter corresponding to a choice of initial time; when t = t
0
we
have q = q
0
. Note also that we have possible signs to make when taking the square root.
This amounts to whether the motion increases q with an increase in time, or decreases q
with an increase in time. Both behaviors are possible in general; the boundary between
these two types of motion is a time where the velocity vanishes. By specifying the sign,
the value of q
0
, and the value of E we have completely specied the solution. Thus these
choices are equivalent to specifying the initial values of q and q (exercise).
By performing the indicated integral we get a formula t = t(q), which also depends
upon the initial position, initial time, and energy. To get the solution of the equations of
motion, q = q(t), we need to invert this formula. This can be done provided
dt
dq
is nite
and non-vanishing (exercise). This is guaranteed by the equations of motion away from
points where E = V , i.e., where the kinetic energy vanishes. As we shall see, these are
isolated points. Away from such points we can solve (in principle) for q = q(t), which
also depends upon the initial position, initial time, and energy. By continuity we get
the solution everywhere. We thus get the solution to the equations of motion depending
upon two constants of integration (q
0
and E). We also have the sign in the solution.
Specifying q
0
and E and a choice of sign is equivalent to specifying the initial position and
velocity (exercise). Thus we obtain the general solution to the equations of motion.
Motion in One Dimension: Examples
For suciently simple forms of a(q) and V (q) a closed form expression for the motion
of the system can be found. Let us look at some elementary examples.
Free Particle
Here V (q) = 0 so that L =
1
2
mv
2
, hence
t t
0
=
_
q
q
0
_
m
2E
dx
=
_
m
2E
(q q
0
).
3
First integrals. Reduction. The 2-body problem.
We thus recover the familiar free particle motion (exercise)
q = q
0
+ v
0
(t t
0
),
where
v
0
=
_
2E
m
is the initial velocity.
Constant Force
We choose V (q) = Fq where F is the constant force on the particle so that L =
1
2
m q
2
+ Fq. We nd (exercise)
t t
0
=
_
q
q
0
_
m
2(E + Fx)
dx
=
1
F

2m(
_
E + Fq
_
E + Fq
0
).
You can easily check that this leads to the usual quadratic dependence of q on time
(exercise).
Harmonic Oscillator
A simple harmonic oscillator is treated via
t t
0
=
_
q
q
0

m
2(E
1
2
kx
2
)
dx
=
_
m
k
_
sin
1
(
_
k
2E
q) sin
1
(
_
k
2E
q
0
)
_
.
We can thus write the solution in the familiar form (exercise)
q(t) = Asin(t + ),
where
=
_
k
m
A =
_
2E
k
and
= sin
1
(
q
0
A
).
4
First integrals. Reduction. The 2-body problem.
More complicated potentials may or may not be amenable to analytic computations,
that is, explicity evaluation of the integral. For example, the simple plane pendulum
has V (q) = mgl cos q and this leads to an elliptic integral (exercise), which is not as
tractable analytically as the previous examples. Still, one can extract whatever information
is needed from the elliptic integral, if only numerically. Of course, in the small oscillation
approximation this integral reduces to our harmonic oscillator example.
Motion near equilibrium
Let us consider the motion near a stable equilibrium. Stable equilibrium occurs at
points q
0
where
V

(q
0
) = 0, V

(q
0
) > 0.
Of course, if we start the system o so that E = V (q
0
), then the velocity is zero and must
remain zero for all time (exercise). If the system has E V (q
0
) = << 1, > 0, then the
particle never goes very far from the equilibrium position. In this case we can approximate
the potential as
V = V (q
0
) +
1
2
V

(q
0
)(q q
0
)
2
and we can approximate
a(q) a(q
0
).
With this approximation, we are back to our harmonic oscillator example with the substi-
tutions
m a(q
0
), k V

(q
0
), E .
In particular, the motion is not just periodic but harmonic (frequency independent of
amplitude) with frequency
=

(q
0
)
a(q
0
)
.
Example: A Falling Rod
Let us work a non-trivial example in a little detail. Consider a thin rigid rod of mass m
and length L standing upright in a uniform gravitational eld. This is a state of unstable
equilibrium, the slightest perturbation of its position or horizontal velocity will cause the
rod to fall. Let us compute the time it takes to fall. We make the simplifying assumption
that the base of the rod does not move relative to the ground. Thus the system has one
degree of freedom corresponding to rotation of the rod about an axis perpendicular to and
through the bottom end of the rod. We choose as our generalized coordinate the angle
5
First integrals. Reduction. The 2-body problem.
between the rod and the ground. The kinetic and potential energies are (exercise)
T =
1
6
mL
2

2
, V =
mgL
2
sin.
The conserved total energy is T +V so that, for initial conditions characterized by
0
and
E, we have
t t
0
=
_

0
dx

_
1
3
mL
2
2(E
mgL
2
sinx)
=

L
3g
_

0
dx
_
1
sinx
,
where
=
2E
mgL
.
This results in an expression involving elliptic integrals of the second kind, which is not
surprising since our model of the falling rod is just a pendulum. Let us consider the time it
takes for the rod to fall to the ground. This means should decrease in time; we need the
minus sign in the above formula. Just to get some numbers, let us choose L = 2m, = 1,
= 0, and
0
=

2
0.1 radian. This corresponds to a very small initial displacement from
the vertical and a very small initial velocity. We then get (exercise) t t
0
= 1.04s.
Qualitative Description of the Motion
We now indicate some general, semi-qualitative techniques for analyzing the motion of
autonomous systems with one degree of freedom We are still assuming that the Lagrangian
has no explicit time dependence. Conservation of energy determines the motion. Let us
assume that a(q) > 0. Because the kinetic energy is positive and the solutions to the
equations of motion must be real, it follows that the value of q must be such that total
energy cannot be less than the potential energy:
V (q) E.
For a given value of the energy E, the isolated points where E = V are called turning points
since the velocity must vanish there. Generically, the velocity changes direction (as time
increases) at a turning point, keeping the particle in the domain where E V (q). If the
energy is such that the allowed region of motion is surrounded by a pair of turning points,
then the motion is bound and periodic (exercise). If the energy is such that a given region
only has one or no turning point, then the particle will move to innity, perhaps after
encountering a turning point. We say that the motion is unbound. Exceptional cases occur
when the energy is such that the turning point is a maximum or minimum of the potential.
In this case the motion is that of a (un)stable equilibrium or is such asymptotically.
6
First integrals. Reduction. The 2-body problem.
For bound motion it is easy to see that the motion must be periodic (exercise).* The
period of the motion can be computed from our integral expression for the solution to
the equations of motion. For a given value E of the energy we have (exercise)
(E) =
_
q
2
q
1

2a(x)
E V (x)
dx,
where the values q
1
and q
2
are the turning points of the bound motion, i.e., for a given E,
V (q
1
) = V (q
2
) = E. Of course, the existence of two roots is guaranteed by our assumption
that the motion is bounded.
As a simple example of this latter formula, let us return to the simple harmonic oscil-
lator. The turning points are (exercise)
q
1
=
_
2E
k
, q
2
=
_
2E
k
.
We thus get (exercise)
(E) =
_
_
2E
k

_
2E
k

2m
E
1
2
kx
2
dx,
= 2
_
m
k
[sin
1
(1) sin
1
(1)]
= 2
_
m
k
.
To summarize, when the Lagrangian for a one-dimensional system is of the form
L =
1
2
a(q) q
2
V (q),
the motion can be completely determined and involves the familiar cases of periodic bound
motion, unbound motion, as well as stable and unstable equilibria. This very regular
type of motion arises because conservation of energy is running the show. If we drop
conservation of energy, i.e., we let t appear in the Lagrangian, then the motion of the
system can be considerably more complicated. Indeed, already in the relatively simple
setting of a system with one degree of freedom and a time dependent Lagrangian, it is
possible to nd chaotic motion.
* Note that periodic does not mean harmonic. In general the period of the motion
depends upon the amplitude, i.e., the turning points.
As a very nice exercise, you should be able to show that unless the turning point is at a
critical point of the potential the period is always nite. This is not immediately obvious
since the integrand blows up as you approach a turning point. What happens physically
when the integral does diverge?
7
First integrals. Reduction. The 2-body problem.
Two Body Central Force Problem
We have already mentioned the two body central force problem several times. Here we
shall begin a systematic study of this important dynamical system. As we shall see, the
conservation laws admitted by this system allow for a complete determination of the motion
(up to quadrature). Many of the topics we have been discussing in previous lectures come
into play here. While this problem is very instructive and physically quite important, it is
worth keeping in mind that the complete solvability of this system makes it an exceptional
type of dynamical system. We cannot solve for the motion of a generic system as we do
for the two body problem.
The two body problem involves a pair of particles with masses m
1
and m
2
described
by a Lagrangian of the form:
L =
1
2
m
1

r
2
1
+
1
2
m
2

r
2
2
V (|r
1
r
2
|).
This Lagrangian is invariant under spatial translations, time translations, rotations, and
boosts.* Thus we will have conservation of energy, momentum and angular momentum
for this system.
The EL equations are equivalent to (exercise)
m
1

r
1
= V

(|r
1
r
2
|)
r
2
r
1
|r
1
r
2
|
,
m
2

r
2
= V

(|r
1
r
2
|)
r
1
r
2
|r
1
r
2
|
.
Here the prime on the potential energy function denotes a derivative with respect to its
argument. The equations of motion consist of 6 coupled ODEs corresponding to the 6
degrees of freedom. In general these equations are non-linear. There is one exceptional
case where the equations are linear. Can you see what this case is?
From these equations we see that the force exerted on each particle (i) is equal in magni-
tude and opposite in direction to that on the other particle (i.e., this force obeys Newtons
third law); (ii) is directed along a line joining the (instantaneous) particle positions, and
(iii) has a magnitude depending only on the distance between the particles.
Of course, the standard example of such a two body problem arises in the motion
of gravitating bodies (Kepler problem), or in the motion of a pair of electric charges
ignoring magnetic and radiative eects. In each case the potential energy is of the form
V
1
|r
1
r
2
|
.
* Boosts are transformations corresponding to changing to a reference frame moving with
constant relative velocity. The totality of these symmetry transformations (spacetime
translations, rotations, boosts) in non-relativistic Newtonian mechanics forms a group
known as the Galileo group.
8
First integrals. Reduction. The 2-body problem.
The constant of proportionality depends upon the masses, charges, gravitational constant,
etc.
Reduction to relative motion
Our rst step toward solving the two body problem is to reduce the problem to that of
the relative motion of the two bodies. The physical idea is simple. Given the homogeneity
of space, the absolute position of either of the particles has no relevant meaning. Only
the relative position should matter. This idea manifests itself as follows. Because the
system is invariant under spatial translations, total momentum will be conserved. This
means the motion of the center of mass will be at constant velocity in any inertial reference
frame (IRF) (exercise). So, this conservation law tells us that 3 degrees of freedom move
as a free particle. The remaining 3 degrees of freedom describe the relative motion of
the particles; it is here where all the interesting dynamics lie. This eective reduction in
degrees of freedom from 6 to 3 as arises from the conservation of total momentum (which
is the center of mass momentum) of the system.
Let us now make the above comments mathematically precise. Make a change of
generalized variables (i.e., a point transformation):

R =
m
1
r
1
+ m
2
r
2
m
1
+ m
2
,
r = r
1
r
2
.
We call

R the center of mass position and we call r the relative position. This is easily seen
to be an invertible transformation (exercise). We can express the Lagrangian as (exercise)
L =
1
2
(m
1
+ m
2
)

R
2
+
1
2
m

r
2
V (|r|),
where
m =
m
1
m
2
m
1
+ m
2
is the reduced mass of the system. We see that, in these coordinates, the Lagrangian is of
the form
L(r,

R,

r,

R) = L
cm
(

R) + L
relative
(r,

r),
indicating that the center of mass motion

R(t) and relative motion r(t) are completely
decoupled (exercise). We can solve for the motion of

R and r separately. The motion of

R is obvious:

R(t) =

R
0
+

V t.
By a suitable choice of inertial reference frame we can even set

R(t) = 0 without aecting
r(t) (exercise). Put dierently, up to a boost which is a symmetry for this system
9
First integrals. Reduction. The 2-body problem.
all solutions to the equations of motion can be obtained by setting

R(t) = 0 and solving
for the relative motion. Once we have determined r(t), we can easily reconstruct r
1
(t) and
r
2
(t) from r(t) and

R(t) (exercise).
In any case, the only non-trivial problem to solve is the relative motion problem. We
henceforth focus on the Lagrangian
L =
1
2
m

r
2
V (|r|),
which describes the relative motion. Note that, mathematically speaking, this dynamical
system is indistinguishable from a single particle moving in three dimensions subject to
a central force eld directed at xed point in space. We have thus used the conservation
of total momentum to reduce the analysis of the motion of two bodies interacting by a
central force to the analysis of a system mathematically equivalent to that of a single
particle moving in a central force eld. All our subsequent analysis will apply to this
physical system. Keep in mind that, in applications to the 2-body problem, the particle
with mass m and position r is ctitious: its position is really the relative position of the
two bodies, and the particle mass is really the reduced mass of the two body system.
Exercise: Show that when m
2
>> m
1
we can work in a reference frame in which m
2
is
(approximately) at rest and we can view the reduced Lagrangian as describing the motion
of m
1
in the xed force eld of m
2
.
A small digression: Lagrangian Reduction
By a clever change of variables in the 2 body problem, (r
1
, r
2
) (

R, r), we are able

to obtain a Lagrangian in which three coordinates, namely (R
x
, R
y
, R
z
), are cyclic. Thus
the center of mass degrees of freedom move according to free particle EL equations and
can be eliminated from the problem leaving us with a reduced problem involving 3 fewer
degrees of freedom. You can see now why cyclic coordinates are also sometimes called
ignorable.
The reduction of the 2 body problem to a problem involving only 3 degrees of freedom
by using conservation of total momentum is an instance of what is usually called Lagrangian
reduction. This term describes the eective reduction in degrees of freedom aorded by
a conservation law. A trivial example of Lagrangian reduction occurs already for motion
in one-dimension q described by a time independent Lagrangian. The conservation of
energy allows us to completely solve for q(t) (up to quadrature)thus leaving no more
degrees of freedom to worry with. As we shall see, the complete integration of the EL
equations for the 2-body problem is nothing but a sequence of Lagrangian reductions.
There is a very general and powerful theory of reduction in mechanics (and eld theory),
10
First integrals. Reduction. The 2-body problem.
but it will take us too far aeld to develop it systematically. The best version of it takes
place in the Hamiltonian formalism (to be discussed later), where the technique is called
symplectic reduction. Still, even though we will refrain from developing the general
theory of Lagrangian reduction, because this is such an important tool in mechanics it is
worth digressing for a moment to look at a couple of other examples of this phenomenon.
Example: Spherical Pendulum
The Lagrangian is
L =
1
2
mR
2
(

2
+ sin
2

2
) + mgRcos .
The coordinate is ignorable it does not appear in L and hence (exercise)
mR
2
sin
2

= l = constant.
For any given l, this equation can be used to eliminate the degree of freedom in terms
of an integral involving (t) (exercise). Thus we have a reduced problem involving a single
degree of freedom . Once we nd (t) we can easily obtain (t) by an integral, and the
problem is solved.
The motion in , i.e., (t), is determined by the equation of motion for with (t)
eliminated using the conservation law above (exercise). This can be done because (t) does
not appear in the equation, only

(t). Alternatively, we can eliminate

in the energy:
E =
1
2
mR
2

2
+
1
2
l
2
mR
2
sin
2
mgRcos ,
and use the conservation of energy to nd (t) as described earlier. Note that it is precisely
the symmetry ( is cyclic) which guarantees that the conservation law can be used to
completely eliminate from the equation of motion. Thus we use the conservation of
angular momentum to reduce the two degrees of freedom of the pendulum to an eective
one-dimensional system which can be solved via conservation of energy.
Example: Plane pendulum with moving point of support.
The Lagrangian is
L =
1
2
(m
1
+ m
2
) x
2
+
1
2
m
2
_
2l x

cos + l
2

2
_
m
2
gl cos .
The coordinate x is ignorable it does not appear L, so that the conservation law is
(m
1
+ m
2
) x + (m
2
l

cos ) = constant p.
11
First integrals. Reduction. The 2-body problem.
This conservation law can be used to eliminate x(t) once p is specied and (t) is known.
Thus we end up with a reduction to a single degree of freedom . As in our previous
example, we can nd the motion of by eliminating x from the energy in terms of (t)
and the constant p and then using conservation of energy to nd (t).
In the two body problem, we likewise used the conservation of total momentum 3
conservation laws to reduce the problem from 6 to 3 degrees of freedom. Unlike our
two examples above, the particular geometry of center of mass and relative motion is such
that this reduction is particularly simple: the two sets of degrees of freedom completely
decouple. In general this decoupling need not occur, instead we get the scenario typied
in the previous two examples. Indeed, we can use conservation laws to further reduce the
relative motion problem, even though we dont get a decoupling.
Two Body Problem: Reduction to 1 radial degree of freedom.
We now return to our study of the 2 body problem, which has been reduced to the
problem of a (ctitious) particle moving in a time independent central force eld described
by the Lagrangian
L(r,

r) =
1
2
m

r
2
V (|r|).
The EL equations are
m

r = V = V

(|r|)
r
|r|
.
To get this far we have used the conservation of total momentum. We still have conservation
of angular momentum and conservation of energy at our disposal. In this section we
implement the reduction procedure associated with conservation of angular momentum.
Note that L(r,

r) given above is rotationally invariant about any axis through the
center of force (exercise). Thus the angular momentum,

M = r p,
is conserved. Note that this the angular momentum of the relative motion. As a good
exercise you should express this in terms of the original degrees of freedom of the two
particles and see what it means.
Naively, since there are 3 conserved components in

M, it is tempting to believe we can
use conservation of angular momentum alone to completely integrate the EL equations.
This turns out not to be the case because of a technical subtlety: at any given point r,
only 2 components of angular momentum are linearly independent this follows from the
fact that
r

M = 0.
12
First integrals. Reduction. The 2-body problem.
For this reason we can use conservation of angular momentum to eliminate only 2 degrees
of freedom. We could do this by writing out the equations and just computing things as we
did in the Lagrangian reduction examples those pendulum examples we shall do this
below. But, rst, a more elegant and useful geometric point of view is worth exploring.
For any motion of the system, the vector

M is xed in space for all time. (Of course
which vector this is depends upon the initial conditions, i.e., upon which solution is being
studied.) Because
r(t)

M = 0 = p(t)

M,
it is easy to see that the motion r(t) of the particle is always in a plane orthogonal to

M,
unless

M = 0. In the

M = 0 case there are three possibilities (1) r = 0, which we dont
allow; (2) p = 0, which cant happen for more than a single instant of time; the acceleration
of r is such that at the next instant of time p = 0 and is parallel to r; (3) r and p are
proportional, in which case the particles are moving along the line joining them. This is,
of course, also motion in a plane. So, let us suppose, without loss of generality, that we
have chosen a plane for the orbit to lie in. This specication amounts to a choice of the
allowed initial conditions. Alternatively, the initial conditions pick out a plane orthogonal
to

M that the motion will lie within. For

M = 0, this choice is equivalent to specifying
the direction of the angular momentum vector. Although we are to imagine having made
a denite choice for this plane, it is completely arbitrary and no special features of the
choice will be used in what follows. Furthermore, the rotational symmetry which spawned
the conservation law means any other choice of plane, i.e., another set of initial conditions,
can be obtained from the one we choose by a rotation. Since rotations map solutions to
solutions,* we can get all solutions from the one obtained by restricting to a given plane
by simply rotating the obtained solutions. So, we can restrict our study to motion in a
plane with no essential loss of generality. Thus we have used conservation of the angular
momentum direction to reduce the system by one degree of freedom, i.e., we now have a
problem of motion of a particle in a plane.
There is available a less elegant, but more straightforward, brute force approach to
understanding this bit of reduction. Suppose we x some initial conditions and want to
see what the resulting motion looks like. Any given initial position and velocity will span a
plane. Call this plane the x-y plane and construct spherical coordinates. The Lagrangian
takes the form
L =
1
2
m( r
2
+ r
2

2
+ r
2
sin
2

2
) V (r).
The EL equations are
m r mr(

2
+ sin
2

2
) + V

(r) = 0,
* Any symmetry of the Lagrangian transforms a solution of the EL equations into another
solution of these same equations. It is a good exercise to prove this.
13
First integrals. Reduction. The 2-body problem.
d
dt
(mr
2

) mr
2
sin cos

2
= 0,
d
dt
(mr
2
sin
2

) = 0.
From these equations it follows that all three components of

M are conserved. The initial
position and velocity lie in the x-y plane; this means that M
x
= M
y
= 0 initially and for
all time. Expressed in spherical coordinates, we have
M
x
= mr
2
(sin

1
2
sin(2) cos

) = 0, M
y
= mr
2
(cos

1
2
sin(2) sin

) = 0.
One solution of these conditions is

=

= 0, which is the

M = 0 case (straight line
motion of the two particles). In this case we can assume our coordinates are chosen so
that (t) =

2
. Assuming

M = 0, we must have

= 0 (exercise) and then the above
two conditions force (t) =

2
. Thus we can always choose coordinates so that motion
associated with any given initial conditions is in the x-y plane.
The equations of motion for r(t) and (t), when restricted to the x-y plane, are (exer-
cise)
m r = mr

2
V

(r),
d
dt
(mr
2

) = 0.
These equations can be derived from the Lagrangian (exercise)
L(r, r,

) =
1
2
m( r
2
+ r
2

2
) V (r).
This Lagrangian can be obtained by simply setting =

2
in the original Lagrangian,
L(r,

r), after expressing the Lagrangian in spherical coordinates (exercise). That this is
the correct procedure in this particular step is actually guaranteed by the general theory
of Lagrangian reduction, but in general it is dangerous to simply eliminate degrees of
freedom in the Lagrangian, since one can obtain incorrect EL equations by this procedure
(see below). The reason it works in this case is that we are free to impose the holonomic
constraint = /2 in the Lagrangian as usual when analyzing dynamical systems with
constraints. Anyway, we have now reduced our analysis to that of a system with 2 degrees
of freedom, namely (r, ).
Let us summarize the situation thus far. We started with 6 degrees of freedom. The
conservation of total momentum allowed us to reduce to 3 (relative) degrees of freedom.
The conservation of the direction of angular momentum allows us to reduce to 2 degrees of
freedom. The reduced system at this stage consists of a particle with the reduced mass
m moving in a plane under the inuence of a central force.
14
First integrals. Reduction. The 2-body problem.
So far we have used the fact that the direction of

M is a constant of motion, we still
can use the fact that the magnitude of

M is conserved. Having chosen our z-axis along

M, the magnitude of

M is, of course, the z-component of angular momentum:
|

M| = mr
2

= constant = M.
That this quantity is a constant of motion is easily seen directly from our latest Lagrangian,
L(r, r,

), since is cyclic. As discussed in our previous examples, we can use this conser-
vation law to reduce our analysis to that of a system with 1 degree of freedom. Specically,
we can use the conservation of M to eliminate (t) in terms of r(t) and the constant M.
We have

=
M
mr
2
, = (t) =
_
t
t
0
ds
M
mr
2
(s)
.
The reduced equation of motion for r(t) is (exercise)
m r =
d
dr
V
e
(r),
where
V
e
= V (r) +
M
2
2mr
2
.
This equation of motion can be derived from the Lagrangian for a system with a single
degree of freedom (exercise):
L =
1
2
m r
2
V
e
(r).
Note that this Lagrangian is not obtained by using conservation of angular momentum
to algebraically eliminate

from L(r, r,

).
We now have reduced the problem to an autonomous system with one degree of free-
dom. As we have discussed, we can handle this problem using conservation of energy. The
total conserved energy of the system is correctly given by simple substitution for

in the
conserved energy for the motion in the plane (exercise):
E =
1
2
m r
2
+ V
e
.
The contribution of the angular momentum to the eective potential energy represents
an eective repulsive force (exercise) due to the angular motion of the particle. For this
reason, the term
M
2
2mr
2
is often called the centrifugal energy.
15
First integrals. Reduction. The 2-body problem.
We have reduced the dynamics of the two body problem to that of a system of one
degree of freedom, r, described by the Lagrangian
L =
1
2
m r
2
V
e
(r),
where
V
e
= V (r) +
M
2
2mr
2
.
Using the conservation of energy, in which
E =
1
2
m r
2
+ V
e
,
we can express the motion of the radial degree of freedom via (exercise)
t t
0
=
_
r
r
0
1
_
2
m
(E V (x))
M
2
m
2
x
2
dx.
Thus using conservation of total momentum, angular momentum, and energy we are
able to solve the equations of motion up to quadrature.
The orbits
The preceding reduction procedure provides a method for analyzing the motion in
time of the two bodies. There is no closed form expression, however, for this motion it
is simply too complex to be expressed using elementary analytic functions. It is possible,
however, to nd the (relative) motion at the expense of dropping all information about
how the path is traced out in time.
To begin, we have the relation between

=
M
mr
2
(t)
,
which means
d =
M
mr
2
(t)
dt.
Provided M = 0, we can parametrize the curves in conguration space with instead of t.
Going back to the conservation of energy formula and trading dt for d we have (exercise)
d =
M
r
2
_
2m(E V (r))
M
2
r
2
dr. (1)
16
First integrals. Reduction. The 2-body problem.
Thus we can express the angular motion as the integral:

0
=
_
r
r
0
M
x
2
_
2m(E V (x))
M
2
x
2
dx. (2)
Here we have absorbed the into the choice of M. Typically, it is far easier to perform this
integral than those needed to get the motion in time. Mathematically, this is because of the
extra
1
x
2
factor in the integrand. Physically this is because, by dropping the dependence
on time, the integral does not have to convey so much information about the motion of
the particle.
We can understand this integral expression in another way. Consider the radial equa-
tion of motion:
r = V

e
(r).
Make the change of variables d =
M
mr
2
(t)
dt. After a little algebra we get the dierential
equation for the orbit r = r():
M
r
2
d
d
_
M
mr
2
dr
d
_
= V

e
(r).
The solution r = r() is determined by the integral expression (2) given above. This
follows from the rst integral for the orbit equation
1
2
M
2
m
2
r
4
r
2
+ V
e
= E,
which is just conservation of energy expressed in terms of the orbit r().
The relation between r and from the last integral gives the path traced out by the
particle via = (r). If desired, we can invert to get r = r(), which is the orbit of
the particle. Knowledge of the orbit (path) carries no information about how the path is
traced out in time. This is obtained from the integral relating r and t, giving r = r(t)
and = (t) = (r(t)). Note that our elimination of the time in favor of the radius is
possible only when r = 0. Consequently we cannot use this trick to get the orbit when
the particles radius is at an equilibrium point of the eective potential. Of course this
is either a turning point, which is easily handled by continuity of the orbit (exercise), or
one has r = constant, i.e., circular motion. In this case it is hardly necessary to use the
integral representation of the path in order to understand the orbit!
For a given choice of the energy E, the values of r where
E = V
e
(r),
dene the turning points of the radial motion. At the turning points the radial velocity
vanishes; it changes sign as the particle passes through the turning point. Of course, the
17
First integrals. Reduction. The 2-body problem.
particle doesnt actually come to rest at the turning point since its angular motion is
monotonic in time, as can be seen from the conservation of angular momentum formula
(exercise). If E is such that there are two turning points then we have a bound state
and the motion is conned to an annular region in the plane of motion; we have orbital
motion. If there is only one turning point the motion is unbound and we have a scattering
situation, which we may discuss later.
For now, let us restrict our attention to bound motion. If the potential energy function
is such that V
e
has a minimum, and if E takes that minimum value, then the radial motion
is that of stable equilibrium. In this case r = constant and the motion is circular and, of
course, periodic. More generally, however, the motion is considerably more complicated.
Even though the bound motion is periodic in the radial variable, this does not mean that
the motion in the plane is periodic. This is because the time it takes for to change by 2
need not be the same as the time it takes for r to pass through one cycle. In detail, during
the time that the radial variable passes through one cycle, the angle changes by (exercise)
= 2
_
r
1
r
0
M
x
2
_
2m(E U(x))
M
2
x
2
dx,
where r
0
and r
1
are the turning points of the radial motion. For the path of the particle
in the plane to dene a closed orbit must be a rational multiple of 2:
=
p
q
(2),
so that after q cycles in the radial motion the angular variable will have made a net change
of 2p, i.e., the particle will have made p revolutions.
Generically, that is, for almost all potentials V (r), the condition for closure of the
orbit will not be satised. Thus, for a typical central force problem the relative motion will
correspond to that of a particle that follows a curve about the origin which is bounded
by r
0
and r
1
and eventually passes through every point in this annular region. Exceptions
to this occur when V
1
r
(Kepler problem; to be discussed shortly) or V r
2
(isotropic
oscillator). We shall see shortly that there is a hidden symmetry and conservation law in
these exceptional cases that explains the closure of the orbits.
Finally we note that thanks to the repulsive eect of the centrifugal energy, the
particle cannot reach the center of force r = 0 (where the 2 bodies collide) unless the
central force is suciently attractive and singular as r 0. To see this, simply
consider the formula for the energy. Assuming that r is changing, and using the fact that
the radial contribution to the kinetic energy is positive, we have the inequality (exercise)
r
2
V (r) +
M
2
2m
< Er
2
18
First integrals. Reduction. The 2-body problem.
or
r
2
(V (r) E) <
M
2
2m
.
This means that as r approaches zero V cr

, up to an additive constant, with c > 0

appropriately chosen, and 2. Of course, if M = 0 this condition has much weaker
implications for the potential.
Exercise: What is the nature of the motion if M = 0?
19