Theoretical Mechanics

________________________________________________
 “I hear and I forget.
I see and I remember.
I do and I understand.”

Confucius

盖耳听为虚，眼见为实，行之为悟，悟而思，思而后得道者也。
 Preface

The fundamental principles of classical mechanics were laid down by

Galileo and Newton in the 16th and 17th centuries. In 1686, Newton published
the Principia in which he wrote down three laws of motion, and pretended he
didn't know calculus. Probably the single greatest scientific achievement in
history, you might think this pretty much wraps it up for classical mechanics.
And, in a sense, it does. Given a collection of particles, acted upon by a
collection of forces, you have to draw a nice diagram, with the particles as
points and the forces as arrows. The forces are then added up and Newton's
famous “F = ma” is employed to figure out where the particle's velocities are
heading next. All you need is enough patience and a big enough computer and
you're done.
From a modern perspective this is a little unsatisfactory on several levels:
it's messy and inelegant; it's hard to deal with problems that involve extended
objects rather than point particles; it obscures certain features of dynamics so
that concepts such as chaos theory took over 200 years to discover; and it's not
at all clear what the relationship is between Newton's classical laws and
quantum physics.
The purpose of this course is to resolve these issues by presenting new
perspectives on Newton's ideas. We shall describe the advances that took
place during the 150 years after Newton when the laws of motion were
reformulated using more powerful techniques and ideas developed by some of
the giants of mathematical physics: people such as Euler, Lagrange, Hamilton
and Jacobi. This will give us an immediate practical advantage, allowing us to
solve certain complicated problems with relative ease. But, perhaps more
importantly, it will provide an elegant viewpoint from which we'll see the
profound basic principles which underlie Newton's familiar laws of motion.
We shall prise open “F = ma” to reveal the structures and symmetries that lie
beneath.
Moreover, the formalisms that we'll develop here are the basis for all of
fundamental modern physics. Every theory of Nature, from electromagnetism
and general relativity, to the standard model of particle physics and more
speculative pursuits such as string theory, is best described in the language we
shall develop in this course. The new formalisms that we'll see here also
provide the bridge between the classical world and the quantum world.
ADVICE TO STUDENTS WHO USE THIS BOOK
Learning physics is an active experience, If you were learning to ride a
bicycle or to play the violin, you would expect to practice the technique until
it became second nature. Falling off the bicycle gives you immediate
feedback- it tells you that you need more practice. Exactly the same thing is
true of learning physics. It must become part of you, something intuitive.
Almost everyone has to work very hard to achieve this. Until you can solve
problems, your understanding is not sufficiently deep. It is one thing to watch
a lecturer solve a problem, where every step seems to be a logical one, and
quite another to tackle a real problem on your own. Do not think of learning
physics as "art appreciation: It is a "do-it-yourself" activity.
The key to success is how one studies the subject outside of class. A last
minute "all-nighter" to solve a problem set is an exercise in self-delusion. You
are strongly advised against trying to learn physics in this way, because it
inhibits the crucial transition from short-term to long-term memory. The new
concepts have to soak into your consciousness. Remember that it took about
150 years to develop Hamiltonian dynamics. It can't be learned adequately in
one night. You should put aside a regular time for studying this material and
concentrate on it without distraction. Do the reading early in the week it is
assigned. Think about the problems more than one day before they are due.
Try to isolate the points you don't understand. Read the material again. Most
important: Discuss it with other students. Don't hesitate to ask others for an
explanation and don't be satisfied until you get one. Another lip: Make the
effort to memorize what the notation means. By experience, we have often
observed that lack of familiarity with the symbols and what they stand for is
one difference between strong and weak students. Memorizing the meaning of
symbols becomes automatic with trained physicists. Acquiring the skill of
learning physics will serve you well in later years, when most learning must
be self-taught.
If you don't fully understand what you are reading, try to construct a
simple example for yourself. But don't let the lack of understanding remain.
Pester someone- your teaching assistant, your colleague, or your professor -
until your questions are answered. And don't assume that it is clear to
everyone except yourself.
 Recommended Books

1. Classical Mechanics, H. Goldstein, C.P. Poole, and J.L. Safko, 3rd edition.
In previous editions it was known simply as \Goldstein" and has been the canonical
choice for generations of students. It is considered the standard reference on the
subject. Goldstein died and the current, third, edition found two extra authors. A
classic on the subject. Covers more material than Marion, and at a higher level. It is
not always easy to follow.
2. Mechanics (Course of Theoretical Physics), L.D. Landau, and E.M. Lifshitz,
3rd edition (Butterworth-Heinemann).
This is a gorgeous, concise and elegant summary of the course in 150 content packed
pages. Landau is one of the most important physicists of the 20th century and this is
the first volume in a series of ten, considered by him to be the “theoretical minimum"
amount of knowledge required to embark on research in physics. In 30 years, only 43
people passed Landau's exam! Another classic, Not the easiest text to follow, but one
from which you seem to learn more each time you go back to it.
3. Mathematical Methods of Classical Mechanics, V. I. Arnold.
Arnold presents a more modern mathematical approach to the topics of this course,
making connections with the differential geometry of manifolds and forms. It kicks off
with "The Universe is an Affine Space" and proceeds from there...
 Contents
1. Newtonian Mechanics………………..…………………………………………………1
1.1 Newton’s Laws of Single Particle.. ………………………………………...………1
1.2 Frames of Reference………………………………………………….…………….2
1.3 Conservation Theorems of Single Particle.....……………………….…………….3
1.4 Conservation Theorems of Systems of Many Particle…..………….………………7
2. D’Alembert Princilpe……………………………………..……………..…………….14
2.0 Important Notes on Notation…………………………..………………………….14
2.1 Constraints………………………………………………..……………………….14
2.2 Degrees of Freedom and Generalized Coordinates……..………….……..……….17
2.3 Virtual Work and D’Alembert Princilpe……………………..……….…………...19
2.4 Lagrange’s Equations………………………………………...…………………....21
3. Lagrangian Mechanics……………………………………..………………………….24
3.1 Why Lagrange’s Equations?.....………………………..………………………….24
3.2 Conservation Theorems ..………………………………..………………….…….31
3.3 Dissipative Forces and Rayleigh’s Disspative Function ....……………………….36
3.4 Velocity-dependent Potentials…...…………………………....………….………..38
4. Variational Principle of Mechanics………….……………………………….....……..39
4.1 Statement of the Problem………………………………………………......……...39
4.2 Euler’s Equation………………………………………………………......……….41
4.3 The  Notation……………………………………………………………..…...…42
4.4 Hamilton’s Principle……………………….………………………………......….43
4.5 Examples………………………………………………………………….…....…44
5. Hamilton’s Principle with Undetermined Multipliers……………….……........……..49
5.1 Euler’s equation with Equations of Constraints………………….……......……...49
5.2 Lagrange's Equations with Undetermined Multipliers……....…………...……….51
5.3 Examples…………………………………………………………….……..…...…51
5.4 Equation of Constraints as Integral………………………………..…...………….53
5.5 The Lagrangian Formulation for Continuous Systems………………………....…55
6. Hamiltonian Dynamics……………………………………………………...…...……58
6.1 The Canonical Equations of Motion……………………………………...…....….58
6.2 Hamiltonian as a Legendre Transform of Lagrangian…………………..……......61
6.3 Deriving Hamilton’Equations from Hamilton’s Principle…………………......63
6.4 The Poisson Bracket……………………………………………………………64
7. Central Force Motion………………………………………………………………69
7.1 Reduction to the Equivalent One-body Problem………………………………69
7.2 The First Integrals of Motion……………………………………………...…..71
7.3 The Equations of Motion………………………………………………………72
7.4 Centrifugal Energy and the Effective Potential………………………………..75
 7.5 Planetary Motion – Kepler’s Problem…………………………………………77
7.6 Scattering in a Central Force Field…………………………………………….81
8. Motion in a Non-inertial Reference Frame……………………………………….91
8.1 Rotating Coordinate Systems…………………………………………………91
8.2 The Centrifugal and Coriolis Forces………………………………………….95
8.3 Lagrangian formulation for noninertial motion……………………………..96
8.4 The Motion Relative to the Earth……………………………………………..96
9. Dynamics of Rigid Bodies………………………………………………………..104
9.1 The Independent Coordinates of a Rigid Body……………………………….104
9.2 The Inertia Tensor…………………………………………………………….105
9.3 Angular Momentum…………………………………………………………..108
9.4 The Principal Axes of Inertia…………………………………………………111
9.5 Similarity Transformations……………………………………………...…...115
9.6 Moments Inertia for Different Body Coordinate Systems………………...…117
9.7 The Euler Angles…………………………………………………………..…119
9.8 Euler’s Equations……………………………………………………….........122
9.9 The Tennis Racket Theorem……………………………..………………..…126
9.10 Lagrangian Method for Rigid-Body Dynamics……………………………..127
10. Coupled oscillations…………………………………………………………….135
10.1 A Simple Example – Two Coupled Oscillators……………………….........135
10.2 The General problem of Coupled Oscillations……………………………..138
10.3 Orthogonality of the Eigenvectors and the Normal Coordinates…………..141
11. Canonical transformation and Hamilton-Jacobi equation………………...…...……148
11.1 Canonical Trasformations………………………………………………….....148
11.2 The Hamilton-Jacobi Theory…………………………………………………..153

Problems………………………………………………………..………………….158
Problem Set 1(Ch.1-2)…….….………………………………………………….158
Problem Set 2(Ch.3)……………………………………..……………………….162
Problem Set 3(Ch.4-5)……….………………………………………………….164
Problem Set 4(Ch.6)……………………………………..……………………….168
Problem Set 5(Ch.7)……………………………………..……………………….169
Problem Set 6(Ch.8)…….……………………………………………………….175
Problem Set 7(Ch.9)…….……………………………………………………….177
Problem Set 8(Ch.10)…………………………………………………………….181
Problem Set 9(Ch.11)…………………………………………………………….183
 Chapter 1 Newtonian Mechanics

Chapter 1
Newtonian Mechanics

Since our course on the subject of Analytical Mechanics could accurately be called “A 1001
ways of writing F=ma”, we will start from the beginning …

1.1 Newton’s Laws of Single Particle

Newton’s Laws are usually simply stated as:
I. A body remains at rest or in uniform motion unless acted upon by a force.
II. A body acted upon by a force moves in such a manner that the time rate of change of the
momentum equals the force.
III. If two bodies exert forces on each other, these forces are equal in magnitude and
opposite in direction.
The First Law is meaningless without the concept of “force”, but conveys a precise meaning
for the concept of “zero force”.
The Second Law is very explicit: Force is the time rate of change of the momentum. But what
is the momentum p …
p  mv, (1.1)
with m the mass, and v the velocity of the particle. We, therefore, re-write the Second Law as

dp d
F   mv  . (1.2)
dt dt
However, we still don't have a definition for the concept of "mass". This is made clear with
the Third Law, which can be rewritten as follows to incorporate the appropriate definition of
mass:
III'. If two bodies constitute an ideal, isolated system, then the accelerations of these bodies
are always in opposite direction, and the ratio of the magnitudes of the accelerations is
constant. This constant ratio is the inverse ratio of the masses of the bodies.
If we have two isolated bodies, 1 and 2, then the Third Law states that
F1  F2 , (1.3)
and from the Second Law, we have

-1-
 Chapter 1 Newtonian Mechanics

dp1 dp
 2, (1.4)
dt dt
or
dv1 dv
m1   m2 2 , (1.5)
dt dt
and using the acceleration a
m1a1   m2a 2
m1 a (1.6)
 2.
m2 a1
If one chooses m1 as the reference or unit mass, m2 , or the mass of any other object, can be
measured by comparison (if it is allowed to interact with m1 ). Incidentally, we can use
equation (1.4) to provide a different interpretation of Newton’s Second Law
d
(p1  p 2 )  0 (1.7)
dt
or
p1  p 2  cste. (1.8)
The momentum is conserved in the interaction of two isolated particles. This is a special
case of the conservation of linear momentum.
One should note that the Third Law is not a general law of nature. It applies when dealing
with central forces (e.g., gravitation (in the non-relativistic limit), electrostatic, etc.), but not
necessarily to other types of forces (e.g., velocity-dependent forces such as between two
moving electric charges).

1.2 Frames of Reference

A reference frame is called an inertial frame if Newton’s laws are valid in that frame. More
precisely,
 If a body subject to no external forces moves in a straight line with constant velocity,
or remains at rest in a reference frame, then this frame is inertial.
 If Newton’s laws are valid in one reference frame, they are also valid in any other
reference frame in uniform motion (or not accelerated) with respect to the first one.
The last point can be expressed mathematically like this. If the position of a free particle of
mass m is represented by r in a first inertial frame, and that a second frame is moving at a
constant velocity V2 relative to the first frame, then we can write the position r' of the
particle in the second frame by
r   r  V2t. (1.9)
The particle’s velocity v' in that same frame is

-2-
 Chapter 1 Newtonian Mechanics

d dr
v   r  V2t    V2  v  V2 , (1.10)
dt dt
where v is the velocity of the particle in the first frame. Similarly, we can calculate the
particle’s acceleration in the second frame (a' ) as a function of its acceleration (a) in the first
one
dv  d dv
a    v  V2    a. (1.11)
dt dt dt
We conclude that the second reference frame is inertial since Newton’s laws are still valid in
it (F' =ma' = ma ). This result is called Galilean invariance, or the principle of Newtonian
relativity.
Newton’s equations do not describe the motion of bodies in non-inertial reference frame (e.g.,
rotating frames). That is to say, in such frames Newton’s Second Law, or the equation of
motion, does not have the simple form F = ma.

1.3 Conservation Theorems of single particle

We now derive three conservation theorems that are consequences of Newton’s Laws of
dynamics.
1.3.1 Conservation of linear momentum
This theorem was derived in section 1.1 for the case of two interacting isolated particles (see
equations (1.7) and (1.8)). We now re-write it more generally from Newton’s Second Law
(equation (1.2)) for cases where no forces are acting on a (free) particle
p  0 (1.12)
where p is the time derivative of p, the linear momentum. Note that equation (1.12) is a
vector equation, and, therefore, applies for each component of the linear momentum. In cases
where a force is applied in a well-defined direction, a component of the linear momentum
vector may be conserved while another is not. For example, if we consider a constant vector
s such that F·s=0 (the force F is in a direction perpendicular to s), then
p  s  F  s  0 (1.13)

If we integrate with respect to time, we find

p  s  cste. (1.14)
Equation (1.14) states that the component of linear momentum in a direction in which the
forces vanishes is constant in time.
1.3.2 Conservation of angular momentum
The angular momentum L of a particle with respect to an origin from which its position
vector r is being measured is given by
L  rp (1.15)
The torque or moment of force N with respect to the same origin is given by

-3-
 Chapter 1 Newtonian Mechanics

N  r  F, (1.16)
where the force F is being applied at the position r. Because the force is the time derivative of
the linear momentum, we can write
dL d
L    r  p    r  p    r  p  , (1.17)
dt dt
but r  p  0 , since r  v and p  mv . We, therefore, find that
L  r  p  N. (1.18)
It follows that the angular momentum vector L will be constant in time ( L  0 ) if no torques
are applied to the particle ( N  0 ). That is, the angular momentum of a particle subject to no
torque is conserved.
1.3.3 Conservation of energy
If we consider the resultant of all forces (i.e., the total force) F applied to a particle of mass m
between two points “1” and “2”, we define the work done by this force on the particle by
2
W12   F  dr (1.19)
1

We can rewrite the integrand as

dv dr dv
F  dr  m  dt  m  vdt
dt dt dt

m d
2 dt
 v  v  dt 
m d 2
2 dt
 v  dt (1.20)

1 
 d  mv 2  .
2 
Since equation (1.20) is an exact differential, we can integrate equation (1.19) and find the
work done on the particle by the total force
2
1  1
W12   mv 2   m(v22  v12 )  T2  T1 , (1.21)
2 1 2
where T= mv2/2 is the kinetic energy of the particle. The particle has done work when
W12<0.
Similarly, we can also define the potential energy of a particle as the work required, from
the force F, to transport the particle from point “1” to point “2” when there is no change in its
kinetic energy. We call this type of forces conservative (e.g., gravity). That is
2
 1
F  dr  U1  U 2 , (1.22)

Where Ui is the potential energy at point “ i ”. The work done in moving the particle is simply
the difference in the potential energy at the two end points. Equation (1.22) can be expressed
differently if we consider F as being the gradient of the scalar function U (i.e., the potential
energy)
F  U . (1.23)

-4-
 Chapter 1 Newtonian Mechanics

The potential energy is, therefore, defined only within an additive constant. Furthermore, in
most systems of interest the potential energy is a function of U=U(r,t) position and time, i.e., ,
not the velocityr .
We now define the total energy E of a particle as the sum of its kineti and potential energies
E  T U. (1.24)
The total derivative of E is
dE dT dU
  (1.25)
dt dt dt
Since we know from equation (1.20) that dT=F·dr, then
dT dr
 F   F  r (1.26)
dt dt
On the other hand
dU U xi U U
    U   r  (1.27)
dt i xi t t t
Inserting equations (1.26) and (1.27) in equation (1.25), we find
U U U
  F   U    r 
dE
 F  r   U   r   (1.28)
dt t t t
The last step is justified because we are assuming that F is a conservative force
(i.e., F  U ). Furthermore, for cases where U is not an explicit function of time, we
have U / t  0 and
dE
0 (1.29)
dt
We can now state the energy conservation theorem as: the total energy of a particle in a
conservative field is a constant in time.
Finally, we group the three conservation theorems that we derived from Newton’s equations:
I. The total linear momentum p of a particle is conserved when the total force on it is
zero.
II. The angular momentum of a particle subject to no torque is constant.
III. The total energy of a particle in a conservative field is a constant in time.

Examples
1-1 (Newton’s equation of motion has a simple form F = ma only in an inertial frame.)
A particle of mass m is constrained to move on the surface of a sphere of radius R by an
applied force F(). Write the equation of motion.
Solution
Using spherical coordinates, we can write the force applied to the particle as

-5-
 Chapter 1 Newtonian Mechanics

F  Fr e r  F e  F e (1.30)
But since the particle is constrained to move on the surface of a sphere, there must exist a
reaction force  Fr er that acts on the particle. Therefore, the total force acting on the particle
is
F  F e  F e  m
r. (1.31)
 total
The position vector of the particle is
r  Re r (1.32)
where R is the radius of the sphere and is constant. The acceleration of the particle is
a  
r  Re r (1.33)
We must now express ër in terms of er, e, and e. Because the unit vectors in rectangular
coordinates e1, e2, and e3, do not change with time, it is convenient to make the calculation in
terms of these quantities. Using Figure 1.1 for the definition of a spherical coordinate system
we get
e r  e1 sin  cos   e 2 sin  sin   e 3 cos
e  e1 cos cos   e 2 cos sin   e 3 sin  (1.34)
e  e1 sin   e 2 cos 
Then
e r  e1 ( sin  sin    cos cos  )  e 2 ( cos sin    sin  cos  )  e 3 sin 
(1.35)
 e sin   e


Figure 1-1 Spherical coordinate system.

Similarly,
e   e r  e cos
(1.36)
e   e r sin   e cos

-6-
 Chapter 1 Newtonian Mechanics

And, further,
e  e r ( 2 sin 2    2 )  e (   2 sin  cos )  e (2 cos  sin  ) (1.37)

r
which is the only second time derivative needed. The total force acting on the particle is
Ftotal  m
r  mR
er (1.38)
and the components are
F  mR(  2 sin  cos  )
(1.39)
F  mR(2   cos   sin  )


It is clear that eq.(1.39) does not has a simple form F = ma. Even if we know the force, it is
usually very difficult to write down the equations of motion directly.

1.4 Conservation Theorems of Systems of Many particles

1.4.1Centre of Mass
For a system composed of n particles, the total mass M is given by
M   m (1.40)


where m is the mass of the th particle, with   1,..., n . If each particle is (mathematically)
connected to the origin of the system through a position vector r, then the centre of mass
vector is defined as
1
R
M


m r (1.41)

For a continuous system, the summation over  is replaced with an integral over an
infinitesimal amount of mass dm such that
1
M
R rdm (1.42)

It is important to realize that the position vector R of the centre of mass depends on the origin
chosen for the coordinate systems.
1.4.2 The Conservation of Linear Momentum
The force acting on particle of a system of particles is composed of the resultant of all forces
external to the system F  e  , and the resultant of the internal forces f stemming from its
interaction with the other particles that are part of the system. If we define these internal
interaction forces as f , the resulting force f acting on particle  is

f   f (1.43)
 

The total force F acting on the particle is

F  F    f
e
(1.44)

-7-
 Chapter 1 Newtonian Mechanics

From Newton’s Second Law we can write

p   F  e   f (1.45)
or
d2
 mar   F  e  f  F  e   f (1.46)
dt 2  

where no summation on repeated index is implied. Summing equation (1.46) over all
particles we get
d2  
m r   F  e     f  F    f  f  
2  a  
(1.47)
dt         ,  pairs

where we have defined the sum over all external forces as

F   F 
e
(1.48)


and the second term on the right of equation (1.47) was replaced by a single summation over
every pair of internal interactions between the particles. However, we know from Newton’s
Third Law that f  f  . We can therefore write, from equation (1.47) that
  F
MR (1.49)
This last equation can also be used to express the conservation of momentum since
d  
P   m r    
m r   MR (1.50)
 dt   
and then
P  MR
  F (1.51)
We can summarize this result as follows

I. The centre of mass of a system moves as if it were a single particle of mass equal to the
total mass of the system, acted upon by the total external force, and independent of the
internal forces (as long as f  f  (Newton’s Third Law) holds).
II. The total linear momentum of a system is the same as that of a singe particle of mass M
located at the position of the centre of mass and moving in the manner the centre of
mass moves.
III. The total linear momentum for a system free of external forces is a constant and equal to
the linear momentum of the centre of mass (the law of conservation of linear momentum
for a system).

1.4.3 The Conservation of Angular Momentum

It is often more convenient to define the positions of the particles composing a system by
vectors

8
 Chapter 1 Newtonian Mechanics

r originating at the centre of mass (see Figure 1-2). The position vector r in the inertial
frame is
r  R  r (1.52)
The angular momentum of the th particle is given by
L  r  p (1.53)
and summing over all particles
L   L   r  p   r  m r 
  

   R  r   m  R
  r   
  (1.54)


    R  r     r  R
  m  R  R  
    r  r   
  


Figure 1-2 Description of the position of a particle using its position vector from the centre
of mass of the system.

The second and third terms on the right hand side equal zero from
d 
 m  R  r    r  R
    m
    R  r    R   r    r  R 
   dt 
(1.55)
d   
  R   m r   2   m r   R  0
dt      
since, from equations (1.41) and (1.42),


m r  0 (1.56)

Equation (1.54) now becomes

9
 Chapter 1 Newtonian Mechanics

      r  m r  
L   R   m R    
    (1.57)
 R  P    r  p 


We, therefore, have this important result

IV. The angular momentum about an origin is the sum of the angular momentum of the
centre of mass about that origin and the angular momentum of the system about the
position of the centre of mass.
The time derivative of the total angular momentum is
L   L   r  p  
 

 

  

  r  F  e     r   f 
  
(1.58)

 
  r  F  e     r  f    r  f   
  

where  means a sum over  and  with  .

 

We know, however, from Newton’s Third Law that f= -f so that equation (1.58) can
be re-written
L   (r  F ( e ) )   [(r  r )  f ] (1.59)
  

If we further limit ourselves to internal forces f that are also directed along the straight line
joining the two interacting particles (i.e., along r -r), we must have the following
(r  r )  f  0 (1.60)
The time derivative of the total angular momentum is then
L   (r  F ( e ) )   (1.61)


or if we express the right hand side as a sum of the external torque applied on the N ( e )
different particles
L   N ( e )  N ( e ) (1.62)


We, therefore, have the following results

V. If the net resultant external torque about an axis vanishes, then the total angular
momentum of the system about that axis remains a constant in time.
Furthermore, since we found that the total internal torque also vanishes, i.e.,


(r   f )   [(r  r )  f ]  0
   
 
(1.63)

10
 Chapter 1 Newtonian Mechanics

and we can state that

VI. The total internal torque must vanish if the internal forces are central (i.e., f=
-fand the internal forces between two interacting particles are directed along the
line joining them), and the angular momentum of an isolated system cannot be altered
without the application of external forces.

1.4.4 The Energy of the System

Consider a system of particles that evolves from a starting configuration “1” to an ulterior
configuration “2” where the positions rof the particles may have changed in the process.
We can write the total work done on the system as the sum of the work done on individual
particles

W12    F  dr
2

1


dv dr dv
   m dt    m   v dt
2 2


1 dt dt 
1 dt
(1.64)
1 dv 2 2 d 1 
  m  dt   
2
 m v 2  dt

1 2 dt  
1 dt 2

2 1 
   d  m v 2   T2  T1

1
2 
where
1
T   T   m v 2 (1.65)
  2

Using equation (1.52) we can write

v 2  r  r   R
  r     R

  r  


R R
  2r   R
  r   r  (1.66)
  
2 
 V  2r   R  v 2
 

 | . Inserting equation (1.66) into (1.65), while using the earlier

where v | r | and V | R
results that states that 

m r  0 ， we find that

1 1
T MV 2   m v 2 (1.67)
2  2

In other words

VII. The total kinetic energy of the system is equal to the sum of the kinetic energy of a
particle of mass M moving with velocity of the centre of mass and the kinetic energy
of motion of the individual particles relative to the centre of mass.

11
 Chapter 1 Newtonian Mechanics

Alternatively, we can rewrite first of equations (1.64) by separating the total force applied on
each particle in its external and internal components

W12    F  e   dr  
2 2
f  dr (1.68)
1 1
   
, 

If the forces involved are conservatives, we can then derive them from potentials such that
F  e   U 
(1.69)
f  U 

where U and U  are independent potential functions. The gradient operator  is a

vector operator meant to apply to the coordinate components of the th particle (i.e., is the
index that specifies a given particle, and does not represent a coordinate such as x, y, or z).
The first term on the right hand side of equation (1.68) can be written as

 F  e   dr    U   dr   U 

2 2

2
1
(1.70)
1 1
  

The last term of the same equation is transformed to

 f  dr     f  dr  f   dr 

2 2

1 1
  
,   
(1.71)
   f   dr  dr 
2

1
 

Before we use the last of equations (1.69) to further transform equation (1.71), we consider
the following differential
dU   U    dr     U    dr
  f   dr   f    dr (1.72)
  f    dr  dr 

Since   U   f   f (note also that U   U  ). Combining this result with equations
(1.68), (1.70), and (1.71), we get
W12   U  1   U
2 2

1
(1.73)
  

If we define the total potential energy as

U  U   U  (1.74)
we get
W12  U |12  U1  U 2 (1.75)
Combining equation (1.75) and the last of equations (1.64), we find that
T2  T1  U1  U 2 (1.76)

12
 Chapter 1 Newtonian Mechanics

or,
T1  U1  T2  U 2 (1.77)
and finally
E1  E2 (1.78)
We have therefore proved the conservation of energy for a system of particles where all the
forces can be derived from a potential that are independent of time; such a system is called
conservative.

VIII. The total energy for a conservative system is constant.

13
 Chapert 2 D’Alembert’s Principles

Chapter 2
D’Alembert’s Principles

Although Newton’s equation F  p correctly describes the motion of a particle (or a system
of particles), it is often the case that a problem will be too complicated to solve using this
formalism. For example, a particle may be restricted in its motion such that it follows the
contours of a given surface, or that the forces that keep the particle on the surface (i.e., the
forces of constraints), are not easily expressible in Cartesian coordinates. It may not even
be possible at times to find expressions for some forces of constraints. Such occurrences
would render it impossible to treat the problem with the Newtonian formalism since this
requires the knowledge of all forces acting on the particles. In this section we will study a
different approach for solving complicated problems in a general manner. The formalism
that will be introduced is called Lagrange’s mechanics.
Lagrange’s equations were originally derived on a sound mathematical foundation by using
the concept of virtual work along with D’Alembert’s Principle.

2.1 Constraints
The complete description of a general mechanical system of n unconstrained particles
requires 3n coordinates. You can think of the state of the system at any time as being
represented by a single point in 3n-dimensional space. If the system consists of molecules
in a gas, or a cluster of stars, or a swarm of bees, the coordinates will be continually
changing, and the point that describes the system will be moving, perhaps completely
unconstrained, in its 3n-dimensional space.
However, in many cases, the motion of bodies considered in mechanics is not free but is
restricted by certain constraint conditions. The constraints can take different forms. For
instance, a mass point can be bound to a space curve or to a surface. The constraints for a
rigid body state that the distances between the individual points are constant. If one
considers gas molecules in a vessel, the constraints specify that the molecules cannot
penetrate the wall of the vessel. Since the constraints are important for solving a
mechanical problem, mechanical systems are classified according to the type of constraints.
A system is called holonomic if the constraints can be represented by equations of the form
f k (r1 , r2 ,..., rn , t )  0, k  1, 2,..., m. (2.1)

The word "holonomic" was from the Greek ő meaning "whole" or "entire" and ó,
meaning "law". It also said "applied to a constrained system in which the equations

- 14 -
 Chapert 2 D’Alembert’s Principles

defining the constraints are integrable or already free of differentials, so that each equation
effectively reduces the number of coordinates by one; also applied to the constraints
themselves."
This form of the constraints is important since it can be used for eliminating dependent
coordinates. We give few typical examples.
1. A simple pendulum of length l. For a pendulum of length l the equation (2.1) reads
x 2  y 2  l 2  0. (2.2)
if we put the coordinate origin at the suspension point. The coordinates x and y can
be expressed by this equation.
2. A rigid body is another simple example of holonomic constraints, where distances
between any two particles making it, remain constant during the motion. Thus
ri  r j  cij  0, (2.3)

where cij is the distance between the particle i and j positioned at ri and r j . In this
case the constraints served to reduce the 3n degrees of freedom of a system of n mass
points to the 6 degrees of freedom of the rigid body.
All constraints that cannot be represented in the form of algebraic equations (2.1) are
called nonholonomic. These are conditions that cannot be described by a closed form or by
equalities. However, frequently they could be expressed as inequalities. An example of this
type of constraint is the system of gas molecules enclosed in a sphere of radius R. Their
coordinates must satisfy the conditions ri  R.

A further classification of the constraint conditions is made based on their time dependence.
If the constraint is an explicit function of the time, then it is called rheonomic. If the time
does not enter explicitly, the constraint is called scleronomic. A rheonomic constraint
appears if a mass point moves along a moving space curve, or if gas molecules are enclosed
in a sphere with a time-dependent radius.

In certain cases the constraints may also be given in differential form, for example if there
is a condition on velocities, e.g., for the rolling of a wheel. The constraints then have the
form
n

 a ( x , x ,..., x )dx
k 1
k 1 2 n k  0, (2.4)

where the xi represent the various coordinates, and the ak are functions of these coordinates.
We now have to distinguish between two cases. If the equation (2.4) represents the total
differential of a function U, we can integrate it immediately and obtain an equation of the
form of equation(2.1). In this case the constraints are holonomic. If equation (2.1) is not a
total differential, we can integrate it only after having solved the full problem. Then
equation (2.4) is not suitable for eliminating dependent coordinates; it is nonholonomic.

From the requirement that equation (2.4) be a total differential, one can derive a criterion
for the holonomity of differential constraints. One must have

- 15 -
 Chapert 2 D’Alembert’s Principles

 a dx
k
k k  dU (2.5)

with ak  U / xk . This leads to

ak  2U a
  i.
xi xi xk xk

Thus, equation (2.4) represents a holonomic constraint if the coefficients obey the
integrability conditions
ak ai
 .
xi xk
For example, an equation of the form
dxi
A i
dt
B0 (2.6)
i

cannot, in general, be integrated to give an equation of the form f ( xi , t )  0 . Such

equations of constraints are non-holonomic. We will not consider this type of constraints
any further. However, if the constants Ai and B are such that equation (2.6) can be
expressed as
f xi f
 x t

t
 0, (2.7)
i i

or more simply
df
 0, (2.8)
dt
then it can be integrated to give

f ( x, t )  cste  0. (2.9)

Example 2.1. Wheel rolls on a plane (Fig.2-1).

An example of a system with differential constraints is a wheel that rolls on a plane without
gliding. The wheel cannot fall over. The radius of the wheel is a. For the calculation, we
use the coordinates xM , yM of the center, the angle  that describes the rotation, and the
angle that characterizes the orientation of the wheel plane relative to the y-axis.

- 16 -
 Chapert 2 D’Alembert’s Principles

Fig.2-1 Wheel rolls on a plane.

The velocity v of the wheel center and the rotation velocity are related by the rolling
condition
v  a.
The components of the velocity are
xM  v sin ,
y M  v cos .
By inserting v, we obtain
dxM  a sin  d  0,
dyM  a cos  d  0.
i.e., a constraint of the type of equation (2.4). Since the angle is known only after
solving the problem, the equations are not integrable. Hence, the problem is nonholonomic,
scleronomic, and conservative.

2.2 Degrees of Freedom and Generalized Coordinates

If a system is made up of n particles, we can specify the positions of all particles with 3n
coordinates. On the other hand, if there are m equations of constraints (for example, if some
particles were connected to form rigid bodies), then the 3n coordinates are not all
independent. There will be only 3n-m independent coordinates, and the system is said to
possess 3n-m degrees of freedom.
Furthermore, the coordinates and degrees of freedom do not have to be all given in
Cartesian coordinates, or any other systems. In fact, we can choose to have different types
of coordinate systems for different coordinates. Also, the degrees of freedom do not even
need to share the same unit dimensions. For example, a problem with a mixture of
Cartesian and spherical coordinates will have “lengths” and “angles” as units. Because of
the latitude available in selecting the different degrees in freedom, the name of generalized
coordinates is given to any set of quantities that completely specifies the state of the
system. The generalized coordinates are usually written as qj.
It is important to realize that there is not one unique way of setting up the generalized
coordinates, there are indeed many different ways of doing this. Unfortunately, there are no
clear rules for selecting the “best” set of generalized coordinates. The ultimate test is

- 17 -
 Chapert 2 D’Alembert’s Principles

whether or not the choice made leads to a simple solution for the problem at hand.
In addition to the generalized coordinates qj, a corresponding set of generalized velocities
q j is defined. In general, the relationships linking the Cartesian and generalized
coordinates and velocities can be expressed as
x ,i  x ,i ( q j , t )
(2.10)
x ,i  x ,i ( q j , q j , t ),   1,..., n; i  1, 2,3; j  1, 2,...,3n  m.
or alternatively
q j  q j (x ,i ,t)
(2.11)
q j  q j (x ,i , x ,i ,t).
We must also include the equations of constraints
f k ( x ,i , t )  f k (q j , t )  0. (2.12)

Example 2.2: Double Pendulum (Fig.2.2).

X
l1
1
m1

l2
2

Y
m2

Fig.2-2 Double Pendulum

We claim that, the two angles 1 and  2 made by the pendula with the vertical are
generalized coordinates capable of describing the motion completely. Coordinates of m1
are ( x1 , y1 , z1 ) and coordinates of m2 are ( x2 , y2 , z2 ). Because, m1 and m2 move in a
fixed plane, say z  0 , the two constraint equations are
z1  0, z2  0.
This reduces independent coordinates to four. Length of one pendulum is l1 giving a
constraint equation
x12  y12  l12  0
and the length of the other pendulum is l2 giving

- 18 -
 Chapert 2 D’Alembert’s Principles

x2  x1    y2  y1 
2 2
 l2 2  0
This further reduces to two independent coordinates giving the degrees of freedom to be
two. Any two coordinates, one each of m1 and m2 can describe the motion. However, the
natural variables are the two angles 1 and  2 as shown above. We use them as
generalized coordinates. It is important that the angles are measured with respect to the
fixed direction. Note - they do not have the dimensions of length.

Example 2.3: Cylinder rolls on an inclined plane (Fig.2-3).

Fig.2-3 Cylinder rolls on an inclined plane.

The position of a cylinder on an inclined plane is completely specified by the distance l

from the origin to the center of mass and by the rotational angle  of the cylinder about its
axis. If the cylinder glides on the plane, both generalized coordinates are significant.

2.3 Virtual Work and D’Alembert’s Principle

A virtual displacement is the result of any infinitesimal change of the coordinates rthat
define a particular system, and which is consistent with the different forces and constraints
imposed on the system at a given instant t. The term virtual is used to distinguish these
types of dt displacement with actual displacement occurring in a time interval, during
which the forces could be changing.
Now, suppose that a system is in equilibrium. In that case the total force F on each particle
that compose the system must vanish, i.e., F  0 . If we define the virtual work done on a
particle as F   r (note that we are using Cartesian coordinates) then we have



F   r  0. (2.13)

Within the context of Newtonian mechanics, we distinguish between two classes of forces,
depending on whether a force is able to do work or not. In the first class, an active force F(a)
is involved in infinitesimal work dW = F(a) · dr evaluated along the infinitesimal
displacement dr; the class of active forces includes conservative and nonconservative
forces. In the second class, a passive force (labeled f) is defined as a force not involved in
performing work, which includes constraint forces such as normal and tension forces. Here,
the infinitesimal work performed by a passive force is f · dr = 0 because the infinitesimal
displacement dr is required to satisfy the constraints.

- 19 -
 Chapert 2 D’Alembert’s Principles

Let’s now decompose the force F as the sum of the applied force F( a ) and the force of
constraint f such that

F  F( a )  f , (2.14)
then equation (2.13) becomes
 F (a)
  r   f   r  0.

(2.15)

In what follows, we will restrict ourselves to systems where the net virtual work of the
forces of constraints is zero, that is


f   r  0, (2.16)

and
 F (a)
  r  0 (2.17)

This condition will hold for many types of constraints. For example, if a particle is forced
to move on a surface, the force of constraint if perpendicular to the surface while the virtual
displacement is tangential. It is, however, not the case for sliding friction forces since they
are directed against the direction of motion; we must exclude them from our analysis. But
for systems where the forces of constraints are consistent with equation (2.16), then
equation (2.17) is valid and is referred to as the principle of virtual work.
The Principle of Virtual Work is one of the oldest principles in Physics that may find its
origin in the work of Aristotle (384-322 B.C.) on the static equilibrium of levers. The
Principle of Virtual Work was finally written in its current form in 1717 by Jean Bernoulli
(1667-1748). It was Jean Le Rond D’Alembert (1717-1783) who generalized the Principle
of Virtual Work (in 1742) by including the accelerating force in the Principle of Virtual
Work, so that the equations of dynamics could be obtained. Hence, D’Alembert’s Principle,
in effect, states that the work done by all active forces acting in a system is algebraically
equal to the work done by all the acceleration forces (taken to be positive).
Now, let’s consider the equation of motion , which can be written as

(2.18)
Inserting this last equation in equation (2.13) we get
(2.19)

and upon using equations (2.14) and (2.16) we find

(2.20)

Equation (2.20) is often called D’Alembert’s Principle.

- 20 -
 Chapert 2 D’Alembert’s Principles

2.4 Lagrange’s Equations

We now go back to our usual coordinate transformation that relates the Cartesian and
generalized coordinates
x ,i  x ,i (q j ,t) or r  r (q j ,t) (2.21)

from which we get

dr r r
 q j  . (2.22)
dt j q j t
Similarly, the components  x ,i of the virtual displacements vectors can be written as

r
 r   qj. (2.23)
j q j
Note that no time variation  t is involved in equation (2.23) since, by definition, a
virtual displacement is defined as happening  t at a given instant t, and not within a time
interval . Inserting equation (2.23) in the first term of equation(2.20), we have
r


F   r  F     q   Q  q
a

 q
a

j
j
j
j j (2.24)
j

and the quantity

r
Q j   F a   (2.25)
 q j

are the components of the generalized forces. It is noted that the principle of virtual work
(2.17) has a perfect form in terms of generalized coordinates
Q j  0, j  1,...,3n  m. (2.26)
since  q j ’s are independent from each other.

Concentrating now on the second term of equation (2.20) we write

r


p   r  m 

r    q .
q j
j (2.27)
j

This last equation can be rewritten as

r d  r  d  r  
 r    q     m r     m r       q .
m 
q dt 
j
q  dt  q 
j (2.28)
,j j  ,j   j   j  
We can modify the last term since

- 21 -
 Chapert 2 D’Alembert’s Principles

d  r  r
  (2.29)
dt  q j  q j
Furthermore, we can verify from equation (2.22) that
r r
 . (2.30)
q j q j
Substituting equations (2.29) and (2.30) into (2.28) leads to

r d  r  r 

 m r  q j
 q j     m r  
, j 
 q j
  m r     q j
q j 
, j  dt  
(2.31)
d    1    1 
               q j .
    2   q  2
m r r  m r r 
j  dt  q  j    
 j

Combining this last result with equations (2.25) and(2.27), we can express (2.20) as

d T 
T     
  F   p  r   Q   dt  q   q    q


a
  j j  0, (2.32)
  j    j  j  
where we have introduced T the kinetic energy of the system, such that
1
T 
2 
ma r2 . (2.33)

Since the set of virtual displacement  q j are independent, the only way for equation
(2.32) to hold is that

d  T  T
   Qj. (2.34)
dt  q j  q j
If we now limit ourselves to conservative systems, we must have
U
F    or F  rU
a
(2.35)
 ,i
x ,i
and similarly,
x ,i U x ,i U
Q j   F ,i   
a
, (2.36)
 ,i q j  ,i x ,i q j q j
with U  U ( x ,i , t )  U (q j , t ) the potential energy. We can, therefore, rewrite equation
(2.34) as


d   T U   T U   
   0, (2.37)
dt  q j  q j

- 22 -
 Chapert 2 D’Alembert’s Principles

since U / q j  0 .

If we now define the Lagrangian function for the system as

L  T U , (2.38)
we finally obtain Lagrange’s equations

d  L  L
   0. (2.39)
dt  q j  q j

These equations are called Lagrange equations, and the quantities L / q j are called
generalized momenta. In Newton's formulation of mechanics, the equations of motions are
established directly. The forces are thus put in the foreground; they must be specified for a
given problem and inserted into the basic dynamic equations p  F .

In the Lagrangian formulation the Lagrangian is the central quantity, and L includes both
the kinetic energy T and the potential energy U. The latter one implicitly involves the
forces. After L is established, the Lagrange equations can be established and solved. Both
methods are equivalent to each other, as can be seen by stepwise inversion.

- 23 -
 Chapter 3 Lagrangian Mechanics

Chapter 3
Lagrangian Mechanics

In the last chapter we obtained Lagrange’s equations from D’Alembert’s principle. In this
chapter we shall show the power of Lagrangian formulation of mechanics in solving
complex mechanical problems and understanding conservation theorems.

3.1 Why Lagrange’s equations ?

If we apply Lagrange’s equation to the problem of the one-dimensional harmonic oscillator

(without damping), we have
1 2 1 2
L  T U  mx  kx , (3.1)
2 2
and
L
 kx
x
(3.2)
d  L  d
    mx   mx.
dt  x  dt
After substitution of equations (3.2) into equation (3.1) we find
mx  kx  0 (3.3)
for the equation of motion. This result is identical than what was obtained using Newtonian
mechanics. This is, however, a simple problem that can easily (and probably more quickly)
be solved directly from the Newtonian formalism. But, the benefits of using the Lagrangian
approach become obvious if we consider more complicated problems. For example, we try
to determine the equations of motion of a particle of mass m constrained to move on the
surface of a sphere under the influence of a conservative force F  F e , with F a constant.
In this case we have
1 2 1 2 1 1
T mv  mv  mR 2 2  mR 2 sin 2   2
2 2 2 2 (3.4)
U   F R ,
where we have defined the potential energy such that U  0 when     0 . The

- 24 -
 Chapter 3 Lagrangian Mechanics

Lagrangian is given by
1 1
L  T U  mR 2 2  mR 2 sin 2   2  F R . (3.5)
2 2
Upon inspection of the Lagrangian, we can see that there are two degrees of freedom for
this problem, i.e.,  and  . We now need to calculate the different derivatives that
compose the Lagrange equations
L
 mR 22 sin   cos    F R

L
0

(3.6)
d  L  d
 
dt    dt
 
mR 2  mR 2

d  L  d
 
dt    dt
 
mR 2 sin 2    mR 2 2  
  sin   cos    sin 2  ),

applying Lagrange’s equations for  and  we find the equations of motion to be


F  mR   2 sin   cos   
(3.7)

0  mR 2 sin   sin    2 
  cos   .

Incidentally, this problem was analyzed at the end of Chapter 1 on Newtonian Mechanics
(problem 1-1, with F  0 ), where Newton’s equation was used to solve the problem. One
can see how simpler the present treatment is. There was no need to calculate relatively
complex equations like er . Furthermore, it is important to realize that the spherical
coordinates  and  are treated as Cartesian coordinates when using the Lagrangian
formalism.
Example 3-1
1) The simple pendulum. Let’s solve the problem of the simple pendulum (of mass m and
length l) by first using the Cartesian coordinates to express the Lagrangian, and then
transform into a system of cylindrical coordinates.

Figure 3-1 A simple pendulum of mass m and length l .

- 25 -
 Chapter 3 Lagrangian Mechanics

Solution. In Cartesian coordinates the kinetic and potential energies, and the Lagrangian
are
1 2 1 2
T mx  my
2 2
U  mgy (3.8)
1 2 1 2
L  T U  mx  my  mgy.
2 2
We can now transform the coordinates with the following relations
x  l sin  
(3.9)
y  l cos  
Taking the time derivatives, we find
x  l cos( )
y  l sin  

L
1
2
 
m l 2 2 cos 2    l 2 2 sin 2    mgl cos   (3.10)

1
 ml 2 2  mgl cos   .
2
We can now see that there is only one generalized coordinates for this problem, i.e., the 
angle. We can use Lagrange’s equation to find the equation of motion
L
  mgl sin  

(3.11)
d  L  d
 
dt    dt
 
ml 2  ml 2,

and finally
g
  sin    0. (3.12)
l
2) The double pendulum. Consider the case of two particles of mass m1 and m2 each
attached at the end of a mass less rod of length l1 and l2, respectively. Moreover, the second
rod is also attached to the first particle (see Figure 3-2). Derive the equations of motion for
the two particles.
Solution. It is desirable to use cylindrical coordinates for this problem. We have two
degrees of freedom, and we will choose 1 and  2 as the independent variables. Starting
with Cartesian coordinates, we write an expression for the kinetic and potential energies for
the system

- 26 -
 Chapter 3 Lagrangian Mechanics

X
l1
1
m1

l2
2

Y
m2
Figure 3-2 The double pendulum.

 m1  x12  y12   m2  x2 2  y 2 2  

1
T
2  (3.13)
U  m1 gy1  m2 gy2
with
x1  l1 sin 1 
y1  l1 cos 1 
(3.14)
x2  l1 sin 1   l2 sin  2 
y2  l1 cos 1   l2 cos  2 
and
x1  l11 cos 1 
y  l  sin  
1 1 1 1
(3.15)
x2  l11 cos 1   l22 cos  2 
y  l  sin    l  sin  
2 1 1 1 2 2 2

Inserting equations (3.14) and (3.15) in (3.13), we get

T
2   
m1l1 1  m2 l1212  l2 22 2  2l1l212 cos 1  cos  2   sin 1  sin  2  
1  2 2

1

  m1l1212  m2 l1212  l2 22 2  2l1l212 cos 1   2  
2
 (3.16)

U    m1  m2  gl1 cos 1   m2 gl2 cos  2  ,

and for the Lagrangian

- 27 -
 Chapter 3 Lagrangian Mechanics

L  T U


1  2 2
2   
m1l1 1  m2 l1212  l2 22 2  2l1l212 cos 1   2   (3.17)

  m1  m2  gl1 cos 1   m2 gl2 cos  2  .

Inspection of equation (3.17) tells us that there are two degrees of freedom for this
problem, and we choose 1 and  2 as the corresponding generalized coordinates. We now
use this Lagrangian with Lagrange’s equation
L
 m2l1l212 sin(1   2 )   m1  m2  gl1 sin 1 
1
d L
dt 1
 
  m1  m2  l121  m2l1l2 2 cos(1   2 )  2 1  2 sin(1   2 ) 
(3.18)
L
 m2l1l212 sin(1   2 )  m2 gl2 sin  2 
 2
d L
dt 2
  
 m2 l2 22  l1l2 1 cos(1   2 )  1 1  2 sin(1   2 )  , 
and
 m1  m2  l121  m2l1l2 2 cos(1   2 )  2 2 sin(1   2 )    m1  m2  gl1 sin 1   0
(3.19)
 
m2 l2 22  l1l2 1 cos(1   2 )  12 sin(1   2 )   m2 gl2 sin  2   0.

We can rewrite these equations as

m2 l2  g
1   2 cos(1   2 )  2 2 sin(1   2 )   sin 1   0

m1  m2 l1 l1
(3.20)
l g
2  1 1 cos(1   2 )  12 sin(1   2 )   sin  2   0.
l2 l2
3) The pendulum on a rotating rim. A simple pendulum of length b and mass m moves on a
mass-less rim of radius a rotating with constant angular velocity  (see Figure 3-3). Get the
equation of motion for the mass.
Solution. If we choose the center of the rim as the origin of the coordinate system, we
calculate
x  a cos t   b sin  
(3.21)
y  a sin t   b cos   ,
and
x   a sin t   b cos  
(3.22)
y  a cos t   b sin   .

- 28 -
 Chapter 3 Lagrangian Mechanics

Figure 3-3 A simple pendulum attached to a rotating rim.

The kinetic and potential energies, and the Lagrangian are

T
1
2

m a 2 2  b 2 2  2ab sin   cos t   sin t  cos    
1

 m a 2 2  b 2 2  2ab sin   t 
2
 (3.23)
U  mg  a sin t   b cos   

L  T U 
1
2
 
m a 2 2  b 2 2  2ab sin   t   mg  a sin t   b cos   

We now calculate the derivatives for the Lagrange equation using  as the sole generalized
coordinate
L
 mab cos   t   mgb sin  

(3.24)
d  L 
dt   
2   
    mb   mab    cos   t  .
Finally, the equation of motion is
a g
   2 cos   t   sin    0
b b (3.25)
.
We recover the standard equation of motion for the pendulum when a or ω vanish.
Note that the terms
m 2 2
a   mga sin t 
2

- 29 -
 Chapter 3 Lagrangian Mechanics

in the Lagrangian (3.23) play no role in determining the dynamics of the system. In fact, as
can easily be shown, a Lagrangian L is always defined up to an exact time derivative, i.e.,
the Lagrangians L and L’ = L − df/dt, where f(q,t) is an arbitrary function, lead to the same
Lagrange equations. In the present case,
m 2 2 mga
f (t )  a  t cos t 
2 
and thus this term can be omitted from the Lagrangian (3.23) without changing the
equations of motion.

Figure 3-4 A bead slides along a smooth wire that rotates about z  cr 2

4) A bead slides along a smooth wire that has the shape of a parabola (see Figure 3-4). At
equilibrium, the bead rotates in a circle of radius R when the wire is rotating about its
vertical symmetry axis with angular velocity . Find the value of c.
Solution. We choose the cylindrical coordinates r, , and z as generalized coordinates. The
kinetic and potential energies are

T
1
2

m r 2  r 2 2  z 2  (3.26)
U  mgz.
We have in this case some equations of constraints that we must take into account, namely
z  cr 2
(3.27)
z  2crr
,
and
  t
(3.28)
  .
Inserting equations (3.27) and (3.28) in equation (3.26), we can calculate the Lagrangian
for the problem

- 30 -
 Chapter 3 Lagrangian Mechanics

m  r 2  4c 2 r 2 r 2  r 2 2   mgcr 2 .
1
L  T U  (3.29)
2
It is important to note that the inclusion of the equations of constraints in the Lagrangian
has reduced the number of degrees of freedom to only one, i.e., r. We now calculate the
equation of motion using Lagrange’s equation
L
 m  4c 2 rr 2  r 2  2 gcr 
r
(3.30)
d L
 m   r  8c 2 rr 2  ,
r  4c 2 r 2 
dt r
and
r 1  4c 2 r 2   r 2  4c 2 r   r  2 gc   2   0.
 (3.31)

When the bead is in equilibrium, we have r  R and r  

r  0 , and equation (3.31)
reduces to
R  2 gc   2   0, (3.32)

or
2
c . (3.33)
2g

3.2 Conservation Theorems

Before deriving the usual conservation theorem using the Lagrangian formalism, we must
first consider how we can express the kinetic energy as a function of the generalized
coordinates and velocities.
3.2.1 The Kinetic Energy
In a Cartesian coordinates system the kinetic energy of a system of particles is expressed as
1
T  m r2 .
2 
(3.34)

In order to derive the corresponding relation using generalized coordinates and velocities,
we go back to the equations, which relates the two systems of coordinates
r  r (q j ,t), j  1,2,...,3n  m. (3.35)

Taking the time derivative of this equation we have

r r
r   q j   , (3.36)
j q j t
and squaring it

- 31 -
 Chapter 3 Lagrangian Mechanics

r r r r r r
r r    q j qk  2    q j     . (3.37)
j ,k q j qk j q j t t t
An important case occurs when a system is scleronomic, i.e., there is no explicit
dependency on time in the coordinate transformation, we then have
r
 0, (3.38)
t
and the kinetic energy can be written in the form
T   a jk q j qk (3.39)
jk

with
1 n r r
a jk   m  g  . (3.40)
2  1 q j qk

Just as was the case for Cartesian coordinates, we see that the kinetic energy is a quadratic
function of the (generalized) velocities. If we next differentiate equation (3.39) with
respect to q j , and then multiply it by q j (and summing), we get
T
 q
j
j
q j
 2 a jk q j qk  2T
jk
(3.41)

since a jk is not a function of the generalized velocities, and it is symmetric in the

exchange of the j and k indices.
3.2.2 Conservation of Energy
Consider a general Lagrangian, which will be a function of the generalized coordinates and
velocities and may also depend explicitly on time (this dependence may arise from time
variation of external potentials, or from time-dependent constraints). Then the total time
derivative of L is
dL L L L
 q j   qj  . (3.42)
dt j q
j j q
j t
But from Lagrange’s equations,
L d  L 
   , (3.43)
q j dt  q j 
and equation (3.42) can be written as
dL d  L  L L d  L  L
   q j   qj    q  .
  j  t
(3.44)
dt  
j dt  q j q
j t j dt  q
j  j 
It therefore follows that

- 32 -
 Chapter 3 Lagrangian Mechanics

d  L  L dH L
 
dt j  q j
q j  L  
 t

dt

t
0 (3.45)

or
dH L
 , (3.46)
dt t
where we have introduced a new function
L
H  q j  L. (3.47)
j q j
In cases where the Lagrangian is not explicitly dependent on time we find that
L
H  q j  L  cste. (3.48)
j q j
If we are in presence of a scleronomic system, where there is also no explicit time
dependence in the coordinate transformation (i.e., x , i  x , i (q j ) ), then U  U (q j ) and
U / q j  0 and
L  (T  U ) T
  . (3.49)
q j q j q j

Equation (3.48) can be written as

L
H  q j  L  2T  L  T  U  E  cste, (3.50)
j q j
where we have used the result obtained in equation (3.41) for the second line.
The function H is called the Hamiltonian of the system and it is equaled to the total energy
only if the following conditions are met:
1. The equations of the transformation connecting the Cartesian and generalized
coordinates must be independent of time (the kinetic energy is then a quadratic
function of the generalized velocities).
2. The potential energy must be velocity independent.
It is important to realize that these conditions may not always be realized.

3.2.3 Noether’s Theorem

We can derive conservation theorems by taking advantage of the so-called Noether’s
theorem, which connects a given symmetry to the invariance of a corresponding physical
quantity.
Consider a set of variations  q j on the generalized coordinates that define a system,
which may or may not be independent. We write the variation of the Lagrangian as

- 33 -
 Chapter 3 Lagrangian Mechanics

 L L 
 L    qj   q j , (3.51)
 q q j 
j  j 

but from the Lagrange equations we have

L d  L 
   , (3.52)
q j dt  q j 
and
 d  L  L 
L      q j   q j 
j  dt  q j  q j 
(3.53)
d  L 
   q j .
dt  j q j 

Noether’s theorem states that any set of variations  q j (or symmetry) that leaves the
Lagrangian of a system invariant (i.e.,  L  0 ) implies the conservation of the following
quantity (from equation (3.53))
L
 q
j
 q j  cste (3.54)
j

3.2.4 Conservation of Linear Momentum

Consider the translation in space of an entire system. That is to say, every generalized
coordinates is translated by an infinitesimal amount such that
qj  qj  qj. (3.55)

Because space is homogeneous in an inertial frame, the Lagrangian function of a closed

system must be invariant when subjected to such a translation of the system in space.
Therefore,
 L  0. (3.56)
Equation (3.54) then applies and
L
 q
j
 q j  cste (3.57)
j

or
L
 cste (3.58)
q j

if  q j   q,  qk  j  0 . Let’s further define a new function

L
pj  . (3.59)
q j

- 34 -
 Chapter 3 Lagrangian Mechanics

Because of the fact that p j reduces to an “ordinary” component of the linear momentum
when dealing with Cartesian coordinates through
1 
   m x , k x , k 
L  T  U  T
    m x ,
2  ,k
p ,i    a  ,i (3.60)
x ,i x ,i x ,i x ,i
they are called generalized momenta. Inserting equation (3.59) into equation (3.58) we
find
p j  cste. (3.61)

The generalized momentum component is conserved. When dealing with Cartesian

coordinates the linear momentum in a given direction is also conserved. That is, from
equation (3.60) we have
p ,i  m x ,i  cste. (3.62)

Furthermore, whenever equation (3.58) applies (or equivalently L / q j  0 ), it is said that

the generalized coordinate q j is cyclic. We then find the corresponding generalized
momentum component p j to be a constant of motion, i.e.,

If the generalized coordinate q j is cyclic, then the corresponding

generalized momentum component p j to be a constant of motion.

3.2.5 Conservation of Angular Momentum

Since space is isotropic, the properties of a closed system are unaffected by its orientation.
In particular, the Lagrangian will be unaffected if the system is rotated through a small
angle. Therefore,

L 
d

p  q  0.
dt j j
 (3.63)

Let’s now consider the case where the Lagrangian is expressed as a function of Cartesian
coordinates such that we can make the following substitutions

q j  x ,i
(3.64)
p j  p ,i  ma x ,i .
Referring to Figure 3-5, we can express the variation  r in the position vector r caused
by an infinitesimal rotation  θ as
 r    r (3.65)
Inserting equations (3.64) and (3.65) in equation (3.63) we have

- 35 -
 Chapter 3 Lagrangian Mechanics

d  d 
 
dt   ,i
p ,i x ,i     p   r 
 dt   
d  d  d 
   p    r       r  p       r  p  ,
dt    dt    dt   
(3.66)
where we used the identity a   b  c   b   c  a   c   a  b  . We further transform equation
(3.66) to
d  d
   
dt  
r  p      L  0
 dt 
(3.67)

with L  r  p is the angular momentum vector associated with the particle identified
with the index . But since  θ is arbitrary we must have
n
L    r  p   ctse. (3.68)
 1

that is, the total angular momentum L is conserved.

Figure 3-5 – A system rotated by an infinitesimal amount  θ .

It is important to note that although we only used Noether’s theorem to prove the
conservation of the linear and angular momenta, it can also be used to express the
conservation of energy. But to do so requires a relativistic treatment where time is treated
on equal footing with the other coordinates (i.e., x ,i or q j ).

3.3 Dissipative Forces and Rayleigh’s Dissipative Function

So far, we have only dealt with system where there is no dissipation of energy. Lagrange’s
equations can, however, be made to accommodate some of these situations. To see how this

- 36 -
 Chapter 3 Lagrangian Mechanics

can be done, we will work our way backward from Lagrange’s equation
d  L  L
    0. (3.69)
dt  q j  q j
If we allow for the generalized forces on the system Q j to be expressible in the following
manner
U d  U 
Qj      , (3.70)
q j dt  q j 
then equation (3.69) can be written as
d  T  T
    Qj. (3.71)
dt  q j  q j

We now allow for some frictional forces, which cannot be derived from a potential such as
expressed in equation(3.70), but for example, are expressed as follows
f j   k j q j . (3.72)
Expanding our definition of generalized forces to include the friction forces
Qj  Qj  f j , (3.73)

Equation (3.71) becomes

d  T  T U d  U 
        f j , (3.74)
dt  q j  q j q j dt  q j 
or, alternatively
d  L  L
    f j. (3.75)
dt  q j  q j

Dissipative forces of the type shown in equation (3.75) can be derived in term of a
function R, known as Rayleigh’s dissipation function, and defined as

1
R 
2 j
k j q j 2 . (3.76)

From this definition it is clear that

R
fj   , (3.77)
q j
and the Lagrange equations with dissipation becomes
d  L  L R
     0, (3.78)
dt  q j  q j q j

- 37 -
 Chapter 3 Lagrangian Mechanics

so that two scalar functions, L and R, must be specified to obtain the equations of motion.

3.4 Velocity-dependent Potentials

Although we exclusively studied potentials that have no dependency on the velocities, the
Lagrangian formalism is well suited to handle some systems where such potentials arise.
This is the case, for example, when the generalized forces can be expressed with equation
(3.70). That is,
U d  U 
Qj      . (3.79)
q j dt  q j 
This equation applies to a very important type of force field, namely, the electromagnetic
forces on moving charges.
Consider an electric charge, q, of mass m moving at velocity v in a region subjected to an
electric field E and a magnetic field B, which may both depend on time and position. As is
known from electromagnetism theory, the charge will experience the so-called Lorentz
force
F  q E   v  B   . (3.80)

Both the electric and the magnetic fields are derivable from a scalar potential  and a
vector potential A by
A
E    , (3.81)
t
and
B    A. (3.82)
The Lorentz force on the charge can be obtained if the velocity-dependent potential energy
U is expressed
U  q  qA  v, (3.83)
so that the Lagrangian is
1 2
L  T U  mv  q  qA  v. (3.84)
2

- 38 -
 Chapter 4 Variational Principle of Mechanics

Chapter 4
Variational Principle of Mechanics

So far, we have studied laws of motion via Newton and Lagrange equations. The form of
the Lagrangian, L(q, q , t ) , naturally depends upon the system under study. It is a difference
in kinetic and potential energy of the system. It is also seen that symmetries of the problem
are also reflected in the Lagrangian. With this in mind an alternate and more general
formulation of mechanics is possible, and is called as a variational formulation or so-called
Hamilton’s Principle. Although the method based on Hamilton’s Principle does not
constitute in itself a new physical theory, it is probably justified to say that it is more
fundamental than Newton’s equations. This is because Hamilton’s Principle can be applied
to a much wider range of physical phenomena than Newton’s theory (e.g., quantum
mechanics, quantum field theory, electromagnetism, relativity). Variational methods unify
many areas of physics using energy as a key concept.

4.1 Statement of the Problem

In preparation for our study of variational principle or Hamilton’s principle, we must first
consider the principles and techniques of the calculus of variations.
Let us first define an integral J such that

J   f  y ( x), y '( x); x dx

x2
(4.1)
x1

where f is a function of the dependent variable y(x) and its derivative y '( x)  dy / dx (the
semicolon in f separates them from the independent variable x). The problem consists of
imposing virtual variations on y(x) and finding which variation brings the functional
integral J to an extremum. The limits of integration are kept fixed during this process. In
order to keep track of the different versions of y(x), we will extend its definition to include
a new parameter  such that
y ( , x)  y ( x)   ( x) (4.2)

where (x) is some function that has a continuous first derivative, and vanishes at the
integration limits. We therefore impose the condition that (x1)(x2)=0. We further define
the function corresponding to =0, y(0,x)=y(x), as the one that yields the extremum for J.
The situation is shown in Figure 4-1. For example, if y(x) brings J to a minimum, then any

- 39 -
 Chapter 4 Variational Principle of Mechanics

other function must make J increase. With the introduction of the parameter in equation
(4.2), the integral also becomes a function of 

J ( )   f  y ( , x), y '( , x); x dx

x2
(4.3)
x1

The condition necessary for equation (4.3) to be an extremum is

J
0 (4.4)
  0

for all function (x).

Figure 4-1 The function y(x) is the path that makes the functional J an extremum. The
neighboring functions y(x)+(x) vanish at the integration limits and may be close to y(x),
but they are not extrema.
Example
Consider the function f =(y’)2, with y=x. Add to y(x) the function (x)=sin(x), and find the
functional J() with integration limits 0 and 2. Show that the extremum of J() happens
when =0 .
Solution. We first extend y(x) to include  and (x)
y ( , x)  x   sin( x) (4.5)
We then calculate
2
J ( )   (1   cos( x)) 2 dx
0
2
  (1  2 cos( x)   2 cos 2 ( x))dx
0
(4.6)
 2
2 2 
  1   2 cos( x)  cos(2 x)  dx
0
 2 2 
 2    .
2

We see that J() is always greater than J(). It can also be verified that J /  | 0  0 .

- 40 -
 Chapter 4 Variational Principle of Mechanics

4.2 Euler’s Equation

We now calculate the derivative of J()(i.e., of equation (4.3)) to meet the condition
expressed by equation (4.4)
J  x2
  x1
 f { y ( x), y '( x); x}dx
(4.7)
x2  f y f y ' 
    dx
x1
 y  y '  
Since y(,x)=y(x)+(x), we have y /    ( x) and y ' /   d / dx . Inserting these
two derivatives in equation (4.7)
J x2  f f  
    dx. (4.8)
 x1
 y y ' x 
We integrate by parts the second term in the integral
x2  f   f x2 d  f 
x1   dx 
 y ' x  y '
 ( x) |xx12  
x1 dx  y ' 
 
 ( x)dx. (4.9)

The first term on the right hand side vanishes since (x1)(x2)=0. Inserting the remaining
term in equation (4.8) we get
J x2  f d  f  
    ( x)     ( x)  dx
 x1
 y dx  y '  
(4.10)
x2  f d  f  
      ( x)dx.
 y dx  y '  
x1

Even if (x) is subjected to the conditions that it vanishes at the integration limits and that it
has a continuous first derivative, it is otherwise completely arbitrary. It must be, therefore,
that the quantity that is between the parentheses in the last equation of (4.10) cancels
when J /  | 0  0 . That is

f d  f 
  0 (4.11)
y dx  y ' 
where y and y’ are now the original functions since  is now set to zero. Equation (4.11) is
the necessary condition for the functional J to be an extremum, and is known as Euler’s
equation.
Euler’s equation (i.e., equation (4.11)) previously derived is the solution of the variation
problem when only one dependent function y(x) was to be found to bring the integral J to
an extremum. When dealing with physical problems, it is more common that several such
functions are involved. We then have
f  { y1 ( x), y1 ' ( x),..., y n ( x), y n ' ( x); x}, (4.12)
or simply

- 41 -
 Chapter 4 Variational Principle of Mechanics

f  { y i ( x), y i ' ( x); x}, i  1,2,..., n. (4.13)

Just as was done for the case of a single function, we write
y i ( , x)  y i ( x)   i ( x), (4.14)
and
J x2  f d f 
     i ( x)dx, (4.15)
 x1
 y i dx y i ' 
where Einstein’s summation convention was used. Because the variation i(x) are all
independent of one another, the vanishing of equation (4.15) when implies that each
equation in parentheses vanishes simultaneously. That is
f d  f 
    0, i  1, 2,..., n. (4.16)
yi dx  yi ' 

4.3 The  Notation

In analysis that use calculus of variations, or in physics, we often encounter a different 

notation than what was presented in the preceding sections. The so-called  notation can be
used to rewrite equation (4.10) as
J x2  f d f  y
d      d dx, (4.17)
 x1
 y dx y '  
which is then expressed as
x2  f d f 
J      ydx (4.18)
x
 y dx y ' 
1

where
J
d   J

(4.19)
y
d   y.

The condition of extremum then becomes

 J    f  y, y '; x dx  0.
x2
(4.20)
x1

We now solve this equation

x2 x2  f f 
 J    fdx     y   y '  dx. (4.21)
x 1 x
 y y '
1

However,

- 42 -
 Chapter 4 Variational Principle of Mechanics

 dy   y ( x  x)  y ( x) 
 y'      lim 
 dx   x  0 x 
 y ( x  x)   y ( x)
 lim (4.22)
x  0 x
d
  y  .
dx
Inserting this equation in equation (4.21), we get
 f f d 
 y   dx
x2
J   y
x
 y
1 y ' dx 
(4.23)
x2  f d f 
     ydx,
x1
 y dx y ' 
where we have integrated by parts to derive the last equation. Since y is arbitrary, J can
only equal zero if the quantity in parentheses vanishes. This quantity is, as we saw before,
Euler’s equation.
It is important to realize that the  notation is only an expedient for the derivation of Euler’s
equation. The varied path represented by y is a virtual displacement from the actual path
that is consistent with the forces and constraint of the problem, but that is operated with the
independent variable x being kept fixed. (The independent variable, when we apply
calculus of variations to physical problems, will become the time. In that context, the varied
path y, in fact, does not even need to correspond to a possible path of motion.)

4.4 Hamilton’s Principle

Hamilton’s Principle is concerned with the minimization of a quantity (i.e., the action) in a
manner that is identical to extremum problems solved using the calculus of variations.
Hamilton’s Principle can be stated as follows:
The motion of a system from time t1 to t2 is such that the line integral (called the
action or the action integral),
t2
I   Ldt , (4.24)
t1

where L  T  U (with T and U the kinetic and potential energies, respectively), has a
stationary value for the actual path of the motion.
Note that a “stationary value” for equation (4.24) implies an extremum for the action, not
necessarily a minimum. But in almost all important applications in dynamics a minimum
occurs.
Because of the dependency of the kinetic and potential energies on the generalized
coordinates q j , the generalized velocities q j , and possibly the time t, it is found that

L  L  q j , q j , t  (4.25)

- 43 -
 Chapter 4 Variational Principle of Mechanics

Hamilton’s Principle can now be expressed mathematically by

  L  q j , q j , t  dt  0.
t2
(4.26)
t1

Equation (4.26) can readily be solved by the technique described above. The solution is
the Lagrange equations of motion
L d L
  0, j  1, 2,...,3n  m. (4.27)
q j dt q j

4.5 Examples

1. The shortest distance between two points in a plane. An element of length in a plane is
ds  dx 2  dy 2 (4.28)
and the total length of any curve going between point (x1,y1) and (x2,y2) is
2
( x2 , y2 ) x2  dy 
I  ds   1    dx. (4.29)
( x1 , y1 ) x1
 dx 
The condition that the curve be the shortest path is that I be a minimum. We, therefore,
define
2
 dy 
f  1   . (4.30)
 dx 
Inserting f in equation (4.11) with
f f y'
 0,  , (4.31)
y y ' 1  ( y ') 2
we have

d  y' 
 0 (4.32)
dx  1  ( y ') 2 

or
y'
 cste (4.33)
1  ( y ') 2
This can only be true if
y'  a (4.34)
where a is a constant. Integrating equation (4.34) finally yields
y  ax  b (4.35)

- 44 -
 Chapter 4 Variational Principle of Mechanics

with b an another constant (of integration). We recognize in equation (4.35) the equation
of a straight line.
2. The problem of the brachistochrone. Consider a particle constrained to move in a force
field (e.g., gravity), starting at rest from a point (x1,y1) to some lower point (x2,y2). Find the
path that allows the particle to accomplish the transit in the least possible time. We assume
that the force on the particle is constant and that there is no friction.
Solution. Let’s choose the point (x1,y1) as the origin, and assume that the force field is
directed along the positive x-axis (see Figure 4-2). Because of the assumptions of a constant
force field (i.e., the field is conservative) and the absence of friction, the total energy T +U
must be conserved. Furthermore, since the particle starts from rest at (x1,y1) and that we
chose this point as the origin (i.e., U=0), we have
T U  0 (4.36)
2
At the arrival point (x2,y2) we have T=mv /2, and U=-mgx. Using equation (4.36), the
velocity of the particle will, therefore, be
v  2 gx (4.37)

Figure 4-2 The brachistochrone problem.

We must find the path that will bring a particle subjected to the force field F from point
(x1,y1) to point (x2,y2) in the least amount of time.
The total time taken by the particle to travel the distance between the two points is given by
the following integral

1   y '
2
( x2 , y2 ) ds ( x2 , y2 ) dx 2  dy 2 x2
I    dx. (4.38)
( x1 , y1 ) v ( x ,
1 1y ) 2 gx x1  0 2 gx
We can, therefore, identify the function to be used in Euler’s equation as
1  ( y' ) 2
f  (4.39)
2 gx
When we apply Euler’s equation (4.11) we find that

- 45 -
 Chapter 4 Variational Principle of Mechanics

f
0 (4.40)
y
which in turn implies that

d  f  d  y' 
      0. (4.41)
dx  y '  dx  2 gx(1  ( y ') 2 ) 

This last equation can be written as

y' 1
 , (4.42)
x 1  ( y ') 2  2a

where a is a constant. If now square equation (4.42), rearrange the equation (we multiply
by x on both sides of the equality), and rewrite the result in the form of an integral we get
xdx
y . (4.43)
2ax  x 2
We now make the following change of variable
x  a (1  cos( ))
(4.44)
dx  a sin( )d ,
and equation (4.43) becomes

y   a 1  cos( ')  d '. (4.45)
0

This last integral is easily solved to give

y  a   sin     cste (4.46)
Setting the constant in equation (4.46) to zero, we now have the parametric equations of a
cycloid, that is
x  a 1  cos   
(4.47)
y  a   sin    .

The path of the particle described by equation (4.47) is shown in Figure 4.3.

- 46 -
 Chapter 4 Variational Principle of Mechanics

Figure 4.3 – The solution of the brachistochrone problem: a cycloid.

3. Minimum surface of revolution. Suppose we form a surface of revolution by taking some
curve passing between two fixed end points (x1,y1) and (x2,y2) defining the (x, y)-plane, and
revolving it around the y-axis (see Figure 4-4). The problem is to find the curve for which
the surface area is a minimum.

Figure 4-4 – Minimum surface of revolution. The geometry of the problem and area dA are
indicated to minimize the surface of revolution around the y-axis.

Solution. The area of a strip of surface is

dA  2 xdx 1   y ' 
2
(4.48)

and that for the total area is

A  2  x 1   y ' dx
x2 2
(4.49)
x1

To find the extremum we define

f  x 1   y ' ,
2
(4.50)

and insert it in equation (4.11). We find

- 47 -
 Chapter 4 Variational Principle of Mechanics

f f xy '
 0,  , (4.51)
y y ' 1   y '
2

which implies that

xy '
a (4.52)
1  ( y' ) 2
with a some constant. Squaring the above equation and factoring terms, we have
( y' ) 2 ( x 2  a 2 )  a 2 (4.53)
or solving,
dy a
 (4.54)
dx x  a2
2

The general solution of this differential equation is

dx x
y  a  a cosh 1    b, (4.55)
x2  a2 a
or
 y b 
x  a cosh  , (4.56)
 a 
where b is another constant of integration. Equation (4.56) is that of a catenary, the curve
of a flexible cord hanging freely between two points of support.

- 48 -
 Chapter 5 Hamilton’s Principle with Undetermined Multipliers

Chapter 5
Hamilton’s Principles with Undetermined
Multipliers

In the last chapter we derived Euler or Lagrange equations from the variational principle.
The variation discussed is free and is not subject to any constraints. In other words, the
variables ( yi or q j ) are independent. The problem of finding an extremum can be further
expanded to include cases where constraints are imposed. For example, determining the
shortest path between two points can be complicated by the fact that we may require the
path to follow the surface of a given shape (e.g., a sphere).

5.1 Euler’s Equation with Equations of Constraints

Let’s consider a variational problem where we have two dependent variables y and z with
one equation of constraint

J   f  y ( x), y '( x), z ( x), z ( x); x dx

x2
(5.1)
x1

We now write equation (4.15) for the case of two variables. There is, however, an equation
of constraint that we must somehow include in the problem. It is of the form
x2  f d f   f d f  
 J      y      z  dx, (5.2)
 y dx y '   z dx z '  
x1

g ( y , z; x )  0 (5.3)
For example, in the case of restricting the problem to a great circle (of radius a) on the
surface of a sphere, we would have
y 2  z 2  a 2  0. (5.4)
Because of equation(5.3), the variations  y and  z are no longer independent, so the
two terms in the parentheses of equation (5.2) do not vanish separately. If we differentiate
the equation of constraint, we have

- 49 -
 Chapter 5 Hamilton’s Principle with Undetermined Multipliers

g g g
dg  dy  dz  dx  0 (5.5)
y z x

In variational form equation (5.5) is written as

g g
g  y z 0 (5.6)
y z
We use the method of undetermined multiplier of Lagrange and introduce  ( x ) as multiplier
which is unknown (undetermined) so far. Then
 g g 
 ( x) g   ( x)   y   z   0. (5.7)
 y z 
We now insert equation (5.7) in equation (5.2) while slightly rearranging the terms
x2  f d f g   f d f g  
 J       ( x)   y      ( x)   z  dx, (5.8)
 y dx y ' y   z dx z ' z  
x1

If we choose  y is independent, we choose  ( x) in such a way that the last bracket in

equation (5.8) becomes zero. That is
f d f g
   ( x)  0. (5.9)
z dx z ' z
Then, the first bracket in equation (5.8) is also zero. Finally we have
 f d f   g 
     ( x)    0
 y dx y '   y  (5.10)
 f d f   g 
     ( x)    0
 z dx z '   z 
We, therefore, need to find a solution to three equations (i.e., y(x) , z(x), and  ( x) ) in order
to completely solve the problem. For this, the two equations (5.10) along with the
equation of constraint (5.3), for a total of three, will be used. The function  ( x) is known as
a Lagrange undetermined multiplier.
The procedure described in this section can readily be generalized to the case of several
dependent variables and more than a single equation of constraints are involved. We then
have the following solution
 f d f   g j 
     j ( x)  0
 yi dx yi   yi  (5.11)
g j { yi ; x}  0.
It is clear that the treatment in this section can be extended to the cases where equation of
constraints can be presented in a differential form such as
dg j  a ji dyi  a jt dt  0 (5.12)

- 50 -
 Chapter 5 Hamilton’s Principle with Undetermined Multipliers

In this case, the constraint equation for virtual displacement is

 g j  a ji yi  0 (5.13)
Then, Euler equation with Lagrange undetermined multiplier becomes
 f d f 
     j ( x)a ji  0
 yi dx yi  (5.14)
 g j  a ji yi  0

5.2 Lagrange’s Equations with Undetermined Multipliers

A system that is subjected to holonomic constraints (i.e., constraints that can be expressed
in the form f ( x ,i , t )  f (q j , t )  0 ) will always allow the selection of a proper set of
generalized coordinates for which the equations of motion will be free of the constraints
themselves. However, we can also integrate holonomic constraints into Lagrange equation
since the holonomic constraint equation can always be written as in the form of a
differential
f f
df  dq j  dt  0 (5.15)
q j t
Problems involving constraints such as the holonomic kind discussed here or in a
differential form like (5.12)can be handled in exactly the same manner as was done in the
last section. This is done by introducing the so-called Lagrange undetermined multipliers.
When this is done, we find that the following form for the Lagrange equations
L d L f
  k (t ) k  0, j  1, 2,...,3n  m; k  1, 2,..., m. (5.16)
q j dt q j q j
Although the Lagrangian formalism does not require the insertion of the forces of
constraints involved in a given problem, these forces are closely related to the Lagrange
undetermined multipliers. The corresponding generalized forces of constraints can be
expressed as
f k
Q j  k  t  . (5.17)
q j

5.3 Examples

1) The rolling disk on an inclined plane. We now solve the problem of a disk of mass m
and of radius R rolling down an inclined plane (see Figure 5-1).
Solution. Referring to Figure 5-1, and separating the kinetic energies in a translational
rotational part, we can write

- 51 -
 Chapter 5 Hamilton’s Principle with Undetermined Multipliers

1 2 1 2 1 2 1
T  my  I  my  mR 2 2 . (5.18)
2 2 2 4
where I  mR 2 / 2 is the moment of inertia of the disk about its axis of rotation. The
potential energy and the Lagrangian are given by
U  mg  l  y  sin  
1 2 1 (5.19)
L  T U  my  mR 2 2  mg  l  y  sin   .
2 4
where l and  are the length and the angle of the inclined plane, respectively. The equation
of constraint given by
f  y  R  0. (5.20)
This problem presents itself with two generalized coordinates (y and  ) and one equation
of constraints, which leaves us with one degree of freedom. We now apply the Lagrange
equations as defined with equation (5.16)
L d L f
   mg sin    my    0
y dt y y
(5.21)
L d L f 1
     mR 2   R  0.
 dt   2

Figure 5-1 -A disk rolling on an incline plane without slipping.

From the last equation we have

1
   mR, (5.22)
2
which using the equation of constraint (5.20) may be written as
1
   my. (5.23)
2
Inserting this last expression in the first of equation (5.21) we find

- 52 -
 Chapter 5 Hamilton’s Principle with Undetermined Multipliers

2
y
 g sin   , (5.24)
3
and
1
   g sin   . (5.25)
3
In a similar fashion we also find that
2 g sin  
  . (5.26)
3R
Equations (5.24) and (5.26) can easily be integrated, and the forces of constraints that
keep the disk from sliding can be evaluated from equation (5.17)
f 1
Qy       mg sin  
y 3
(5.27)
f 1
Q     R   mRg sin   .
 3
Take note that Qy and Q are a force and a torque, respectively. This justifies the
appellation of generalized forces.

5.4 Equation of Constraints as Integrals

It is also possible that an equation of constraints can be stated in the form of an integral. For
example, suppose that we want to find an extremum for a functional integral J by
determining the corresponding curve which satisfies the boundary conditions y(a) =A and
y(a) =B

J  y    f  y , y '; x dx.
b
(5.28)
a

Suppose also that the constraint is stated as a functional (or integral) K that fixes the value
for the length of the curve

K  y    g  y, y '; x dx  l
b
(5.29)
a

where l is the aforementioned length of y. Similarly as was done before, we now consider
virtual variations on the curve such that y ( , x)  y ( x)   ( x) . Since these variations
apply to both J and K, we can write
J  J ( )
K  K ( ),
and,

- 53 -
 Chapter 5 Hamilton’s Principle with Undetermined Multipliers

J
0
  0
(5.30)
K
0
  0

Furthermore, since K and K /  are not a function of x, we can introduce a Lagrange

undetermined multiplier such that
 J K  
     J   K  0  0. (5.31)
    0 
Take note that in this case  is truly a constant; it is not a function of the independent
variable x. We now rewrite equation (5.31) as

  f   g  dx 
b
0 (5.32)
 a 0

We can find the extremum for equation (5.32) in the same way as was done before, and
we get
f d f  g d g 
     0, (5.33)
y dx y '  y dx y ' 
with the constraints
y (a)  A, y (b)  B, and K [ y ]  l.

Example
We want to maximize the area of a closed curve y(x) that has a length l.
Solution. The area inside a closed curve and its length are
J   dxdy
(5.34)
K   ds  l.

Since the curve is closed, it is appropriate to change coordinates and consider the curve
r ( ) instead of y ( x) , with  as the independent variable instead of x. Equations (5.34)
can be rewritten as
2 r ( ) 1 2 2
J   r   d
2 0
r ' dr ' d 
0 0
(5.35)
2
K   r   d
0

We can now identify the two functions f and g

1 2
f  r ( ), g  r ( ). (5.36)
2
Using equation (5.33) we find

- 54 -
 Chapter 5 Hamilton’s Principle with Undetermined Multipliers

r ( )    0. (5.37)

which implies that r is a constant. Using equation (5.35) we find

l
r   
2
(5.38)
l2
J   r2  .
4
We therefore find that the area is maximized if the curve is that of a circle.

5.5 The Lagrangian Formulation for Continuous Systems

5.5.1 The Transition from a Discrete to a Continuous System

Let’s consider the case of an infinite elastic rod that can undergo small longitudinal
vibrations. A system composed of discrete particles that approximate the continuous rod is
an infinite chain of equal mass points spaced a distance a apart and connected by uniform
mass less springs having force constants k (see Figure 5-4).
Denoting the displacement of the jth particle from its equilibrium position by  j , the
kinetic and potential energies can be written as
1
T 
2 j
m j 2
(5.39)
U   k  j 1   j  .
1 2

2 j

Figure 5-4 A discrete system of equal mass springs connected by springs, as an

approximation to a continuous elastic rod.

The Lagrangian is then given by

m j 2  k  j 1   j  .
1 

2
L  T U   (5.40)
2 j  
which can also be written as

- 55 -
 Chapter 5 Hamilton’s Principle with Undetermined Multipliers

1 m   j 1   j  
2

L   a   j  ka 
2
    aL j , (5.41)
2 j a  a  
 j

where a is the equilibrium separation between the particles and L j is the quantity
contained in the square brackets. The particular form of the Lagrangian given in equation
(5.41) was chosen so that we can easily go to the limit of a continuous rod as a approaches
zero. In going from the discrete to the continuous case, the index j becomes the continuous
position coordinate x, and we therefore have
 j 1   j  ( x  a)    x  d
lim   , (5.42)
a 0 a x dx
where a takes on the role of dx. Furthermore, we have
m
lim   , lim ka  Y , (5.43)
a 0 a a 0

where  is the mass per unit length and Y is Young’s modulus (note that in the
continuous case Hooke’s Law becomes F  Yd / dx ). We can also impose the same
limit to the Lagrangian of equation (5.41) while taking equations (5.42) and (5.43) into
account. We then obtain
1  2 d  
2

L      Y   dx. (5.44)
2   dx  
This simple example illustrates the main features of passing from a discrete to a continuous
system. The most important thing to grasp is the role played by the position coordinates x.
It is not a generalized coordinates, but it now takes on the role of being a parameter in the
same right as the time is. The generalized coordinate is the variable    ( x, t ) . If the
continuous system were three-dimensional, then we would have    ( x, y , z , t ) , where x, y,
z and t would be completely independent of each other. We can generalize the Lagrangian
for the three-dimensional system as
L   L dxdydz , (5.45)

where L is the Lagrangian density. In the example of the one-dimensional continuous

elastic rod considered above we have
1 2 d  
2

L     Y   . (5.46)
2   dx  
5.5.2 The Lagrange Equations of Motion for Continuous Systems
We will now derive the Lagrange equations of motion for the case of a one-dimensional
continuous system. The extension to a three-dimensional system is straightforward. The
Lagrangian density in this case is given by
 d d 
L  L  , , , x, t  . (5.47)
 dx dt 

- 56 -
 Chapter 5 Hamilton’s Principle with Undetermined Multipliers

We now apply Hamilton’s Principle to the action integral in a way similar to what was
done for discrete systems
t2 x2
I   L dxdt  0. (5.48)
t1 x1

We then propagate the variation using the shorthand  notation

t2 x2
I   L dxdt
t1 x1

 
t2 x2  L L  d  L  d 
(5.49)
           dxdt
t1 x1
   d   dx   d   dt  
   
  dx   dt  

and since   d / dx   d   / dx and   d / dt   d   / dt , we have

 

t2 x2 L L d   L d   
 I        dxdt (5.50)
t1 x1
   d  dx  d  dt 
   
  dx   dt  

Integrating the last two terms on the right hand side by parts we finally get
t2 x2
I   L dxdt
t1 x1

    

t2 x2 L d  L  d  L  
(5.51)
        dxdt  0.
t1 x1
  dx    d   dt    d   

   dx     dt   

Since the virtual variation  is arbitrary, we have for the Lagrange equations of motion
of a continuous system
   
L d  L  d  L 
    0 (5.52)
 dx   d   dt   d  
 
  dx     dt  

Applying equation (5.52) to our previous Lagrangian density for the elastic rod (i.e.,
equation (5.46), we get
d 2 d 2
  Y  0. (5.53)
dt 2 dx 2
This is the so-called one-dimensional wave equation, which has for a general solution
 ( x, t )  f ( x  vt )  g ( x  vt ), (5.54)

where f and g are two arbitrary functions of x+vt and x-vt , and v  Y /  .

- 57 -
 Chapter 6 Hamiltonian Dynamics

Chapter 6
Hamiltonian Dynamics

The variables of the Lagrangian are the generalized coordinates and the accompanying
generalized velocities. In Hamilton's theory, the generalized coordinates and the
corresponding momenta are used as independent variables. In this theory the position
coordinates and the "momentum coordinates" are treated on an equal basis. Hamiltonian
theory leads to an essential understanding of the formal structure of mechanics and is of
basic importance for the transition from classical mechanics to quantum mechanics.

6.1 The Canonical Equations of Motion

As we saw before, the generalized momentum is defined by

L
pj  (6.1)
q j
We can rewrite the Lagrange equations of motion
d  L  L
   0 (6.2)
dt  q j  q j
as follow
dp j L
 (6.3)
dt q j
We can also express our earlier equation for the Hamiltonian using equation (6.1) as
H   p j q j  L (6.4)
j

Equation (6.4) is expressed as a function of the generalized momenta and velocities, and
the Lagrangian, which is, in turn, a function of the generalized coordinates and velocities,
and possibly time. There is a certain amount of redundancy in this since the generalized
momenta are obviously related to the other two variables through equation (6.1). We can,
in fact, invert this equation to express the generalized velocities as
q j  q j  qk , pk , t  (6.5)

- 58 -
 Chapter 6 Hamiltonian Dynamics

This amounts to make a change of variables from  q , q , t 

j j to  q , p , t  ; we, therefore,
j j

express the Hamiltonian as

H (qk , pk , t )   p j q j  L(qk , qk , t ) (6.6)
j

where it is understood that the generalized velocities are to be expressed as a function of the
generalized coordinates and momenta through equation (6.5). The Hamiltonian is,
therefore, always considered a function of the (q j , p j ,t) set, whereas the Lagrangian is a
function of the (q j , q j , t ) set:
H  H (qk , pk , t )
(6.7)
L  L(qk , qk , t )
Let’s now consider the following total differential
 H H  H
dH    dqk  dpk   dt. (6.8)
k  qk pk  t
but from equation (6.6) we can also write
 L L  L
dH    qk dpk  pk dqk  dqk  dqk   dt (6.9)

k  qk q j  t

Replacing the third and fourth terms of equation (6.9) by inserting equations (6.3) and
(6.1), respectively, we find
L
dH    qk dpk  p k dqk   dt (6.10)
k t
Equating this last expression to equation (6.8) we get the so-called Hamilton’s or
canonical equations of motion
H
qk 
pk
(6.11)
H
p k  
qk
and
dH L
 (6.12)
dt t
Furthermore, if we insert the canonical equations of motion in equation (6.8) we find
(after dividing by dt on both sides)
dH  H H H H  H
    (6.13)
dt k  qk pk pk qk  t
or

- 59 -
 Chapter 6 Hamiltonian Dynamics

dH H
 (6.14)
dt t
Thus, the Hamiltonian is a conserved quantity if it does not explicitly contain time.
We can therefore choose to solve a given problem in two different ways: i) solve a set of
second order differential equations with the Lagrangian method or ii) solve twice as many
first order differential equations with the Hamiltonian formalism. It is important to note,
however, that it is sometimes necessary to first find an expression for the Lagrangian and
then use equation (6.6) to get the Hamiltonian when using the canonical equations to solve
a given problem. It is not always possible to straightforwardly find an expression for the
Hamiltonian as a function of the generalized coordinates and momenta.

Example
Use the Hamiltonian method to find the equations of motion for a spherical pendulum of
mass m and length b.

Figure 5-1 A spherical pendulum with generalized coordinates

Solution. The generalized coordinates are  and  . The kinetic energy is therefore
1
T m( x 2  y 2  z 2 )
2
1   d  
2 2 2
 d  d
 m    b sin   cos(      b sin   sin(      b cos(   
2   dt   dt   dt  
1
2 
 mb 2  cos   cos     sin   sin    
2
(6.15)
2

 cos   sin     sin   cos      sin   
2


2

1 2  2 2 2
mb    sin   
With the usual definition for x, y, and z, the potential energy is
U  mgb cos   (6.16)

- 60 -
 Chapter 6 Hamiltonian Dynamics

The generalized momenta are then

L
p   mb 2

(6.17)
L
p    mb 2 sin 2   

And using these two equations we can solve for  and  as a function of the generalized
momenta.
Since the system is conservative and that the transformation from Cartesian to spherical
coordinates does not involve time, we have
H  T U
2
2
1 2  p  1 2 2  p 
 mb  2   mb sin    2  2   mgb cos   (6.18)
2  mb  2  mb sin   
p 2 p 2
   mgb cos  
2mb 2 2mb 2 sin 2  
Finally, the equations of motion are
H p
   2
p mb
H p
  
p mb sin 2  
2

(6.19)
H p 2 cos  
p     2 3  mgb sin  
 mb sin  
H
p    0

Because  is cyclic (or ignorable), the generalized momentum p about the symmetry
axis is a constant of motion. p is actually the component of the angular momentum along
the z-axis.

6.2 Hamiltonian as a Legendre Transform of Lagrangian

6.2.1 Legendre Transform

Consider a function of two kinds of variables
f  f ( x1 , x2 ,..., xn ; y1 , y2 ,..., yn ) (6.20)
Here t is a parameter. F is differentiable. The set of variables  xi  is called active variable
and the set of  yi  is called passive variable. We define,

- 61 -
 Chapter 6 Hamiltonian Dynamics

f f
 ui ,  vi i  1, 2,..., n (6.21)
xi yi
Then
f
df    ui dxi  vi dyi   dt (6.22)
i t
Define
g   ui xi  f (6.23)
i

The function g is called Legendre transform of f. Then,


dg   ui dxi  xi dui  df 
i

f

  ui dxi  xi dui  ui dxi  vi dyi   t
dt (6.24)
i

f

  xi dui  vi dyi   t
dt
i

Thus,
g g g f
 xi ,  vi ,  . (6.25)
ui yi t t
Also,
g  g (u1 , u2 ,..., un ; y1 , y2 ,..., yn ; t ) (6.26)
That is, g is a function of ui and yi . Comparing (6.25) with (6.21), we see that, they are
“converse like”.
Now, comparing (6.6) with (6.23), we see that Hamiltonian H is just the Legendre
transform of Lagrangian L
H (qk , pk , t )   p j q j  L(qk , qk , t ) (6.27)
j

with
L L
 p j ,  pj (6.28)
q j q j
and
H H H L
  p j ,  q j ,  (6.29)
q j p j t t
Equations (6.28) and (6.29) are Lagrange’s equations and Hamilton’s equations
respectively.

- 62 -
 Chapter 6 Hamiltonian Dynamics

6.3 Deriving Hamilton’s Equations from Hamilton’s Principle

In the preceding chapter we derived Lagrange’s equation from Hamilton’s Principle

  L  q j , q j , t  dt  0
t2
(6.30)
t1

As it turns out, it is also possible to obtain the canonical equations of motions using the
same variational principle. To do so, we start from equation (6.6) to express the
Lagrangian as a function of the Hamiltonian
L(q j , q j , t )   pk qk  H (q j , p j , t ) (6.31)
k

Inserting equation (5.21) in (5.20) we have

 
    pk qk  H (q j , p j , t )  dt  0
t2
(6.32)
t 1
 k 
and carrying the variation on the integrand
 H H 
   p  q
t2
 qk  pk   qk   pk dt  0 (6.33)
qk pk
k k
t1
k 
But since  qk  d  qk  / dt , we can write

  p  q dt    p
t2

t1
k
k k
t2

t1
k
k
d
dt k
 k

 qk dt   pk qk |tt12  t1 p k qk dt   t1 p k qk dt
t2 t2

(6.34)
Using equation (6.34) into equation (6.33)
 H   H  
    q
t2
   pk   pk    qk dt  0 (6.35)
pk  qk 
k
  
t1
k

Since the  qk and  pk are independent variations, we retrieve Hamilton’s (or the
canonical) equations
H
qk 
pk
(6.36)
H
p k  
qk
Important points about Hamilton’s equations are:
1. H, the Hamiltonian is a function of q j and p j .
2. Hamilton’s equations are the first order ones while Lagrange’s equations are the second
order.
3. If qk is cyclic or ignorable in Lagrangian, it is also cyclic in Hamiltonian.

- 63 -
 Chapter 6 Hamiltonian Dynamics

Corresponding momentum pk is constant, equal to  . Then,

H  H (q1 ,..., qk 1 , qk 1 ,..., qn , p1 ,..., pk 1 ,  , pk ,..., pn ) (6.37)
and,
H H
  p j ,   q j jk (6.38)
q j p j
and there is no equation corresponding qk . But in Lagrangian formulation we have an
equation
d L
0 (6.39)
dt qk
In this sense, ignorable coordinates is truly ignored in Hamiltonian formulation.
4. Hamiltonian is total energy=T+U when the system is conservative. If L is independent
of time t, H is conserved because H is also independent of t.

6.4 The Poisson Bracket

6.4.1 The Poisson Bracket and Canonical Transformations

The Poisson bracket of two functions u and v with respect to the canonical variables
q j and p j is defined as

 u v u v 
u, vq, p      (6.40)
j  q j p j p j q j 
where a summation on the repeated index is implied. Suppose that we choose the functions
u and v from the set of generalized coordinates and momenta, we then get the following
relations

q ,q   p , p 
j k q, p j k q, p
 0, (6.41)

and
q , p 
j k q, p
   p j , qk 
q, p
  jk (6.42)

where  jk is the Kronecker delta tensor. Alternatively, we could choose to express u and v
as functions of a set Q j and Pj , which are canonical transformations of the original
generalized coordinates qk and pk , and calculate the Poisson bracket in this coordinate
system. To simplify calculations, we will limit our analysis to so-called restricted
canonical transformations where time is not involved in the coordinate transformations.
That is,

- 64 -
 Chapter 6 Hamiltonian Dynamics

Q j  Q j (qk , pk )
(6.43)
Pj  Pj (qk , pk ),

or, alternatively
 
q j  q j Qk , Pk
(6.44)
pj  p Q , P .
j k k

In order to carry our calculation more effectively, we will introduce a new notation. Let’s
construct a column vector η , which for a system with n degrees of freedom is written as
 j  q j ,  j  n  p j , j  1,..., n. (6.45)

Similarly, another column vector involving the derivatives of the Hamiltonian relative to
the component of η is given by
 H  H  H  H
    q ,     p , j  1,..., n. (6.46)
j j jn j

Finally, we define a 2n  2n matrix J as

 0 1
J   (6.47)
 1 0 
where the matrices 0 and 1 are the n  n zero and unit matrices, respectively. Using
equations (6.46) and (6.47), we can write the canonical equations as
H
  J (6.48)

It can be easily verified that equation (6.40) for the Poisson bracket can also be rewritten as
T
 u  v
u, v   J (6.49)
   

 u / η is the transpose of u / η and, therefore, a row vector. We can also

T
where
compactly write the results of equations (6.41) and (6.42) as
 ,  J (6.50)

Equations (6.50) are called the fundamental Poisson brackets.

Just as we did for the generalized coordinates qk and pk with equation (6.45), we introduce
a new column vector ζ such that
 j  Q j ,  j  n  Pj , j  1,..., n. (6.51)

Relations similar to equations (6.46), (6.48) and (6.49) exist for ζ just as they do for η .
We now introduce the Jacobian matrix M relating the time derivatives between the two
systems

- 65 -
 Chapter 6 Hamiltonian Dynamics

  M (6.52)
where
 j
M jk  . (6.53)
k
Transformation of derivatives of the Hamiltonian can also be expressed as a function of the
Jacobian matrix
H H  k H
  M kj , (6.54)
 j  k  j  k
or
H H
 MT (6.55)
 
Using equations (6.52), (6.48) and (6.55) we can write
H H
  M  MJMT J . (6.56)
 
This last result implies that
MJMT  J. (6.57)
We are now in a position to obtain a few important results. First, from equations (6.49),
(6.53) and (6.57), we have
T
   
 ,    J  MJMT  J. (6.58)
   
But me must also have a similar result for the fundamental Poisson bracket in the ζ system,
that is
 ,   J. (6.59)

We have, therefore, the result that the fundamental Poisson brackets are invariant under
canonical transformation. We will now show that this result also extends to any arbitrary
functions u and v. We first note that
u v
 MT . (6.60)
 
Inserting this equation in (6.49), we have
T T T
 u  v  u   v   u   v 
u, v   J    MJMT      J   (6.61)
               
and therefore
u,v  u,v  u,v.
 
(6.62)

- 66 -
 Chapter 6 Hamiltonian Dynamics

We thus find that all Poisson brackets are invariant to canonical transformations. This is
the reason why we dropped any index in the last expression in equation (6.62). Just as the
main characteristic of canonical transformations is that they leave the form of Hamilton’s
equations invariant, the canonical invariance of the Poisson brackets implies that equations
expressed in term of Poisson brackets are also invariant in form to canonical
transformations. It is important to note that, although we have limited our analysis to
restricted canonical transformations, the results obtained here can be shown to be equally
valid for coordinate transformations that explicitly involve time.
The algebraic properties of the Poisson bracket are therefore of considerable importance.
Here are some important relations
{u , u}  0
{u , v}  {v, u}
{au  bv, w}  a{u, w}  b{v, w}
(6.63)
{uv, w}  u{v, w}  {u, w}v
{u , vw}  v{u, w}  {u , v}w
{u ,{v, w}}  {v,{w, u}}  {w,{u, v}}  0.
The last of equations (6.63) is called Jacobi’s identity.

6.5.2 The Equations of Motion

Practically the entire Hamiltonian formalism can be restated in terms of Poisson brackets.
Because of their invariance to canonical transformations, the relations expressed with
Poisson brackets will be independent of whichever set of generalized coordinates and
momenta are used. Let’s consider, for example the total time derivative of a function u
du  u u  u
  q j  p j  
dt 
j  q j p j  t
(6.64)
 u H u H  u
     ,

j  q j p j p j q j  t
or
u
du
dt
 
 u, H  .
t
(6.65)

Equation (6.65) is indeed independent of the system of coordinates. Special cases for
equation (6.65) are that of Hamilton’s equations of motion
q j  q j , H 
(6.66)
p j   p j , H 

Equations (6.66) can be combined into a single vector equation using the notation defined
earlier

- 67 -
 Chapter 6 Hamiltonian Dynamics

   , H  (6.67)
We can also verify that
dH H H
 H , H    . (6.68)
dt t t
Finally, if a function u is a constant of motion, that is u  0 , then
u
{H , u}  (6.69)
t
and for function that are independent of time, we find that
{H , u}  0 (6.70)

- 68 -
 Chapter 7 Central Force Motion

Chapter 7
Central Force Motion

In this chapter we will study the problem of two bodies moving under the influence of a
mutual central force.

7.1 Reduction to the Equivalent One-body Problem – the Reduced Mass

We consider a system consisting of two point masses, m1 and m2 , when the only forces are
those due to an interaction potential U . We will assume that U is a function of the distance
between the two particles r = |r| = |r2-r1|. Such a system has six degrees of freedom. We
could choose, for example, the three components of each of the two vectors, r1 and r2 (see
Figure 7-1). However, since the potential energy is solely a function of the distance
between the two particles, i.e., U =U(r), it is to our advantage if we also express the kinetic
energy as function of r (that is, of r ). Let’s first define the center of mass R of the system
as
 m1  m2  R  m1r1  m2r2 . (7.1)
We also consider the distance between each particle and the center of mass
m1r1  m2r2 m2
r1  R  r1   r
m1  m2 m1  m2
(7.2)
m r  m2r2 m1
r2  R  r2  1 1  r
m1  m2 m1  m2
with r = r2-r1.

Figure 7-1 – The different vectors involved in the two-body problem.

69
 Chapter 7 Central Force Motion

We now calculate the kinetic energy of the system

1 1
T m1r12  m2r2 2
2 2
(7.3)
1    r  R
   1 m R
   r  R
 
2 2
 m1  R 1  2 2 2 
2
and inserting equations (7.2) in equation (7.4) we get
2 2
1  m2  1  m1 
T  m1  R  r   m2  R  r 
2  m1  m2  2  m1  m2 

1 2  
2
m2   m2
 m1  R  2 r  R   r   
2  m1  m2  m1  m2  
  (7.4)
1 2 
2

m2  R  2
m1    m1 r 
r  R 
2  m1  m2  m1  m2  

1  2  1 m1m2 r 2
  m1  m2  R
2 2 m1  m2
We introduce a new quantity  , the reduced mass, defined as
m1m2
 (7.5)
m1  m2
which can alternatively be written as
1 1 1
  (7.6)
 m1 m2
We can use equation (7.5) to write the kinetic energy as
1 1
T  m1  m2  R 2  r 2 (7.7)
2 2
We have, therefore, succeeded in expressing the kinetic energy as a function of r . We are
now in a position to write down the Lagrangian for the central force problem
1 1
L  T U   m1  m2  R 2  r 2  U (r ) (7.8)
2 2
where the potential energy U(r) is yet undefined, except for the fact that it is solely a
function of the distance between the two particles. It is, however, seen from equation (7.8)
that the three components of the center of mass vector are cyclic. More precisely,
L L L
  0 (7.9)
X Y Z
with R  Xe x  Ye y  Ze z . The center of mass is, therefore, either at rest or moving
uniformly since the equations of motion for X, Y, and Z can be combined into the following

70
 Chapter 7 Central Force Motion

vector relation
 m1  m2  R  0 (7.10)
Since the center of mass vector (and its derivative) does not appear anywhere else in the
Lagrangian, we can drop the first term of the right hand side of equation (7.8) in all
subsequent analysis and only consider the remaining three degrees of freedom. The new
Lagrangian is therefore
1 2
L  r  U (r ) (7.11)
2
What is left of the Lagrangian is exactly what would be expected if we were dealing with a
problem of a single particle of mass  subjected to a fixed central force. Thus, the central
force motion of two particles about their common center of mass is reducible to an
equivalent one-body problem.

7.2 The First Integrals of Motion

Since we are dealing with a problem where the force involved is conservative, where the
potential is a function of the distance r of the reduced mass to the force center alone, the
system has spherical symmetry. From our discussion of Noether’s theorem in Chapter 3 (cf.,
section 3.2.3) we know that for such systems the angular momentum is conserved. That is,
L  r  p  cste (7.12)
From the definition of the angular momentum itself we know that L is always parallel to
vectors normal to the plane containing r and p. Furthermore, since in this case L is fixed, it
follows that the motion is at all time confined to the aforementioned plane. We are,
therefore, fully justified to use polar coordinates as the two remaining generalized
coordinates for this problem (i.e., we can set the third generalized coordinate, say z, to be a
constant since the motion is restricted to a plane). Since r  re r , we have
r  re r  re r  re r  re (7.13)
In the last equation we have used the following relation (with the usual transformation
between the (r,) polar and the (x, y) and the Cartesian coordinates)
e r  cos   e x  sin   e y
(7.14)
e   sin   e x  cos   e y
from which it can be verified that
er e
e r  r  r   e (7.15)
r 
We can now rewrite the Lagrangian as a function of r and

L
1
2
 
 r 2  r 2 2  U  r  (7.16)

We notice from equation (7.16) that  is a cyclic variable. The corresponding generalized

71
 Chapter 7 Central Force Motion

momentum is therefore conserved, that is

L
p    r 2  ctse (7.17)

The momentum p is a first integral of motion and is seen to equal the magnitude of the
angular momentum vector. It is customarily written as
l   r 2  ctse (7.18)
As the particle (i.e., the reduced mass) moves along its trajectory through an infinitesimal
angular displacement d within an amount of time dt, the area dA swept out by its radius
vector r is given by
1 2 1
dA  r d  r 2dt (7.19)
2 2
Alternatively, we can define the areal velocity as
dA 1 2  l
 r  cste (7.20)
dt 2 2
Thus, the areal velocity is constant in time. This result, discovered by Kepler for planetary
motion, is called Kepler’s Second Law. It is important to realize that the conservation of
the areal velocity is a general property of central force motion and is not restricted to the
inverse-square law force involved in planetary motion.
Another first integral of motion (the only one remaining) concerns the conservation of
energy. The conservation is insured because we are considering conservative systems.
Writing E for the energy we have

E  T U 
1
2
 
 r 2  r 2 2  U  r  (7.21)

or using equation (7.18)

1 2 1 l2
E  r  U r  (7.22)
2 2 r 2

7.3 The Equations of Motion

We will use two different ways for the derivation of the equations of motion. The first one
consists of inverting equation (7.22) and express r as a function of E, l, and U(r) such
that
dr 2 l2
r    E  U  r    2 2 (7.23)
dt   r
or alternatively

72
 Chapter 7 Central Force Motion

dr
dt   (7.24)
2 l2
 E  U  r    2 2
  r
Equation (7.24) can be solved, once the potential energy U(r) is defined, to yield the
solution t  t  r  , or after inversion r = r (t ). We are, however, also interested in
determining the shape of the path (or orbit) taken by the particle. That is, we would like to
evaluate r  r   or     r  . To do so we use the following relation

d dt 
d  dr  dr (7.25)
dt dr r
Inserting equations (7.18) and (7.23) into equation (7.25), we get

  r   
 l / r  dr
2

 cste. (7.26)
 l2 
2  E  U  r   
 2 r 2 
It is important to note that the integral given by equation (7.26) can be solved analytically
only for certain forms of potential energy. Most importantly, if the potential energy is of the
form U  r   r n , for n = 2, -1, and -2 the solution is expressible in terms of circular
functions.
The second method considered here for solving the equations of motion uses the Lagrange
equations
L d L
 0
r dt r
(7.27)
L d L
 0
 dt 
The second of these equations was already used to get equation (7.18) for the conservation
of angular momentum. Applying the first of equations (7.27) to the Lagrangian (equation
(7.16)) gives
U
  
r  r 2     F r  (7.28)
r
We now modify this equation by making the following change of variable
1
u (7.29)
r
We calculate the first two derivatives of u relative to 

73
 Chapter 7 Central Force Motion

du 1 dr 1 dr dt
 2  2
d r d r dt d
1 (7.30)
r  l  
  2  2    r
r  r  l

where we have used the fact that   l /  r 2 , and for the second derivative
d 2u d    dt d     
r 2 2
   
r     
r      r 
r (7.31)
d 2 d  l  d dt  l  l  l2
From this equation we have
l2 d 2u
r 
 u2 (7.32)
 2
d 2
and from equation (7.18)
2
 l  l2
r 2  r  2   2 u 3 . (7.33)
 r  
Inserting equations (7.32) and (7.33) in equation (7.28) yields
d 2u  1 1
u   2 2 F  (7.34)
d 2
l u u
which can be rewritten as
d2  1 1 r 2
     F r (7.35)
d 2  r  r l2
Equation (7.35) can be used to find the force law that corresponds to a known orbit
r  r   .

Examples
1. Let’s consider the case where the orbit is circular, i.e., r  cste . Then from equation
(7.35) we find that
r 2
F  r   cste. (7.36)
l2
Equation (7.36) implies that
1
F r   (7.37)
r2
which is the functional form of the gravitational force.
2. Let’s assume that we have an orbit given by
r  ke (7.38)

74
 Chapter 7 Central Force Motion

From the second derivative of equation (7.38) relative to  　　 we have

d 2  1  d 2  e   d  
2
   
d 2  r  d 2  k 
  
k d
 e    2 e

k

r
(7.39)

and using equation (7.35), we get

l2   2 1  1
F r      (7.40)
r2  r r  r3

7.4 Centrifugal Energy and the Effective Potential

In equations (7.23) to (7.26) for dr, dt, and  , respectively, appeared a common term
containing different energies (i.e., the total, potential, and rotational energies)
 l2 
 E  U  r    (7.41)
 2 r 2 

The last term is the energy of rotation since

l2 1
  r 2 2 (7.42)
2 r 2
2
It is interesting to note that if we arbitrarily define this quantity as a type of “potential
energy” U c , we can derive a conservative force from it. That is, if we set

l2
Uc  (7.43)
2 r 2
then the force associated with it is
U c l2
Fc    3   r 2 (7.44)
r r
The force defined by equation (7.44) is the so-called centrifugal force. It would, therefore,
be probably better to call U c is the centrifugal potential energy and to include it with the
potential energy U(r) to form the effective potential energy V(r) defined as
l2
V (r )  U  r   (7.45)
2 r 2
If we take for example the case of an inverse-square-law (e.g., gravity or electrostatic), we
have
k
F (r )   (7.46)
r2
and

75
 Chapter 7 Central Force Motion

k
U (r )    F  r  dr   (7.47)
r
The effective potential is
k l2
V (r )    (7.48)
r 2 r 2
It is to be noted that the centrifugal potential “reduces” the effect of the inverse-square-law
on the particle. This is because the inverse-square-law force is attractive while the
centrifugal force is repulsive. This can be seen in Figure 7-2.
It is also possible to guess some characteristics of potential orbits simply by comparing the
total energy with the effective potential energy at different values of r. From Figure 7-3 we
can see that the motion of the particle is unbounded if the total energy (E1 ) is greater than
the effective potential energy for r  r1 when E1  V  r1  . This is because the positions for
which r  r1 are not allowed since the value under the square root in equation (7.41)
becomes negative, which from equation (7.23) would imply an imaginary velocity. For the
same reason, the orbit will be bounded with r2  r  r4 for a total energy E2 , where
V  r2   E2  V  r4  . The turning points r2 and r4 are called the apsidal distances. Finally,
the orbit is circular with r  r3 when the total energy E3 is such that E3  V  r3  .

Figure 7-2 -Curves for the centrifugal, effective, and gravitational potential energies.

76
 Chapter 7 Central Force Motion

Figure 7-3 – Depending on the total energy, different orbits are found.

7.5 Planetary Motion – Kepler’s Problem

The equation for a planetary orbit can be calculated from equation (7.26) when the
functional form for the gravitational potential is substituted for U(r)

  r   
 l / r  dr
2

 cste (7.49)
 k l2 
2  E   
 r 2 r 2 
This equation can be integrated if we first make the change of variable u = 1/r , the integral
then becomes
ldu
 u     cste
 l 2u 2 
2  E  kuU  r   
 2 
(7.50)
du
   cste
2
 E  ku   u 2

l2
the sign in front of the integral has not changed as it is assumed that the limits of the
integral were inverted when going from equation (7.49) to (7.50). We further transform
equation (7.50) by manipulating the denominator

77
 Chapter 7 Central Force Motion

du
 u     cste
 k 2 E  k 
2 2 2

  u  2 
 l 
4 2
l l
du
   cste (7.51)
2
 k  2 El 2   ul 2 
1  
2  
 1
l 2
 k   k 2

 du
   cste,
  u 1 
2

1  
  
with
l2 2 El 2
 , and   1  (7.52)
k k
Now, to find the solution to the integral of equation (7.51), let’s consider a function f (u)
such that
f  u   cos   u   . (7.53)

Taking a derivative relative to u we get

df d
  sin   (7.54)
du du
or
d 1 df 1 df 1 df
   (7.55)
du sin   du 1  cos 2   du 1  f 2 du
Returning to equation (7.51), and identifying f (u) with the following
u 1
f u   (7.56)

We find that
 u 1 
  u   cos 1   (7.57)
  
Or since cos     cos   ,

 1   cos     (7.58)
r
where  is some constant. If we choose r to be minimum when   0 , then   0 and we
finally have

78
 Chapter 7 Central Force Motion


 1   cos   (7.59)
r
with
l2 2 El 2
 , and   1  (7.60)
k k
Equation (7.59) is that of a conic section with one focus at the origin. The quantities  and
2 are called the eccentricity and the latus rectum of the orbit, respectively. The
minimum value of r (when   0 ) is called the pericenter, and the maximum value for the
radius is the apocenter. The turning points are apsides. The corresponding terms for
motion about the sun are perihelion and aphelion, and for motion about the earth, perigee
and apogee.
As was stated when discussing the results shown in Figure 7-3, the energy of the orbit will
determine its shape. For example, we found that the radius of the orbit is constant
when E  Vmin . We see from, equation (7.60) that this also implies that the eccentricity is
zero (i.e.,   0 ). In fact, the value of the eccentricity is used to classify the orbits
according to different conic sections (see Figure 7-4):
 1 E0 Hyperbola
 1 E0 Pararbola
0   1 Vmin  E  0 Ellipse
 0 E  Vmin Circle
For planetary motion, we can determine the length of the major and minor axes (designated
by 2a and 2b) using equations (7.59) and (7.60). For the major axis we have
  2 k
2a  rmin  rmax     (7.61)
1  1  1  2
|E|
For the minor axis, we start by defining rb and b as the radius and angle where

b  rb sin b  . (7.62)

Accordingly we have
rmin  a  rb cos  b  (7.63)
which can be written
1    
a  rmin    (7.64)
2 1   1    1   2


79
 Chapter 7 Central Force Motion

Figure 7-4 – The shape of orbits as a function of the eccentricity.

But from equation (7.59) we also have

 cos b  
rb cos b    (7.65)
1   cos b  1  2
or
cos b    (7.66)
and

rb  (7.67)
1  2
Inserting equations (7.66) and (7.67) into equation (7.62) we finally get
 l
b  (7.68)
1  2 2 | E |
or
b  a (7.69)
The period of the orbit can be evaluated using equation (7.20) for the areal velocity
2
dt  dA. (7.70)
l
Since the entire area enclosed by the ellipse will be swept during the duration of a period
 , we have
 2 A

0
dt 
l 
0
dA (7.71)

or

80
 Chapter 7 Central Force Motion

2 2
 A  ab (7.72)
l l
where we have the fact that the area of an ellipse is given by  ab . Now, substituting the
first of equations (7.60) and equation (7.69) in equation (7.72) we get
4 2  3
 
2
a (7.73)
k
In the case of the motion of a solar system planet about the Sun we have
k  Gm p ms (7.74)
where G, m p and ms are the universal gravitational constant, the mass of the planet, and the
mass of the Sun, respectively. We therefore get
4 2 m p ms 3 4 2 4 2 3
2  a  a3  a (7.75)
Gm p ms m p  ms G  m p  ms  Gms

since ms  m p . We find that the square of the period is proportional to the semi-major axis
to the third, with the same proportionality constant for every planet. This (approximate)
result is known as Kepler’s Third Law. We end by summarizing Kepler’s Laws
I. Planets move in elliptical orbits about the Sun with the Sun at one focus.
II. The area per unit time swept out by a radius vector from the Sun to a planet is
constant.
III. The square of a planet’s period is proportional to the cube of the major axis of the
planet’s orbit.

7.6 Scattering in a Central Force Field

Figure 7-5 Scattering process in CM frame

In the previous sections, we investigated two types of orbits (bounded and unbounded) for
two-particle systems evolving under the influence of a central potential. In the present
section, we focus our attention on unbounded orbits within the context of elastic scattering

81
 Chapter 7 Central Force Motion

theory. In this context, a collision between two interacting particles involves a three-step
process: Step I – two particles are initially infinitely far apart (in which case, the energy of
each particle is assumed to be strictly kinetic); Step II – as the two particles approach each
other, their interacting potential (repulsive or attractive) causes them to reach a distance of
closest approach; and Step III – the two particles then move progressively farther apart
(eventually reaching a point where the energy of each particle is once again strictly kinetic).
For the deflection of a single particle moving through a field U(r) whose centre is at rest,
the angle of scattering  is given by
    2 0 (7.76)
 is the angle between the scattered and incident paths
 dr
0   (7.77)
rmin
2 1
2 
r 2
E U   2
l r
It is convenient to use the constant E and the impact parameter b, instead of E and l. Impact
parameter b is defined as the perpendicular distance between the center of force and the
incident path. If v0 is the incident speed of the particle, then

2E
l  r  mv 0  mbv0  mb  b 2mE (7.78)
m
and
 dr
0   (7.79)
rmin
1 U 1
r 2
2
 2  2
b b E r
So, this gives     b, E  .
7.6.1 Scattering cross section for a beam of identical particles with the same energy
In practical scattering experiments, we use a beam of identical particles with the same
energy instead of just one particle. Let dN be the number of particle scattered per unit time
through the ring between angle  and   d  or into the solid angle d at the angle  .  0
is the angle between rmin and infinity. Intensity I of the incident particles is defined as the
number of particles pass in unit time through a unit area normal to the beam.

Figure 7-6 Particles in a shell of width db are scattered in the angular range d .

82
 Chapter 7 Central Force Motion

Figure 7-7 Solid angle.

The cross section     for scattering in a given direction  is defined by

dN
   d   (7.80)
I
where  is solid angle. From Figure 7-7, we have a small area dS on a sphere given by
dS  r sin   d  rd   r 2 sin   d d (7.81)
dS
d   sin   d d (7.82)
r2
The scattering is symmetric, so
d   2 sin    d  (7.83)
Suppose the relation between  and b is one to one
b  , b  db    d  (7.84)
Then, the number dN scattered between  and   d  must be equal to the number of the
incident particles with impact parameter b  b  db ,
I 2 bdb  dN (7.85)
From the definition of    

dN  I    d  (7.86)
we get
I 2 bdb  I    d    I    2 sin    d  (7.87)
(Negative sign here means that when b increase,  decreases) and
b db
    (7.88)
sin    d 

    is also designated as differential cross section. Total cross section in atomic physics
is

83
 Chapter 7 Central Force Motion


       d    2    sin    d  (7.89)
0

7.6.2 Rutherford formula

The scattering force field (Coulomb field) is produced by a fixed charge Ze in this case.
The incident particles have Z e , so the force and the potential are
ZZ e 2
U (7.90)
r
Let the potential of the charged particle in a Coulomb field be
k
U r   (7.91)
r
where k  q1q2 / 4 0 , with q1 and q2 the amount of charge that the two particles carry (k
may be either positive or negative, depending on whether the charges are of the same or
opposite sign: k  0 corresponds to a repulsive force and k  0 to an attractive force). Then
using the equation

  
rmax  l / r  dr
2

(7.92)
rmin
 l2 
2  E  U  
 2 r 2 

Note that rmax  , l  b 2  E ,

0  
  b / r  dr
2

(7.93)
rmin
 b2 U 
1  2  
 r E
After integration, one obtains

cos  0  
 / b  (7.94)
1   / b 
2

where   k /  2 E  . From equation (7.94), we can write

b 2   2 tan 2  0  (7.95)

Making use the relation     2 0 , i.e.,  0   / 2   / 2 , so

b   cot   / 2  (7.96)
Thus
db  1
 (7.97)
d 2 sin   / 2 
2

Use equation (7.88), we have

84
 Chapter 7 Central Force Motion

2 cot   / 2 
   (7.98)
2 2sin    sin 2   / 2 
Now,
sin     2sin   / 2  cos   / 2  (7.99)
Hence
2 1
   (7.100)
4 sin   / 2 
4

or
2
 k  1
     (7.101)
 4 E  sin 4   
 
2
which is the Rutherford scattering cross section. Note that the scattering section  does not
depend on the sign of k. That is, the form of scattering distribution is the same for attractive
or repulsive force.
7.6.3 Transformation of the scattering problem to laboratory coordinates.

v1

 
v2

Figure 7-8 Scattering in laboratory frame

In experiments, the scattering angle is actually measured in laboratory system. This angle
 is different from the angle  calculated from equivalent one-body problem (or in CM
system)(Figure 7-8). We want to find relationship between  and  . Scattering of two
particles as viewed in the centre of mass system has
MR   mi ri
i (7.102)
ri  R  ri
where R is the position vector of center of mass and ri and ri are position vectors of
particle i in laboratory frame and MC frame respectively. From (7.102), we have

85
 Chapter 7 Central Force Motion

MR   mi  R  ri  R  mi   mi r (7.103)
i i i

This means
 m r  0
i
i i (7.104)

Thus we get
 m v  0
i
i i (7.105)

That is, in a coordinate system moving with the centre of mass, the total linear momentum
is zero. Two particles always move with equal and opposite momenta.
The angle  , which is the angle between initial and final directions of the relative vector r ,
must be the same as the scattering angle of either particle in the centre of mass system.
Therefore the relationship between the two scattering angle  and  can be obtained by the
transformation between the centre of mass system and the laboratory system. Consider
  V the velocity of
particle 1: r1  v1 in the lab system, r1  v1 in center of mass system, R
C.M.in the lab system. Then,
v1  V  v1 (7.106)
Total linear momentum is conserved, P  MV and so V is a constant vector. So after
scattering the V still lies along the initial direction. From Figure 7-9
v1 sin    sin   
tan     (7.107)
v1 cos     V cos    
V
v1

Figure 7-9 V is supposed on final velocities in CM frame to get their Lab values.

At the beginning of the discussion of the central force problem, we have derived
m2 
r1  r r (7.108)
m1  m2 mi
where r  r1  r2 . So

86
 Chapter 7 Central Force Motion


v1  r (7.109)
m1
Assume the initial velocity r0  v 0 , the energy is conserved.

1 2 1 l2
E mr  V r  (7.110)
2 2 mr 2
At t  0, r   ,
1
E mv0 2 (7.111)
2
After scattering and when r  
1 1
E mrf 2  mv0 2 (7.112)
2 2
and so
r f  v 0 (7.113)

And,
 
v1  rf  v0 (7.114)
m1 m1
The total linear momentum
 m1  m2 V  m1v0  m2 0 (7.115)
where particle 2 is at rest initially. The linear momentum is conserved
m1 
V v0  v0 (7.116)
m1  m2 m2
Substitute v1 and V into tan   　

sin    sin   
tan     (7.117)
V m
cos     cos     1
v1 m2
If m2  m1 , tan    sin    / cos   

tan    tan    or l   (7.118)

then m2 acts as a fixed centre of mass.

Example
A particle of mass m is acted on by attractive force whose potential is given
by U  r 4 (Figure 7-10). Find the effective potential in terms of r and the maximum of it.
Find also the total cross section for the capture of this particle coming from infinity with

87
 Chapter 7 Central Force Motion

initial velocity v .

Figure 7-10 Effective potential for U  r 4

Solution.
k
U 
r4
(7.119)
1 l2
E  mr 2  U
2 2mr 2
Effective potential is
l2 k
V 2
 4 (7.120)
2mr r
Let us find the maximum of V first. From
dV 4k l2
 5  3 0 (7.121)
dr r mr
we get
4mk
r0  (7.122)
l2
i.e., at r  r0 , we obtain Vmax . So

l4 l4 l4
Vmax     V0 (7.123)
8m 2 k 16m 2 k 16m 2 k
Only particles with total energy E  V0 will “fall” to the force centre.

88
 Chapter 7 Central Force Motion

1 l4
E  mv 
2

2 16m 2 k
 mbv   l  mbv
4
1
mv 2   (7.124)
2 16m 2 k
1
 8k  4
b 2 
 mv 
So, we obtain the impact parameter
1
 8k  4
bmax  2 
(7.125)
 mv 
Therefore the total capture cross section is
2k
   bmax 2  2 (7.126)
mv 2
7.6.4 Elastic and inelastic collisions
For elastic collision, total kinetic energy before and after collisions are equal, e.g.,
1 1 1 1
m1u12  m2u2 2  m1v12  m2 v2 2 (7.127)
2 2 2 2
For inelastic collision, total kinetic energy before and after collisions are not the same.
Sometimes we call endoergic collision, e.g.,
1 1 1 1
Q  m1u12  m2u2 2  m1v12  m2 v2 2 (7.128)
2 2 2 2
where Q is negative and it represent the energy loss due to, e.g., fiction, sound, or heat. A
measure of the inelasticity of two bodies colliding may be considered by referring to a
direct head-on collision in which no rotations are involved (translational kinetic energy
only).Newton found experimentally that the ratio of the relative initial velocities to the
relative final velocities was nearly constant for any two bodies. This ratio, called the
coefficient of restitution  , is defined by
v2  v1
v2  v1    u2  u1  or   (7.129)
u2  u1
For a perfectly elastic collision,   1 , and for a totally inelastic collision,   0 . Values of 
have the limits 0 and 1.
Example.
For an elastic head-on collision, show that   1 .The mass m2 is initially at rest.
Solution
Because the final velocities are along the same direction as u1 , we state the conservation of
linear momentum and energy as

89
 Chapter 7 Central Force Motion

m1u1  m1v1  m2 v2
1 1 1 (7.130)
m1u12  m1v12  m2v2 2
2 2 2
Solving the first equation of (7.130) for v2 and substitute into the equation for 
m1u1  m1v1
 v1
v2  v1 m2 m m v v
   1 1 1 1 (7.131)
u1 u1 m2 m2 u1 u1
The ratio v1 / u1 can be found from the second equation of (7.130) after substituting for v2
from the first equation of (7.130)
2
1 1 1  m u  m1v1 
m1u12  m1v12  m2  1 1 
2 2 2  m2  (7.132)
m2
m1u12  m1v12  1  u12  v12  2u1v1 
2

m2
Dividing by m1u12 and letting x  v1 / u1 gives

1  x2 
m1
m2
1  x2  2 x  (7.133)

Collecting terms,
 m1  2 2m1 m 
1  x  x   1  1  0 (7.134)
 m2  m2  m2 
Solving for x, we find
 m1 
  1
x  1, x   m2 
(7.135)
 m1 
  1
 m2 
The solution x  1 is trivial ( v1  u1 , v2  0 ), so we substitute the other solution for x into
equation(7.131)
 1 (7.136)

90
 Chapter 8 Motion in a Noninertial Reference Frame

Chapter 8
Motion in a Noninertial Reference Frame

We have so far dealt only with problems situated in inertial reference frame, or if not,
problems that could be solved with enough accuracy by ignoring the noninertial nature of
the coordinate systems. There are, however, many problems for which it is necessary, or
beneficial, to treat the motion of the system at hand in a noninertial reference frame. In this
chapter, we will develop the mathematical apparatus that will allow us to deal with such
problems, and prepare the way for the study of the motion of rigid bodies that we will
tackle in the next chapter.

8.1 Rotating Coordinate Systems

Let’s consider two coordinate systems: one that is inertial and for which the axes are fixed,
and another whose axes are rotating with respect to the inertial system. We represent the
coordinates of the fixed system by xi and the coordinates of the rotating system by xi . If we
choose some point P in space (see Figure 8-1) we have
r  R  r (8.1)
where R locates the origin of the rotating system in the fixed system. We assume that P is
at rest in the inertial frame so that r is constant. If during an infinitesimal amount of time
dt the rotating system undergoes an infinitesimal rotation d about some axis, then the
vector r will vary not only as measured by an observer co-moving with the rotating system,
but also when measured in the inertial frame.

Figure 8-1 – The xi ’s are coordinates in the fixed system, and xi are coordinates in the
rotating system. The vector R locates the origin of the rotating system in the fixed system.

91
 Chapter 8 Motion in a Noninertial Reference Frame

This problem is the same as was treated in Chapter 3 when considering the conservation of
angular momentum using Noether’s Theorem, and we can write
 dr  fixed  d  r (8.2)

where the designation “fixed” is included to indicate that dr is measured in the fixed or
inertial coordinate system. We can obtain the time rate of change of dt in the inertial system
by dividing both sides of equation (8.2) by
 dr 
    r (8.3)
 dt  fixed
with
d
 (8.4)
dt
If we allow the point P to have some velocity  dr / dt rotating with respect to the rotating
system, equation (8.3) must be correspondingly modified to account for this motion. Then,
we have
 dr   dr 
     r (8.5)
 dt  fixed  dt  rotating
Example
We have a vector r  x1e1  x2e2  x3e3 in a rotating system, which share a common origin
with an inertial system. Find r  in the fixed system by direct differentiation if the angular
velocity of the rotating system is ω in the fixed system.
Solution.
We have
 dr  d 
r        xi ei     xi ei  xi e i  (8.6)
 dt  fixed dt  i  i
The first term on the right hand side of equation (8.6) is simply the velocity as measured
in the rotating system (i.e., we have the components xi along the corresponding axes,
which form the basis vectors of the rotating system). We therefore rewrite equation (8.6)
as
 dr   dr 
      xi e i (8.7)
 dt  fixed  dt rotating i
We need to evaluate e i for i=1,2, and 3. To do so, consider three infinitesimal rotations
d1 , d 2 , and d3 along e1 , e 2 , and e3 , respectively. If we first calculate the effect of the
first rotation about e1 on the other two basis vectors (using Figure 8-2) we have

92
 Chapter 8 Motion in a Noninertial Reference Frame

de2  cos  d1  e 2  sin  d1  e3   e 2

 e 2  d1e3   e 2 (8.8)
 d1e3
since d1 is infinitesimal. Similarly, if we calculate the effect of each infinitesimal
rotations on every basis vectors we find, after a division by dt, that
de1
 3e 2  2e3
dt
de 2
 1e3  3e 2 (8.9)
dt
de3
 2e1  1e 2
dt
with i  di / dt . Alternatively, we can combine equations (8.9) into one vector equation
as
e i    ei (8.10)

with    i e i . Inserting equation (8.10) into equation (8.7), we get

 dr   dr   
 dt         xi e i 
fixed
 dt  rotating  i 
(8.11)
 dr 
 
 dt 
 
  r ,
rotating

which is the same result as equation (8.5).

Figure 8-2 – With this definition for the set of axes, and with i  di / dt , for i=1, 2, and 3,
we can determine the effect of the rotations on the different basis vectors.

8.1.1 Generalization to arbitrary vectors

93
 Chapter 8 Motion in a Noninertial Reference Frame

Although we used the position vector r for the derivation of equation (8.11) (or (8.5)), this
expression applies equally well to an arbitrary vector Q, that is
 dQ   dQ 
 dt  
 dt  rotating

  Q  (8.12)
fixed

 is the same in both systems of

For example, we can verify that the angular acceleration ω
reference
 d   d   d 
 dt   
 dt  rotating

     
 dt  rotating
(8.13)
fixed

We can also use equation (8.12) to find the velocity of point P (in Figure 8-1) as measured
in the fixed system
 dr    dR   dr 
 dt     
fixed
 dt  fixed  dt  fixed
(8.14)
 dR   dr 

 dt 
 
 dt 

  r 
fixed rotating

If we define the following quantities

 dr 
v f  r f   
 dt  fixed
   dR 
VR (8.15)
f  
 dt  fixed
 dr 
v r  rr   
 dt rotating
we can rewrite equation (8.14) as
v f  V  vr    r (8.16)

where
v f  the velocity relative to the fixed axes
V  the linear velocity of the moving origin
v r  the velocity to the rotating axes (8.17)
  the angular velocity of the totating axes
  r  the velocity due to the rorating of moving axes

94
 Chapter 8 Motion in a Noninertial Reference Frame

8.2 The Centrifugal and Coriolis Forces

We know that Newton’s Second Law (i.e., F = ma ) is valid only in an inertial frame of
reference. In other words, the simple form F = ma for the equation of motion applies when
the acceleration is that which is measured in the fixed referenced system, i.e., a  a f . Then,
we can write
 dv f 
F  ma f  m   (8.18)
 dt  fixed
where the differentiation is carried out in the fixed system. Differentiating equation (8.16)
we get
 dv f   dV   dv   dr 
    r     r      (8.19)
 dt   dt  fixed  dt  fixed  dt  fixed
Using equation (8.12) we can transform this equation as follows
   dr  
   dv r 
af  R    v r     r        r
f  
 dt  rotating   dt rotating  (8.20)
R  a    r  2  v      r 
f r r

   dV / dt  . Correspondingly, the force on the particle as measured in the

where R f fixed

inertial frame becomes

  ma  m  r  m    r   2m  v
F  mR (8.21)
f r r

Alternatively, the effective force on the particle as seen by an observer co-moving with the
rotating system is
F  ma  F  m R   m  r  m    r   2m  v (8.22)
eff r f r

The first term is the total force acting on the particle as measured in the inertial frame. The
second (  mR ) and third (  m  r ) are due to the translational and angular accelerations,
f

respectively, of the moving noninertial system. The fourth term ( m    r   ) is the
so-called centrifugal force (directed away from the centre of rotation), and finally, the last
term ( m  v r ) is the Coriolis force. It is important to note that the Coriolis force arises
because of the motion of the particle in the rotating system, i.e., it disappears if v r  0 .
Equation (8.22) is a mathematical representation of what is meant by the statement that
Newton’s Second Law does not apply in a noninertial reference frame. It is not that the
physics dealt with Newtonian mechanics cannot be analyzed in a noninertial frame, but that
 and ω
the form of the equations of motion is different. More precisely, if we set R  in
f

equation (8.22) to zero to simplify things, we have in the rotating frame a more
complicated equation of motion

95
 Chapter 8 Motion in a Noninertial Reference Frame

F  ma r   noninertial terms  (8.23)

where the “noninertial terms” are the centrifugal and Coriolis forces, than in an inertial
frame where the equation of motion is simply
F  ma f (8.24)

8.3 Lagrangian formulation for noninertial motion

We can recover the expression (8.22) for the acceleration in a rotating (non-inertial) frame
from a Lagrangian formulation as follows. The Lagrangian for a particle of mass m moving
in a non-inertial rotating frame (with its origin coinciding with the fixed-frame origin) in
the presence of the potential U(r) is expressed as
m
L  r, r    r    r   U  r 
2
(8.25)
2
where ω is the angular velocity vector and we use the formula
 r    r    r    r    r    r 
2

 r 2  2r    r     r     r  (8.26)
 r 2  2   r  r    2 r 2    r  
2
 
Using the Lagrangian (8.25), we now derive the general Lagrange equation for r. First, we
derive an expression for the canonical momentum
L
p  m  r    r  (8.27)
r
and
d L
 m 
r    r    r  (8.28)
dt r
Next, we derive the partial derivative
L
 m  r  m    r   U  r  (8.29)
r
so that the Euler-Lagrange equations are
r  m  r  2m  r  m    r   U  r   0
m (8.30)
or
ma r  U  r   m  r  m    r   2m  r (8.31)
Here, the potential energy term generates the fixed-frame acceleration, U  ma f , and
thus the Lagrange equation (8.31) yields Eq. (8.22).

8.4 Motion relative to the Earth

We can apply the results obtained in the previous section to motion near the surface of the
Earth. If we set the origin of the inertial (fixed) system xyz  to be at the center of the Earth,
and the moving (rotating) noninertial frame xyz on the surface of the Earth, we can describe

96
 Chapter 8 Motion in a Noninertial Reference Frame

the motion of a moving object near its surface using equation (8.22). We denote by
F  S  mg 0 the total force acting on the object (of mass m ) where S represent any external
forces (except gravity) and g 0 is the gravitational acceleration
GM 
g0   eR (8.32)
R2

Figure 8-3 – The inertial reference system xyz has its origin Q at the centre of the
Earth, and the moving frame xyz has its centre near the Earth’s surface. The vector R gives
the Earth’s radius.

In equation (8.32) G  6.67  1011 N  m 2 / kg 2 is the universal gravitational constant,

M   5.98 1024 kg is the mass of the Earth, and R  6.38 106 m its radius (see Figure
8-3). We assume that the Earth’s radius and gravitational field are independent of latitude.
The effective force Feff as measured in the moving frame near the surface of the Earth
becomes
  m  r  m    r   2m  v .
Feff  S  mg 0  mR (8.33)
f r

The Earth’s angular velocity vector is given by ω  7.3 105 e z rad/s (i.e., it is directed
along the z  -axis ), and we assume that it is a constant. The fourth term on the right hand
side of equation (8.33) therefore equals zero. Also, from equation (8.12) we have
  
   dR f
R     R
 
f  dt f
  rotating
 d  dR      dR  
        R             R    (8.34)
 dt  dt  rotating   
      dt  rotating  
     R 
since R is a constant. Inserting equation (8.34) in equation (8.33) we get

97
 Chapter 8 Motion in a Noninertial Reference Frame

Feff  S  mg 0  m    r  R   2 m  v r (8.35)

The second and third terms on the right hand side of this equation can be combined into a
single term for the effective gravitational acceleration g that is felt near the surface of the
Earth (i.e., on the surface of the Earth we cannot discerned between gravity g 0 and the
centrifugal acceleration     r  R  , we can only feel the resulting acceleration g )

g  g 0       r  R   (8.36)

It is to be noted that because of the presence of the centrifugal acceleration

    r  R  in this equation for the effective gravity, g and g 0 will in general not
point exactly in the same direction. This effect is rather small, but measurable as
 R / g 0  0.0035 . It should also be clear from the equation (8.36) that the magnitude of
the effect is a function of latitude. The equation for the effective force is then rewritten as
Feff  S  mg  2 m  v r (8.37)
As was pointed out earlier, the last term on the right hand side of equation (8.37) is
responsible for the Coriolis effect. This effect is the source for some well-known motions
of the air masses. To see how this happens, let’s consider the xyz coordinate system to be
located at some latitude where the angular velocity vector ω (which represents the Earth’s
rotation) has a component  z e z along the vertical at the specified latitude. If a particle is
projected such that its velocity vector v r is located in the xy plane, then the Coriolis force
will have a component directed to the right of the particle’s motion (see Figure 8-4). The
size of this effect will be a function of the latitude, as the amplitude of also exhibits such a
dependency. So, consider a region where, for some reason, the atmospheric pressure is
lower than it is in its surrounding (see Figure 8-5). As the air flows into this low-pressure
spot from regions of higher pressure all around, the Coriolis effect will deflect the air
motion to the right (in the Northern Hemisphere), resulting into counterclockwise, or
cyclonic, motions in the atmosphere.
As the following example will show, the Coriolis effect generally only becomes important
for the motion of bodies near the surface of the Earth when large enough distance scales are
considered.

Figure 8-4 – In the Northern hemisphere, a particle projected in a horizontal plane will be
directed to the right of its motion. The opposite will happen in the Southern Hemisphere.

98
 Chapter 8 Motion in a Noninertial Reference Frame

Figure 8-5 – The Coriolis effect deflects the air in the Northern Hemisphere to the right
producing cyclonic motion.
Examples
1. Free-falling object. Find the horizontal deflection caused by the Coriolis effect acting on
a free-falling particle in the Earth’s effective gravitational field from a height h(<<R) above
its surface.
Solution.
From equation (8.37), with S=0 and Feff  ma r ,we have

a r  g  2  v r (8.38)
We choose the z-axis attached (virtually) to the surface of the rotating Earth as directed
outward along -g. We also choose the e x and e y bases vectors such that they are in the
southerly and easterly direction, respectively. The latitude is once again denoted by  (see
Figure 8-6). With these definitions we can decompose the Earth’s angular velocity vector
as
 x   cos   
y  0 (8.39)
 z   sin   
Even though the Coriolis effect produces velocity components along e x and e y , we will
neglect these since they will be significantly smaller than the velocity along e z . Then,
x  y  0, z   gt (8.40)
where we assume that the particle is free-falling from rest.

99
 Chapter 8 Motion in a Noninertial Reference Frame

Figure 8-6 – The coordinated system “attached” to the Earth’s surface, for finding the
horizontal deflection of a free-falling particle. The e x and e y bases vectors are, respectively,
in the southerly and easterly direction.

We now calculate the apparent acceleration component a c due to the Coriolis term in
equation (8.38)

  
a c ; 2    cos  e x  sin  e y     gte z  

; 2 gt cos   e x  e z  (8.41)



; 2 gt cos  e y

Inserting equation (8.41) in equation (8.38) we find the apparent acceleration of the
particle as seen from the Earth’s surface
a r  2 gt cos    e y  ge z (8.42)

If we assume that the initial conditions for the position of the particle are x0  y0  0 and
z0  h , we have after twice integrating equation (8.42)
1  1 
r  t    gt 3 cos    e y   h  gt 2  e z (8.43)
3  2 
When the particle reaches the Earth’s surface we will have t  2h / g , and finally for the
horizontal deviation
1 8h 3
d   cos    (8.44)
3 g

100
 Chapter 8 Motion in a Noninertial Reference Frame

Thus, if an object is dropped from a height of 100 m at latitude 45° north, it is deflected
approximately by only 1.55 cm (we neglected any friction brought up by the presence of
the atmosphere).
2. Foucault’s pendulum. In 1851, Jean Bernard L′eon Foucault (1819-1868) was able to
demonstrate, in a classic experiment demonstrating Earth’s rotation, the role played by
Coriolis effects in his investigations of the motion of a pendulum in the rotating frame of
the Earth. His analysis showed that, because of the Coriolis acceleration associated with the
rotation of the Earth, the motion of the pendulum exhibits a precession motion whose
period depends on the latitude at which the pendulum is located.
We set the origin of the noninertial xyz coordinate system at the equilibrium point of the
pendulum and the z-axis along the local vertical. Describe the motion of the pendulum of
length l and mass m in the small angle limit, taking into account the rotation of the Earth.

Solution. The equation of motion is

T
ar  g   2  v r (8.45)
m
where T is the tension in the pendulum. If we restrict ourselves to small oscillations, we
can write
x y
T  T e x  T e y  Te z (8.46)
l l
where we neglected second and higher order terms in x/l and y/l. As in the previous
example, we write
g   ge z (8.47)
and
 x   cos   
y  0 (8.48)
 z   sin   
Again limiting ourselves to small angular displacements, we can write for the velocity of
the pendulum
 v r  x  x,  v r  y  y ,  v r  z  0 (8.49)

101
 Chapter 8 Motion in a Noninertial Reference Frame

Figure 8-7 – Geometry of Foucault’s pendulum. The acceleration vector is along the z-axis,
and the tension T is broken down into components along the x-, y-, and z-axes.

Using equations (8.48) and (8.49) to evaluate the Coriolis effect in equation (8.45), we
can find the apparent acceleration of the pendulum as seen near the surface of the Earth (i.e.,
in the noninertial system) to be
 T x   T y 
ar    2 y sin     e x     2 x sin     e y
 ml   ml 
(8.50)
T 
   2 y cos     g  e z
m 
If we concentrate on the motion in the xy plane, and make the following substitutions
T  mg , 0 2  T / ml  gl , and  z   sin    we find from equation (8.50)

x  0 2 x  2 z y

(8.51)
y  0 2 y  2 z x

which is a system of two coupled second order differential equations. In order to facilitate
the solution of the system, we multiply the second these equations by the unit imaginary
number I and add it to the first equation. Then, defining the following complex variable
q  x  iy (8.52)
we have from equations (8.51) that
q  2i z q  0 2 q  0 (8.53)
Equation (8.53) describes the motion of a damped oscillator (with the difference that the
damping factor is, in this case, purely imaginary). We can write the solution to equation
(8.53) to be


q  t   Ae iz t cos t  z 2  0 2    (8.54)

We see that if the rotation of the Earth were ignored, we would retrieve the usual motion of
a harmonic oscillator motion with

102
 Chapter 8 Motion in a Noninertial Reference Frame

q  t   A cos 0t    ,  z  0 (8.55)

and 0 is thus identified with the oscillation frequency of the pendulum. This frequency is
much greater that the angular frequency of rotation of the Earth, which performs one
complete rotation in approximately 24 hours. So, using the fact that 0   z in equation
(8.54) we have
q  t   Ae  iz t cos 0t   
(8.56)
 A cos  z t   i sin  z t   cos 0t   
which implies, using equation (8.52), that (assuming we chose the initial condition such
that A is real)
x  t   A cos  z t  cos 0t   
(8.57)
y  t    A sin  z t  cos 0t   
It now becomes easy to see that as the pendulum is oscillating at a frequency 0 , it also
performs a precession, or rotation in the xy plane at a frequency of  z . The position angle
made by the axis of oscillation in the xy plane will change with time as the pendulum
rotates, and it is given by
 y t   1  sin   z t 
 
  t   tan 1    tan     z t  t sin    (8.58)
 x t    cos  z t  

103
 Chapter 9 Dynamics of Rigid Bodies

Chapter 9
Dynamics of Rigid Bodies

Wolfgang Pauli and Niels Bohr stare in wonder at a spinning top.

Having now mastered the technique of Lagrangians, this section will be one big application
of the methods. The systems we will consider are the spinning motions of extended objects.
As we shall see, these can often be counterintuitive. Certainly Pauli and Bohr found
themselves amazed!

9.1 The Independent Coordinates of a Rigid Body

The simplest extended-body model that can be treated is that of a rigid body, one in which
the distances ri  r j between points are held fixed. A general rigid body will have six
degrees of freedom (but not always, see below). To see how we can specify the position of
all points in the body with only six parameters, let us first fix some point r1 of the body,
thereafter treated as its “centre” or origin from which all other points in the body can be
referenced from ( r1 can be, but not necessarily, the centre of mass). Once the coordinates
of r1 are specified (in relation to some origin outside of the body), we have already used

- 104 -
 Chapter 9 Dynamics of Rigid Bodies

up three degrees of freedom. With r1 fixed, the position of some other point r2 can be
specified using only two coordinates since it is constrained to move on the surface of a
sphere centered on r1 . We are now up to five degrees of freedom. If we now consider any
other third point not located on the axis joining r1 and r2 , its position can be specified
using one degree of freedom (or coordinate) for it can only rotate about the axis connecting
r1 and r2 . We thus have used up the six degrees of freedom. It is interesting to note that in
the case of a linear rod, any point r3 must lay on the axis joining r1 and r2 ; hence a
linear rod has only five degrees of freedom.
Usually, the six degrees of freedom are divided in two groups: three degrees for translation
(to specify the position of the “centre” r1 , and three rotation angles to specify the
orientation of the rigid body (normally taken to be the so-called Euler angles).

9.2 The Kinetic Energy and The Inertia Tensor

Let’s consider a rigid body composed of n particles of mass m ,   1, 2,..., n . If the body
rotates with an angular velocity ω about some point fixed with respect to the body
coordinates (this “body” coordinate system is what we used to refer to as “noninertial” or
“rotating” coordinate system in Chapter 9), and if this point moves linearly with a velocity
V with respect to a fixed (i.e., inertial) coordinate system, then the velocity of particle is
given by equation (8.16) of derived in Chapter 8
v  V  ω  r (9.1)
where we omitted the term
 dr 
v     0 (9.2)
 dt  rotating
since we are dealing with a rigid body. We have also dropped the f subscript, denoting the
fixed coordinate system, as it is understood that all the non-vanishing velocities will be
measured in this system; again, we are dealing with a rigid body.
The total kinetic energy of the body is given by
1
T   T   m v 2
 2 
1
  m  V  ω  r 
2
(9.3)
2 
1 1
  mV 2   m V   ω  r    m  ω  r 
2

2   2 
Although this equation for the total kinetic energy is perfectly general, considerable
simplification will result if we choose the origin of the body coordinate system to coincide
with the centre of mass. With this choice, the second term on the right hand side of the last

- 105 -
 Chapter 9 Dynamics of Rigid Bodies

of equations (9.3) can be seen to vanish from

  
 m V   ω  r   V  ω    m r    0
  
(9.4)
  
since the centre of mass R of the body, of mass M, is defined such that


m r  0 (9.5)

The total kinetic energy can then be broken into two components: one for the translational
kinetic energy and another for the rotational kinetic energy. That is,
T  Ttrans  Trot (9.6)
with
1 1
Ttrans  
2 
mV 2  MV 2
2
(9.7)
1
Trot   m  ω  r 
2

2 
The expression for Trot can be further modified, but to do so we will now resort to tensor
(or index) notation. So, let’s consider the following vector equation
 ω  r    ω  r    ω  r 
2
(9.8)
and rewrite it using the Levi-Civita and the Kronecker tensors
 ijk  j x ,k    imnm x ,n    ijk  imn j x ,k m x ,n
  jm kn   jn km   j x ,k m x , n (9.9)
  j j x , k x , k   j x , jk x , k

Inserting this result in the equation for Trot in equation (9.7) we get
1
 m  2 r2   ω  r  
2
Trot  (9.10)
2   
Alternatively, keeping with the tensor notation we have
1
Trot   m  j j x ,k x ,k  i x ,i j x , j 
2 
1
  m i j ij  x ,k x ,k  i x ,i j x , j  (9.11)
2 
1
 i j   m  ij x ,k x ,k  x ,i x , j 
2 

We now define the components Iij of the so-called inertia tensor I by

I ij   m  ij x ,k x ,k  x ,i x , j  (9.12)


- 106 -
 Chapter 9 Dynamics of Rigid Bodies

and the rotational kinetic energy becomes

1
Trot  i I ij j (9.13)
2
or in vector notation
1
Trot  ωIω (9.14)
2
For our purposes it will be usually sufficient to treat the inertia tensor I as a
regular 3  3 matrix. Indeed, we can explicitly write I using equation (9.12) as
 
  m  x ,2  x ,3   m x ,1 x ,2  m x ,1 x ,3
2 2

  

I    m x ,2 x ,1  m  x ,1
2
 x ,32   m x ,2 x ,3  (9.15)
    
 2 
  m x ,3 x ,1  m x ,3 x ,2  m  x ,1  x ,2  
2

    
It is easy to see from either equation (9.12) or equation (9.15) that the inertia tensor is
symmetric, that is,
I ij  I ji (9.16)
The diagonal elements I11 , I 22 , and I 33 are called the moments of inertia about the x1 -,
x2 -, and x3 -axes , respectively. The negatives of the off-diagonal elements are the
products of inertia. Finally, in most cases the rigid body is continuous and not made up of
discrete particles as was assumed so far, but the results are easily generalized by replacing
the summation by a corresponding integral in the expression for the components of the
inertia tensor
I ij     r   ij xk xk  xi x j  dx1dx2 dx3 (9.17)
V

where   r  is the mass density at the position r, and the integral is to be performed over
the whole volume V of the rigid body.
Example
Calculate the inertia tensor for a homogeneous cube of density  , mass M, and side length
b. Let one corner be at the origin, and three adjacent edges lie along the coordinate axes
(see Figure 9-1).
Solution.
We use equation (9.17) to calculate the components of the inertia tensor. Because of the
symmetry of the problem, it is easy to see that the three moments of inertia I11 , I 22 , and
I 33 are equal and that same holds for all of the products of inertia. So,

- 107 -
 Chapter 9 Dynamics of Rigid Bodies

   x  x32  dx1dx2 dx3

b b b
I11   2
2
0 0 0

   dx3   x2 2  x32  dx2  dx1

b b b

0 0 0

b b   b4 b4 
3 (9.18)
  b  dx3   bx32   b   
0
 3   3 3
2b5 2
  Mb 2
3 3

Figure 9-1 – A homogeneous cube of sides b with the origin at one corner.

And for the negative of the products of inertia

b b b
I12     x1 x2 dx1dx2 dx3
0 0 0

 b 2  b 2 
       b  (9.19)
 2  2 
b5 1
     Mb 2
4 4
It should be noted that in this example the origin of the coordinate system is not located at
the centre of mass of the cube.

9.3 Angular Momentum

Going back to the case of a rigid body composed of a discrete number of particles; we can
calculate the angular momentum with respect to some point O fixed in the body coordinate
system with
L   r  p (9.20)


- 108 -
 Chapter 9 Dynamics of Rigid Bodies

Relative to the body coordinate system the linear momentum of the th particle is
p  m v  m ω  r (9.21)
and the total angular momentum becomes
L   m r   ω  r  (9.22)


Resorting one more time to tensor notation we can calculate r   ω  r  as

 x
ijk  , j   l x ,m    kij  klm x ,ll x ,m
klm

  il jm   im jl  x ,ll x , m (9.23)

 x , j x , ji  x , j j x ,i
or alternatively in vector notation
r   ω  r   r 2ω  r  r  ω  (9.24)
Then, the total angular momentum is given by
L   m  r 2ω  r  r  ω   (9.25)


Using the tensor notation the component of the angular momentum is

L   m  x ,k x ,k i  x , j j x ,i 

(9.26)
  j  m  ij x ,k x ,k  x , j x ,i 


and upon using equation (9.12) for the inertia tensor

Li  I ij j (9.27)
or in tensor notation
L  Iω (9.28)
Finally, we can insert equation (9.28) for the angular momentum vector into equation
(9.14) for the rotational kinetic energy to obtain
1
Trot  L  ω (9.29)
2

- 109 -
 Chapter 9 Dynamics of Rigid Bodies

Figure 9-2 – A dumbbell connected by masses m1 and m2 at the ends of its shaft. Note
that the angular velocity ω is not directed along the shaft.
Example
The dumbbell. A dumbbell is connected by two masses m1 and m2 located at distances
r1 and r2 from the middle of the shaft, respectively. The shaft makes an angle  with a
vertical axis, to which it is attached at its middle (i.e., the middle of the shaft). Calculate the
equation of angular motion if the system is forced to rotate about the vertical axis with a
constant angular velocity ω (see Figure 9-2).
Solution. We define the inertial and the body coordinate systems such that their respective
origins are both connected at the point of junction between the vertical axis and the shaft of
the dumbbell. We further define the body coordinate system as having its x3 -axis
orientated along the shaft and its x1 -axis perpendicular to the shaft but located in the plane
defined by the axis of rotation and the shaft. The remaining x2 -axis is perpendicular to this
plane and completes the coordinate system attached to the rigid body. For the inertial
system, we chose the x3 -axis to be the vertical, and the other two axes such that the basis
vectors can expressed as
e1  cos   cos t  e1  sin t  e 2  sin   cos t  e3
e2  cos   sin t  e1  cos t  e 2  sin   sin t  e3 (9.30)
e3  sin   e1  cos   e3
From equation (9.12), we can evaluate the components of the inertia tensor. We can in the
first time identify the components that are zero (because x1,1  x1,2  x2,1  x2,2  0 )

I12  I 21  I13  I 31  I 23  I 32  I 33  0 (9.31)

The only two remaining components are

- 110 -
 Chapter 9 Dynamics of Rigid Bodies

I11  I 22  m1 x1,32  m2 x2,32  m1r12  m2 r2 2 (9.32)

The components of the angular velocity in the coordinate of the rigid body are
1   sin  
2  0 (9.33)
3   cos  
Inserting equations (9.32) and (9.33) in equation (9.27) we find for the components of L
L1  I1ii  I111   sin    m1r12  m2 r2 2 
L2  I 2ii  I 222  0 (9.34)
L3  I 3ii  I 333  0
If should be noted from equations (9.33) and (9.34) that the angular velocity and the
angular momentum do not point in the same direction. To calculate the equations of motion,
we express the angular momentum with the inertial coordinates instead of the coordinates
of the rigid body system. From equation (9.34) we can write
L   sin    m1r12  m2 r2 2  e1  I11 sin   e1 (9.35)

but we can transform the basis vector e1 using equation (9.30) (or its inverse)

L  I11 sin   cos   cos t  e1  cos   sin t  e2  sin   e3  (9.36)

We also know that

dL
N (9.37)
dt
where N is the torque. Assuming that the angular speed is constant, we find
N1   I11 2 sin   cos   sin t 
N 2  I11 2 sin   cos   cos t  (9.38)
N3  0
An interesting consequence of the fact that the angular momentum and angular velocity
vectors are not aligned with each other is that we need to apply a torque to the dumbbell to
keep it rotating at a constant angular velocity.

9.4 The Principal Axes of Inertia

We now set on finding a set of body axes that will render the inertia tensor diagonal in form.
That is, given equation (9.15) for I, we want to make a change in the body basis vectors
(i.e., a change of variables) that will change the form of the inertia tensor to

- 111 -
 Chapter 9 Dynamics of Rigid Bodies

 I1 0 0
I   0 I2 0  (9.39)
 0 0 I 0 
We will then require that all the products of inertia be zero. Carrying this program will
provide a significant simplification for the expressions of the angular momentum and the
kinetic energy, as measured in the inertial reference frame. That is, these two quantities will
be given by
L1  I11 , L2  I 22 , L3  I 33 (9.40)
and
1
Trot 
2
 I112  I 222  I332  (9.41)

The set of axes that allow this transformation is called the principal axes of inertia. When
the equations for the components of the angular momentum can be expressed as in equation
(9.39), then L and ω are directed along the same axis.
The problem of finding the principal axes is mathematically equivalent to solving a set of
linear equations. More precisely, we have from equation (9.28) that
L  Iω (9.42)
but we are actually looking for a way to reduce this equation to the following form
L  Iω  Iω (9.43)
Mathematically speaking, I, which is called a principal moment of inertia, is an
eigenvalue of the inertia tensor, and ω , which will give us the corresponding principal axis
of inertia, is an eigenvector. The system of equations (9.43) can be written as
L1  I 1  I111  I122  I133
L2  I 2  I 211  I 222  I 233 (9.44)
L3  I 3  I 311  I 322  I 333
or, after some rearranging
 I11  I  1  I122  I133  0
I 211   I 22  I  2  I 233  0 (9.45)
I 311  I 322   I 33  I  3  0
The mathematical condition necessary for this set of equation to have a nontrivial solution
is that the determinant of the coefficient vanishes
 I11  I  I12 I13
I 21  I 22  I  I 23 0 (9.46)
I 31 I 32  I 33  I 
The expansion of this determinant leads to the so-called secular or characteristic equation

- 112 -
 Chapter 9 Dynamics of Rigid Bodies

for the eigenvalues I (i.e., I1 , I 2 and I 3 in equation (9.40)); it is a third order

polynomial. Once the characteristic equation has been solved, the principal axes can be
determined by inserting the eigenvalues back in equation (9.45) and evaluating the ratios
of the angular velocity components ( 1 : 2 : 3 ), therefore, determining the corresponding
eigenvectors.
It is important to realize that in many cases, the rigid body under study will exhibit some
symmetry that will allow one to guess what the principal axes are. For example, a cylinder
will have one of its principal axes directed along the centre axis of the cylinder. The two
remaining axes will be directed at right angle to this axis (and to each other).
Finally, here are a few definitions: a body that has i) I1  I 2  I 3 is called a spherical top,
ii) I1  I 2  I 3 is a symmetric top, iii) I1  I 2  I 3 is an asymmetric top, and finally, if
I1  0, I 2  I 3 the body is a rotor.

Example
Find the principal moment of inertia and the principal axes of inertia for the cube of Figure
9-1.
Solution.
In the previous example solved, we found that the products of inertia did not equal zero.
Obviously, the axes chosen were not the principal axes. To find the principal moments of
inertia, we must solve the characteristic equation
2  1 1
  I    
3  4 4
1 2  1
    I   0 (9.47)
4 3  4
1 1 2 
      I
4 4  3 
where   Mb 2 . Before trying to evaluate the determinant from equation (9.47), it is
always good to see if we can bring it to a simpler form with a subtraction of a column or a
row to another column or row. This is permitted since the determinant is not affected by
such operations. In our case, we see that, for example, the third element of the first two
rows is the same; we will therefore subtract the second row to the first. The determinant
then becomes which implies that
 11  11
  I   0
 12  12
1 2  1
    I   0 (9.48)
4 3  4
1 1 2 
      I
4 4 3 

- 113 -
 Chapter 9 Dynamics of Rigid Bodies

which implies that

1 1 0
 11  1 2  1
  I     I   0 (9.49)
 12  4 3  4
1 1 2 
      I
4 4 3 
Expanding this equation gives

  2   1  
2 2 2
 11   1   1  2
   I     I            I 
     0
 12   3   4   4  3   4  
(9.50)
  2   1  
2 2
 11   1  2
   I     I          I   2      
 12   3   4  3   4  

The square-bracketed part of the last expression is a second order polynomial in  2  / 3  I  .

If we solve for the roots of this polynomial we get

 1  1  1  
2 2
2 1 
   I         8   
3  2  4 4   4   (9.51)
1
  1  3
8
The last of equation (9.50) can now be factored as
 11  11  1 
   I    I     I   0 (9.52)
 12  12  6 
and the three principal moments of inertia are
1 11
I1   , I 2  I3   (9.53)
6 12
To find the direction of the principal axis of inertia, we insert the eigenvalues of equation
(9.53) into equation (9.45). For I1 , we have (after some manipulation)
21  2  3  0
(9.54)
1  22  3  0
which implies that 1  2  3 and the corresponding eigenvector is directed along
e1  e 2  e3 . For I 2 and I 3 , because they are equal, the orientation of their corresponding
principal axes is arbitrary; they need only to lie in a plane perpendicular to the main
principal axis determined for I1 .

- 114 -
 Chapter 9 Dynamics of Rigid Bodies

9.5 Similarity Transformations

The diagonalization of the inertia tensor (or any other matrix for that matter) discussed in
the previous section can sometimes be achieved in a different manner. For example, in the
case of the cube discussed in the last example, because of the symmetry of the problem we
could have guessed what the principal axes were. When this can be done, it is then
straightforward (if sometimes tedious) to determine the transformation (rotation) matrix
that brings us from the initial set of coordinate axes to the principal axes, and use it to
render the inertia tensor diagonal. In a way, this technique follows a path that is reversed
from what was done in the previous section. That is, instead of, first, rendering the inertia
tensor diagonal and then, determining the principal axes (i.e., the eigenvectors), we now
guess the orientation of the principal axes and then diagonalize the inertia tensor.
So, let’s consider the angular momentum
L  Iω (9.55)
in the initial set of coordinate axes, which we now transform to a new set of axes. We will
have a new equation for the angular momentum as measured in this new system; we define
the angular momentum with an equation similar to equation (9.55)
L  Iω (9.56)
If we denote the transformation matrix that links the two coordinate systems by λ such
that
L  λL (9.57)
and
ω  λω (9.58)
We have relations similar to equations (9.57) and (9.58) for any corresponding vectors
between the two systems of coordinates. For example, the position vectors are related
through
r  λr (9.59)
Combining equations (9.57) to (9.58) we can write
L  λL  Iω  I  λω  (9.60)
or
L   λ 1Iλ  ω (9.61)

Comparison with equation (9.55) reveals that

I  λIλ 1 (9.62)
A transformation of this type is called a similarity transformation. In instances where we
deal with orthogonal transformations (which will often be our case), we have λ 1  λ T ,
with λ T the transpose of λ , and
I  λIλ T (9.63)

- 115 -
 Chapter 9 Dynamics of Rigid Bodies

Finally, if the transformation matrix is such that it takes the initial coordinate axes into the
principal axes, then, the transformed inertia tensor will be diagonal.
Example
Use the results of the preceding example of the cube (equations (9.58) and the following
paragraph) to render its inertia tensor (using equations (9.18) and (9.19)) diagonal.
Solution.
From equations (9.18) and (9.19), and the corresponding set of coordinate axes defined in
Figure 9-1, we can write the inertia tensor of the cube as
 2 1 1 
 3  
4
 
4
 
1 2 1
I       (9.64)
 4 3 4 
 
 1  1
 
2 

 4 4 3 
where   Mb 2 . If the three initial basis vectors are denoted by e1 , e 2 and e3 , we know
from previous results that the main principal axis can be chosen to be
1
e1   e1  e2  e3  (9.65)
3
This is equivalent to bringing the x1 -axis along the main diagonal of the cube with a 45°
rotation. More precisely, to achieve this we first make rotation a λ1 of 45° about the x3 -axis,
 
and a second rotation λ 2 of  tan 1 1/ 2  35 about the new x2 -axis resulting from
the initial rotation. The two remaining vectors necessary to complete the new basis are then
1
e2   e1  e2 
2
(9.66)
1
e3   e1  e2  2e3 
6
and are seen to satisfy the following (required) relations
e1  e2  e2  e3  e1  e3  0 (9.67)
From equations (9.65) and (9.66), the total transformation matrix can be written as
 
 1 1 1 
 
1  3 3 
λ  0  (9.68)
3 2 2 
 1 1 
  2
 2 2 

- 116 -
 Chapter 9 Dynamics of Rigid Bodies

It can be verified that λ  λ 2 λ1 with

 1 1   2
 2 0 1 
2  0 
   3 3
 1 1 
λ1    0 , λ 2   0 1 0  (9.69)
2 2  
   1 2
 0 0 1
 3 0
   3 

Using equation (9.63) with equations (9.64) and (9.68), we find

1 
6  0 0 
 
11
I  λIλ T   0  0  (9.70)
 12 
 
 0 11 
0 
 12 
this is the same result as was obtained in equation (9.53).

9.6 Moments Inertia for Different Body Coordinate Systems

We consider two sets of coordinate axes that are oriented in the same direction, but have
different origins. The xi -axes have their origin O located at the centre of mass of the rigid
body, and the X i -axes have their origin Q located somewhere else inside, or outside, of the
body (see Figure 9-3).
The elements of the inertia tensor J relative to the X i -axes are

J ij   m  ij X  ,k X  ,k  X  ,i X  , j  (9.71)


If the vector a connects the origin Q to the centre of mass (and origin) O, then the general
vector R for the position of a point within the rigid body is written as R  a  r , or using
components
X  ,i  ai  x ,i (9.72)

- 117 -
 Chapter 9 Dynamics of Rigid Bodies

Figure 9-3 The X i -axes are fixed in the rigid body and have the same orientation as the
xi -axes, but its origin Q is not located at the same point O, which is the centre of mass of
the body.
Inserting equation (9.76) into equation (9.75) we get
J ij   m  ij  ak  x , k  ak  x , k    ai  x ,i   a j  x , j  


  m  ij x , k x , k  x ,i x , j  

(9.73)
 m  ij  2 x , k ak  ak ak    ai x , j  a j x ,i  ai a j  


 I ij   m  ij ak ak  ai a j    m  ij 2 x , k ak  ai x , j  a j x ,i 
 

But from the definition of the centre of mass itself, the last term on the right hand side of
the last of equations (9.73) equals zero since


m x ,i 0 (9.74)

We then find the final result that

J ij  I ij  M  ij a 2  ai a j  (9.75)

with M   m and a 2  ak ak .


We see from equation (9.75) that the inertia tensor components are minimum when
measured relative to the centre of mass.

Example
Find the inertia tensor of a homogeneous cube of side b relative to its centre of mass.
Solution.
We previously found that the components of the inertia tensor of such a cube relative to one
of its corner are given by

- 118 -
 Chapter 9 Dynamics of Rigid Bodies

2 1
J11   , J12    (9.76)
3 4
for the moments and the (negative of) the products of inertia, respectively. Since the centre
of mass of the cube is located at relative to a corner, we can use equation (9.75) to
calculate the components I ij of the inertia tensor relative to the centre of mass

3 b2  2 1 1
I11  J11  M  b 2         (9.77)
4 4 3 2 6
and
b2
I12  J12  M 0 (9.78)
4
The inertia tensor I is, therefore, seen to be diagonal and proportional to the unit tensor
with 1
1
I Mb 2 1 (9.79)
6

9.7 The Euler Angles

We stated in section 10.1 that of the six degrees of freedom of a rigid body, three are
rotational in nature (the other three are for the translation motion of the centre of mass). In
this section, we set to determine the set of angles that can be used to specify the rotation of
a rigid body.
We know that the transformation from one coordinate system to another can be represented
by a matrix equation such as
x  λx (9.80)
If we identify the inertial (or fixed) system with x and the rigid body coordinate system
with x, then the rotation matrix λ describes the relative orientation of the body in relation
to the fixed system. Since there are three rotational degrees of freedom, λ is actually a
product from three individual rotation matrices; one for each independent angles. Although
there are many possible choices for the selection of these angles, we will use the so-called
Euler angles  ,  and .
The Euler angles are generated in the following series of rotation that takes the
fixed x system to the rigid body x system (see Figure 9-4).
1. The first rotation is counterclockwise through an angle  about the x3 -axis. It transforms
the inertial system into an intermediate set of xi -axes. The transformation matrix is

 cos   sin   0 
 
λ     sin   cos   0  (9.81)
 0 0 1 

- 119 -
 Chapter 9 Dynamics of Rigid Bodies

with 0    2 , and
x  λ  x (9.82)

Figure 9-4 – The Euler angles are used to rotate the fixed x system to the rigid body x
system. (a) The first rotation is counterclockwise through an angle  about the x3 -axis .(b)
The second rotation is counterclockwise through an angle  about the x1 -axis . (c)The
third rotation is counterclockwise through an angle  about the x3 -axis.

2. The second rotation is counterclockwise through an angle  about the x1 -axis (also
called the line of nodes). It transforms the inertial system into an intermediate set of
xi -axes. The transformation matrix is

1 0 0 
 
λ  0 cos   sin    , (9.83)
0  sin   cos   

with 0     , and
x  λ x (9.84)
3. The third rotation is counterclockwise through an angle  about the x3 -axis. It
transforms the inertial system into the final set rigid body xi -axes . The transformation
matrix is
 cos   sin   0 
 
λ    sin   cos   0  , (9.85)
 0 0 1 

with 0    2 , and
x  λ x (9.86)

- 120 -
 Chapter 9 Dynamics of Rigid Bodies

Combining the three rotations using equations (9.81), (9.83), and (9.85) we find that the
complete transformation is given by
x  λ λ λ  x (9.87)

and the rotation matrix is

λ  λ λ λ  (9.88)

Upon calculating this matrix, we find that its components are

11  cos   cos    cos   sin   sin  
21   sin   cos    cos   sin   cos  
31  sin   sin  
12  cos   sin    cos   cos   sin  
22   sin   sin    cos   cos   cos   (9.89)
32   sin   cos  
13  sin   sin  
23  cos   sin  
33  cos  
with 0    2 , 0     , and 0    2 .

Correspondingly, the rate of change with time (i.e., the angular speed) associated with each
of the three Euler angles are defined as  ,  , and  . The vectors associated with  ,  ,
and  can be written as
φ  e3
θ  e
1 (9.90)
ψ   e3   e3
Taking the projections of the unit bases vectors appearing in equation (9.90) on the rigid
body bases vectors, we find
φ   sin   sin   e1  sin   cos   e2  cos   e3 
θ   cos   e1  sin   e 2  (9.91)
ψ   e3

Combining the three equations (9.91), we can express the components of the total angular
velocity vector ω as a function of  ,  , and 

- 121 -
 Chapter 9 Dynamics of Rigid Bodies

1   sin   sin     cos  

2   sin   cos     sin   (9.92)
3   cos   

9.8 Euler’s Equations

To obtain the equations of motion of a rigid body, we can always start with the fundamental
equation
 dL 
  N (9.93)
 dt  fixed
where N is the torque, and designation “fixed” is used since this equation can only applied
in an inertial frame of reference. We also know from our study of noninertial frames of
reference in Chapter 9 that
 dL   dL 
     ωL (9.94)
 dt  fixed  dt body
or
 dL 
   ωL  N (9.95)
 dt body
Using tensor notation we can write the components of equation (9.95) as
L    L  N
i ijk j k i (9.96)
Now, if we chose the coordinate axes for the body frame of reference to coincide with the
principal axes of the rigid body, then we have from equations (9.40)
L1  I11 , L2  I 22 , L3  I 33 (9.97)
Since the principal moments of inertia I1 , I 2 and I 3 are constant with time, we can
combine equations (9.96) and (9.97) to get
I11   I 2  I 3  23  N1
I 2 2   I 3  I1  31  N 2 (9.98)
I 3 3   I1  I 2  12  N 3
Alternatively, we can combine these three equations into one using indices
I i  I j  i j    ijk  I k  k  N k   0 (9.99)
k

where no summation is implied on the I and j indices. Equations (9.99) are the so-called
Euler equations of motion for a rigid body.
Examples

- 122 -
 Chapter 9 Dynamics of Rigid Bodies

1. The dumbbell. We return to problem of the rotating dumbbell that we solved earlier.
Referring to equation (9.35) we found that the angular momentum was given by (using the
rigid body coordinate system)
L   sin    m1r12  m2 r2 2  e1 (9.100)

with the principal moments of inertia from the system given by

I1  I 2  m1r12  m2 r2 2
(9.101)
I3  0
and the angular velocity components by
1   sin  
2  0 (9.102)
3   cos  
Since the system of axes chosen correspond to the principal axes of the dumbbell, then we
can apply Euler’s equations of motion (i.e., equations (9.98). With the constraint that   0 ,
we find
N1  0
N 2  I131   2 sin   cos    m1r12  m2 r2 2  (9.103)
N3  0
or
N   m1r12  m2 r2 2   2 sin   cos   e 2 (9.104)

Upon using equations (9.30) we can rewrite the torque in the inertial coordinate system
(that shares a common origin with the body system) as
N   m1r12  m2 r2 2   2 sin   cos     sin t  e1  cos t  e2  (9.105)

This is the same result as what was obtained with equations (9.38) without resorting to
Euler’s equation.

2. Force-free motion of a symmetric top.

We consider a symmetric top, whose principal moments of inertia are I1  I 2  I 3 , when
no forces or torques are acting on it. In this case, the Euler’s equations of motion are
 I 2  I 3  23  I11  0
 I3  I1  31  I 2 2  0 (9.106)
I 3 3  0
where I1 (as opposed to I 2 ) was used throughout. Because there are no forces involved,
the top will be either at rest or in uniform motion with respect to the inertial frame of

- 123 -
 Chapter 9 Dynamics of Rigid Bodies

reference. We will assume the top to be at rest, and located at the origin of the inertial
coordinate axes. We want to find equations for the time evolution of the components of the
angular velocity vector ω .
From the third of equations (9.106) we find that  3  0 , or

3  t   cste (9.107)
The equations for the other two components of the angular velocity yield
1  2  0
(9.108)
 2  1  0
with
I I 
   3 1  3 (9.109)
 I1 
Equations(9.108) form a system of coupled first order differential equations that can be
solved in a similar fashion as was done for the problem of the Foucault pendulum in the
previous chapter. So, we multiply the second equation by i and add it to the first to get
  i  0 (9.110)
with   1  i2 . The solution to equation (9.110) is of the form
  t   Aeit (9.111)
or, alternatively,
1  t   A cos  t 
(9.112)
2  t   A sin  t 
Since 3 is a constant, we find that the speed of rotation is also a constant. That is,

 2  12  2 2  32  A2  32  cste (9.113)

Because equations (9.112) are that of a circle, we find that the angular velocity vector ω
precesses about the top’s axis of symmetry with a constant angular frequency  . To an
observer attached to the rigid body coordinate system, ω traces a cone around the x3 -axis,
called the body cone. Also, since the top is not subjected to any forces or torques, the
angular momentum L and the energy (which in this case reduces to the kinetic energy of
rotation Trot ) are conserved. Therefore,
1
Trot  L  ω  cste (9.114)
2
with
 dL 
  0 (9.115)
 dt  fixed

- 124 -
 Chapter 9 Dynamics of Rigid Bodies

Equations (9.114) and (9.115) imply that the angular velocity vector ω makes a constant
angle with the (also constant) angular momentum vector L. Therefore, not only ω does
precesses about the body x3 -axis, but it also precesses the axis that specifies the direction of
L. The angular velocity vector ω traces a cone around the L-axis, called the space cone.

Figure 9-5 The relative orientation of L, ω and the x3 -axis that result from the force-free
motion of a symmetric top. We let L to lie along the x3 -axis in the fixed coordinate system
and I1  I 3 . We can imagine the body cone rolling around the space cone.

We can find out the relative orientation of L, ω and the x3 -axis by calculating the
following double product
L   ω  e3   Li ijk  j 3k   3ij Li j
(9.116)
 L12  L21  I112  I 221  0
since I1  I 2 for a symmetric top. We therefore have that L, ω and the x3 -axis all lie on
the same plane. An example of this is shown in Figure 9-5 for the case where we let L to lie
along the x3 -axis in the fixed coordinate system and I1  I 3 .
Finally, we can calculate the rate at which the rigid body will precess (about the x3 -axis) in
the inertial system. We know from the definition of the Euler angles that the rotation rate
about the x3 -axis is given by  . Furthermore, the second Euler angle  is that made
between the x (or L) and x -axes, with   0 since this angle is constant. Then, using the
3 3

first two of equations (9.92) we have

12  2 2  2 sin 2   (9.117)
If we define the components of the angular velocity vector ω as a function of the angle 
it makes to the x3 -axis, we have

12  2 2   sin  
(9.118)
3   cos  
Since I1  I 2 we can also express the components of the angular momentum as

- 125 -
 Chapter 9 Dynamics of Rigid Bodies

L12  L2 2  I1 sin  

(9.119)
L3  I 3 cos  
or, alternatively,
L12  L2 2  L1 sin  
(9.120)
L3  L3 cos  
Combining equations (9.117) to (9.120) we find that
 sin   L
   (9.121)
sin   I1

9.9 The Tennis Racket Theorem

The so-called tennis racket theorem is concerned with the stability of the rotational motion
of a rigid body about its principal axis. We want to find out if, when a small perturbation is
applied to the body, the motion either returns to its initial state or perform small oscillations
about it (i.e., if it does not do this, then it is unstable).
We consider a general rigid body with principal moments of inertia such that I 3  I 2  I1
We assume that the body coordinate axes are aligned with its principal axes, and first
consider an initial rotation about the x1 -axis, and then apply small perturbations about the
other two axes such that the angular velocity vector becomes
ω  1e1   e2   e3 (9.122)
with   1 and   1 . Using equations (9.98) we can write the equations of motion
for the system
I11   I 2  I 3    0
I    I  I    0
2 3 1 1 (9.123)
I 3    I1  I 2  1  0
If we will only keep terms of no higher than first order in  and  , the first of equations
(9.123) imply that
1  cste (9.124)
while the other two can be rewritten as
 I 3  I1 
   1  
 I2 
(9.125)
I I 
   1 2 1  
 I3 

- 126 -
 Chapter 9 Dynamics of Rigid Bodies

where the quantities in parentheses are constant. Taking the time derivative of the first of
equations (9.125)(9.129) and inserting the second in it we get
 I 3  I1    I  I  I  I  
   1     1 3 1 2 12   (9.126)
 I2   I 2 I3 
The solution to this second order differential equation is of the following form
  t   Aeit  Bd  it (9.127)
with

1  1
 I1  I3  I1  I 2  (9.128)
I 2 I3
Since I1  I 3 and I1  I 2 , 1 is real and the motion resulting from the perturbation is a
bounded oscillation. Furthermore, upon inserting the result in equations (9.125) we find a
similar result for   t  . The rotation motion about the x1 -axis is therefore stable.
If we study next rotations about the x2 -axis and x3 -axis, we obtain similar results and
identify the frequency of oscillations, stemming form the perturbations, by permutation of
the indices in equation (9.128). That is,

 2  2
 I 2  I1  I 2  I 3 
I 3 I1
(9.129)
3  3
 I3  I 2  I3  I1 
I1 I 2
But because I 3  I 2  I1 , we find that 3 is also real, while  2 is imaginary. Just as the
motion about the x1 -axis was found to be stable, so is the motion about the x3 -axis. On
the other hand, because  2 is imaginary a perturbation will increase exponentially with
time when the initial rotation is about the intermediate x2 -axis. Motion about this axis is
thus unstable. We can readily test this result by spinning a rigid body (like a tennis racket!)
about each of its principal axis.

9.10 Lagrangian Method for Rigid-Body Dynamics

9.10.1 Rotational Kinetic Energy of a Symmetric Top
The rotational kinetic energy for a symmetric top can be written as

Trot 
1
2
I 332  I1 12  2 2   (9.130)


or explicitly in terms of the Euler angles  , ,  and their time derivatives , ,  as

- 127 -
 Chapter 9 Dynamics of Rigid Bodies

Trot 
1
2  
I 3    cos  
2

 I1  2  2 sin 2   (9.131)

We now briefly return to the case of the force-free symmetric top for which the Lagrangian
 
is simply L  , , ,  Trot . Since  and  are ignorable coordinates, their canonical
angular momenta
L
p 

 
 I 3    cos  cos   I1 sin 2 
(9.132)
L
p 


 I 3    cos   I 33 
are constants of the motion. By inverting these relations, we obtain

 
p  p cos 
,   3 
p   p cos   cos 
(9.133)
I1 sin 2  I1 sin 2 
and the rotational kinetic energy (9.131) becomes
   
2
1  2 p  p cos  
Trot   I1  I 332   (9.134)
2 I1 sin 
2
 
The motion of a force-free symmetric top can now be described in terms of solutions of the
Lagrange equation for the Euler angle θ:
d  L   L   sin  I cos   p
    I1 
dt    
1   
(9.135)
p  p cos   p  p cos  
2 2
 

I1 sin 2  sin 2 
Once θ(t) is solved for given values of the principal moments of inertia I1  I 2 and I 3
and the invariant canonical angular momenta p and p , the functions   t  and   t 
are determined from the time integration of Eqs.(9.133).

9.10.2 Symmetric Top with One Fixed Point

We now consider the case of a spinning symmetric top of mass M and principal moments
of inertia ( I1  I 2  I 3 ) with one fixed point O moving in a gravitational field with constant
acceleration g. The rotational kinetic energy of the symmetric top is given by equation
(9.131) while the potential energy for the case of a symmetric top with one fixed point is
U(θ) = Mgh cos θ, where l is the distance from the fixed point O to the center of mass (CM)
of the symmetric top. The Lagrangian for the symmetric top with one fixed point is
I1  2 2 2 I
L (   sin  )  3 ( cos   ) 2  Mgh cos  (9.136)
2 2
A normalized form of the Euler equations for the symmetric top with one fixed point (also

- 128 -
 Chapter 9 Dynamics of Rigid Bodies

known as the heavy symmetric top) is expressed as

b  cos 
 
sin 2 
(9.137)
   a sin  
1  b cos   b  cos  
sin 3 

where time has been rescaled such that ...   I1 / p  ... and the two parameters a and
b are defined as
MghI1 p
a 2
, b  (9.138)
p p

Figure 9-6 Symmetric top with one fixed point

The normalized heavy-top equations (9.137) have been integrated for the initial conditions
 
0 , 0 ; 0  1, 0;0  . The three Figures shown below correspond to three different cases (I,
II, and III) for fixed value of a (here, a = 0.1), which exhibit the possibility of azimuthal
reversal when   changes sign for different values of b; the azimuthal precession motion is
called nutation.
The Figures on the left show the normalized heavy-top solutions in the  ,   -plane while
the Figures on the right show the spherical projection of the normalized heavy-top solutions
 ,     sin  cos  ,sin  sin  , cos   , where the initial condition is denoted by a dot (•). In
Case I  b  cos   , the azimuthal velocity  never changes sign and azimuthal
0

precession occurs monotonically. In Case II  b  cos  0  , the azimuthal velocity 

vanishes at    (where  also vanishes) and the heavy symmetric top exhibits a cusp at
0

   0 . In Case III  b  cos  0  , the azimuthal velocity  vanishes for    0 and the
heavy symmetric top exhibits a phase of retrograde motion. Since the Lagrangian (9.136)
is independent of the Euler angles  and , the canonical angular momenta p and p ,
respectively, are constants of the motion. The solution for θ(t) is then most easily obtained

- 129 -
 Chapter 9 Dynamics of Rigid Bodies

by considering the energy equation

   
2
1  2 p  p cos  
E   I1  I 33 
2
  Mgh cos  (9.139)
2 I1 sin 
2


where p and p  I 33 are constants of the motion. Since the total energy E is itself a
constant of the motion, we may define a new energy constant
1
E '  E  I 33 (9.140)
2
and an effective potential energy

p  p cos  
2

V  

  Mgh cos  (9.141)
2 I1 sin 2 

Figure 9-7 Orbits of heavy top – Case I

Figure 9-8 Orbits of heavy top – Case II

Figure 9-9 Orbits of heavy top – Case III

- 130 -
 Chapter 9 Dynamics of Rigid Bodies

so that Eq. (9.139) becomes

1 2
E  I1  t   V   (9.142)
2
which can be formally solved as
d
t      (9.143)
 2 / I1   E   V  
Note that turning points tp are again defined as roots of the equation E   V   .
A simpler formulation for this problem is obtained as follows. First, we define the
following quantities
2Mgh 2E E p p
2  ,   ,   ,   (9.144)
I1 I1 2 mgh I1 I1
so that Eq. (9.143) becomes
du du
 u     (9.145)
1  u    u      u 
2 2
1  u    W  u 
2

where τ (u) = Ωt(u), u = cos θ, and the energy equation (9.142) becomes
1  du  2
2 2
2 1  du 
      u    u  1  u     W  u  (9.146)
1  u 2   d    d 
where the effective potential is

   u 
2

W u  u (9.147)
1  u  2

We note that the effective potential W(u) is infinite at u = ±1 and has a single minimum at
u  u0 (or    0 ) defined by the quartic equation
2
    u0      u0 
W   u0   1  2u0  2 
 2  2 
0 (9.148)
 1  u0   1  u0 
This equation has four roots: two roots are complex roots, a third root is always greater than
one for α > 0 and β > 0 (which is unphysical since u = cosθ ≤ 1), while the fourth root is
less than one for α > 0 and β > 0; hence, this root is the only physical root corresponding to
a single minimum for the effective potential W(u) (see Figure 9-10). Note how the linear
gravitational-potential term u is apparent at low values of α and β.
We first investigate the motion of the symmetric top at the minimum angle  0 for which
  W  u0  and u  u0   0 . For this purpose, we note that when the dimensionless azimuthal
frequency

- 131 -
 Chapter 9 Dynamics of Rigid Bodies

d    u
  v u  (9.149)
d 1 u2
is inserted in Eq. (9.148), we obtain the quadratic equation 1  2u0 v0 2  2 v0  0 , which
has two solutions for v0  v  u0  :

  2u0 
v  u0   1  1  2  (9.150)
2u0   
Here, we further note that these solutions require that the radicand be positive, i.e.,
 2  2u0 , or I 33  I1 2u0 (9.151)

if u0  0 (or  0   / 2 ); no conditions are applied to  3 for the case u0  0 (or

 0   / 2 ) since the radicand is strictly positive in this case.

Figure 9-10 Effective potential for the heavy top.

Hence, the precession frequency 0  v  u0   at    0 has a slow component and a fast
component
  I1 
2 
I 33 
 
0 slow

2 I1 cos  0 
1 1 2
I 
 cos  0 

  3 3 
(9.152)
  I1 
2 
I 33 
 
0 fast

2 I1 cos  0 
1 1 2
I 
 cos  0 

  3 3 
We note that for  0   / 2 (or cos  0  0 ) the two precession frequencies have the same
sign while for  0   / 2 (or cos  0  0 ) the two precession frequencies have opposite
 
signs   0 and 
0 slow  0 0.
fast

- 132 -
 Chapter 9 Dynamics of Rigid Bodies

Figure 9-11 Turning-point roots.

Next, we investigate the case with two turning points u1  u0 and u2  u1 (or 1   2 )
where   W  u  (see Figure 9-11), where the θ-dynamics oscillates between 1 and  2 .
The turning points u1 and u2 are roots of the function

F  u   1  u 2    W  u    u 3      2  u 2  1  2  u      2  (9.153)

Although a third root u3 exists for F(u) = 0, it is unphysical since u3  1 . Since the
azimuthal frequencies at the turning points are expressed as
d1    u1 d2    u2
 ,  (9.154)
d 1  u12 d 1  u2 2

where    u1     u2 , we can study the three cases for nutation numerically

investigated below Eqs.(9.137); here, we assume that both   b and β are positive. In
Case I (    u2 ), the precession frequency d / d is strictly positive for u1  u  u2 and
nutation proceeds monotonically. In Case II (    u2 ), the precession frequency d / d
is positive for u1  u  u2 and vanishes at u  u2 ; nutation in this Case exhibits a cusp at
 2 . In Case III (    u2 ), the precession frequency reverses its sign at ur   /   b or
 2   r  arccos  b   1 .

9.10.3 Stability of the Sleeping Top

Let us consider the case where a symmetric top with one fixed point is launched with initial
conditions  0  0 and     0 , with   0 . In this case, the invariant canonical
momenta are
p  I 3 0 , p  p cos  0 (9.155)
These initial conditions ( u0   /  , u0  0 ), therefore, imply from Eq. (9.146) that
  u0 and that the energy equation (9.146) now becomes

- 133 -
 Chapter 9 Dynamics of Rigid Bodies

2
 du 
   1  u     u0  u    u0  u 
2 2
(9.156)
 d 
Next, we consider the case of the sleeping top for which an additional initial condition is
 0  0 (and u0  1 ). Thus Eq. (9.156) becomes
2
 du 
   1  u   1  u 
2 2
(9.157)
 d 
The sleeping top has the following equilibrium points (where u  0 ): u1  1 and
u2   2  1 .We now investigate the stability of the equilibrium point u1  1 by writing u =
1−δ (with   1 ) so that Eq. (9.157) becomes
2
 d 
1

  2    
2 2
 (9.158)
 d 
The solution of this equation is exponential (and, therefore, u1 is unstable) if  2  2 or
oscillatory (and, therefore, u1 is stable) if  2  2 . Note that in the latter case, the
condition  2  2 implies that the second equilibrium point u2   2  1 >1 is unphysical.
We, therefore, see that stability of the sleeping top requires a large spinning frequency 3 ;
in the presence of friction, the spinning frequency slows down and ultimately the sleeping
top becomes unstable.

- 134 -
 Chapter 10 Coupled Oscillations

Chapter 10
Coupled Oscillations

We studied systems that exhibit oscillations in their response, either naturally or when
driven by an external force. We now generalize the problem to include situations where not
only multiple oscillation modes or frequencies are possible, but also when there are
interactions amongst the different oscillating components of a given system. This leads us
to the study of the more complicated topic of coupled oscillations.

10.1 A Simple Example – Two Coupled Oscillators

We consider the problem of two particles of similar mass M connected by a spring of

constant 12 , and further each particle connected to fixed points with springs of constant
 . The motion of particles is restricted to direction along the x-axis, so the system has two
degrees of freedom x1 and x2 that give the displacement of the masses from their respective
equilibrium position (see Figure 10-1). The kinetic and potential energies of the system is
given by

M  x12  x2 2 
1
T (10.1)
2
and

U    x12  x2 2   12  x2  x1 
1 1 2
(10.2)
2 2
respectively. Using L=T-U for the Lagrangian, we can easily calculate the equations of
motion to be
Mx1    12  x1  12 x2  0
(10.3)
Mx2    12  x2  12 x1  0
Because we expect oscillatory motions for the systems response, we attempt a solution of
the form
xk  t   Bk eit , k  1, 2 (10.4)

with Bk the complex amplitudes and and  a frequency of oscillation. As we will see,

135
 Chapter 10 Coupled Oscillations

Bk and  can take different values depending on the mode of oscillation.

Figure 10-1 – Two masses connected by a spring to each other and by other springs to
fixed points.

xk   2 xk , we can transform equations (10.3) to

Using equations (10.4) along with 
 M  2 B1eit    12  B1eit  12 B2 eit  0
(10.5)
 M  2 B2 eit    12  B2 eit  12 B1eit  0
Regrouping terms and simplifying (by dropping the common exponential term), this
equation can be written in a matrix form as
  12   2 M  12  B 
  1  0 (10.6)

 12   12   M   B2 
2

As usual, for this system of equations to have a non-trivial solution the determinant of the
matrix on the left side of equation (10.6) must vanish. That is,

   12  2M  12
0 (10.7)
12    12   M 
2

The expansion of this determinant yields the so-called characteristic equation of the
system

     2 M   12 2  0
2
12 (10.8)

or, if we take the square root,

  12   2 M  12 (10.9)
Solving for  , we find the characteristic frequencies (or eigenfrequencies, or
eigenvalues) of the system. In this case, there are four frequencies: 1 and 2 , with

  212 
1  , 2  (10.10)
M M
If we set   1 in equations (10.4) and insert it in equation (10.6), we find that

136
 Chapter 10 Coupled Oscillations

B1   B2 . Similarly, if we set   2 in equations (10.4) and insert it in equation

(10.6), we find that B1  B2 . If we associate one amplitude constant for each
eigenfrequency, i.e., Bi  for i , we can write the complete solution to the system of
equations (10.6) as
x1  t   B1 ei1t  B1 e  i1t  B2  ei2t  B2  e  i2t
(10.11)
x2  t   B1 ei1t  B1 e i1t  B2  ei2t  B2  e  i2t
We see from this last set of equations that the position of the particles are both functions of
the two frequencies 1 and 2 , The two degrees of freedom x1  t  and x2  t  are not,
therefore, independent of each other. We would like to find out if there exists a
transformation that will lead to a new set of coordinates that would be decoupled along the
different modes of oscillation. Inspection of equations (10.11) suggests an obvious
candidate. That is, if we introduce the following new coordinates
1  x1  x2
(10.12)
2  x1  x2
or,
1
x1  1  2 
2 (10.13)
1
x2  2  1 
2
and we substitute this last set of equations into equations (10.5) we find
   
M 1  2    2 12 1  2  0
(10.14)
M        2    
1 2 12 1 2
0
By adding and subtracting the last two equations, we easily solve this system to obtain

M 1    2 12 1  0 (10.15)
M 2  2  0
We can proceed as was done for x1  t  and x2  t  to find that

1  t   C1 ei t  C1 e i t

1 1

(10.16)
2  t   C2  ei t  C2  e  i t
2 2

where the frequencies 1 and 2 are as defined by equations (10.10). We see from
equations (10.15) and (10.16) that 1  t  and 2  t  are decoupled and independent.

The constants Ci  are to be determined from the initial conditions. For example, if we
   
have x1  0    x2  0  and x1 0   x2 0 , then 2 0   2 0  0 and C2   C2   0 ; that
is,  2  t   0 at all times. We find that in this case the particles oscillate out of phase with

137
 Chapter 10 Coupled Oscillations

each other at frequency 1 ; this is the anti-symmetrical mode of oscillation. Conversely,

 
if we set x1  0   x2  0  and x1 0  x2 0 , we find that 1  t   0 at all times. The
particles then oscillate in phase with each other at frequency 1 ; this is the symmetrical
mode of oscillation. These modes are illustrated in Figure 10-2.

Figure 10-2 – The two modes of oscillation. The anti-symmetrical mode is shown on the
left, and the symmetrical mode on the right.

10.2 The General Problem of Coupled Oscillations

We now consider a general problem of a conservative system with and a corresponding set
of generalized coordinates qk , with k=1,2,…,n. We suppose that there exists a
configuration where the system is at equilibrium, with the generalized coordinates having
values qk 0 . We expand the potential energy U of the system with a Taylor series around this
configuration of equilibrium
U
 
U q1 ,q2 ,...,qn  U 0  
qk
q k
 qk 0 
k 0
(10.17)
1
 
 2U
2 jk q j qk
 q  q q
j j0 k 
 qk 0  ...
0

where we neglected any terms of higher than second order. We can arbitrarily set the first
term on the right hand side U 0 (the potential energy at equilibrium) to zero since the
potential energy can only be defined up to a constant; therefore, U 0  0 . Moreover, the
existence of an equilibrium configuration implies that the first derivative of the potential
energy relative to each generalized coordinate evaluated at the corresponding positions of
equilibrium (i.e., at qk 0 ) is also zero. That is,

U
0 (10.18)
qk 0

and U is at a minimum when qk  qk 0 . Finally, if we further simplify the notation by

setting qk 0  0 , we can approximate the potential energy by

138
 Chapter 10 Coupled Oscillations

1
U A q q
2 jk jk j k
(10.19)

with
 2U
Ajk  (10.20)
q j qk
0

It is obvious from the form of equation (10.20) that Ajk is symmetric (i.e., Ajk  Akj ).
If the potential energy is a quadratic function of the generalized coordinates, as is evident
from equation (10.19), we can use already derived results for the kinetic energy of the
system when the equations connecting the generalized coordinates and the Cartesian
coordinates do not explicitly involve time. That is, if
x ,i  x ,i  qk  or qk  qk  x ,i  (10.21)

then the kinetic energy is given by

1
T  m q q
2 jk jk j k
(10.22)

with
n 3 x ,i x ,i
m jk   m (10.23)
 1 i 1 q j qk
As was the case for Ajk , m jk is symmetric (i.e., m jk  mkj ). Just as we did for the
potential energy, we can expand the expression for the quantities m jk about the position of
equilibrium; we then get
m jk
m jk  q1 , q2 ,..., qn   m jk  ql 0     qi  ql 0   ... (10.24)
i qi 0

However, in order to be consistent in the accuracy kept for both the potential and kinetic
energies, we only keep the first term on the right hand side of equation (10.24). This way,
both expressions are valid to the second order (in velocities for the kinetic energy, and in
displacement for the potential energy). We then write
1
T  m q q
2 jk jk j k
(10.25)
1
U   Ajk q j qk
2 jk
with the understanding that m jk consist only of the first term in the expansion on the right
side of equation (10.24).
We are now interested in solving for the equations of motion of the system, using the
Lagrangian formalism. That is,

139
 Chapter 10 Coupled Oscillations

L d  L 
 0
qk dt  qk 
(10.26)

which, in this case simplifies to

U d  T 
 0
qk dt  qk 
(10.27)

Using equations (10.25), the equations of motion are reduced to the following

A jk
q j  m jk qj  0 (10.28)
j

Equations (10.28) represent a set of coupled second-order differential equations with

constant coefficients. Since we expect oscillatory motions, we propose a solution of the
form
q j  t   a j eit   (10.29)

where the amplitudes a j are real. Inserting this equation in equations (10.28), we find for
the equations of motion

 A jk 
  2 m jk a j  0 (10.30)
j

Alternatively, the system of equations (10.30) can be written in a matrix form

 A   m a  0
2
(10.31)

where the matrices A and m are composed of the elements Ajk and mjk ,
respectively(remember that A and m are symmetric). In order to get a non-trivial solution
to this equation, the determinant of the quantity in parentheses must vanish
A   2m  0 (10.32)

This determinant is called the characteristic or secular equation and is an equation of

degree n in  2 . The corresponding n roots r 2 are the characteristic frequencies or
eigenfrequencies of the system. The eigenvector associated with a given root r can be
evaluated by inserting it back in equations (10.30) to determine the ratios
a1 : a2 :...: an (this is similar to what we did in Chapter 10 when determining the principle
axes of the inertia tensor). If we represent by a jr the jth component of the rth eigenvector,
we can write the generalized coordinate q j as a linear combination of the solutions for
each root
q j  t    a jr eir t  r  (10.33)
r

It is, however, understood that the actual solution must be real (in a mathematical sense)
and we must, therefore, take real part of equation (10.33). That is,

140
 Chapter 10 Coupled Oscillations

q j  t    a jr cos r t   r  (10.34)

r

Example
We apply the formalism just developed to the previous problem of the two masses
connected by springs to find the characteristics frequencies.
Solution.
We know from equation (10.2) that

U    x12  x2 2   12  x2  x1 
1 1 2
(10.35)
2 2
The elements of the matrix A are therefore (from equation(10.20))
 2U
A11     12
x12 0

 2U
A12  A21   12 (10.36)
x1x2 0
 2U
A22     12
x2 2 0
The kinetic energy of the system is

T
1
2

M x12  x22  (10.37)

and the elements of the matrix m are, from equation (10.23)

m11  m22  M , m12  m21  0. (10.38)
The determinant is then given by

   12  2M  12
0 (10.39)
12    12   M 
2

which is the same equation (10.7) that we obtained before. The eigenfrequencies are
therefore unchanged at
  212 
1  , 2  (10.40)
M M

10.3 Orthogonality of the Eigenvectors and the Normal Coordinates

According to equation (10.30), we can write for the sth root s

 2
s
m jk aks   Ajk aks (10.41)
k k

141
 Chapter 10 Coupled Oscillations

and a similar one for the another root, say r

 2
r
m jk a jr   Ajk a jr (10.42)
j j

where we used the symmetry of the m and A matrices. We now multiply equation (10.41)
by a jr and the equation (10.42) by aks , and subtract the two. We then find that

 2
r
  s2  a jk
jr
m jk aks  0 (10.43)

If r  s and r 2  s 2 , then we must have that

a jr
m jk aks  0, r  s (10.44)
jk

If r=s (and, therefore, r 2  s 2  0 ) then, the double product a jr

m jk aks may not vanish.
jk

This can, in fact, be verified by calculating the kinetic energy using equation (10.34)
1
T  m q q
2 jk jk j k
  

1
  
m jk    r a jr sin  r t   r     s a js sin  st   s    (10.45)
2 jk  r  s 
1
  
   r s sin  r t   r sin  st   s   a jr m jk aks 
2 r ,s

jk

But inserting equation (10.44) when r=s and r 2  s 2 , equation (10.45) reduces to

T
1
2 r r

  2 sin 2  r t   r   a jr
m jk akr  (10.46)
jk

and since both T and r 2 sin 2 r t   r  are greater than zero, it must be that

a jr
m jk akr  0 (10.47)
jk

Moreover, because we can only measure the ratios of the components a jr , we arbitrarily
normalize them according to

a jr
m jk akr  1 (10.48)
jk

On the other hand, if we have a case of degeneracy for one eigenvalue r 2 (i.e. r  s ,
but r 2  s 2 ), we cannot outright say that equation (10.44) is satisfied (since r 2  s 2  0
in equation (10.43)). However, it turns out that we can always ensure that this is so (i.e.,
that  a jr m jk aks  0 ). For example, if we assume that two different vectors a r and a s share
jk

142
 Chapter 10 Coupled Oscillations

the same eigenfrequency  2 , then we can also say from equation (10.31) that
Aai   2mai (10.49)
with i=r,s. These two vectors bring six unknowns (one per component), for which we can
match two equations of motions (i.e., from equation (10.49)), one equation to ensure the
sought after orthogonality of the vectors (i.e.,  a jr m jk aks  0 ), and two equations to
jk

normalize the vectors (i.e., a jr

m jk akr  1, and a similar equation for s ). This gives us
jk

five equations for six unknowns. The fact that we have fewer equations than unknowns
ensures non-trivial solutions (we can, for example, arbitrarily set the value of one of the
unknowns and solve for the remaining five unknowns using the five aforementioned
equations). Thus, we can show that the orthgonality between eigenvectors is respected even
in the case of degeneracy in the values of the eigenfrequencies.
We can, therefore, combine equations (10.44) and (10.48) to get, for any r and is, the
following relation

a jr
m jk aks   rs (10.50)
jk

We say that the eigenvectors are orthonormal, in the sense defined in this last equation. It
is customary to refer to the frequencies r as being the different modes of oscillation of
the system. The vectors a r are the corresponding eigenvectors associated with these modes.
It is important to note that we have now imposed a normalization constraint on the
eignvectors that was not assumed when we initially solved the problem with equation
q j  t  (10.29) for the generalized coordinates. We must therefore introduce a new set of
quantities  r that are to be multiplied to the vectors a r . More precisely, we now have

q j  t     r a jr e 
i r t  r 
(10.51)
r

which we promptly simplify by introducing yet another constant (but, this time complex)
such that
q j  t     r a jr eir t (10.52)
r

We further define the so-called normal coordinates  r as

r  t    r ei t r
(10.53)
so that
q j  t    a jr r  t  (10.54)
r

Finally, it fairly straightforward to show, using the orthonormality condition of equation

(10.50), that the forms of the kinetic and potential energies are significantly simplified by
the use of the normal coordinates. A few lines of calculations reveal that

143
 Chapter 10 Coupled Oscillations

1
T   2
2 r r
(10.55)
1
U    r2r2
2 r
Applying the Lagrange equations to these two equations we get
r   r2r  0 (10.56)
We have the interesting result that this new system of second-order differential equations is
completely decoupled, i.e. we have n independent equations of motions.
Example
Three linearly coupled plane pendula. Three identical pendula of mass M and length are
suspended from a slightly yielding rod, which brings a certain amount of coupling between
each pair of pendula (see Figure 10-3). Find the eigenfrequencies, eigenvectors, and the
normal modes of oscillation. Consider only the case of small oscillations.
Solution.
We start by evaluating the kinetic and potential energies; we have

T
2

1 2 2 2 2
Ml 1   2   3 
    
U  Mgl 1 cos 1   1  cos  2   1  cos  3   (10.57)

  
   1   2  1   3   3   2 
1
  
2 2 2

2  
In the case of a small oscillation, we have
1
1  cos     2 (10.58)
2

Figure 10-3 – Three identical pendula that are coupled through a lightly yielding rod.

so we can re-write the potential energy as

144
 Chapter 10 Coupled Oscillations

U
1
2
  1
 
Mgl 12   22   32   212  2 22  232  21 2  213  2 3 2
2 (10.59)
1

  12   22   32  21 2  21 3  2 3 2
2

with
 
  Mgl  2 ,    . (10.60)
Mgl  2 
The transformation between the Cartesian to the polar coordinates is given by
x ,1  l sin    l
1 (10.61)
x ,2  l 1  cos     l 2
2
for   1, 2,3 (depending on the pendulum). We must use equation (10.23) to determine
the elements of the matrix m, that is
n x ,i x ,i
m jk   m (10.62)
 1  j  k
Inserting equations (10.61) in equation (10.62) we find
m11  Ml 2
m22  Ml 2
(10.63)
m33  Ml 2
m12  m13  m23  0
However, we must remember from the discussion following equation (10.24) that since
we want to keep the precision of the potential energy to the second-order in the generalized
coordinates, we must only keep the lowest order terms in equations (10.63). We therefore
make the following approximation
m11  m22  m33  Ml 2 (10.64)
and
1 0 0 
m  Ml 0 1 0 
2
(10.65)
0 0 1 
Using equation (10.20) we can directly evaluate the matrix A from equation (10.59) for
the potential energy
1   
A     1   (10.66)
   1 

145
 Chapter 10 Coupled Oscillations

We must now evaluate the following determinant to determine the eigenfrequencies

   2 Ml 2  
A  m 
2
    Ml 2 2
 (10.67)
     Ml
2 2

Expanding this determinant, we have

   Ml 
2 2 3
 2 3 3  3 2 2     2 Ml 2   0 (10.68)

which can be factored into

 Ml       2 Ml 2    2   0
2 2 2
(10.69)

The roots are therefore

 1    Mgl  3
 12   22  2

Ml Ml 2 (10.70)
 1  2 
g
 32  2

Ml l
We notice from the first of equations (10.70) that the system is degenerate, since
12  2 2
Having evaluated the eigenfrequencies, we can insert them back into the equations of
motion to find the eigenvectors a r . That is, starting with 32 ,

 A jk 
  32 m jk a j3  0 (10.71)
j

or
2 a13   a23   a33  0
(10.72)
 a13  2 a23   a33  0
We only used two of the three equations of motion since we only have two unknowns (i.e.,
two ratios taken from the components of a3 ). The third equation of motion will
automatically be satisfied. From equations (10.72) we find
1
a13  a23  a33  (10.73)
3Ml 2
where the vector was normalized.
For the degenerate case, we insert  12   22   in the equations of motion to calculate
a1 and a 2 . This yields

146
 Chapter 10 Coupled Oscillations

  a11  a21  a31   0

(10.74)
  a12  a22  a32   0
The orthogonality condition a j1m jk ak 2  0 , and the normalization conditions (imposed on
the eigenvectors) respectively give

Ml 2 a11a12  a21a22  a31a32  0 

Ml 2 a112  a21
2
 a312  1  (10.75)
Ml  a
2 2
12
 a22
2
 a 1
2
32

We therefore have five equations for six unknowns, which guaranties non-trivial solutions.
If we arbitrarily set a11  0 , we have from the first of equations (10.74) and the second of
equations (10.75) that
1
a21  a31  (10.76)
2 Ml
Then, from the first of equations (10.75) and the second of equations (10.74)
1
a22  a32   a12 (10.77)
2
and, finally, from the last of equations (10.75)
2
a12  (10.78)
3Ml 2
The three eigenvectors can be written as

a1 
1
 0,1,1
2 Ml 2

a2 
1
 2,1,1 (10.79)
6 Ml 2

a3 
1
1,1,1
3Ml 2
We see that the third normal mode corresponds to an in-phase oscillation with the three
pendula moving in the same direction with the same amplitude. On the other hand, the first
two normal modes have an out-of-phase character. In the first mode the second and third
pendula are moving in opposite directions with equal amplitude, while in the second mode
the first pendulum is moving in opposite direction of the other two, and at twice their
amplitude.

147
 Chapter 11 Hamiltonian-Jacobi Equation

Chapter 11
Canonical Transformation and
Hamilton-Jacobi Equation

11.1 Canonical Transformations

As we saw in our previous example of the spherical pendulum, there is one case when the
solution of Hamilton’s equation is trivial. This happens when a given coordinate is cyclic;
in that case the corresponding momentum is a constant of motion. For example, consider
the case where the Hamiltonian is a constant of motion, and where all coordinates q j are
cyclic. The momenta are then constants
pj   j (11.1)
and since the Hamiltonian is not be a function of the coordinates or time, we can write
H  H  j  (11.2)
where j=1, …, n with n the number of degrees of freedom. Applying the canonical equation
for the coordinates, we find
H
q j   j (11.3)
 j
where the  j ’s can only be a function of the momenta (i.e., the  j ’s) and are, therefore,
also constant. Equation (11.3) can easily be integrated to yield
q j   jt   j (11.4)

where the  j ’s are constants of integration that will be determined by the initial conditions.
It may seem that such a simple example can only happen very rarely, since usually not
every generalized coordinates of a given set is cyclic. But there is, however, a lot of
freedom in the selection of the generalized coordinates. Going back, for example, to the
case of the spherical pendulum, we could choose the Cartesian coordinates to solve the
problem, but we now know that the spherical coordinates allow us to eliminate one
coordinate (i.e.,  ) from the expression for the Hamiltonian. We can, therefore, surmise
that there exists a way to optimize our choice of coordinates for maximizing the number of

148
 Chapter 11 Hamiltonian-Jacobi Equation

cyclic variables. There may even exist a choice for which all coordinates are cyclic. We
now try to determine the procedure that will allow us to transform form one set of variable
to another set that will be more appropriate.
In the Hamiltonian formalism the generalized momenta and coordinates are considered
independent, so a general coordinate transformation from the  q j , p j  to, say, Q j , Pj  
sets can be written as
Q j  Q j (qk , pk )
(11.5)
Pj  Pj (qk , pk ).

Since we are strictly interested in sets of coordinates that are canonical, we require that
there exits some function K (Qk , Pk , t ) such that
K
Q k 
Pk
(11.6)
K
Pk  
Qk
The new function K plays the role of the Hamiltonian. It is important to note that the
transformation defined by equations (11.5)-(11.6) must be independent of the problem
considered. That is, the pair  Q j , Pj  must be canonical coordinates in general, for any
system considered.
We know from our study of the modified Hamilton’s Principle in section 6.2 that the
canonical equations resulted from the condition
 
    pk qk  H (q j , p j , t )  dt  0
t2
(11.7)
t
1
 k 
which we can similarly write for equations (11.6)
 
    Pk Q k  K (Q j , Pj , t )  dt  0
t2
(11.8)
t
1
 k 
Simultaneous validity of (11.7) and (11.8) does not imply that, the integrands are
identical, but they can differ by a total time derivative of some function F. Now consider an
arbitrary function F (with a continuous second order derivative), which can be dependent
on any mixture of q j , p j , Q j and Pj , and time. Such a function has the following
characteristic
dF
 
dt    dFdt   F |tt12    F (t1 )  F  t2    0
t2 t2
 (11.9)
t1 dt t1

The last step is valid because of the fact the no variations of the generalized coordinates are
allowed at the end points of the system’s trajectory. The function F is called the generating
function of the transformation. It has four types and depending on convenience, one of
them is used

149
 Chapter 11 Hamiltonian-Jacobi Equation

F1 (q j , Q j , t ) , F2 (q j , Pj , t ) , F3 ( p j , Q j , t ) and F4 ( p j , Pj , t )
Equations (11.7)-(11.9) imply that
  dF
   pk qk  H (q j , p j , t )    Pk Q k  K (Q j , Pj , t )  (11.10)
 k  k dt
where  is some constant. But since it will always be possible to scale the new variables
Q j and Pj , it is sufficient to only consider the case when   1 . We, therefore, rewrite
equation (11.10) as
dF
 p q
k
k k  H (q j , p j , t )   Pk Q k  K (Q j , Pj , t ) 
k dt
(11.11)

Let’s consider the case where

F  F1 (q j , Q j , t ) (11.12)
Equation (11.11) then takes the form
dF
 p q
k
k k  H   Pk Q k  K  1
k dt
(11.13)
F  F F 
  Pk Q k  K  1    1 qk  1 Q k 
k t k  qk Qk 
This equation will be satisfied if
F1 F F
pj  , Pj   1 , K  H  1 (11.14)
q j Q j t
The first of equations (11.14) represent a system of n equations that define the p j as
functions of the qk , Qk , and t. If these equations can be inverted, we then have established
relations for the Q j in terms of the qk , pk , and t, which can then be inserted in the second
of equations (11.14) to find the Pj as functions of qk , pk , and t. Finally, the last of
equations (11.14) is used to connect K with the old Hamiltonian H (where everything is
expressed as functions of the Qk , and Pk ).
From the definition, F2 (q j , Pj , t ) is Legendre transform of F1 (q j , Q j , t ) with respect to Q.
Thus,
F2 (q j , Pj ,t)   Pk Qk  F1 (q j ,Q j ,t) (11.15)
k

The condition (11.11) is

d  dF
 p q
k
k k  H   Pk Q k  K    Pk Qk   2
k dt  k  dt
(11.16)
  
F
 F F
  Pk Q k  K   Pk Q k  Pk Qk  2   2 qk  2 Pk
K k t k qk Pk


150
 Chapter 11 Hamiltonian-Jacobi Equation

The transformation equations are

F2 F F
pj  , Qj  2 , K  H  2 (11.17)
q j Pj t
The third type of generating function F3 ( p j , Q j , t ) is Legendre transform of
F1 (q j , Q j , t ) with respect to q j .

F3 ( p j , Q j , t )   pk qk  F1 (q j , Q j , t ) (11.18)
k

Then, using the similar procedure above, we get the transformation equation as,
F3 F F
qj   , Pj   3 , K  H  3 . (11.19)
p j Q j t

The fourth type of generating function F4 ( p j , Pj ,t) is a double Legendre transform

of F1 (q j ,Q j ,t) .

 
F3 (q j ,Q j ,t)   Pk Qk  pk qk  F1 (q j ,Q j ,t) (11.20)
k

The transformation equations are,

F4 F F
qj   , Qj  4 , K  H  4 . (11.21)
p j Pj t
In all these transformations, time remains an untransformed quantity. It is simply a
parameter. For relativistic formulation, this cannot be so because the space and the time is
treated on equal footings.
Since t does not appear directly except in K  H  F / t , we can say at every instant of
time that, a transformation generated by the function F is canonical if the differential dF is
exact and equals,

dF   pk dqk  Pk dQk  (11.22)
k

This is a simple criterion to test canonicity. Since the differentials dq and dQ are involved
in dF, the generating function is of the type F1 (q j , Q j , t ) .

Example
Let’s consider the problem of the simple harmonic oscillator. The Hamiltonian for this
problem can be written as
p 2 kq 2
H  T U 
2m

2

1
2m
 p 2  m 2 2 q 2  (11.23)

with k  m 2 . The form of the Hamiltonian (a sum of squares) suggests that the following
transformation

151
 Chapter 11 Hamiltonian-Jacobi Equation

p  f  P  cos  Q 
f ( P) (11.24)
q sin  Q 
m
will render the Q cyclic. Indeed, we find that
f 2  P f 2  P
H
2m
  
cos 2
Q  sin 2
 
Q 
2m
(11.25)

and Q is indeed cyclic. We must now find the form of the function f(P) that makes the
transformation canonical. Consider the generating function
m q 2
F cot  Q  (11.26)
2
Using equations (11.14) we get
F
p  m q cot  Q 
q
(11.27)
F m q 2
P 
Q j 2sin 2  Q 
and we can solve for q, and p using these two equations
2P
q sin  Q  , p  2 Pm cos  Q  (11.28)
m
We can now identify f(P) with the help of equations (11.24)
f  P   2m P (11.29)
It follows that
K  H  P (11.30)
and once again, Q is indeed cyclic. Identifying the Hamiltonian with the energy E, which is
a constant, we find that the generalized momentum P is also a constant
E
P (11.31)

The equations of motion for Q is given by
K
Q   , (11.32)
P
which yields
Q  t   (11.33)
where  is an integration constant determined by the initial conditions. Finally, we have
from equation (11.28)

152
 Chapter 11 Hamiltonian-Jacobi Equation

2P
q sin t    , p  2 Pm cos t    (11.34)
m
Although the solution of such a simple problem does not require the use of canonical
transformations, this example shows how the Hamiltonian can be brought into of form
where all generalized coordinated are cyclic.

11.2 The Hamilton-Jacobi Theory

Referring back to section 6.4 on canonical transformations, where

Q j  Q j (qk , pk )
(11.35)
Pj  Pj (qk , pk ), j , k  1,..., n
we consider the case where the generating function is now a function of the old coordinates
and the new momenta, i.e., F  F2  q j , Pj , t   S . The transformation equations are

S S S
pj  , Qj  , KH (11.36)
q j Pj t
We now seek a transformation that ensures that both the new coordinates and momenta are
constant in time. From the canonical equations we then have
K
Q k  0
Pk
(11.37)
K
Pk   0
Qk
Equations (11.37) will be satisfied if we simply force the transformed Hamiltonian K to
be identically zero, or alternatively
S
H qj , pj ,t   0 (11.38)
t
We can use the first of equations (11.36) to replace the old momenta p j in equation

 S  S
H qj, ,t  0
 q  t
(11.39)
 j 
with j , k  1,..., n . Equation (11.39), know as the Hamilton-Jacobi equation, constitutes a
partial differential equation in (n+1) variables, q j and t, for the generating function S. The
solution S is commonly called Hamilton’s principal function.
The integration of equation (11.39) only provides the dependence on the old coordinates
q j and time; it does not appear to give information on how the new momenta Pj are
contained in S. We do know, however, from equations (11.37) that they are constants.
Mathematically, once equation (11.39) is solved it will yield n+1 constants of integration

153
 Chapter 11 Hamiltonian-Jacobi Equation

1 ,...,  n 1 . But one of the constants of integration, however, is irrelevant to the solution
since S itself does not appear in equation(11.39); only its partial derivatives to q j and t are
involved. Therefore, if S is a solution, so is S   , where  is anyone of the (n+1)
constants. We can then write for the solution

S  S (q j ,  j , t ), j  1,..., n (11.40)
and we choose to identify the new momenta to the n (non-additive) constants, that is
Pj   j (11.41)
An interesting corollary is that the new coordinates and momenta can now be expressed as
functions of the initial conditions for the old coordinates and momenta, i.e.,
q j  t0  and p j  t0  . This is because we have in the first two equations of (11.36)

S  qk ,  k , t 
pj 
q j
(11.42)
S  qk ,  k , t 
Qj   j,
 j
relations that can be evaluated at time t0 . The first equation allows us to connect the
constants  k (which are, in fact, the new momenta) with the initial conditions when t  t0
is set in equation(11.42), and the second equation completes the connection by defining the
new coordinates in terms of the same initial conditions (through, in part, the  k ’s). The new
coordinates Q j are also identified with a new set of constants  j (see equation(11.37).
The problem can finally be solved by, first inverting the second of equations (11.42) to get
q j  q j  k ,  k , t  , (11.43)

and using the first of equations (11.42) to replace q j in the relation for p j to get

p j  p j  k ,  k , t  (11.44)
To get a better understanding of the physical significance of Hamilton’s principal function,
let’s calculate it total time derivative
dS  S S  S S S
  q j   j     q j  (11.45)
dt 
j  q j  j  t j q j t

since the new momenta Pj   j are constants. But from the first of equations (11.36) and
the fact that we set K=0, we can rewrite equation (11.45) as
dS
  p j q j  H  L (11.46)
dt j

Alternatively, we have

154
 Chapter 11 Hamiltonian-Jacobi Equation

S   Ldt  cste (11.47)

where the integral is now indefinite. We know that by applying Hamilton’s Principle to the
definite integral form of the action, we could solve a problem through Lagrange’s equations
of motion. We now further find that the indefinite form of the same action integral
furnishes a different way to solve the same problem through the formalism presented in this
section.
A simplification arises when the Hamiltonian is not a function of time. We can then write
H (q j , p j )  a (11.48)
where a is a constant. From equation (11.39) we can write
S  q j ,  j , t   W  q j ,  j   at (11.49)

and
 W 
H  H qj,  (11.50)
 q j
 
The function W (q j , j ) is called Hamilton’s characteristic function. Its physical
significance is understood from its total time derivative
dW  W W  W
  q j   j   (11.51)
dt 
j  q j  j  t
But since
S W
pj   (11.52)
q j q j
we find
dW
  p j q j (11.53)
dt j

Equation (11.53) can be integrated to give

W    p j q j dt    p j dq j (11.54)
j j

a quantity usually called the abbreviated action.

Example
We go back to the one-dimensional harmonic oscillator problem. The Hamiltonian for this
problem is

H
1
2m
 p 2  m2 2 q 2   E (11.55)

with

155
 Chapter 11 Hamiltonian-Jacobi Equation

k
 (11.56)
m
We now write the Hamlton-Jacobi equation for Hamilton’s principal function S by setting
p  S / q in the Hamiltonian

1  S   S
2

   m 2 2 q 2   0 (11.57)
2m  q   t
Since the Hamiltonian is not a function of time, we can use equation (11.49) for S and
write

1  W  
2

   m 
2 2 2
q   (11.58)
2m  q  
It is clear that, in this case, the constant of integration  is to be identified with the total
energy E. Equation (11.58) can be integrated to
m 2 q 2
W  2mE  dq 1  (11.59)
2E

and
m 2 q 2
S  2mE  dq 1   Et (11.60)
2E
Although we could solve equation (11.60) for S, we are mostly interested in its partial
derivatives such as
S S m dq
2E 
    t
 E m 2 q 2
1
2E (11.61)
1  m 2 
 arcsin  q t
  2 E 
 
Equation (11.61) is easily inverted to give
2E
q sin t    (11.62)
m 2
where
    (11.63)
The momentum can be evaluated with

156
 Chapter 11 Hamiltonian-Jacobi Equation

S W
p   2mE  m 2 2 q 2
q q
 2mE 1  sin 2 t     (11.64)

 2mE cos t   

Finally, we can connect the constants E (or  ) and  to the initial conditions q0 and p0 at
t=0. Therefore, setting t=0 in equations (11.62) and(11.64), and rearranging them we have
2mE  p0 2  m 2 2 q0 2 (11.65)
Using the same equations, we can also easily find that
q0
tan     m (11.66)
p0
Thus, Hamilton’s principal function is the generator of a canonical transformation to a new
coordinate that measures the phase angle of the oscillation and to a new momentum
identified as the energy.

157
Problems
Problem Set 1 for Chapter 1 and 2
1.1 Spherical coordinates. Many of the problems which we will solve will possess
spherical symmetry, so that the most natural coordinates are spherical coordinates. We'll
need to know how to express the position, velocity, acceleration, and kinetic energy of a
particle in this coordinate system.
The spherical coordinates ( r ,  ,  ) are related to the rectangular coordinates ( x, y , z )
through x  r sin  cos  , y  r sin  sin  , z  r cos  . The unit vectors are

er  sin  cos  e x  sin  sin  e y  cos  e z

e  cos  cos  e x  cos  sin  e y  sin  e z
e   sin  e x  cos  e y

(a) Draw a careful figure which illustrates the meaning of the spherical coordinates and
their relation to the rectangular coordinates.
(b) Show that this is indeed an orthogonal coordinate system. In other words, show that
er  e  0, er  e  0, e  e  0 . Also show that each of the vectors e r , e , e has a
length of one. Finally, show that the coordinate system is “right-handed”, so that
er  e  e , e  e  e r , e  er  e .
(c) Use these relations to show that

e r  e  sin  e

e  e  cos  e
 r 

e    sin  er  cos  e

(d) In spherical coordinates the position vector is r  re r . Use your results above to
calculate the velocity v  r , the kinetic energy T  mv 2 / 2 , and the acceleration a  r
of a particle in spherical coordinates.

Answer:
v  re r  re  r sin e

T
1 2
2

r  r 2 2  r 2 sin 2  2 
   
r  r 2  r sin 2  2 e r  r  2 r  r sin  cos  2 e 
a  
  r sin   2sin  r  2r cos  e

1.2 Central forces. A central force is a force directed along the line connecting the two

158
interacting particles which only depends on the separation between the particles.
Assuming that the center of the force is at the origin (r = 0), then the central force can
be written as F(r) = f(r) er, with r the distance from the origin. Important examples of
central forces are the gravitational force, the electrostatic force (Coulomb's law), and the
force exerted by a spring on a mass.
(a) Assume that a particle of mass m is acted upon by a central force. The natural
coordinates to describe the motion of the particle are spherical coordinates ( r ,  ,  ).
Separate Newton's Second Law in spherical coordinates, and write down the three
equations of motion (one for each component).
(b) Show that for any f(r), a solution of the  and  components of these equations has
   / 2 and mr 2  l , with l a constant which is independent of time. What is the
physical significance of these results?
(c) Use the conditions above to simplify the equation for the radial coordinate r. Show
that
l2
mr   f r 
mr 3
Answer:
 
r  r 2  r sin 2  2  f (r)
(a) m 
m  r  2 r  r sin  cos    0
2

m  r sin   2 sin  r  2r cos    0



1.3 Charged particle in a uniform magnetic field. A particle of mass m and charge q
moves in a uniform magnetic field B  Be z . The particle is deflected due to the Lorentz
force F  qv  B .
(a) Working in cylindrical coordinates  ,  , z , write down the equations of motion for

the particle. You should find that

 
m   2  qB 
m   2 
    qB 
mz  0

(b) Show that one solution of this set of equations is   R,   t  0 , z  v0 z t  z0 ,

where R,  , 0 , v0 z , z0 are all constants. What is  ? What type of curve is the
particle's trajectory?

1.4 Spherical pendulum. A mass m is attached to the ceiling by an inextensible (i.e.,

fixed length), massless string of length b, as shown in the figure.

159
(a) Draw a free-body diagram for the mass, showing all forces acting upon it.
(b) Use F = ma in spherical coordinates to derive the equations of motion for  and  .
(c) Use your equations of motion to show that the quantity
p  mb 2 sin 2  

is independent of time (i.e., show that dp / dt  0 ).

(d) Use your result from part (c) to eliminate the dependence
upon  in your equation of motion for  . You should find
that

 g p 2 cos 
  sin   2 4 3  0
b m b sin 
Answer:

 
(b) m b 2  b sin 2  2  T  mg cos 
m  r  b sin  cos    mgsin 
2

m  bsin   2b cos    0

1.5 Bead on a spoke. A bead moves outward with a constant

speed u along the spoke of a wheel. The angular position of the
spoke is given by    t , where  is constant.
(a) Find the velocity v and the acceleration a in the polar
coordinates r ,  .
(b) Find the velocity and acceleration in the Cartesian
coordinates (x, y).
(c) Now suppose that the radial position of the bead
is r  t   r0 e  t , with r0 and  constants, and that    t .
Find the velocity and acceleration in polar coordinates. Show that if you
chose    the radial component of the acceleration is zero.
(d) In part (c) above, how is it possible to have ar  0 , while having vr  r0 2 e t ?

Answer:
(a) v  ue r   ut   e
 
a   2 ut e r   2u  e

(b) v  u cos  t   ut sin  t   e x  u sin  t   ut cos  t   e y

a   ut 2 cos  t   2u sin  t   e x   ut 2 sin  t   2u cos  t   e y

160
 (c) v  r0  e t e r  r0 e t e
 
a  r0  2   2 e t e r  2r0  e t e

1.6 Moving inclined plane. A block of mass m is placed on a frictionless plane which
is initially horizontal (i.e.,   0 at t = 0). The plane is lifted at one corner (see the figure)
at a constant rate so that    .
(a) Draw a free-body diagram for the mass, showing all
of the forces acting upon it.
(b) Use F = ma in polar coordinates r ,   (r is the
distance up the plane) to derive the equations of
motion for the block. Show that
r   2 r   g sin t 

and that the normal force acting on the block is
N  2m r  mg cos t 
(c) Solve the equation of motion to find r(t) for initial conditions r(0) = R, r 0   0.
[Hint: try a solution of the form r t   A sinh t   B cosh t   C sin t , with A,
B, and C constants.]

Answer:
g
(c) r(t)  R cosh  t   sin  t  sinh  t 
2 2

1.7 Bead on a hoop I. A bead of mass m slides without

friction on a wire hoop of radius R. The hoop spins with
a constant angular velocity    about its vertical
diameter.
(a) Draw a free body diagram for the bead (note that the
normal force will have a component directed toward
the center of the hoop as well as a component
normal to the plane of the hoop). Resolve the forces
into components in spherical coordinates, and use
Newton's Second Law to find the equations of
motion for the bead.
(b) Find the equilibrium position(s) of the bead on the wire (i.e., values  0 of  for

which     0 ).
(c) Investigate the stability of each equilibrium position. Do this by perturbing about
each position; i.e., let    0   t , with   1 and find the equation for  ; for

161
 unstable equilibria  will grow with time.

Answer:
(a)  
m R 2  R 2 sin 2   mg cos   N r
m  R  R sin  cos    mgsin 
2

m  2R cos    N 


g g
(b) sin  0  0, cos  0  2
with  1.
R R 2
(c) Stable when g / R  2 , unstable when g / R   2 .

Problem Set 2 for Chapter 3

3.1 Lagrangian vs. Newtonian mechanics. Construct the Lagrangians and derive the
equations of motion for (a) the spherical pendulum and (b) the bead on the hoop. In both
cases compare your results with the results which you obtained in previous assignments
using Newton's Second Law.

3.2 Moving pendulum. A plane pendulum of length l has a bob of mass m, which is
attached to a cart of mass M which can move on a frictionless, horizontal table.
(a) Express the rectangular coordinates xb , yb of the pendulum
bob in terms of the coordinates  X ,  .
(b) Using X and  as your generalized coordinates, construct
the Lagrangian for this system.
(c) Using Lagrange's equations, derive the equations of motion
for the coupled pendulum and cart system.
(d) Find all first integrals of the motion (i.e., find the conserved quantities).

Answer:
1 1
(b) L (m  M ) X 2  ml cos  X   mgl cos   ml 2 2 .
2 2
g 1
(c)   sin   cos X  0
l l
X 
ml
mM
 
sin  2  cos 

3.3 Pendulum. A pendulum consisting of a rigid massless rod of

length l with a mass m at one end moves in a uniform gravitational
field g. The point of support is not stationary but moves in the vertical
direction with a displacement y(t) which is a given function of time.

162
(a) Find the Lagrangian and the equations of motion.
(b) Show that the equation is the same as that for pendulum with fixed support, with g
replaced by g eff t . What is g eff t ?
[Hint: Denote the Cartesian coordinates of the mass m by xm , ym  , where
ym t   y t ]

Answer:
1
(a) L m(l 2 2  y 2  2ly sin  )  mgy  mgl cos 
2
( g  
y)
  sin   0
l

3.4 Double pendulum. A double pendulum consists of

two simple pendula, with one pendulum suspended
from the bob of the other (see the figure below). The
two pendula have equal lengths l and have bobs of
equal mass m. Both pendula are confined to move in the
same plane.
(a) Write down the relationship between the the
coordinates x1 , y1  and  x2 , y2  of the two
pendula and the angles 1 and  2 .
(b) Derive the Lagrangian for this system, expressed in terms of the generalized
coordinates 1 and  2 .
(c) Use Lagrange's equations to derive the equations of motion for the two pendula.

3.5 Block on a cylinder. A block of mass m

starts at rest at the top of a cylinder of mass M
and radius R which is placed on the floor, as
shown in the figure. At time t = 0 the block is
given a gentle tap and it begins sliding down the
surface of the cylinder. Ignore the friction
between the block and the cylinder and the
cylinder and the floor.
(a) Using X and  as the generalized coordinates, construct the Lagrangian for this
system.
(b) Use Lagrange's equations to derive the equations of motion. Use any conservation
laws to simplify these equations (in other words, find the first integrals of the
motion).
(c) Try to reduce the equations to the form   f  , where f   is some function of

163
  which you'll need to determine. Is your result sensible in the limit that M  m ?

Answer:
1 2 1
(a) L
2 2
 
MX  m X 2  R 2 2  2 R cos  X   mgR cos  .

g 1
(b)   sin   cos  X  0
R R
2 g 1  cos   m
(c )    , 
R 1   cos  
2
mM

3.6 Cycloid pendulum (from Greiner p306). A mass point glides

without friction on a cycloid, which is given by x  a(  sin  ) and
y  a(1  cos  ) (with 0    2 ). Determine (a) the Lagrangian
and (b) the equation of motion. (c) Solve the equation of motion.

Answer:
(a) L  ma 2 (1  cos  ) 2  mga(1  cos  )
1 g
(b) (1  cos )   sin    sin   0
2 2a
 g g
(c) u  cos    C1 cos t  C2 sin t
 2 4a 4a

3.7 String pendulum. A mass m is suspended by a spring with spring

constant k in the gravitational field. Besides the longitudinal spring
vibration, the spring performs a plane pendulum motion. Find the
Lagrangian, derive the equations of motion, and discuss the meaning of
the resulting terms.
Answer:
1
(a) L 
2
  k
m r 2  r 2 2  mgr cos    r  r0 
2
2

(b) mr  mg sin   2mr

  
r  mr  mg cos   k  r  r0 
m

Problem Set 3 for Chapter 4 and 5

4.1 The variation of the integral. Consider the functional given by the definite integral
I x    f  y, y, y, x dx
b

which contains the first and second derivatives of y(x). To find the conditions for a

164
stationary value,  I  0 , construct the variation of the integral and apply integration by
parts twice. The imposed boundary conditions are y xa   ya and y xb   yb .What is the
resulting differential equation, and what additional boundary conditions must be
satisfied for y(x) to be an extremal solution?

4.2 The brachistochrone. In this problem you will work out some of the details of the
brachistochrone problem.
(a) Let's suppose that the bead on the wire is to go from the origin (x = y = 0) to the
point x = 1 m, y = 1 m (remember that our convention for this problem is that
positive y is down). Find the parameter a which appears in the solution, and plot the
resulting curve. Remember that “plot” means that you should either use a computer
plotting package of your choice or make a careful plot by hand on ruled graph
paper---a freehand sketch is not sufficient.
(b) For the parameters above, find the time of descent for the bead. Compare this result
with the time it would take if the curve were a straight line.

4.3 Projectile motion, the hard way. Jacobi's

Principle (which is a variation of Hamilton's Principle)
states that motion occurs in such a way that the
quantity
 x2 , y2 
A mv  x, y  ds
 x1 , y1 

is minimized. Here m is the mass of a particle, v is the

speed, and ds is an infinitesimal displacement. We
want to use Jacobi’s Principle to study projectile motion in a uniform gravitational field.
A projectile is launched from the position (0, 0) with an initial velocity v 0 , and strikes
the ground at x = R.
(a) Use energy conservation to express the speed v in terms of the initial speed v0 and the
height y of the projectile.
(b) Use the result of part (a) to show that A can be written as a functional of y:

 f  y, y; x dx
R
A  y  
0

Find the function f.

(c) Use Euler's equation to derive the differential equation which y must satisfy in order
to make A stationary. You needn't solve this equation.
(d) Finally, show either by direct substitution or by integrating the differential equation
that y(x) = ax2 +bx+c is a solution of this differential equation for suitably chosen
coefficients a, b, and c. You should find algebraic equations which determine these
coefficients, but you do not need to solve the equations.

165
4.4 Fermat's Principle. Fermat's Principle states that a ray
of light in a medium will follow the path which takes the
least amount of time.

(a) Suppose that a ray of light connects the points x1 , y1 
 
and x2 , y2 in a medium in which the speed of light v is
a function only of x; i.e., v = v(x). Construct the
functional t[y] which must be minimized to find the path
y = y(x).
(b) Now find the equation which y(x) must satisfy so that t[y] is stationary; i.e., what is
Euler's equation for this functional?
(c) As one application of your result in part (b), suppose that light is passing from one
medium with speed v1 (on the left) into a second medium with speed v2 (on the
right), as shown in the figure below. Derive Snell's Law, n1 sin 1  n2 sin  2 , where
n = c=v is the index of refraction of the medium and c is the speed of light in
vacuum.

(d) Suppose instead that v x   c / 1  x / l , where l is some characteristic length. If

dy=dx = 1 at x = y = 0, find y(x).

4.5 Electrostatics in variational form I. Many of the laws in physics may be cast into
a variational form, which is often important for further theoretical development and for
approximation methods. Let's consider electrostatics as an example. The electric field E
can be determined from Gauss's Law, which in differential form is
  E   / 0

where  is the charge density and  0 is the permittivity of the vacuum. For electrostatic
fields, we also have   E  0 ; this equation implies that an electrostatic field can be
written as the gradient of a scalar field, E   , with  the electrostatic potential.
Combining this with the equation above, we have
 2    /  0
which is Poisson's equation.
(a) Now consider the following functional:
  0           
2 2 2

                dV
 2  x   y   z   
where dV = dx dy dz and the integrals are to be taken over all space. Show that E is
minimized when  is a solution of Poisson's equation.
(b) Show that the functional can be written in the more compact form
0
           dV
2

2 
166
4.6 Electrostatics in variational form II. A simplification occurs if we consider
regions of space free of charge; in this case  2   0 , so that the electrostatic potential
satisfies Laplace's equation.
(a) Suppose we have two concentric spheres with radii a and b, a < b. The potential on
the inner sphere is V , while the potential on the outer sphere is held at 0. Show that
the potential in the region a < r < b is
 ab 1 a 

 r V  
 b  a r b  a 
In other words, show that this is a solution of Laplace's equation and that it satisfies
the boundary conditions.
(b) Now let's suppose that we didn't know how to solve Laplace's equation. Instead, let's
seek an approximate solution by trying trial solutions in the functional
0

2
      dV
2
Try a potential of the form
  r  a  r  a 
2


 r  V 1  A 
  b  a 

 1 A  
 b  a  


and find the optimal value of the parameter A. Pick some values of a and b and
compare your trial solution with the exact solution in part (a). [Hint: the gradient
and the integrals are to be evaluated in spherical coordinates; when doing the
integrals, you should only integrate over the region between the spheres.]
In working this problem you might want to read Ch. 19 of The Feynman Lectures on
Physics, Vol. II.

5.1 Block on a cylinder II. Let's return to the block on a cylinder problem. A block of
mass m starts at rest at the top of a cylinder of mass M and radius R which is placed on
the floor, as shown in the figure. At time t = 0 the block is given a gentle tap and it
begins sliding down the surface of the cylinder. Ignore the friction between the block
and the cylinder and the cylinder and the floor. Use the method of Lagrange multipliers
to find the normal force which the cylinder exerts upon the block.
Determine the angle  0 (as a function of the mass ratio m/M) at which the block loses
contact with the cylinder. Check your result in the limit M  m .

5.2 Pendulum attached to springs. We have a pendulum

(mass m at the end of the massless rod of length l). The top
support of the pendulum is attached to two springs of spring
constant k (which are fixed at the opposite ends), and slides
back and forth along the x axis. Gravity acts downward.

167
(a) Assuming the top support of the pendulum is massless, find the constraint equations.
(b) For small oscillations of the system, write down the Lagrangian using the Lagrange
method of undetermined multipliers, retaining terms up to and including second
order in the small quantities. (Assume all other physical parameters, such as k/m and
g/l are of order unity.)
(c) Obtain the corresponding equations of motion, and show that the system acts like a
simple pendulum with an effective length, leff , and find that effective length.

5.3 Particle moving along inverted cone. A particle of mass m

slides without friction on the surface of a stationary inverted
cone in a uniform gravitational field. The cone axis is aligned
with the gravitational acceleration, g, and the cone angle is  .
(a) Find the Lagrangian using generalized spherical coordinates.
(b) Identify the cyclic coordinate(s) and associated conserved
quantities.
(c) Suppose the particle has very large energy E and fixed angular momentum, l. What
are the approximate maximum and minimum possible values of the distance from
the vertex, r ?
(d) Find the angular speed for motion having constant r = r0. Express your answer in
terms of g, r0 and  .
(e) Using the Lagrange method of undetermined multipliers to find the normal force on
the particle, and verify your answer using Newton’s second law.

Problem Set 4 for Chapter 6

6.1 Cartesian or cylindrical coordinates (from Greiner p349). A mass point m shall
move in a cylindrically symmetric potential V(z). Determine the Hamiltonian and
the canonical equations of motion with respect to a coordinate system that rotates
with constant angular velocity  about the symmetry axis: (a) in Cartensian
coordinates, and (b) in cylindrical coordinates.
Answer:

(a) H 
1
2m

 px2  p y2  pz2     xp y  ypx   V x 2  y 2 , z 
1  2 1 2 
(b) H   p  ,2 p  pz2    p  V   , z 
2m   

6.2 Small Particle in Bowl (Stony Brook) A small particle of mass m

slides without friction on the inside of a hemispherical bowl, of radius R,
that has its axis parallel to the gravitational field g. Use the polar
angle  (see Figure) and the azimuthal angle  to describe the location of
the particle (which is to be treated as a point particle).

168
(a) Write the Lagrangian for the motion.
(b) Determine formulas for the generalized momenta p and p
(c) Write the Hamiltonian for the motion.
(d) Develop Hamilton's equations for the motion.
(d) Combine the equations so as to produce one second order differential equation for as
a function of time.
(e) If    and   0 independent of time, calculate the velocity(magnitude and
0

direction).
(f) If at t=0,    0 ,   0 and   0 , calculate the maximum speed at later times.

6.3 Rotating Hollow Hoop (Boston). A thin hollow cylindrical pipe is bent
to form a hollow circular ring of mass m and radius R. The ring is attached
by means of massless spokes to a vertical axis, around which it can rotate
without friction in a horizontal plane. Inside the ring, a point mass P of mass
m is free to move without friction, but is connected to a point H of the ring
by a massless spring which exerts a force k s where s is the length of the
arc HP (see Figure). Take as variables the angles  and  of CH and CP with the axis.

(a) Write the Lagrangian and the Hamiltonian, and rewrite them in terms of the
variables
     / 2,       / 2
(b)Find an integral of motion other than the energy, and show that its Poisson bracket
with H is zero.
(c) Integrate the equations of motion with these initial conditions at
 
   ,   ,     0
4 4

6.4 Practice with Poisson brackets. Consider the angular momentum for a point
particle, L  r  p .
 
(a) Compute the Poisson brackets Li , L j of components of L with one another.
 
(b) Compute the Poisson brackets L2 , Li for each component i.
(c) Assume H , Lx   H , Ly  0 for some Hamiltonian H. Without using the explicit
form of Lx and Ly, show that also H , Lz   0 .

Problem Set 5 for Chapter 7

7.1 A particle of mass m is constrained to move under gravity without friction on the
inside of a paraboloid of revolution whose axis is vertical. Find the one-dimensional
problem equivalent to its motion. What is the condition on the particle’s initial velocity

169
to produce circular motion? Find the period of small oscillations about this circular
motion.

7.2 A particle moves in a central force field given by the potential

e ar
V  k
r
where k and a are positive constants. Using the method of the equivalent one-
dimensional potential discuss the nature of the motion, stating the ranges of l and E
appropriate to each type of motion. When are circular orbits possible? Find the period of
small radial oscillations about the circular motion.

7.3 Two particles move about each other in circular orbits under the influence of grav-
itational forces, with a period τ. Their motion is suddenly stopped, and they are then
released and allowed to fall into each other. Prove that they collide after a time τ/4√2.

7.4 (a) Show that if a particle describes a circular orbit under the influence of an
attractive central force directed toward a point on the circle, then the force varies as
the inverse-fifth power of the distance.
(b) Show that for the orbit described the total energy of the particle is zero.
(c) Find the period of the motion.
(d) Find x , y and v as a function of angle around the circle and show that all
three quantities are infinite as the particle goes through the center of force.

7.5 (a) For circular and parabolic orbits in an attractive 1/r potential having the
same angular momentum, show that the perihelion distance of the parabola is
one-half the radius of the circle.
(b) Prove that in the same central force as in part (a) the speed of a particle at am point
in a parabolic orbit is 2 times the speed in a circular orbit passing through the same
point.

7.6 A uniform distribution of dust in the solar system adds to the gravitational attraction
of the sun on a planet an additional force
F = −mCr
where m is the mass of the planet, C is a constant proportional to the gravitational
constant and the density of the dust, and r is the radius vector from the sun to the planet
(both considered as points). This additional force is very small compared to the direct
sun-planet gravitational force.
(a) Calculate the period for a circular orbit of radius r0 of the planet in this combined
field.
(b) Calculate the period of radial oscillations for slight disturbances from this circular

170
orbit.
(c) Show that nearly circular orbits can be approximated by a precessing ellipse and find
the precession frequency. Is the precession the same or opposite direction to the orbital
angular velocity?

7.7 Dusty solar system. Suppose that the sun were surrounded by a dust cloud of
uniform density  which extended at least as far as the orbital radius of the Earth. The
effect of the dust cloud is to modify the gravitational force experienced by the Earth, so
that the potential energy of the Earth is (neglecting the effects of the planets)
GMm 1 2
U r     kr
r 2
where M is the mass of the sun, m is the mass of the Earth, G is the gravitational
constant, and k  4 mG / 3 (note that k > 0, so this additional term is attractive). The
effect of the dust cloud is to cause elliptical orbits about the sun to precess slowly; your
job is to find the precession frequency.
(a) From the potential, find the force F acting upon the Earth.
(b) Make a careful sketch of the effective potential V r  . On your sketch, indicate the (i)
energy E0 and radius r0 of a circular orbit, and (ii) the energy E1 and turning points r1
and r2 of an orbit which is not circular.
(c) Assume that the Earth is in a circular orbit of radius r0 about the sun. Derive the
equation satisfied by r0 in terms of the angular momentum l, and the constants m, M,
G, and k. You do not need to solve this equation.
(d) Find the frequency of small oscillations  about the the circular orbit of radius r0 .
You should find that your result can be written as
3k
 r  0 2 
m

where 0 is the angular velocity of revolution about the sun for a circular orbit.

(e) Finally, by assuming that k is small, find the precession frequency for a nearly
circular orbit.

7.8 A particle on the torus. A mass m is constrained to move

without friction on the surface of a torus shown below. The large
radius of the torus is R. The small radius is r. A point on the torus
may be labeled by the coordinates (  ,  ), as shown.  is the angle
about the z-axis, and  is the angle measured about the small circle,
relative to the radically outward vector in the x -y plane. Suppose
first, that there are no external forces (such as the force of gravity
on the particle).
(a) Find the coordinates x, y and z of the mass in terms of the generalized coordinates

171
  and  (and constants r,R). Write down the Lagrangian for the particle.
(b) Express the energy of the particle in terms of  ,  and the conserved component of
the angular momentum, M z .
(c) Plot the “effective potential” for motion in the  -variable, and describe the motion
for M z  0 .
(d) Now gravity is turned on, F=mgz. How is the effective potential of part (c) changed?
What is the relationship between the frequency and orbital radius for uniform
circular motion about the z-axis?

7.9 Show that the motion of a particle in the potential field

k h
V (r)   
r r2
is the same as that of the motion under the Kepler potential alone when expressed
in terms of a coordinate system rotating or precessing around the center of force. For
negative total energy, show that if the additional potential term is very small compared
to the Kepler potential, then the angular speed of precession of the elliptical orbit is

  2 mh

l 2
The perihelion of Mercury is observed to precess (after correction for known
planetary perturbations) at the rate of about 40" of arc per century. Show that this
precession could be accounted for classically if the dimensionless quantity
k

ka
(which is a measure of the perturbing inverse-square potential relative to the
gravita- gravitational potential) were as small as 7 × 10-8. (The eccentricity of
Mercury's orbit is 0.206, and its period is 0.24 year).

7.10 Laplace-Runge-Lenz vector . Show from the Poisson bracket condition for
conserved quantities that the Laplace-Runge-Lenz vector
mkr
A  pL 
r
is a constant of motion for a mass m moving under the Kepler potential. Also, show this
by taking the time derivative of this quantity.

7.11 Orbits on a cone. A particle of mass m is constrained to move on the surface of a

cone of half-angle  . Work in cylindrical coordinates  ,  , z .
(a) Show that the effective potential for this problem is

172
 l2
V     mg  cot 
2m  2
where l is the angular momentum.
(b) Using the effective potential, discuss the types of possible orbits for the particle,
assuming that it has a total energy E and angular momentum l. Make sure to include
a carefully labeled figure. What is the equation which determines the turning points?
What is the condition for a circular orbit?
(c) Investigate the stability of circular orbits on the surface of the cone.

7.12 Extensions of the Bohr model of the hydrogen atom. The Bohr model was
developed by assuming that the electron orbit is circular. However, we know that bound
orbits for attractive inverse square forces are generally ellipses. Sommerfeld (1916)
attempted to generalize Bohr's work to the case of elliptical orbits, by introducing the
quantization condition

 p dq  nh
where p is the generalized momentum conjugate to the generalized coordinate q,
defined as
L
p
q
the integral is over one period of motion, n is an integer, and h is Planck's constant.
For the hydrogen atom the Lagrangian is
1

L   r 2  r 2 2 
2
k
r

where k  e 2 / 4 0 , and the total energy (relative to the center of mass) is
1
E
2
 
 r 2  r 2 2 
k
r

with   me m p / me  m p  the reduced mass of the system.
(a) Start with the angular momentum,

 p d  mh
Since the angular momentum is conserved, show that this implies that p  m ,
where   h / 2 and m is a positive integer.
(b) Substitute your result from part (a) into the energy, and show that
pr 2 m 2  2 k
E  
2 2 r 2 r
(c) Sketch the effective potential for this problem, and draw the energy and turning
points for a bound orbit. Show that the turning points rmin and rmax can be determined
by solving a quadratic equation, and find expressions for the turning points as a

173
 function of the energy E.
(d) Now we need to quantize the radial motion. Solve the equation for E to obtain pr as a
function of r. Substitute the result into

 p r
dr  nh
with n an integer, and perform the integral. You might find the following integral to
be useful:
r  rmin rmax  r  r r 

rmax
dr    min max   rmin rmax
rmin r  2 
Substitute your expressions for rmin and rmax as a function of the energy E into this
expression, and solve the resulting equation for E to show that the allowed energy
levels are
k 2
En , m  
2 2 m  n 
2

Note that now we can have circular orbits (n = 0) or elliptical orbits ( n  0 ). Notice
that we can have many different orbits with the same energy; for instance, the
circular orbit with m = 3 and n = 0 has the same energy as the elliptical orbits with m
= 2 and n = 1 and m = 1 and n = 2. This is a case of degeneracy.
The Sommerfeld quantization condition is only an approximation; it can be obtained
from the full Schrodinger equation in the “semiclassical” (   0 ) limit. Note that it
requires that the orbits be periodic; what happens for nonperiodic, or chaotic, motion?
That is the subject of an area of current research called “quantum chaos” (the quantum
mechanics of classically chaotic systems).

7.13 Diatomic molecules. The two atoms in a diatomic molecule (of masses m1 and m2)
interact through a central force which is derived from the potential energy
a2 b2
U r   
4r 4 3r 3
where r is the separation between the atoms.
(a) Find the equilibrium separation of the atoms and the frequency of small oscillations
about the equilibrium point assuming that the molecule does not rotate. How much
energy must be supplied to the molecule in order to break it up?
(b) Determine the maximum angular momentum which the molecule can have without
breaking up, assuming that the motion is in circular orbits. [Hint: break up occurs
when effective potential no longer has a minimum.] Find the particle separation at
the break up angular momentum.
(c) Calculate the velocity of each particle in the LAB system at break up, assuming that
the center of mass is at rest.

7.14 Scattering. A fixed force center scatters a particle of mass m according to the force
law F r   k / r 3 . If the initial speed of the particle is u0 , show that the differential

174
scattering cross section is
k 2    
   
mu0 2 2 2    sin 
2

Problem Set 6 for Chapter 8

8.1 Bead on a hoop yet again. Consider again the problem of a bead on a frictionless
hoop which is rotating at a constant angular velocity about its vertical diameter (sound
familiar?). This time we want to derive the equation of motion using old fashioned
Newtonian mechanics, in the frame of reference rotating with the hoop.
(a) Draw a careful force diagram(s) showing the centrifugal force, Coriolis force, and
forces of constraint acting upon the bead.
(b) By balancing these forces find the condition for steady motion of the bead (such that
    0 ).
(c) Next, apply Newton's Second Law in the rotating frame to derive the equation of
motion for the bead. [Hint: resolve the forces into components which are normal and
tangent to the hoop.]
(d) Finally, find the forces of constraint acting upon the bead. Do this using Newtonian
mechanics, not Lagrange's equations with undetermined multipliers. Your results
will involve  and its derivatives.

8.2 Clever sailor. At the Earth's equator there is very little wind (this is known as the
“equatorial doldrums”). Suppose that a small sailboat is becalmed at the equator. The
captain, a former physics major, decides to put the boat into motion by raising the
anchor (m = 200 kg) to the top of the mast (h = 20 m). The rest of the boat has a mass of
M = 1000 kg.
(a) Why will the boat begin to move?
(b) In which direction will the boat move?
(c) What is the boat's final velocity (with respect to the Earth)?

8.3 Billiard shots. A homogeneous billiard ball of mass

m and radius R moves on a horizontal table. The
coefficient of kinetic friction between the table and the
ball is  . The ball is struck by a cue, which delivers an
impulsive force F of duration t ; the point of impact is
a height h above the table, and the cue makes an
angle  with respect to the horizontal.
(a) Find the moment of inertia of the billiard ball about an axis passing through its
center of mass.
(b) Start by assuming that the cue is horizontal (   0 ) and that the cue ball is struck at
175
 the “equator” (h = R). Find the time it takes for the ball to roll without slipping, and
the center of mass speed when this rolling occurs.
(c) Next, assume that   0 . What is the value of h required so that the cue ball rolls
without slipping immediately after being struck?

8.4 Rolling spheres. A hollow spherical shell has a mass m and radius R.
(a) Calculate the inertia tensor for a set of coordinates whose origin is at the center of
mass of the shell.
(b) Now suppose that the shell is rolling without slipping toward a step of height h as
shown in the figure. The shell has a linear velocity v. What is the angular
momentum of the shell relative to the point C?

(c) The shell now strikes the point C inelastically (so that the point of contact sticks to
the step, but the shell can still rotate about point C). What is the angular momentum
of the shell immediately after contact? [Hint: think about the torque produced by the
normal force exerted by the step on the shell.]
(d) Finally, find the minimum velocity which enables the shell to surmount the step.
Express your result in terms of m, g, R, and h. [Hint: think about energy
conservation.]

8.5 Coriolis deflection. A particle is thrown up vertically with initial speed v0, reaches
a maximum height and falls back to the ground. Show that the Coriolis deflection when
it again reaches the ground is opposite in direction and four times greater in magnitude
than the Coriolis deflection when it is dropped at rest from the same height. You may
ignore contributions of second order and higher in the angular speed of the Earth about
its axis.

8.6 Circular disk on an inclined plane. You have a uniform circular disk of mass m
and radius R which rolls without slipping on an inclined plane of angle  . One end of a
spring of spring constant k is attached to the center of the wheel, while the other end is
held fixed at the top of the incline. At t = 0, the wheel starts from rest at the equilibrium
point of the spring, x = 0, where x is measured along the length of the ramp. Gravity
acts downward.
(a) Construct the Lagrangian, and find the equation of motion for the disk.
(b) Solve the equation of motion. Show that the motion along the x-axis is harmonic,
and find the frequency of oscillations.

176
Problem Set 7 for Chapter 9

9.1 Tension of inertia for various bodies

(a) A thin spherical shell of radius R and mass M which is homogeneously distributed
over the shell. The origin of the coordinate system is chosen to be at center of the shell.
(b) A sphere of radius R and mass M. The origin of the coordinate system is chosen to
be at center of the sphere.
(c) A thin rod of radius L and mass M. The origin of the coordinate system is chosen to
be in the middle of the rod. The rod is assumed to be positioned along x-axis.
(d) A thin rod of radius L and mass M. The origin of the coordinate system is chosen to
coincide with one end of the rod.
(e) A thin ring of radius R and mass M. The origin of the coordinate system is located in
the center of the ring and z-axis is the symmetry axis.
(f) A thin hollow cylinder of radius R, length L and mass M. The coordinate system is
chosen to have its origin in the center of the hollow cylinder with z-axis being the
symmetry axis.
(g) A cylinder of radius R, length L and mass M. The coordinate system is chosen to
have its origin in the center of the cylinder with z-axis being the symmetry axis.
(h) A thin disc of radius R, length L and mass M. The coordinate system is chosen to
have its origin in the center of the disc with z-axis to go through the center and to be
orthogonal to the extension of the disc.
Answer:
 1 0 0   1 0 0   0 0 0 
2 2  2 2  1
(a) MR 0 1 0 , (b) MR 0 1 0 ; (c) 2
ML 0 1 0 
3   5   12  
 0 0 1   0 0 1   0 0 1 

 0 0 0   1 0 0   1 0 0 
2 2
1 2  2
  2
 
(d) ML 0 1 0 ; (e) MR  0 2 0  ; ( f ) MR  0 12 0 
1

3  
 0 0 1     
 0 0 1   0 0 1 
 R 2  1 L2 0 0   1 0 0 
3
1   2
1  
(g) M  0 R 2  13 L2 0  ; (h) MR  0 12 0 
2

4   2  0 0 1 
 0 0 2R 2   

9.2 Moments of inertia. A two-dimensional square of

uniform density has a smaller square removed, as shown.
The total mass of the resulting object is M.
The e x and e y axes are shown, and e z points out of the page.
(a) Find the moments of inertia of this object for rotation

177
 about the e x , e y and e z axes.
(b) The system undergoes torque-free rotation about each axis. Describe whether the
rotation is stable or unstable with respect to rotation about each of the axes.

9.3 Rolling circular top (from Greiner p207).

(a) Find the tensor of inertia of a homogeneous circular
top(density , mass m, height h, vertex angle 2) with its
vertex fixed (Fig.9.xa);
(b) Find its kinetic energy rolling on a plane (Fig.9.xb) and the
vertex is fixed at a point.
(c) Find its kinetic energy if its base circle rolls on a
plane while its longitudinal axis is parallel to the plane
and the vertex is fixed at a point (Fig.9.xc).

Answer:
3 3
(a) I xx  I yy 
20
 
mh 2 tan 2   4 , I zz  mh 2 tan 2 
10
I xy  I xz  I yz  0

3
(b) T 
40

mh 2 2 1  5 cos 2  
3  R2 
(c) T  mh 2 2  6  2 
40  h 

9.4 A system of three masses(from Greiner p212).

A rigid body consists of three mass points that are
connected to the z-axis by rigid massless bars (see
Fig. 9x):
(a) Find the elements of the tensor of inertia relative
to the xyz-system.
(b) Calculate the ellipsoid of inertia with respect to
the origin 0, and the moment of inertia of the entire
body with respect to the axis 0a.
Answer:
(a) I xx  125.7(kgcm 2 ), I yy  117.5(kgcm 2 ), I zz  104.75(kgcm 2 ),
I xy  19.1(kgcm 2 ), I xz  44.8(kgcm 2 ), I yz  4.8(kgcm 2 )
(b) cos   0.268, cos   0.358, cos   0.895

9.5 Euler angles. You make a series of rotations through Euler angles with
  45 ,  30 ,  30 .

178
(a) Draw a picture of the location of the rotated axes e1 , e2 , e3  following these
rotations.
(b) Find the rotation matrix A.
(c) Demonstrate that A is orthogonal.

9.6 A spherical bowling ball inside a wedge. A spherical bowling

ball of mass m and radius r rolls without slipping inside a wedge
which has a circular bowl of radius R. Gravity acts downward, and
the bowling ball rolls without slipping on the surface of the circular
bowl. The ball is released at an initial angle from rest.
(a) Derive the equations of motion for the ball. (Hint: If the angle of
rotation of the ball through any axis through its center is  , what is the “rolling
without slipping constraint”?)
(b) For small oscillations of the ball (   1), show that the ball undergoes simple
harmonic motion, and find the frequency  of the harmonic motion.
(c) The wedge which has mass M is now allowed to move on the frictionless horizontal
surface on which it rests. Derive the new equations of motion for the wedge-ball
system.
(d) Find the frequency of harmonic motion for small oscillations of the wedge-ball
system of part (c).
(e) For small oscillations, this system is harmonic and hence could be used to tell time
like a clock. Does this “clock” speed up or slow down when M is allowed to move?

9.7 Symmetric top. Investigate the motion of the symmetric top for the case in which
the axis of rotation is vertical (i.e., the x3 - and x3 -axes coincide). Show that the motion
is either stable or unstable depending on whether the quantity 4 I1Mhg / I 33 is less than
or greater than unity. Sketch the effective potential V   for the two cases, and point
out the features of these curves that determine whether the motion is stable. If the top is
set spinning in the stable configuration, what is the effect as friction gradually reduces
the value of 3 ? (This is the case of the “sleeping top”.)

9.8 Noncentral force. As a result of the Earth's aspherical mass distribution, its
gravitational potential has the form
1 3z 2  r 2 
U r       
r r5 

where the Earth's axis has been taken as the z-axis,

and  and  are constants. The corresponding
gravitational force contains a small noncentral
component.

179
(a) Determine the equations of motion in rectangular coordinates of a particle in this
field .
(b) Consider a particle moving initially in an approximately circular orbit, with radius a
and angular velocity    . Let the normal to the plane of the orbit lie in the x - z
plane, at angle  with respect to the z-axis, as indicated in the figure below.
Determine the rate of change of the angular momentum vector L of the particle.
Then find the time average of these rates over one period of the circular orbit.
(c) Assuming dL/ dt is sufficiently small, evaluate the rate of precession of L.

9.9 A dumbbell-shaped satellite. A nonspherical orbiting satellite can stabilize its

attitude automatically: given enough time, it will assume a fixed orientation wit h
respect to the radius vector from the Earth. This involves merely the Earth's
gravitational pull, plus a small amount of damping. In the following, assume the
gravitational field of the Earth is that of a sphere.

(a) Consider a dumbbell-shaped satellite, consisting of point

masses m connected by a massless rod of length l. It is
placed into a circular orbit so that it s center of mass (CM)
remains at a fixed distance R from the center of the Earth.
The two mass points are assumed to lie in the orbital
plane of the CM, and the satellite is free to rotate about
the CM. Using the notation for the coordinates of the two
masses and the CM shown in the figure, express
r1 , r2 ,1 , 2 in terms of  , R, l ,  . Make use of the fact that l / R   1.
2

(b) Express the total mechanical energy of the satellite in terms of m, R, l ,  ,  , and
  0 . Neglect any terms of order l / R  and higher.
3

(c) Show that a stable equilibrium is possible and find the orientation angle 0 for this
equilibrium.
(d) Explain how the above results can be used to understand why we on Earth always
see the same face (that is Elvis' ) on the Moon.

9.10 A thumbtack. A rigid body in the shape of a thumbtack formed from a thin disk of
mass M and radius a and massless stem is placed on an inclined plane that makes an
angle  with the horizontal. The point of the tack remains stationary at the point P, and
the head rolls along a circle of radius b. Introduce a set of laboratory coordinates
whose e30 axis is perpendicular to the inclined plane and whose e 02 axis points down the
plane. Also, introduce body coordinates with origin at the center of mass, whose e3 axis
is perpendicular to the head of the tack pointing outward, whose e 2 axis passes through

180
the point of contact with the plane, and whose e1 axis is parallel to the surface and
tangent to the circle. The set of angles  ,  ,   shown in the figure specify the
orientation of the tack.

(a) Find the angular velocity of the tack in terms of body-frame unit vectors.
(b) Find the kinetic and potential energy of the tack.
 
(c) Construct the Lagrangian, L  L  , , , , and write Lagrange's equations for
 and  , incorporating the constraint    0  sin 1 a / b  with a Lagrange
multiplier,  .
(d) Show that  t is described by the simple harmonic oscillator equation, and find the
frequency.
(e) Interpret  .
.

Problem Set 8 for Chapter 10

10.1 A bead on hoop. One point of a uniform circular hoop of

mass m and radius a is fixed by a pivot, and the hoop is free to
move in its vertical plane about this point. A bead of mass m
slides without friction on the hoop.
(a) Write down the Lagrangian for the system in terms of the
angles  and  .
(b) Simplify the Lagrangian for the case of small oscillations
about the stable equilibrium position, and find the equations of motion.
(c) Find the frequencies of the two normal modes, and determine the ratios of the
amplitudes of oscillation for each mode.

10.2 A double pendulum is constructed from two uniform

rods, each of length L and mass M, as shown. The rods are
attached to each other with a frictionless pin at A, and hang
from a fixed, frictionless pivot at B. Find the normal mode
frequencies of oscillations, for small angular deviations about

181
the equilibrium position of this system. The moment of inertia of a rod of mass M and
length L about an axis through its center-of-mass and perpendicular to the rod
is ML2 /12 .

10.3 Normal mode. Consider three equal masses m that slide

without friction on a circular wire of radius R. The masses are
connected by identical springs of spring constant k. The
angular positions of the three masses, 1 ,  2 , and  3 are

measured from their respective equilibrium positions.

(a) Write down the Lagrangian for this system and derive the
equations of motion.
(b) Find the normal modes and frequencies, and describe qualitatively the motion to
which each normal mode corresponds.

10.4 Coupled oscillators. An infinite string of mass points, each of mass m, are
connected by springs. The spring constant k1 and k 2 alternate as shown. Guess for the
displacement of the jth mass point
 j t   Ae eit  jka 
for j even, and
 j t   Ao eit  jka 

for j odd. Here a is the equilibrium spacing between the masses, and Ae , Ao are

constants. Find the frequency  of small amplitude oscillations as function of the

wavevector, k.

10.5 Electric ossiliators. Consider the circuit consisting of

three equal capacitors and two different inductors shown in the
figure. For charges Qi on the capacitors and currents I i

through the components, write down Kirchhoff’s laws for the

total voltage change around each of two complete circuit loops.
Note that to within an (unimportant) constant, conservation of current
implies Q3  Q1  Q2 . Find the normal mode frequencies. For L1  L2  L find the

normal mode eigenvectors and describe the patterns of current flow in the circuit.

182
Problem Set 9 for Chapter 11
11.1 One of the attempts at combining the two sets of Hamilton's equations into one
tries to take q and p as forming a complex quantity. Show directly from Hamilton's
equations of motion that for a system of one degree of freedom the transformation Q
= q + ip, P = Q* is not canonical if the Hamiltonian is left unaltered. Can you find
another set of coordinates Q', P' that are related to Q, P by a change of scale only, and
that are canonical?

11.2 Show directly that the transformation

1 
Q  log  sin p  , P  q cot p
p 
is canonical.

11.3 Show directly that for a system of one degree of freedom the transformation
q  q2  p2 
Q  arctan , P 1
p 2   2 q 2 
is canonical, where α is an arbitrary constant of suitable dimensions.

11.4 (a) For a one-dimensional system with the Hamiltonian

p2 1
H  2
2 2q
show that there is a constant of the motion
pq
D  Ht
2
(b) As a generalization of part (a), for motion in a plane with the Hamiltonian

H | p |n ar  r
where p is the vector of the momenta conjugate to the Cartesian coordinates,
show that there is a constant of the motion
pr
D  Ht
n
(c) The transformation Q = λq, p = λP is obviously canonical. However, the
same transformation with t time dilatation, Q = λq, p = λP, t' = λ2t, is not. Show that,
however, the equations of motion for q and p; for the Hamiltonian in part (a) are
invariant under this transformation. The constant of the motion D is said to
be associated with this invariance.

11.5 Canonical transformation. Construct the Hamiltonian for the one-dimensional

183
harmonic oscillator of mass m and spring constant k. Determine the value of the
constant C such that the following equations define a canonical transformation from the
old variables (q, p) to the new variables (Q, P):
Q  C  p  im q 
P  C  p  im q 
where   k / m . What is the generating function, F(q, P), for this transformation?
Find Hamilton’s equations of motion for the new variables and integrate them. Hence
find the solution to the original problem.

11.6 Canonical transformation. Find the values of  and  for which the
transformation
Q  q cos  p 
P  q sin  p 
is canonical. For these values of  and  , construct a generating function F3  p, Q 
for the canonical transformation.

11.7 Solve the problem of the motion of a point projectile in a vertical plane, using
the Hamilton-Jacobi method. Find both the equation of the trajectory and the
dependence of the coordinates on time, assuming the projectile is fired off at time t = 0
from the origin with the velocity v0 , making an angle  with the horizontal.

11.8 A charged particle is constrained to move in a plane under the influence of a

central force potential (nonelectromagnetic) V= kr2/2, and a constant magnetic field B
perpendicular to the plane, so that
1
A  Br
2
Set up the Hamilton-Jacobi equation for Hamilton's characteristic function in
plane polar coordinates. Separate the equation and reduce it to quadratures. Discuss
the motion if the canonical momentum p is zero at time t= 0.

11.9 (a) A single particle moves in space under a conservative potential. Set up
the Hamilton-Jacobi equation in ellipsoidal coordinates u, v, ϕ defined in terms of the
usual cylindrical coordinates r, z, ϕ by the equations
r  asinh v sin u, z  a cosh v cosu

For what forms of V(u, v, ϕ) is the equation separable?

(b)Use the results of part (a) to reduce to quadratures the problem of a point particle of
mass m moving in the gravitational field of two unequal mass points fixed on the z
axis a distance 2a apart.

184
11.10 Suppose the potential in a problem of one degree of freedom is linearly
dependent upon time, such that the Hamiltonian has the form
p2
H  mAtx
2m
where A is a constant. Solve the dynamical problem by means of Hamilton's
principal function, under the initial conditions t = 0, x = 0, p = mv0. Set the plane
Kepler problem in of the generalized coordinates.

11.11 A particle moves in periodic motion in one dimension under the influence of a
poten- potential V(x) = F|x|, where F is a constant. Using action-angle variables, find
the period of the motion as a function of the particle's energy.

11.12 A particle of mass m moves in one dimension under a potential V= -k/|x|.

For energies that are negative, the motion is bounded and oscillatory. By the method
of action-angle variables, find an expression for the period of motion as a function of
the particle's energy.

11.13 A particle of mass m is constrained to move on a curve in the vertical plane

defined by the parametric equations
y  l(1  cos 2 )
x  l(2  sin 2 )

There is the usual constant gravitational force acting in the vertical y direction. By the
method of action-angle variables, find the frequency of oscillation for all
initial conditions such that the maximum of ϕ is less than or equal to /4.

11.14 A three-dimensional harmonic oscillator has the force constant k1 in the x- and
y- directions and k3 in the z-direction. Using cylindrical coordinates (with the axis
of the cylinder in the z direction), describe the motion in terms of the
corresponding action-angle variables, showing how the frequencies can be obtained.
Transform to the "proper" action-angle variables to eliminate degenerate frequencies.

185