CM-Kaist Lecture Note (2013)

Classical Mechanics
Hyoungsoon Choi
Spring, 2013
Contents
1 Introduction 4
1.1 Kinematics and Kinetics . . . . . . . . . . . . . . . . . . . . . . . 5
1.2 Kinematics: Watching Wallace and Gromit . . . . . . . . . . . . 6
1.3 Inertia and Inertial Frame . . . . . . . . . . . . . . . . . . . . . . 8
2 Newtons Laws of Motion 9
2.1 The First Law: The Law of Inertia . . . . . . . . . . . . . . . . . 9
2.2 The Second Law: The Equation of Motion . . . . . . . . . . . . . 10
2.3 The Third Law: The Law of Action and Reaction . . . . . . . . . 11
3 Laws of Conservation 13
3.1 Conservation of Momentum . . . . . . . . . . . . . . . . . . . . . 13
3.2 Conservation of Angular Momentum . . . . . . . . . . . . . . . . 14
3.3 Conservation of Energy . . . . . . . . . . . . . . . . . . . . . . . 15
3.3.1 Kinetic energy . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3.2 Potential energy . . . . . . . . . . . . . . . . . . . . . . . 16
3.3.3 Mechanical energy conservation . . . . . . . . . . . . . . . 17
4 Solving Equation of Motions 18
4.1 Force-Free Motion . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Constant Force Motion . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2.1 Constant Force Motion in One Dimension . . . . . . . . . 20
4.2.2 Constant Force Motion in Two Dimensions . . . . . . . . 21
4.3 Varying Force Motion . . . . . . . . . . . . . . . . . . . . . . . . 23
4.3.1 Drag Force . . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.3.2 Harmonic Oscillator . . . . . . . . . . . . . . . . . . . . . 27
5 Lagrangian Mechanics 28
5.1 The Euler-Lagrange Equation . . . . . . . . . . . . . . . . . . . . 29
5.2 Lagrangian Mechanics . . . . . . . . . . . . . . . . . . . . . . . . 30
5.2.1 Generalized Coordinates . . . . . . . . . . . . . . . . . . . 31
5.3 DAlemberts Principle . . . . . . . . . . . . . . . . . . . . . . . . 34
5.4 Conjugate Variables . . . . . . . . . . . . . . . . . . . . . . . . . 35
6 Hamiltonian Mechanics 36
6.1 Conguration Space and Phase Space . . . . . . . . . . . . . . . 36
6.2 Hamiltons Equations . . . . . . . . . . . . . . . . . . . . . . . . 37
1
CONTENTS 2
7 Central Force Motion 39
7.1 Conservation Laws in Central Force Field . . . . . . . . . . . . . 39
7.2 The Path Equation . . . . . . . . . . . . . . . . . . . . . . . . . . 41
8 System of Multiparticles 43
8.1 Weighted Average . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.2 Center of Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
8.3 Linear Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . 45
8.4 Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . . 46
8.5 Energy of the System . . . . . . . . . . . . . . . . . . . . . . . . 47
9 Collision Theory 49
9.1 Elastic Collisions of Two Particles . . . . . . . . . . . . . . . . . 49
9.2 Kinematics of Elastic Collisions . . . . . . . . . . . . . . . . . . . 51
9.3 Inelastic Collisions . . . . . . . . . . . . . . . . . . . . . . . . . . 51
9.4 Scattering Cross Section . . . . . . . . . . . . . . . . . . . . . . . 52
9.5 Rutherford Scattering Formula . . . . . . . . . . . . . . . . . . . 56
10 Rigid Body Kinematics 57
10.1 Rotation and Linear Velocity . . . . . . . . . . . . . . . . . . . . 57
10.2 Inertia Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
10.3 Angular momentum . . . . . . . . . . . . . . . . . . . . . . . . . 59
10.4 Principal Axes of Inertia . . . . . . . . . . . . . . . . . . . . . . . 60
10.5 Moments of Inertia for Dierent Body Coordinate Systems . . . 62
10.6 Eulers Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
10.7 Free Body Rotation of a Rigid Body with Axial Symmetry . . . 63
10.8 Precession of a Symmetric Top due to a Weak Torque . . . . . . 64
10.9 Steady Precession of a Symmetric Top under a Uniform Torque . 65
10.10Eulerian Angles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
10.11Motion of a Symmetric Top with One Point Fixed . . . . . . . . 68
CONTENTS 3
Disclaimer: The world view represented in this section only holds
for Newtonian physics and Galileian relativity. However, it is still
very useful in developing physical intuitions relevant to classical me-
chanics. Basically, I will be treating you like a person born pre 1900,
before quantum mechanics and special relativity.
Chapter 1
Introduction
Unlike quantum mechanics, in which our intuitive world view breaks down com-
pletely from the get go, there is no big secret in classical mechanics. The objects
you are interested in are mostly visible and they respond to a push or pull, tech-
nically known as the force. The question here is how do the objects move
under given set of forces? At the end of the day, your answer will be given as
the position of an object as a function of time.
To achieve this goal, everything you learn in classical mechanics boils down
to understanding the consequences of a single equation, F = ma. We will not
pretend that this is a completely new concept you are yet to learn. Instead,
we will deal with it exactly as what it is, something you have learned already,
of which you do not understand the full consequences yet. In other words, we
will learn how to interpret the equation and its consequences more carefully,
thoroughout this course.
But before we seriously delve into mechanics, lets take a brief trip to France.
In S`evres, France, there is an underground vault that holds big chunks of metal
that are made out of 90% platinum and10% iridium. There are three keys to
this vault and you need all three to get in. Each year three people holding
the keys to this vault gather together, go down to the vault open it up and
make sure these metal chunks are still there unaltered. This seemingly bizarre
behavior that can be mistaken as a ritual by platinum-iridium worshipping cult
is actually of utmost importance to us that study science, especially physics.
The vault actually belongs to the International Bureau for Weights and
Measures (Bureau International des Poids et Mesures, BIPM), and those metal
pieces are the international prototype kilogram (IPK) and the international
prototype meter (IPM). The sole purpose of BIPMs existence is to dene a
kilometer, a meter (and a second) and IPK and IPM are exactly that. There
isnt any international prototype second (IPS) in that vault only because there
is simply no physical object that can represent time. Now, what does this have
anything to do with what we are about to study? As we will soon see, more
than what you can imagine.
4
CHAPTER 1. INTRODUCTION 5
Figure 1.1: Galileos thought experiment on inertia.
1.1 Kinematics and Kinetics
It should be noted that what we learn is usually dubbed as mechanics which
encompasses kinetics (or dynamics) and kinematics. Kinematics is an older eld
of study, in the sense, that a motion of a particle, a body, or a collection of them
are studied without regard to why it happens, whereas kinetics deals with the
motion in relation to its causes, i.e. forces and torques.
For example, a particle shot up at an angle under a constant vertical down-
ward acceleration follows a parabolic trajectory. In kinematics, this is the only
relevant information. Position, time and their relations are all there is to it.
Although it is clear from Galileos thought experiment of a body rolling down
a slope that he understood the concept of inertia, his contribution to physics
leaned more on kinematics than mechanics.
Kinetics is distinguished from kinematics by recognizing that the constant
acceleration is a result of constant force acting on the object, and also the
magnitude of acceleration depends on the magnitude of the force acting on it.
This relation was established by Newton through his Laws of Motion, and only
then did mechanics truly form by integrating kinematics and kinetics. Mechanics
is after all understanding the implications of Newtons Second Law of Motion,
F = ma. This adds extra dimension to mechanics that lacked in kinematics,
and that last missing dimension is mass.
We now have three physical quantities, mass, distance and time to fully
grasp the concept of classical mechanics, the very starting point of almost every
scientic endeavour, and it is not by chance that the standard units, also known
as SI units, have three elements: meter, kilogram and second.
Bear in mind that vision is, by far, the most far reaching sensory system
among the humans ve primary senses, and kinematics is a visual interpretation
of motion and kinetics a more abstract interpretation. In other words, the
position of an object is visible where as force or mass is not. Noting where the
Sun, the Moon and stars everyday were a relatively simple task. Since time
is, as we shall soon see, a measure of change, any change in position that we
observe already includes the notion of time. In other words, observation of
positional change is the kinematics itself. When we are observing objects, say
stars in the sky, move, we are probing into its kinematics already. Why they
were there when they were there was a lot more tricky business. And exactly
for this reason, we start with kinematics.
1.2 Kinematics: Watching Wallace and Gromit
Figure 1.2: Wallace and Gromit.
Wallace and Gromit is one of the most beloved stop motion claymations,
created by Nick Park in 1989. In a claymation, a malleable material, such
as plasticine clay, is formed into desired shapes, forming characters and back-
grounds and a still shot is taken. Then these characters and backgrounds are
deformed to represent changes and movements, and another still shot is taken.
By repeating this process thousands of times, kinematics of clay objects are,
well, fabricated. Nonetheless, it does allow us to peek into the key concepts in
kinematics.
Suppose, the plasticine clay is formed into shapes and you take a still shot.
After a few hours, you take another still shot, another after a few hours, and
another and another and so on. The clay wont move itself and when you play
a movie out of these still shots, you will end up with a very boring movie. No
one would notice, if you pause the video during the showing of that movie, that
is of course, if there is anyone coming to see this. Just by staring at this, no
one would be able to tell what happens before what. In that little claymation
universe, the time doesnt exist.
Time takes any meaning only if there are changes; something has to move.
Lets take a look at two images in Fig. 1.2. We can tell by looking at them that
they are not the same. Then how far did Wallaces cup move? By about the
width of the cup. But how far is it? Is it big or small? If you try to answer this
question, what you learn is that there is no absolute scale of length in physics.
When you say the cup moved by the width of the cup, you are setting the size
of the cup as the reference scale. Once you decide to use the size of the cup as
your unit of length measurements, you can now say something has moved by
three cups or eleven cups, etc.
However, spatial change alone is not sucient to describe motion. The fact
that the position of some objects changed self-creates this other dimension,
time. Unless you are willing to accept the notion that an object can be at
two dierent places at the same time, which you cant in classical physics, the
fact that there is a change in position allows you to conclude only that these
two images represent two dierent points in time. But, because you can never
gure out how far apart in time these images are by simply looking at them,
the concept of time exists, but not in truly meaningful way. Just like length,
there is no absolute measure of time. To quantify, how much time has elapsed,
you need to be able to compare changes.
Imagine an independently moving object, say a red ball. We cannot tell in
any absolute sense how long it took for the cup to move, we can say whether
it took longer than the red ball to move from point A to point B. However,
because the point of reference vanishes once the red ball stops moving, it is
better to pick a repetitive motion as a reference of time, such as the Earths
rotation or orbital motion around the Sun. We can then use one cycle of this
motion as a unit of time and describe other objects motion based on it. For
example, we can say that Wallaces cup has moved by one and a half cup in
1/864000 of the Earths rotation whereas Gromits cup has moved by 0.8 cup
in the same period of time, thereby Wallaces cup moved at a higher velocity.
This is precisely why we need IPK and IPS. Only after having a reference
point for distance and time, such as the width of the cup and the duration of
the Earths one full rotation, can we dene what motions are and kinematics
starts to make sense. IPM and IPS (if there was one) are the internationally
agreed upon basis for this relativeness, and that is why it is so important to
have a precise, easily reproducible denition of them and keep them unaltered.
Currently, one meter is dened as the length of the path travelled by light in
vacuum during a time interval of 1/299,792,458 of a second. One second is
dened as the duration of 9,192,631,770 periods of the radiation corresponding
to the transition between the two hyperne levels of the ground state of the
caesium 133 atom. We will learn how IPK is related to mechanics, shortly,
but at this point, it is not too big a stretch to say that our understanding of
mechanics is kept in that vault in France.
Now that we have established the concept of time and space, motion has
a meaning. We can dene velocity and acceleration, as the amount change in
position in a given time and the amount of change of that change in a given time.
The agreement on that time frame is what allows you to create claymations.
In real life, it can take a really long time to create two successive shots. But
because you set the time dierence between any two successive shots equal when
you are playing it in video, you can create a controlled motion, and suddenly
the movie gets a life and becomes a form of entertainment, and also a subject
of kinematics.
1.3 Inertia and Inertial Frame
Unlike length and time, mass is not a visually identiable quantity. So the
question follows: how do we identify mass? But this question is in fact, missing
a more important point entirely: why do we even care about mass? Kinemat-
ics, such as the famous Keplers Laws of planetary motion, can be suciently
described by space and time. Mass and motion at rst glance are unrelated.
However, if we want to describe the origin or root cause of such a motion, that
is when we get into trouble. To answer this question adequately, we have to
rst understand what inertia is.
The concept of force goes back to ancient Greece, when Archemedes al-
ready seemed to grasp the vector nature of force with Eucledian geometry and
trigonometry. Such methods were very eective at describing statics. The
problem was that their understanding of kinetics (dynamics) was deeply awed.
Aristotle argued that for something to move, force has to be applied and the
speed of an objects motion is proportional to the applied force and inversely
proportional to the viscosity of the surrounding medium. It is obvious that
they knew motion and force were related, but they simply had no clue what
that somehow was.
The problem with this reasoning is that, well there are just so many that
I dont even know where to start. It is Galileos brilliant insight on inertia
(Fig. 1.1) that allowed us to view the relationship between motion and force on
a completely dierent light. He argued that every object has a natural tendency
to maintain its motion, and a motion of the object will be unaltered unless there
is net force acting on it. This tendency to maintain its motion is called inertia,
and this is why we care about mass. But we will get to this point in the next
chapter.
Another important contribution of Galileo is that he singled out constant
velocity motion from all other motions, yet recognized that all constant velocity
motion can be grouped as one indistinguishable set regardless of what that
velocity is. From this, the concept of inertial frame was born.
Imagine a reference frame S that we can declare as absolutely not moving.
Within this frame, an object, say a hockey puck, is moving at a constant velocity
v
p
. An observer within the reference frame would see the motion of the hockey
puck as a force-free motion. Now imagine a moving frame S
at a velocity v
S
relative to the xed frame S. Then to an observer within the S
frame, the
hockey puck slides at a velocity v
p
v
S
, which is also a constant velocity. Thus
the observer in the S
frame would see the motion of the puck also as force-free.

Despite the dierence in velocity, both observers would see the eect of inertia,
that is the hockey puck continues its original motion. For this reason, frames
moving with a constant velocity is referred to as inertial frame. We will discuss
the importance of this in more detail in the next chapter.
Chapter 2
Newtons Laws of Motion
With his concept of inertia, Galileo emphasized the importance of constant ve-
locity motion. What Newton, in turn, did was to build upon that and emphasize
the role of force, the physical quantity that is required to resist inertia. He then
came up with three Laws of Motion. In its original form, they read:
Lex I: Corpus omne perseverare in statu suo quiescendi vel movendi unifor-
miter in directum, nisi quatenus a viribus impressis cogitur statum illum
mutare.
Lex II: Mutationem motus proportionalem esse vi motrici impressae, et eri
secundum lineam rectam qua vis illa imprimitur.
Lex III: Actioni contrariam semper et qualem esse reactionem: sive corporum
duorum actiones in se mutuo semper esse quales et in partes contrarias
dirigi.
Now, in a language that we can understand, what they are saying are these:
Law I: Every body persists in its state of being at rest or of moving uniformly
straight forward, except insofar as it is compelled to change its state by
force impressed.
Law II: The change of momentum of a body is proportional to the impulse
impressed on the body, and happens along the straight line on which that
impulse is impressed.
Law III: To every action there is always an equal and opposite reaction: or
the forces of two bodies on each other are always equal and are directed in
opposite directions.
OK, this may not be so understandable either. So, we will delve into them one
by one more carefully.
2.1 The First Law: The Law of Inertia
Law I: Every body persists in its state of being at rest or of moving uniformly
straight forward, except insofar as it is compelled to change its state by
force impressed.
9
CHAPTER 2. NEWTONS LAWS OF MOTION 10
All three laws that are seemingly stating dierent facts can be compressed
into a single equation:
F =
dp
dt
= ma (2.1)
This equation is called the equation of motion and it is, without a doubt, the
single most important equation in classical mechanics. In fact, it can be said
that this equation alone is all of classical mechanics. Throughout the year, we
will learn a number of dierent applications of this single equation.
Then the question arises: why did of all the people, Newton, who probably
knew better than anyone else that the laws can be expressed as a single equation,
bothered to elaborate in such detail? For example, the rst law seems to be, at
rst glance, a reiteration of the second law, that is, from F = ma, it is quite
obvious that when no force is applied (F = 0), the object cannot accelerate nor
decelerate.
However, it deserves to stand on its own for one very important reason we
have already discussed in Chapter 1. It allows us to identify inertia and, equally
importantly, set up a useful reference frame, the inertial frame. We can dene
inertial frame as
Denition: Inertial Frame is a reference frame in which the First Law holds.
As we shall soon see, only within the inertial frame, is the equation of motion
meaningful.
Also, even though the rst law seems to be a subset of more general second
law, it is a very distinct subset in the sense that at zero acceleration, the concept
of mass is completely meaningless. Because force-free motion is by denition,
well, force-free, the object has no net-interaction with anything. Because the
object is interactionless, you cannot distinguish a more massive object from a
less massive object in this force free environment.
When you are staring at two dierent objects moving at two dierent veloc-
ities, it maybe tempting to conclude that one is lighter than the other because
this is moving slower than that. However tempting it may be, you simply can-
not.
Imagine you, your friend and two balls constitute the entire universe. You
are sitting still (or so you think) and your friend is moving at a velocity of 5
m/s away from you to your right. Two balls are moving in opposite directions,
a red one to your left at 1 m/s, a blue one to your right at 4 m/s. To you, a
blue ball moves faster than the red. But for your friend, the red ball moves at 6
m/s and the blue one at 1 m/s, both to his left. For him, the red ball is moving
faster.
Unless you are willing to accept the notion that two balls can have dierent
mass for two dierent people, we now have a problem when you try to interpret
physical world based on velocity. Velocity just allows you to establish relative
reference frame against one another but the role of velocity stops right there
(until you learn Einsteins relativity, but that is a whole new story).
2.2 The Second Law: The Equation of Motion
Law II: The change of momentum of a body is proportional to the impulse
impressed on the body, and happens along the straight line on which that
impulse is impressed.
The rst law hints at the concept of force, but the second law is the one
that explicitly denes force. Mathematically expressing the original statement
of the above Law II, we get
P = p (2.2)
Newton appropriately dened momentum p of an object to be a quantity pro-
portional to its velocity, i.e. p = mv and an impulse P occurs when a force F
acts over an interval of time t, i.e. P =
_
t
Fdt. Then the Eq. (2.2) can be
rewritten as
_
t
Fdt = (mv)
F =
d(mv)
dt
= m
dv
dt
= ma (2.3)
Eq. (3.9) is where the equation of motion, Eq. (2.1) directly comes out of,
and we can see that the net force applied to a body produces a proportional
acceleration.
Also, the proportionality constant, m, between the velocity and momentum
is of great signicance here. From Eq. (2.1), we can see that for a given amount
of force, acceleration is inversely proportional to m. In other words, the larger
m is, the greater the tendency to stay its course of motion. To put it another
way, m is the physical quantity that represents inertia i.e. mass.
From this denition of mass, we can see that measurement of force can be
used to measure mass. A balance is an excellent example. Any balance uses
balancing between the known amount of force and a force acting on an object
of unknown mass, . e.g. gravitational force.
For what we will learn throughout this course, however, there is a more
complicated use for the Newtons Second Law. If we assume that the mass m
does not vary with time, Newtons equation of motion, F = ma = mr, is simply
a second-order dierential equation that may be integrated to nd r = r(t) if
the function F is known. Specifying the initial values of r and r = v then allows
us to evaluate the two arbitrary constants of integration. We then determine
the motion of a particle by the force function F and the initial values of position
r and velocity v.
2.3 The Third Law: The Law of Action and Re-
action
Law III: To every action there is always an equal and opposite reaction: or
the forces of two bodies on each other are always equal and are directed in
opposite directions.
The third and nal of Newtons laws is also known as the action-reaction
law. In some sense, the rst two laws were discovered already by Galileo and
in fact none other than Newton himself gave credit to Galileo for the rst law.
However, the third law is the product of Newtons great insight that all forces
are interactions between dierent bodies, and thus that there is no such thing
F
1
F
2
F
1
F
2
(a) (b)
Figure 2.1: The Third Law of Motion in its (a) weak form and (b) strong
form. In (a), there is no net force, but net torque is present.
as a unidirectional force or a force that acts on only one body. Whenever a
rst body exerts a force F
1
on a second body, the second body exerts a force
F
2
= F
1
on the rst body.
We can categorize the third law into two, a weak law of action-reaction and
a strong law. In a weak form of the law, the action and reaction forces only
has to be equal in magnitude and opposite in direction. They need not lie on a
straight line connecting the two particles or objects. As a result, there could be
a net torque acting on the system. On the other hand, the strong form of the
law states that the forces have to be lined up. This distinction may be trivial,
but it will become important in the next chapter.
For the time being, we will rewrite the third law, using the denition of force
given by the second law,
dp
1
dt
=
dp
2
dt
m
1
a
1
= m
2
a
2
m
2
m
1
=
a
1
a
2
(2.4)
and from this we can give a practical denition of mass. If we set m
1
as the
unit mass, then, by comparing the ratio of accelerations when m
1
is allowed to
interact with any other body, we can determine the mass of the other body. So
it is the Newtons Laws that allows us to dene relative mass to our selection
of unit mass m
1
.
Of course, to measure the accelerations, we must have appropriate clocks and
measuring rods. Lets now go back to IPK, IPM and IPS, these are basically the
unit mass, a measuring rod and an appropriate clock required. From this, we
can see that the job of BIPM and classical mechanics are not separable. Having
IPM and IPS is what lets us to describe motion in a quantitative manner, and
that in turn, according to the Newtons Laws, gives us the denition of mass.
Chapter 3
Laws of Conservation
In physics, there are a number of conservation laws, laws that state certain
properties of a system does not change in time. Laws of conservation is related
to dierentiable symmetries of physical systems, as Emmy Noether pointed out,
and can be a subject of intense study. However, we will not delve into this
point to deeply, at least not yet, and will cover three widely used conservation
laws that are direct consequences of Newtons Laws of Motion. These three are
conservation of momentum, angular momentum and energy. We will look into
one by one.
3.1 Conservation of Momentum
Since the momentum is dened as the product of mass and velocity, for a system
that conserves its mass, constant velocity is equivalent to constant momentum.
Therefore, the concept of inertia stated in the First Law of Motion, in and
of itself, is declaration of momentum conservation. Momentum conservation,
however, extends beyond the force-free motion.
Imagine a particle that is accelerating. For this particle to accelerate, force
has to be exerted. From the third law, when a force is acting on an object, there
has to be an entity that is exerting the force, and the rst object in question
has to apply the same amount of force in opposite direction on to that entity.
In a mathematical form: F
2
= F
1
.
The equation can be rewritten as follows:
F
1
+F
2
= 0
dp
1
dt
+
dp
2
dt
=
d
dt
(p
1
+p
2
) = 0
p
1
+p
2
= constant (3.1)
In other words, in the absence of external force, the total momentum of the two
objects that are exchanging force is constant. The same logic can be applied
to a multi-object system beyond two objects. It follows that, at least classical
mechanically, total momentum of the universe is conserved. However, this is not
a practically useful statement. What this implies is that an external force on
a system is always an internal force of a larger system, and in reality, studying
mechanics comes down to isolating an observable systems appropriately.
13
CHAPTER 3. LAWS OF CONSERVATION 14
3.2 Conservation of Angular Momentum
p
p
r
p
t
r
x
y
p
r
x
y
r
||
r
Figure 3.1: Visually interpreting angular momentum.
For any moving particle, one can dene angular momentum, a representation
of circular motion, with respect to a reference point in space as
L r p (3.2)
where r is the position of the particle with respect to the reference point, and
p is the linear momentum of the particle.
One thing that should be emphasized at this point is that, unlike linear
momentum, angular momentum has to be dened about a reference point. Once
the reference point is dened, any linear motion can be decomposed into two
components, a radial component, p
r
, and a tangential component, p
t
, around
a circle with radius r about the reference point as shown in Fig. 3.2(a). Then
the denition of angular momentum is equivalent to
L = r p
t
(3.3)
An alternative is to decompose the position vector into a component that is
parallel to the momentum, r
and a perpendicular component, r
as shown in
Fig. 3.2(b). Breaking down the position vector yields to
L = r
p (3.4)
From these relations, we can tell that angular momentum is a quantity that
represents with how much momentum around how big a circle is the particle
moving?
When there is more than one particle in the system, the total angular mo-
mentum of the system is a sum of angular momentum of individual particles.
L =
i
r
i
p
i
(3.5)
If we dene torque as the time derivative of the angular momentum,
N
dL
dt
=
i
_
dr
i
dt
p
i
+r
i
dp
i
dt
_
(3.6)
Since p
i
= m
dr
i
dt
, the rst term in Eq. (3.6) is zero. From Newtons Second Law,
dp
i
dt
is the net force acting on the i-th particle, which is the sum of external force
and all the internal forces among the constituent particles, i.e.
dp
i
dt
= F
i
= F
ext
i
+
j
F
int
ij
(3.7)
where F
int
ij
is the internal force acting on the i-th particle from the j-th particle.
(In the Gregorys textbook, the notation G
ij
is used for the internal forces.)
Combining this result with the Newtons Third Law, i.e. F
int
ij
= F
int
ji
, Eq. (3.6)
can be rewritten as
N
dL
dt
=
i
r
i
F
ext
i
+
i,j
r
i
F
int
ij
=
i
r
i
F
ext
i
+
i<j
(r
i
F
int
ij
+r
j
F
int
ji
)
=
i
r
i
F
ext
i
+
i<j
(r
i
r
j
) F
int
ij
=
i
N
ext
i
+
i<j
(r
i
r
j
) F
int
ij
(3.8)
The second term in the Eq. (3.8) is an internal torque. What is interesting here
is that the internal torque vanishes only if either there no internal forces F
int
ij
= 0
or r
i
r
j
is in the same direction as F
int
ij
, that is the internal forces are along
the line connecting the two particles. In other words, if the internal forces are
present, the internal torque vanishes only if the strong form of Newtons Third
Law holds. For almost all the problems that we will deal with throughout this
course, the strong form actually holds, so you need not to worry about this
subtle distinction.
In such cases, the total angular momentum of a particle system does not
change with time in the absence of an external torque. In other words, the
angular momentum is conserved.
3.3 Conservation of Energy
Conservation energy is one of the fundamental conservation laws of nature. It
should be a concept you must have learned since middle school and be familiar
with by now. However, the fact that energy, unlike momentum or angular
momentum, takes many forms makes the law of conservation of energy not as
trivial as one might think at rst. In light of these facts, it should come as no
surprise that Newton never mentioned energy in his work.
Nonetheless, one cannot simply look over the importance of the energy con-
servation. Within the framework of classical mechanics, conservation of energy
refers to lossless transfer of energy between two very specic forms of energy:
kinetic energy and potential energy. So the rst step here would be to dene
these two types of energy.
3.3.1 Kinetic energy
Let us start from the Second Law of Motion:
F = m
dv
dt
(3.9)
We need not dene or restrict the type of force F, and it can be considered simply
as the net force acting on the particle of interest. Take the scalar product of
Eq. (3.9) with v, and we get
F v = m
dv
dt
v =
d
dt
_
1
2
mv v
_
(3.10)
By dening T =
1
2
mv
2
, we can rewrite the above equation as
T
2
T
1
=
_
t
2
t
1
F vdt (3.11)
When F is a force eld F(r), that is the amount of force F acting on the particle
is given for each position r, the right hand side of the Eq. (3.11) becomes
_
t
2
t
1
F vdt =
_
t
2
t
1
F(r)
dr
dt
dt =
_
C
F(r) dr = W
12
(3.12)
which is the work done by moving the particle across the xed path ( from point
r
1
to r
2
during the time between t
2
and t
1
. (See Fig. 3.3.1.) Since the right
hand side of the Eq. (3.11) represents work done on a particle, the left hand
side must represent a change in energy in the form of T =
1
2
mv
2
. Because this
energy T represents the energy stored in a particles motion, that is T depends
on m and v, we dene T as the kinetic energy, the energy of motion.
F(r)
dr
r
1
r
2
Figure 3.2: Work done on a particle can be calculated through a line integral
of F(r) dr over a path C of the motion from point r
1
to point r
2
.
3.3.2 Potential energy
If the force eld F(r) can be expressed in the form
F(r) = V (r) (3.13)
where V (r) is a scalar function only of position, then F is said to be a conser-
vative eld. In such a case, Eq. (3.11) is reduced to
_
C
F(r) dr =
_
C
V dr
=
_
C
V
x
dx +
V
y
dy +
V
z
dz
=
_
C
dV = V (r
1
) V (r
2
) (3.14)
This is a striking result. This states that when the force eld is conservative,
the work done on a particle by the force eld across the path ( is simply the
dierence between a scalar eld V (r) at the initial and nal points of the path
(. The function V (r) is a form of energy that is stored in its position, and is
usually referred to as the potential energy.
3.3.3 Mechanical energy conservation
From the relation between the kinetic energy and work, and the potential energy
and work, we can nd the following relationship for a particle under the inuence
of a conservative force:
W
12
= T
2
T
1
= V
1
V
2
(3.15)
and rewriting this equation, we get
T
1
+V
1
= T
2
+V
2
(3.16)
Because the choice of initial and nal time and position, t
1
, t
2
, r
1
and r
2
is
arbitrary, the above equation is equivalent to saying that the mechanical energy,
i.e. the sum of kinetic and potential energy, is constant in time:
T +V = E = constant (3.17)
Simply put, the mechanical energy is conserved for a particle moving in a con-
servative force eld.
An important and very useful property of this energy conservation is that
because the potential energy is a function only of the position of the particle,
the path it follows becomes irrelevant. Put in another way, even if the particle
is under a geometrical constraint, e.g. a pendulum attached to a string or a
marble rolling down a guide rail, if the constraint forces on the particle do no
work on the particle, the mechanical energy is still conserved.
Chapter 4
Solving Equation of
Motions
The starting point of classical mechanics is the equation of motion given by
F = ma (4.1)
Since, at the end of the day, what we want to nd out in classical mechanics is
time evolution of position of a physical object, r(t), the above equation turns
out to be a dierential equation of the form
m
d
2
r
dt
2
F(r, t) = 0 (4.2)
In some rare occasions, the force F(r, t) may contain higher order derivatives of
r with respect to t. But, as I said, this is very rare, especially in undergraduate
physics level, and for the most part, the equation of motion is a second order
dierential equation.
If F(r, t) takes the form
F(r, t) = f
2
(t)
d
2
r
dt
2
+f
1
(t)
dr
dt
+f
0
(t)r +f(t) (4.3)
we can rewrite Eq. (4.2) as
a
2
(t)
d
2
r
dt
2
+a
1
(t)
dr
dt
+a
0
(t)r = f(t) (4.4)
where a
2
(t) = m f
2
(t), a
1
(t) = f
1
(t) and a
0
(t) = f
0
(t). Aside from the
fact that it is a vector equation, Eq. (4.4) has the same form as a linear n-th
order dierential equation
a
n
(x)
d
n
y
dx
n
+a
n1
(x)
d
n1
y
dx
n1
+ +a
1
(x)
dy
dx
+a
0
(x)y = f(x) (4.5)
where a
i
(x)s and f(x) depend only on x.
18
CHAPTER 4. SOLVING EQUATION OF MOTIONS 19
4.1 Force-Free Motion
Let us rst consider a one dimensional force-free motion with F(x, t) = 0. From
Eq. (4.2), we get the equation of motion
m
d
2
x
dt
2
= 0 (4.6)
By dening v dx/dt, the equation reduces to a rst order dierential equation
dv
dt
= 0 (4.7)
A fairly straightforward integration yields the following result
v =
_ _
dv
dt
_
dt =
_
dv = v
0
(4.8)
Since v = dx/dt, we can rewrite the above equation as
dx
dt
= v
0
(4.9)
Following the same procedure as Eq. (4.8), we get
x =
_ _
dx
dt
_
dt =
_
v
0
dt = v
0
t +x
0
(4.10)
Here, x
0
and v
0
are unspecied constants. The position changes with time along
a straight line as shown in Fig. 4.1.
Time
P
o
s
i
t
i
o
n
Figure 4.1: One dimensional force-free motion.
In this case, its starting point (y-intercept on the graph) and velocity (slope)
are unknown, however, if the initial conditions x
0
and v
0
are specied, we can
uniquely pick out a single line among innitely many possibilities.
This method can be generalized to higher dimensional space, and one will
get a solution of the form
r = v
0
t +r
0
(4.11)
In this case, although unfortunate choice of coordinate system might force you
to have to solve a set of three identical dierential equations for this problem, in
fact, an appropriate choice of coordinate system can always reduce the problem
to a one dimensional problem.
4.2 Constant Force Motion
We now turn on a constant external force, F(r, t) = F
0
on to this particle. But,
we will once again start with one dimensional motion.
4.2.1 Constant Force Motion in One Dimension
One dimensional constant force motion can be expressed by the equation of
motion:
d
2
x
dt
2
=
F
0
m
= a
0
(4.12)
Same method used in the previous section can be utilized to nd the solution
to this equation, and one can easily nd that the solution is
x =
1
2
a
0
t
2
+v
0
t +x
0
(4.13)
Time
P
o
s
i
t
i
o
n
Figure 4.2: A general solution to one dimensional constant force motion.
The solution has an additional term
1
2
a
0
t
2
, which is the eect of acceleration
due to the force. The solution is no longer a straight line. Instead, we get a
parabola and if the initial condition is not specied, the parabola can take any
of the curves in Fig. 4.2.1. Here, note that the acceleration a
0
is xed by the
force applied to the given mass in the problem, a
0
= F
0
/m and is not aected
by the initial values of v
0
and x
0
.
We can rewrite the Eq. (4.13)
x =
1
2
a
0
(t +
v
0
a
0
)
2
+x
0
_
v
0
a
0
_
2
(4.14)
which is an equation obtained by moving x =
1
2
a
0
t
2
by
v
0
a
0
along the time-axis
and by x
0

_
v
0
a
0
_
2
along the position-axis. From this perspective, the initial
value problem simply becomes at what point along the parabolic space-time
trajectory you start observing the system as shown in Fig. 4.2.1.
Figure 4.3: A dierent way of interpreting the initial value problem of one
dimensional constant force motion.
4.2.2 Constant Force Motion in Two Dimensions
The two dimensional problem with constant force is a bit trickier than the
force-free case where one could always, in principle, reduce the problem to a
one dimensional problem. Here, one has to examine the direction of the force
(acceleration) and the direction of the initial velocity carefully.
If the velocity vector and acceleration vectors line up, either in the same
direction or opposite direction, as in Fig. 4.3.1(a), the problem can be reduced
to a one dimensional problem. This is because acceleration is dened as the
change of velocity with respect to time and when the velocity component is
all in the direction of acceleration, the particle velocity always remains in that
direction, and we need not care about the motion in any other directions. Note
that the problems with zero initial velocity also falls into this category. Such
problems include the case of a bullet shot straight up in the air, or a free fall
problem.
v
0
a
0
v
0
a
0
(a) (b)
Figure 4.4: Solving a two dimensional constant force problem.
On the other hand, if the velocity and acceleration do not line up, then we can
divide the velocity into two components, one along the direction of acceleration
and the other perpendicular to the acceleration. Then we end up with two
dierent equations. Along the direction of acceleration, we have to solve for a
problem of constant force motion, whereas in the perpendicular direction, the
motion is force-free.
For example, assume that the direction of acceleration is along the y-axis,
and there is no acceleration along the x-axis. Then the motion along the x-axis
can be described with the Eq. (4.10).
x = v
x0
t +x
0
(4.15)
and the motion along the y-axis with the Eq. (4.13)
y =
1
2
a
0
t
2
+v
y0
t +y
0
(4.16)
where v
x0
and v
y0
are the initial velocity along the x- and y-axis, respectively.
The vector r = (x
0
, y
0
) marks the initial position. Combining Eq. (4.15) and
Eq. (4.16), we get
y =
1
2
a
0
_
x x
0
v
x0
_
2
+v
y0
_
x x
0
v
x0
_
+y
0
=
1
2
a
0
v
2
x0
x
2
+
_
v
x0
v
y0
a
0
x
0
v
2
x0
_
x +
_
1
2
a
0
v
2
x0
x
2
0
v
y0
v
x0
x
0
+y
0
_ (4.17)
This is what we call a trajectory. The equation spatially traces how the particle
moves without specically knowing where the particle would be at a given time.
This is a useful way of visualizing a motion only in two or higher dimensions.
(In one dimensional motion, the trajectory is always a straight line and is thus
boring.) From this, one can see that the trajectory is a parabolic equation for
a general constant force motion, and this is exactly what we see for a baseball
thrown in the air or a red canonball with an arbitrary angle.
It is interesting to note that, just as any force-free motion could be reduced
to a one dimensional problem with an appropriate choice of coordinates, any
constant force motion can be reduced to a two dimensional motion. This is
because two vectors, a
0
and v
0
form a two dimensional plane and thus no
motion can exist outside the plane.
4.3 Varying Force Motion
In this section, we will nally deal with forces that are not constant in time.
However, we will not deal with forces that changes explicitly with time. Instead,
we will look at the problems with forces that depend on velocity or position.
Because we are studying mechanics, that is a eld of physics where we inherently
deal with particles in motion, forces that depend on velocity or position results
in implicit change in their magnitudes or directions with time.
4.3.1 Drag Force
Anyone who has swum or rowed a boat understands that motion in water is
considerably more resistive than the motion in air. This is due to the drag
force of a medium and the concept of drag force goes back no later than ancient
Greece, when Aristotle argued that the speed of an objects motion is inversely
proportional to the viscosity of the surrounding medium. As we have mentioned
earlier, Aristotles world view on mechanics was deeply awed, however, his
observation that the viscosity is related to particles speed contains elements of
truth.
The diculty in estimating the drag force lies in understanding the micro-
scopic origin of it. When an object passes through a medium, the object has
to push the medium out of its way. It requires force acting on the medium,
which in turn creates force acting back on the object. To theoretically derive
how this reaction of the medium turns into the net force acting on the moving
object, it requires full understanding of interaction among particles constitut-
ing the medium and between the particles and the moving object. This is an
extremely dicult task, and our understanding of the drag force, for the most
part, relies on phenomenological analysis. Even this phenomenological analysis
is well outside the scope of this course, and we will just take the result of such
analysis here.
An ideal particle with zero cross section does not have its place in a phe-
nomenological description of mechanics, so we will assume an object with maxi-
mum cross sectional radius a perpendicular to the direction of its motion. If the
velocity of the object is given by v = ve
v
, the drag force acting on the object
is given by
D = F(R)a
2
v
2
e
v
(4.18)
where F(R) is a single variable function of R whose value must be obtained
phenomenologically. is the density of the medium and e
v
is the unit vector
along the direction of motion. The minus sign signies that the drag force is in
the opposite direction to the motion, causing the resistance. R is what is known
as the Reynolds number named after Osborne Reynolds, a 19th century expert
in uid dynamics, and is dened as
R
av
(4.19)
where is the viscosity of the medium. Its exact denition and meaning is,
once again, outside the scope of this course and a subject of more detailed uid
dynamics. We will just state here that the function F(R) does not strongly
depend on R over a wide range of values (about 1000 < R < 100, 000), and thus
can be treated as a constant close to unity. Therefore, for an object moving
through a medium that satises the condition of 1000 < R < 100, 000, the drag
force can be represented as
D a
2
v
2
e
v
(4.20)
(Exact value of F(R) depends on the geometry of the object and other factors,
but for the purpose of this class, we will treat it as unity.) If a particle with mass
m happens to have a velocity v in the medium, and the only force acting on it
is the drag force caused by its motion, the equation of motion can be written as
m
d
2
r
dt
2
= m
dv
dt
= F = D = a
2
v
2
e
v
(4.21)
Because the drag force is proportional to the square of velocity, this is called
quadratic law of resistance. Here the direction of the force and the direction of
the velocity is matched and thus we can reduce the problem into one dimension
m
dv
dt
= a
2
v
2
(4.22)
The equation contains the square of velocity v
2
, and is a non-linear dierential
equation. Fortunately, this equation turns out to be one of a few cases where
non-linear dierential equation can be solved. We will leave it as an exercise.
The nal result turns out to be
v(t) =
v
0
1 +t/
(4.23)
x(t) = v
0
ln(1 +t/) +x
0
(4.24)
where 1/ = v
0
a
/
m2. What is remarkable here is that the velocity of this
object reaches zero as time goes on, however, the particle nonetheless travels
across innite distance. This is because the drag force quite large at large
velocities, but drops rapidly as the particle slows down due to the drag. The
drag force becomes so small at small velocities that the velocity cannot quite
come to zero fast enough, and the particle never comes to a complete stop.
This, however, seems unphysical as no object can travel through a medium for
indenite distances.
A closer inspection of the condition required for this problem solves this
dilemma. The equation holds for a nite range of the Reynolds number and from
the way Reynolds number is dened, R = av/, one can see that the quadratic
law of resistance cannot hold as the velocity reaches zero. It is reasonable to
think that the density and viscosity of the medium and the size of the object
traveling through the medium does not change with the objects velocity, and
thus R must approach zero as v approaches zero.
v
0
t
x-x
0
t
Figure 4.5: Velocity and position evolution of the motion under the quadratic
law of resistance. The area under the velocity curve (light orange region)
represents the distance traveled by the object.
It turns out, in small R limit, the F(R) has 1/R dependence. A more careful
analysis shows that F(R) reaches a value close to 6/R, and the drag force is
then represented by the equation
D 6av (4.25)
Because the drag force changes linearly with velocity, this is called the linear
law of resistance. Again, solving this equation will be left as an exercise. The
nal result in this case is
v(t) = v
0
e
t/
(4.26)
x(t) = v
0
(1 e
t/
) +x
0
(4.27)
where 1/ = 6a/m. In this case, the velocity decays exponentially, and as
a result the traveling distance is also limited to a nite value of x = v
0
,
asymptotically approaching it. In other words, a particle moving in a linearly
resistive medium will slowly come to a stop after traveling the distance equal
to the product of the initial velocity, v
0
, and the time constant, inversely
proportional to the viscosity.
v
0
t
x-x
0
t
Figure 4.6: Velocity and position evolution of the motion under the linear
law of resistance. The area under the velocity curve (light orange region)
represents the distance traveled by the object.
We can now think of a case where an object is pushed through a viscous
medium. We will here consider the linear resistance, and leave the quadratic
resistance case as an exercise. One such example is a falling object through a
viscous medium such as air. The push, here, is the gravitation, and when the
object is in motion, it will experience a drag force in addition to gravity. We will
set the upward direction as positive direction, so the gravitational force is then
F
g
= mg. The drag force is always in the opposite direction of the direction
of the motion and thus D = mv/. So the equation of motion is
m
dv
dt
= mg
mv
dv
dt
= g
v
(4.28)
Solve the equation for initial values v
0
and x
0
, and we get
v(t) = (v
0
+g)e
t/
g
x(t) = x
0
+ (v
0
+g
2
)(1 e
t/
) gt (4.29)
Here, the nal velocity approaches v = g, which is the velocity at which the
net force becomes zero, that is
F
g
+D = mg
mv
= 0 (4.30)
This is what is known as the terminal velocity of a motion in a drag medium.
x-x
0
t
v
0
t
Figure 4.7: Velocity and position evolution of the motion under a constant
force plus linear resistance. The area under the velocity curve (light green
region subtracted from light orange region) represents the distance traveled
by the object. The grey dashed line marks the point in time at which the par-
ticle returns to its original position. Note that the velocity curve approaches
non-zero value due to the shift in the force equilibrium point. Accordingly,
the position of the object approaches a straight line with non-zero slope repre-
senting a constant velocity motion. Because the terminal velocity is negative
in this case, the slope is also negative.
Suppose there is no drag force. Then the motion will indenitely accelerate,
provided that the object does not hit the wall or ground. Since there is a force
opposing the motion that grows with velocity, that acceleration has to stop
once the velocity reaches the point where the constant pushing (or pulling) of
the object matches the drag force. At this point, there is no net force and the
inertia takes over and the object continues to move at the speed of g. Therefore,
instead of asymptotically approaching zero velocity as in the no outside force
case, the object approaches the terminal velocity. It is notable that in this
case the velocity does not oscillate around the terminal velocity, but approaches
asymptotically.
4.3.2 Harmonic Oscillator
There are two types of position dependent forces that play an important role in
introductory classical mechanics. One is the gravitational force that is inversely
proportional to distance and the other is the restoring force that is responsible
for harmonic oscillation. Such restoring force follows Hookes law , i.e. force is
proportional to the displacement. We will take a closer inspection of them in
Chapter 8 and Chapter 11, respectively. For now, we will briey look at the
simple harmonic motion in one dimension.
m
d
2
x
dt
2
= F = kx
d
2
x
dt
2
+
2
x = 0 (4.31)
where
2
= k/m. This is a second order linear dierential equation with con-
stant coecients, thus the solution should take the form x(t) = e
t
. We get
auxiliary equation
2
+
2
= 0 (4.32)
and the general solution is then
x(t) = A
1
e
it
+A
2
e
it
(4.33)
where A
1
and A
2
are arbitrary complex numbers that should be determined by
the initial condition. By representing A
1
and A
2
with a set of amplitude and
phase,
A
1
= Ae
i(+/2)
and A
2
= Ae
i(/2)
(4.34)
we can rewrite the solution given by Eq. (4.33) as
x(t) = Acos(t ) (4.35)
This is an oscillatory solution, and one can easily check that the solution satises
the Eq. (4.31). Here the amplitude, A, and the phase are two arbitrary
constants that should be determined by the initial condition. The phase
determines during what part of the oscillatory cycle, ones observation begins.
For example, if the initial condition is given by x
0
= 0 and v
0
= A
0
, the
amplitude and the phase are xed to be A = A
0
and = /2. Then the
solution is a simple sine function x(t) = A
0
sin t.
What is interesting about this motion is that there is a force equilibrium
point at x = 0. Unlike the drag force case where the velocity approaches its
force equilibrium point asymptotically, here the position oscillates around the
force equilibrium point. This is an interesting point to ponder about, and I will
leave it to the readers to think about it.
Chapter 5
Lagrangian Mechanics
So far, we have looked at a few examples in which we can solve the Newtons
equation of motion analytically. Even if it doesnt seem straightforward to solve
the equations of motion in some cases, in theory, they can always be worked
out. It may not be easy, but it can be done. In a sense, that is all there is to it
in classical mechanics. However, there are cases where the equations of motion
are just to cumbersome and impractical to work out.
For example, when the system of interest is presented with a cylindrical
symmetry or a spherical symmetry, solving the problem in Cartesian coordinate
turns out to be extremely cumbersome. (Dont just take my word for it. Try
and solve the equation of motion for a simple pendulum with a small oscilla-
tion. You will not enjoy it.) A natural choice of coordinate system would be
a cylindrical coordinate or a spherical coordinate in such cases. However, such
choice of coordinate system has its own downside to it. The unit vectors in such
coordinate systems are dened by three orthogonal orientations with respect to
the position vector in the space, which means in a mechanical system where the
position vector is bound to change, the unit vectors will change their directions.
This challenge is not limited to these two well known coordinate systems.
Almost any constraints in motion such as arbitrarily curved surfaces will pose
even greater threats. This is where Lagrangian mechanics enters. It is an
alternative way of looking at the mechanics problem through the variational
principle.
In physics, almost any problems can reformulated with variational principle.
The principle itself simply states that a quantity that is expressed as an integral
can vary with the integrand and, if the integrand is tuned to correct value, the
integrated quantity will have an extremum value. In physics, the way variational
principle is commonly used is that physical processes happen in a way such
that some relevant physical quantities are either maximized or minimized. For
example, in mechanics, a motion of an object follows the path that minimizes
what is known as the action.
Figuring out how it can be applied to physics or why it works as well as it
does in physics is more of a philosophical question than a physical one. It is an
interesting question to ponder upon, but we will leave it to philosophers. Here,
instead, we will take the fact that it works as a given and try to work out how
it is mathematically implemented.
28
CHAPTER 5. LAGRANGIAN MECHANICS 29
5.1 The Euler-Lagrange Equation
Let us imagine a set of functions x(t)s whose values are known at two points
t = a and t = b as x(a) = A and x(b) = B. We can dene an integral
J[x] =
_
b
a
F(x, x, t)dt (5.1)
where F is a function of three variables x, x, t. Among the set of functions x(t)
in the given range of a t b, suppose x
(t) minimizes the functional J[x].

This means that
J[x] J[x
] (5.2)
for all functions of x(t). Any functions other than x
(t) can be expressed as

x(t) = x
(t) +(x) (5.3)

with (a) = (b) = 0 to satisfy the condition x(a) = A and x(b) = B for
all x(t)s. Here is a tunable parameter. We can rewrite the Eq. (5.1) and
construct a functional J as a function of instead of x,
J() =
_
b
a
F(x
+, x
+ , t)dt (5.4)
Dierentiating the above equation with respect to , we get
dJ
d
=
_
b
a
F
dt =
_
b
a
_
x
F
x
+
x
F
x
_
dt
=
_
b
a
_
F
x
+
F
x
_
dt
=
_
b
a
_
F
x

d
dt
_
F
x
__
dt +
F
x
(t)
b
a
(5.5)
where integration by parts were performed for the second term of the second
line to obtain the third line. Because (a) = (b) = 0, the last term of the third
line in the above equation vanishes. We have set up J() such that it has an
extremum at = 0, which means the following must be true for an arbitrary
choice of (t):
dJ
d
=0
=
_
b
a
_
F
x

d
dt
_
F
x
__
(t)dt
=0
(5.6)
Therefore, the integrand must vanish irrespective of (t) and we get
F
x

d
dt
_
F
x
_
= 0 (5.7)
which is known as Eulers equation or Euler-Lagrange equation.
When the function F is not an explicit function of t, there is a simple way
to reduce a second order dierential equation given in the form of Eq. (5.7) into
a rst order dierential equation.
d
dt
_
x
F
x
F
_
= x
F
x
+ x
d
dt
_
F
x
_
_
x
F
x
+ x
F
x
_
= x
_
d
dt
_
F
x
_
F
x
_
= 0
(5.8)
Therefore
x
F
x
F = constant (5.9)
5.2 Lagrangian Mechanics
In the previous section we have discussed the potential usefulness of minimiza-
tion principle in mechanics and how it can be implemented through calculus
of variation. The minimization principle in mechanics is usually referred to as
Hamiltons principle, and from this principle, we can obtain Lagrangian me-
chanics.
Why a mechanical analysis followed from Hamiltonian principle is called La-
grangian mechanics, not Hamiltonian mechanics, may be puzzling. It is related
to the historical development of Lagrangian mechanics, that is, Lagrangian me-
chanics in its original form did not evolve based on calculus of variation.
Lagrange was mainly interested in developing a method that is more gen-
erally applicable than Newtons equation of motion. Naturally, the advantage
of Lagrangian mechanics over Newtonian mechanics lies in the fact that it can
be used for generalized coordinate system. Another advantage is that we do not
need to know all the forces acting on the systems or particles within a system.
An appropriate choice of generalized coordinates takes care of most of the im-
posed constraints. This point will become more clear once we look at a few
examples.
Hamiltons insight that the same method can be derived from the minimiza-
tion principle came later. Here, we will not follow Lagranges development of
Lagrangian mechanics in detail. Instead, we will start with Hamiltons principle.
Hamiltons principle in its original form states that
Of all the kinematically possible motions that take a mechanical
system from one given conguration to another within a given time
interval, the actual motion is the one that minimizes the time inte-
gral of the Lagrangian of the system.
In order to dissect this statement, we need to know what Lagrangian is. We
will start by dening it, for an unconstrained one dimensional system, as the
dierence between the kinetic energy, T, and potential energy, V , that is
L(x, x) = T(x, x) V (x) (5.10)
Once the Lagrangian is dened, the Hamiltons principle becomes,
Of all the kinematically possible motions that take a mechanical
system from one given conguration to another within a given time
interval, the actual motion is the one that minimizes the integral
S[x] =
_
t
1
t
0
L(x, x)dt. (5.11)
By applying the Euler-Lagrangian equation(Eq. (5.7)) to minimize the above
integral, we get
L
x

d
dt
_
L
x
_
= 0 (5.12)
which is referred to as the Lagrange equation. With the kinetic energy and
potential energy given by
T =
1
2
m x
2
and V =
_
F(x)dx (5.13)
Eq. (5.12) reduces to
x
(T V )
d
dt
_

x
(T V )
_
=
dV
dt

d
dt
(m x)
= F(x) m x = 0
(5.14)
which is identical to the Newtons equation of motion. For a system in higher
dimensional space or with multiple particles, we can redene Lagrangian by
calculating the total kinetic energy and potential energy,
T =
n
i
T
i
=
n
i
1
2
m x
2
i
V = V (x
1
, x
2
, , x
n
)
L = T V = L(x
1
, x
2
, , x
n
, x
1
, x
2
, , x
n
) (5.15)
One can set Euler-Lagrangian equation for each coordinate and Newtonian equa-
tions of motion will be recovered. This seems like a convoluted way of reaching
the same conclusion. However, this method becomes extremely powerful in more
complicated problems.
5.2.1 Generalized Coordinates
Imagine an unconstrained single particle in three dimensional space. Using
Newtonian method, we need to set up three equations of motion, one for each
coordinate axis.
F
i
(r) = m x
i
(5.16)
where the index i = 1, 2, 3 denotes three coordinate axes x, y, and z, respectively.
Now suppose that the motion is constrained, say along a ring on the xy-
plane. This adds two constrains to the motion
x
2
+y
2
= R
2
and z = 0 (5.17)
where R is the radius of the ring. One of the constraints, z = 0, is a solution to
the Eq. (5.16) with i = 3, and thus simply reduces the three dimensional problem
into a two dimensional one. Similarly, the rst constraint, x
2
+y
2
= R
2
can be
used, in principle, to reduce the second equation of motion from Eq. (5.16) so
that we are left with only one equation to solve.
In practice, this turns out to be a near impossible task, because substituting
for y =
R
2
x
2
and solving for F
y
= m y will give you a headache that will
last for days. What we want to do here is to pick a coordinate system that
reects the given constraints more intuitively. That coordinate system, in this
case, is the cylindrical coordinate system (3D) or polar coordinate system (2D).
By setting x = r cos and y = r sin , we can replace the constraint as r = R.
Then we get x = Rcos and y = Rcos , which are parametrized with a single
variable . However, this doesnt help us solve Newtonian equation of motion
all that much better.
We are still left with a problem of expressing the equation of motion in a
polar coordinate system, which is not trivial. Also the constraint forces that keep
the particle in the circular ring is never specied, that is equations of motion
are not fully specied. It seems pretty obvious already, there is something very
undesirable to Newtons method.
Now, lets extend this to a N particle system in three dimensions. For such
a system, there are maximally 3N equations of motion, three for each particles.
With s number of constraints, you get another s numbers of equations. These
are not independent. In fact, using the constraints, one can reduce the number
of equations to 3N s. But how to get rid of s number of equations of motion
is not clear.
This is where Lagrangian mechanics enters. Instead of sticking to the Carte-
sian coordinates, one can select new coordinates q = (q
1
, q
2
, , q
n
). This new
coordinates are what is known as the generalized coordinates, and they have to
meet two conditions:
(i) The generalized coordinates must be independent of each other, that is,
there can be no functional form connecting two dierent coordinates.
(ii) They must fully specify the conguration of the system, that is for any given
values of q
1
, q
2
, , q
n
, the position of all the N particles r
1
, r
2
, , r
N
must be identiable. In other words, the position vectors of N particles
must be known functions of the n independent generalized coordinates:
r
i
= r
i
(q
1
, q
2
, , q
n
) (5.18)
From the example of the single particle moving on a ring, that coordinate is the
-coordinate. You only need one such coordinate. Likewise, there are n = 3Ns
number of generalized coordinates to fully specify the conguration of a system
with N particles in three dimensional space under s constraints. The number
of generalized coordinates required is also referred to as degrees of freedom.
With Eq. (5.18) available, we can reconstruct Lagrangian in the generalized
coordinate system. A mechanical motion implies time dependence in r
i
s and
this naturally leads to time dependence in q. Then, we can take time derivative
of q and obtain
q = ( q
1
, q
2
, , q
n
) (5.19)
which is called the generalized velocity, since it represents the velocity of the
point q as it moves through the conguration space. One can easily derive the
relationship between the real velocity of a particle and the generalized velocity
through a simple chain rule for dierentiation,
r
i
=
r
1
q
1
q
1
+ +
r
n
q
n
q
n
(5.20)
and kinetic energy of the total system, in terms of the generalized velocity,
becomes
T =
n
j
n
k
a
jk
(q) q
j
q
k
(5.21)
where
a
jk
(q) =
1
2
N
i
m
i
_
r
i
q
j
r
i
q
k
_
(5.22)
The kinetic energy T is then a function of q = (q
1
, , q
n
) and q = ( q
1
, , q
n
).
Similarly, the potential energy V can be rewritten as a function of q. Then the
Lagrangian becomes a function of the generalized coordinates and generalized
velocities:
L = L(q
1
, , q
n
, q
1
, , q
n
) (5.23)
We are then left with n number of Lagrange equations to solve
L
q
j
d
dt
_
L
q
j
_
= 0 (5.24)
where i = 1, 2, , n.
By rewriting the equation
d
dt
_
T
q
j
_
T
q
j
=
V
q
j
= Q
j
(5.25)
we can see that there is a term, Q
j
, that is dened in the generalized coordinate,
corresponding to the concept of force in the real space. For this reason, Q
j
is
known as the generalized force. This quantity is equivalent to
Q
j
=
i
F
i
r
i
q
j
(5.26)
where F
i
is the specied force acting on the i-th particle.
So far, we have only considered a constraint that can be expressed in the form
of r
i
= r
i
(q). Such constraints are referred to as geometric constraints, because
the coordinate transformation between the real coordinates and generalized co-
ordinates is purely geometric, that is there is no explicit time dependence or
velocity dependence in the coordinate transformation. Such a system is said to
be holonomic. Establishing Lagrangian mechanics for a non-holonomic system
in general is outside the scope of this course and we will not treat the problem
here.
However, time dependent constraint can be relatively easily integrated into
the Lagrange equations. Also, there are some cases where the generalized force
can be expressed in the form
Q
j
=
d
dt
_
U
q
j
_
U
q
j
(5.27)
for some function U(q, q, t). Then the function U(q, q, t) is called the velocity
dependent potential of the system and Lagrangian is given by
L(q, q, t) = T(q, q, t) U(q, q, t) (5.28)
and one can still use Lagrange equation to solve mechanics problems. Note
that the velocity dependent potential produces non-conservative force as the
resulting force is not strictly position dependent.
5.3 DAlemberts Principle
It seems pretty obvious, by now, that with the help of Hamiltons principle and
the denition of Lagrangian L(q, q, t) = T(q, q, t) U(q, q, t), one can easily
come up with a mechanical equation in generalized coordinates whose solution
can be transformed later into a real space solution. However, there is a nagging
question regarding the denition of Lagrangian, that is how did someone come
up with such a physical quantity that produces the Lagrangian equation? The
answer to this question lies, as stated earlier, in the fact that the Hamiltons
principle came after Lagrange came up with his denition of Lagrangian. For
Lagrange, the starting point was virtual work and DAlemberts principle.
Suppose a system with N particles where an individual particle is subject
to a net force,
F
i
= F
S
i
+F
C
i
= m
i
v
i
(5.29)
where F
S
i
denotes the specied force and F
C
i
denotes the constraint force. In
other words, we know explicitly how the F
S
i
s act on each particle, but the
eect of F
C
i
s is not known to us, except for the fact that the constraint force
acts as a boundary condition. For most cases, the direction of such constraints
are orthogonal to the directions of particles allowed motion. The eect of the
constraint can be thus written as
N
i
F
C
i
v
i
= 0 (5.30)
In other words, we can rewrite the equation of motion for the total system as
N
i
m
i
v
i
v
i
=
N
i
F
S
i
v
i
+
N
i
F
C
i
v
i
=
N
i
F
S
i
v
i
(5.31)
where the constraint force is no longer explicit.
However, one can consider a virtual path that is also subject to the same
constraint force, and by denoting the velocity along that virtual path as v
i
, we
can get the identical relation
N
i
m
i
v
i
v
i
=
N
i
F
S
i
v
i
(5.32)
which is known as the DAlemberts principle.
For a holonomic system with constraint force doing no virtual work, there
must be a generalized coordinate system q that links the real space position
of the particles r
i
s to q. With such generalized coordinate system, we can
construct a virtual motion v
i
v
i
=
r
q
1
(5.33)
that corresponds to the motion generated by generalized velocities
q
1
= 1, q
2
= 0, , q
n
= 0 (5.34)
From dAlemberts principle, we get
N
i
m
i
v
i
r
i
q
1
=
N
i
F
S
i

r
i
q
1
(5.35)
We can consider dierent virtual motion generated by q
1
= q
2
= = q
n
= 0
for all the generalized coordinates except for q
j
= 1. For all the possible values
of j, we get n equations
N
i
m
i
v
i
r
i
q
j
=
N
i
F
S
i

r
i
q
j
(5.36)
The left hand side of the equation can be rewritten in terms of q
j
s
N
i
m
i
v
i
r
i
q
j
=
d
dt
_
T
q
j
_
T
q
j
=
N
i
F
S
i

r
i
q
j
(5.37)
which is identical to the Lagrange equation given by Eq. (5.25). Now one can
work backwards what we treated in the previous sections and dene Lagrangian
and come up with Hamiltons principle.
5.4 Conjugate Variables
Up to this point, we have treated how an appropriate choice of generalized
coordinates can simplify solving mechanical problems. According to such choice,
we have constructed generalized velocities and generalized forces. It is only
natural to suspect that there must be generalized momenta. In Lagrangian
mechanics, the generalized momentum corresponding to the coordinate q
j
is
dened as
p
j

L
q
j
(5.38)
It is also called the momentum conjugate to q
j
. In this case, the generalized co-
ordinate and the generalized momentum are said to be conjugate variables. The
Lagrangian equation can be rewritten in terms of the generalized momentum
L
q
j
=
dp
j
dt
(5.39)
From this, one can easily see that the generalized momentum, p
j
, is conserved
if the corresponding coordinate, q
j
is cyclic, i.e. absent from the Lagrangian.
In general, the Lagrangian is a function of q, q and t. The case where q
j
is
absent from Lagrangian is a special case of q
j
being cyclic with zero generalized
momentum. What is interesting here is that if the time t is cyclic, we get the
result that the energy of the system is conserved. In other words, energy and
time are conjugate variables.
Chapter 6
Hamiltonian Mechanics
We now turn our attention to Hamiltons formalization of mechanics. It should
be made clear from the very beginning that in terms of solving problems in
classical mechanics, Hamiltonian mechanics provide almost no advantage over
Lagrangian mechanics. In fact, in most cases, things appear to be more dicult
to solve using Hamiltonian mechanics. The benets of Hamiltonian mechanics
lie in interpretation of mechanical motion and as a bridge towards statistical
mechanics and quantum mechanics.
6.1 Conguration Space and Phase Space
In formulating the Lagrangian mechanics, we emphasized the importance of
nding n independent generalized coordinates, q = (q
1
, q
2
, , q
n
). We can vi-
sualize n-dimensional vector space spanned out by these generalized coordinates,
and such space is called the conguration space. The Lagrange equation yields
time dependent solutions for the q
j
(t)s. Since we can trace out the motions
of a mechanical system of interest in real space with an appropriate coordinate
transformations, r
i
(t) = r
i
(q
1
(t), q
2
(t), , q
n
(t)), for all N particles, any me-
chanical motion can be represented by a point q(t) = (q
1
(t), q
2
(t), , q
n
(t))
moving through the conguration space. As powerful as this method is, this is
not an eective method of visually representing a mechanical motion, as this
involves time evolution. Put in another way, a point in conguration space
cannot uniquely dene the subsequent motion.
To illustrate this point, let us look at an example of a simple harmonic
oscillator in one dimension. Here, there is only one generalized coordinate x,
which is identical to the real coordinate of the particle. Thus the conguration
space is simply one dimensional space represented by x. Its Lagrangian can be
written as
L =
1
2
m x
2
1
2
kx
2
(6.1)
and the Lagrange equation is
L
x

d
dt
_
L
x
_
= kx m x = 0 (6.2)
This is a second order dierential equation, and to determine the solution
uniquely, two initial conditions have to be specied, typically position and ve-
36
CHAPTER 6. HAMILTONIAN MECHANICS 37
locity of the particle at some moment in time. Here lies the disadvantage of
conguration space. Imagine at some moment in time t
0
, the particle occupies
the position x
0
in the conguration space. Just with this information alone, one
cannot tell where the particle is going to be after an innitesimal time interval,
t. One also needs to know how fast the particle is moving in which direction
at that moment t
0
.
To overcome this shortcoming, Hamilton formulated Hamiltons equation
and introduced the concept of phase space, a 2n-dimensional vector space spanned
out by generalized coordinates and generalized momenta. By doing so, one can
fully specify the motion of a system, if any one point in the phase space, (q
0
, p
0
),
is known for at some moment in time t
0
.
6.2 Hamiltons Equations
Earlier, we have shown that the by dening generalized momentum according
to Eq. (5.38), Lagrange equation can be given by Eq. (5.39)
p
j
=

q
j
L(q, q, t) (6.3)
Here the right hand side of the equation is, in general, a function of q, q and t,
whereas the left hand side of the equation contains the derivative of p
j
. To solve
this dierential equation, one needs to convert the RHS as a function of q, p,
and t. This can be achieved by obtaining the inverse function of the generalized
momentum, Eq. (5.38)
p
j

L
q
j
= f
j
(q, q, t) q
j
= g
j
(q, p, t) (6.4)
Then, we are basically left with n pairs of rst order dierential equations
p
j
=

q
j
L(q, g(q, p, t), t) and q
j
= g
j
(q, p, t) (6.5)
The proof is more of a mathematics problem than a physics problem, so we
will leave the proof either as an exercise for the readers, but it turns out that,
by dening Hamiltonian, H, using the Legendre transformation
H(q, p, t) = q p L(q, q, t) (6.6)
one can rewrite the above equations as
q
j
=
H
p
j
p
j
=
H
q
j
(6.7)
which are known as the Hamiltons equations. In other words, we have just
swapped out Lagrange equations, which are n second order dierential equa-
tions, with Hamiltons equation, 2n rst order dierential equations. Note that
the denition of Hamiltonian is identical to the energy function h from the
previous chapter.
Following the mathematics, it is not obvious at all why one would do this. As
a matter of fact, unless the Hamiltonian is given a priori, the process described
CHAPTER 6. HAMILTONIAN MECHANICS 38
above is not any less painful than the Lagrangian mechanics. To construct the
Hamiltonian from the Lagrangian, the function q
j
= g
j
(q, p, t) still needs to be
identied from Eq. (5.38) separately. Thus constructing Hamiltons equations
turns out to be a circular process, which seems to add no value. However, there
is a clear advantage in presenting the motion of a mechanical system in the
phase space over the conguration space.
Once again, we will return to the example of a simple harmonic oscillator.
We start from the Lagrangian
L =
1
2
m x
2
1
2
kx
2
(6.8)
and the generalized momentum is then p = m x, from which we can get x = p/m.
With this, we will formulate Hamiltonian
H = x
L
x
L =
p
2
2m
+
1
2
kx
2
(6.9)
The Hamiltons equations are, then,
x =
p
m
and p = kx (6.10)
One ends up solving the equation
x +
k
m
x = 0 (6.11)
and the general solutions are given by
x = Acos(t )
p = mA sin(t ) (6.12)
Since the phase space is a 2-dimensional x-p space, the above solutions can be
used as a set of parametric equations dening the paths in phase space. The
phase space representation of the above solutions is an ellipse given by
_
x
A
_
2
+
_
p
mA
_
2
= 1 (6.13)
Note that for any point in the phase space, there is only one ellipse that passes
through that point, which means that if any phase point is given at any moment
in time, subsequent and previous motions are uniquely dened.
Chapter 7
Central Force Motion
From the historical point, central force motion is one of the most important
problems in classical mechanics. It was peoples desire to understand planetary
motions and stellar objects that gave rise to much of Newtons nest works, and
it turns out that the planets and stellar objects are governed by gravity, one
of the two most representative cases of the central forces, along with Coulomb
force in electrodynamics.
Central force is a force with spherical symmetry that only depends on the
distance from the center of the force eld. In other words, a force eld F(r) is
said to be a central force eld with center O if it has the form
F(r) = F(r)r (7.1)
where r = [r[ and r = r/r. Gravity falls in this category as the force between
two massive objects with masses m and M is given by the relation
F(r) =
GmM
r
2
(7.2)
Note that central force is a function of position and thus is a conservative force
with potential, V (r), satisfying
F(r) =
dV (r)
dr
(7.3)
Also, due to the symmetry of the force eld, an objects motion under central
force eld is always planar, that is, the object is bound in a plane formed by
the center of the force eld, the position and the velocity vector of the object
under the inuence of the central force. Thus two dimensional polar coordinate
is sucient to describe the motion in its full detail.
7.1 Conservation Laws in Central Force Field
The central force conserves the systems angular momentum. This is because
there is no net torque acting on the particle as the force is always along the
direction of the moment arm,
N = r F = F(r)r r = 0 (7.4)
39
CHAPTER 7. CENTRAL FORCE MOTION 40
This can be also seen from the Lagrangian
L = T V =
1
2
m
_
r
2
+r
2

2
_
V (r) (7.5)
which is cyclic about , and the angular momentum p
is conserved,
p
=
L
= mr
2

= constant = l (7.6)
The angular momentum conservation can be rewritten as
=
l
mr
2
(7.7)
(t) =
_
t
0
l
m[r()]
2
d, (7.8)
which means that once the time dependence of r(t) is known, the time depen-
dence of (t) can be easily found out.
The second Lagrange equation is
F(r) +mr
2
m r = F(r) +
l
2
mr
3
m r = 0 (7.9)
This is an ordinary dierential equation for the radial distance r(t) which is
generally not solvable.
Instead of solving the dierential equation, we can obtain qualitatively de-
scription of the motion based on another feature of the Lagrangian. The La-
grangian is not an explicit function of t, and thus the energy function h must
be conserved.
h = r
L
r
+

L =
1
2
m
_
r
2
+r
2

2
_
+V (r) = constant = E (7.10)
Using Eq. (7.7), we can rewrite the above equation as
1
2
m r
2
+V (r) +
l
2
2mr
2
= E (7.11)
In fact, this is identical to Eq. (7.9) as taking the time derivative of the above
equation yields Eq. (7.9). One important feature of the above equation is that
the angular velocity component of the kinetic energy is reformulated as a func-
tion of r,
1
2
mr
2

2
=
l
2
2mr
2
(7.12)
and this allows us to dene the eective potential
V
eff
(x) = V (x) +
l
2
2mx
2
(7.13)
We can then treat the system as a one dimensional particle moving with kinetic
energy T =
1
2
m x
2
under the potential energy V
eff
(x):
1
2
m x
2
+V
eff
(x) = E (7.14)
with x > 0. Even if we cannot solve the equation exactly, one can easily dene
the range of motion in r by plotting the eective potential V
eff
(x) and E.
7.2 The Path Equation
Solving the Eqs. (7.8) and (7.9) can give us a full picture of the time dependent
motion. However, with r(t) and (t) given, we can use them to get parametric
equation r(), which is typically dubbed the orbit of the system.
Instead of having to nd r(t) and (t) to nd the orbit, one can combine
Eq. (7.9) and (7.7) to eliminate time. By introducing a new variable u = 1/r,
r =
l
2
u
2
m
2
d
2
u
d
2
(7.15)
r
2
=
l
2
u
3
m
2
(7.16)
and we can rewrite Eq. (7.9) as
d
2
u
d
2
+u +
mF(1/u)
l
2
u
2
= 0 (7.17)
This is the path equation. Only for inverse square law and inverse cube laws,
the path equation becomes a linear equation. Otherwise, this is a non-linear
equation. It is rather fortunate that the gravity is an inverse square law, which
can be solved analytically.
We will, in fact, take a look at the attractive inverse square law where
F(r) = /r
2
= u
2
. Then the path equation is a linear equation of the form
d
2
u
d
2
+u
m
l
2
= 0 (7.18)
The general solution of the equation is
1
r
= Acos( ) +
m
l
2
=
m
l
2
(1 +e cos( )) (7.19)
Since we are interested in the path of orbital motion, we can set to be zero
and, without losing generality, rewrite the equation
1
r
=
m
l
2
(1 +e cos ) (7.20)
The eccentricity e =
Al
2
m
is related to the initial condition and if e = 0, the path
is a circle. If e < 1, then the path is an ellipse. For e = 1 and e > 1, the results
are parabola and hyperbola respectively.
Also, by comparing the Eq. (7.20) to the standard polar equation of a conic
1
r
=
a
b
2
(1 +e cos ) (7.21)
we get the relation between the angular momentum l and the conic parameters
a and b,
l
2
= m
b
2
a
(7.22)
This relation is known as the L-formula. The total energy E and the conic
parameter a can be also shown to have the following relation, E-formula
Ellipse : E =
2a
< 0
Parabola : E = 0
Hyperbola : E =

2a
> 0
(7.23)
For a particle moving under the central force eld obeying the repulsive
inverse square law, the force is given by F(r) = /r
2
, and the path equation is
correspondingly,
d
2
u
d
2
+u +
m
l
2
= 0 (7.24)
whose solution can only be a parabola or hyperbola,
1
r
=
m
l
2
(1 +e cos ) . (7.25)
Chapter 8
System of Multiparticles
Unless youve been living in a cave all by yourself until you came to KAIST, Im
sure youve all seen people high jump at some point in your life. Now, when you
see high jumping events, theres that one question you cant get shaken o your
head. Why in the hell are these people jumping over a bar in that awkward
posture, twisting your body up in the air, facing backwards not being able to
look at where the bar is, and nally ipping backwards? The secret under this
rather bizarre collective behavior of a group of seemingly normal people in any
other regards lies within the center of mass. We will get back to high jumping
in a minute, and in the mean time lets talk about center of mass.
So far, weve been talking about a single particle with its mass all concen-
trated at a single point. But most of matters in real life are not point-like and
composed of many particles. How are we supposed to deal with such systems?
One rather intuitive (or counter-intuitive depending on your intuition) way is
to nd a single-particle-like representation of systems of particles. And that
representation is what we call center of mass.
8.1 Weighted Average
Lets suppose youve taken ve classes last semester: dierential equations (4),
classical mechanics (3), electromagnetism (3), physics lab (2), and English writ-
ing (2). The numbers in the parentheses are number of credits for each class.
Imagine youve got B, A, A, B, and C in each of these ve classes. What would
be your GPA for the semester? No one in the right mind would calculate your
GPA like (3.0 + 4.0 + 4.0 + 3.0 + 2.0)/5 = 3.2. Instead, what you do is the
following: (3.04+4.03+4.04+3.02+2.02)/(4+3+3+2+2) = 3.29.
This is a way of emphasizing classes that are more important, and the process
is known as calculating weighted-average.
When you have a collection of numbers x
i
s with each of their weight equal
to w
i
, the weighted average can be calculated following the equation,
x =
n
i=1
w
i
x
i
n
i=1
w
i
. (8.1)
This gives us an average value of a set of numbers after taking into account
the weight or importance of these values. By doing so, we can get a single-
43
CHAPTER 8. SYSTEM OF MULTIPARTICLES 44
valued representation of a set of numbers. When your GPA is 3.29 and someone
elses GPA is 3.01, we can safely assume that youve been a better student than
the other person without knowing the grades of every single class youve taken.
Of course, the person with lower GPA could have gotten a better grade than
you in some classes, but overall youve done better. It is a way of compressing
information.
8.2 Center of Mass
We can think of your grade as pure numbers, but we can also present it on a
number line. It gives us a visual and spatial representation, and your grade
becomes a position along a one-dimensional number line. See Fig. 9.1. Such
representation is helpful in understanding what center of mass is.
0 4.0 2.0 3.0 1.0
0 4.0 2.0 3.0 1.0
Figure 8.1:
Now, instead of taking an average of your grades, lets consider a system
of particles. For now lets just focus on the position of the particles along the
x-axis. There are n particles with each particle having mass m
, and position
x
where denotes each particle, i.e. = 1 through n. Instead of having to

specify each particles position, is there a good way to compress this information?
Can we treat this system of particles with a single number? If we want to treat
a collection of particles as a single entity, the sensible thing to do is think of a
particle with the total mass M =
n
=1
m
.
What about the position of the particle. Is there a single point representation
of all these particles? The answer is yes and that is quite literally in this case the
weighted-average position of all the particles. In other words, we can think of
this system of many particles as a single particle with total mass M positioned
at
x =
n
=1
m
n
i=1
m
=
1
M
n
=1
m
. (8.2)
We can do the same for y- and z-axes and write it in vector representation.
R =
1
M
n
=1
m
. (8.3)
This can further be generalized for a system with total mass M and a continuous
mass distribution, i.e. a system with mass density, (r), at the point r. The
summation then becomes an integral and the center of mass is
R =
_
(r)rdV
_
(r)dV
=
1
M
_
(r) r dV. (8.4)
We have thus far calculated the center of mass. Now the natural question that
follows is how is this useful? We will deal with this question in the next few
following sections.
8.3 Linear Momentum
Lets rst think about the linear momentum of a system of particles. Each
particle in the system would have its own momentum, and we will denote the
momentum of the th particle as p
= m
. If we are to calculate the total

momentum of the system with all n-particles in it, we can just sum over and
we get
P =
=
d
dt
=
d
dt
MR = M

R, (8.5)
which is to say:
I. The linear momentum of the system is the same as if a single particle of mass
M were located at the position of the center of mass and moving in the manner
the center of mass moves.
What about the time derivative of the momentum? Before we answer this
question, lets actually impose a couple of assumptions. When we have a col-
lection of particles, it is entirely possible that there is some type of force acting
on one particle by another. For example, if we have a bunch of electrons, there
will be Coulomb force acting between each pair of electrons. This is what we
call internal force and we will assume the following about the internal force:
1. The internal force acting on a particle by another particle is equal in
size and opposite in direction to the force acting on the particle by the
particle . That is f
= f
, where f
denotes the force acting on

by . This is in fact Newtons Third Law.
2. These two forces f
and f
lies on the straight line joining the two

particles. In the language that weve learned, the internal forces are central
forces.
With these assumptions in mind, we now turn to external force. It is con-
ceivable that there is a force acting on a particle that is not exerted by other
particles that constitute the system. Any force acting on a particle that is not
coming from within the system is what we call external force. We will denote
the external force as F
(e)
. In such a case, the total amount of force acting on

the particle , F
is F
= F
(e)
+ f
= F
(e)
. Here, f
is
the internal force from all the other particles s in the system. The equation
of motion for the particle is F
= m
, and by summing over we get the

equation of motion for the total system,
F =
F
(e)
=
f
= F
(e)
+
<
(f
+f
) = F
(e)
(8.6)
=
=
d
2
dt
2
=
d
2
dt
2
MR = M

R =

P (8.7)
The weak form of Newtons Third Law, f
= f
is invoked in Eq. (9.6).

What this equation of motion is saying is that:
II. The center of mass of a system moves as if it were a single particle of
mass equal to the total mass of the system, acted on by the total external force,
and independent of the nature of the internal forces (as long as they follow
f
= f
, the weak form of Newtons Third Law).

Combining the points we learned from I and II the following can be said:
III. The total linear momentum for a system free of external forces is constant
and equal to the linear momentum of the center of mass (the law of conservation
of linear momentum for a system).
This is precisely why your dad has to pull your sleigh on the ice. There is no
way for you to both sit on the sleigh and steer it into a certain direction without
you going opposite way. You and the sleigh as a whole is a system of two bodies
(not particles) and whatever forces you exert on the sleigh are, however you do
it, all internal forces and thus you just cannot move the sleigh in one direction
with you on it, which requires the center of mass to move.
8.4 Angular Momentum
Same type of analysis can be performed for angular momentum and energy
of these multi-particle system. For the case of angular momentum, the total
angular momentum, L, is sum of angular momentum of all the particles, L
=
r
, in the system. Lets rewrite the position vector of a particle, r
in
terms of the center-of-mass position vector, R: r
= R+r
. Then,
L =
(r
) =
(r
) =
[(R+r
) (

R+ r
)] (8.8)
=
[(R

R) + (r
) + (R r
) + (r

R)]. (8.9)
The last two terms in Eq. (9.9) vanishes, because
(r
R) =
= MRMR = 0. (8.10)
This just means that the position of the center of mass in the center-of-mass
coordinate system is a null vector.
The total angular momentum is then simply
L =
[(Rm

R) + (r
)] = RP+
(r
). (8.11)
Our fourth important point is then:
IV. The total angular momentum about an origin is the sum of the angular
momentum of the center of mass about that origin and the angular momentum
of the system about the position of the center of mass.
Now we turn to the time derivative of angular momentum. From L
=
r
= r
+r
= r
= r
_
_
F
(e)
_
_
(8.12)
L =
(r
F
(e)
) +
=
(r
) (8.13)
=
(r
F
(e)
) +
<
(r
+r
) (8.14)
=
(r
F
(e)
) +
<
(r
) f
. (8.15)
r
is a position vector connecting the th particle and th particle, and

therefore is in the same direction as f
if the internal force is a central force.

In such a case, the second term in Eq. 9.15 vanishes and we are left with
L =
(r
F
(e)
) = N
(e)
= N
(e)
. (8.16)
N
(e)
is just an external torque exerted on the th particle, r
F
(e)
, and from
this we can conclude that:
V. If the net resultant external torques about a given axis vanish, then the total
angular momentum of the system about that axis remains constant in time.
8.5 Energy of the System
In chapter 2, weve learned that the total energy, E = T + U, of a particle in
a conservative force eld is a constant, where T is the kinetic energy and U is
the potential energy. Is this true for a system of many particles? If the system
gets rearranged from a certain conguration 1 to a dierent conguration 2 by
the total force acting on the system, that means some work, W
12
, is done on
the system,
W
12
=
_
2
1
F
dr
_
2
1
F
(e)
dr
=
_
2
1
f
dr
(8.17)
=
_
2
1
F
(e)
dr
<
__
2
1
f
dr
+
_
2
1
f
dr
_
(8.18)
=
_
2
1
dr
<
__
2
1
dr
+
_
2
1
dr
_
(8.19)
=
_
2
1
dU
<
_
2
1
d
=
_
_
(
<
U
)
_
_
2
1
(8.20)
Going from Eq. 9.18 and Eq. 9.19, we used the assumption that the force eld
is conservative, that is, the forces F
(e)
and f
are derivable from potential

functions U
and

U
, respectively. We can dene the total potential energy

U = (
<
U
), and we get
W
12
= U
1
U
2
. (8.21)
In terms of kinetic energy, the work done on the system is
W
12
=
_
2
1
d
_
1
2
m
v
2
_
= T
2
T
1
. (8.22)
where T =
1
2
m
v
2
. Combining Eq. 9.21 and Eq. 9.22, we get W

12
=
U
1
U
2
= T
2
T
1
and thus E = U
1
+T
1
= U
2
+T
2
. This means:
VI. The total energy for a conservative system is constant.
The kinetic energy, when written out, is
T =
1
2
m
v
2
1
2
m
[ r
[
2
=
1
2
m
(

R+ r
) (

R+ r
)(8.23)
=
1
2
m
(

R

R+ r
+ 2

R r
) (8.24)
=
1
2
m
V
2
+
1
2
m
v
2
+

R
d
dt
(
) (8.25)
=
1
2
MV
2
+
1
2
m
v
2
(8.26)
which can be stated:
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 the velocity of the center of mass
and the kinetic energy of motion of the individual particles relative to the center
of mass.
Finally we return to the problem of high jumping. A human body, of course,
is a system of particles. When we high jump, we kick o the ground and gain
some momentum in vertical direction. The jumper will have kinetic energy
associated with that momentum and it will allow the jumpers center of mass
to reach a certain height. Ignoring small amount of kinetic energy associated
with moving body parts here and there, the nal height of the jumpers center
of mass is more or less given by the amount of momentum he or she picks up
at the time of kicking o the ground. So the most eective way to high jump
would be to nd a way to put your center of mass outside your body and have
your body parts actually role around the center of mass. That is exactly what
these high jumpers are doing.
The most traditional method known as the scissors technique looks quite
natural and easy, but the center of mass actually has to clear the height of the
bar and thus is not very eective. Straddle technique or Fosbury op are more
advanced techniques and lets you clear higher bars because the center of mass
actually goes under the bar, and therefore widely used these days.
Chapter 9
Collision Theory
9.1 Elastic Collisions of Two Particles
For the time being, we return to two body problems and discuss elastic collisions
between two particles. Although there are rarely strictly two particle collision
in our real world, there are plenty of collision events that can be approximated
to two particle collision, starting with billiard, a number of sports including
baseball, car crashes, and etc. In all of these, there is a coordinate system that
we are sitting in and observing these events. This is what we can be labelled
as a laboratory coordinate system (LAB system) in practice, however, for this
particular exercise, we will refer to the LAB system as the coordinate system in
which the target particle is at rest.
In contrast to the LAB system, we will also consider a coordinate system
commonly known as the center-of-mass system (CM system). This is a coor-
dinate system that moves with the center of mass of two particles so that the
center-of-mass position is at rest.
Lets suppose a particle with mass m
1
is moving towards a particle with
mass m
2
in the LAB system. The particle 1 is moving with velocity u and the
particle 2 is at rest. After the collision, the particle 1 and particle 2 ies o
with velocities u
1
and u
2
, respectively. This process is shown in Fig. 9.1(a).
The center-of-mass velocity is not zero in the LAB system and we will denote
it with V.
The interaction between the two particles, either by direct contact or some
force eld between them, are strictly internal and therefore the center of mass
momentum does not change before and after the collision. From Eq. (9.5) and
conservation of momentum, we get the relation
P = MV = (m
1
+m
2
)V = m
1
u = m
1
u
1
+m
2
u
2
. (9.1)
where u
1
and u
2
are the after-the-collision velocities of particle 1 and 2, respec-
tively. The center-of-mass velocity is thus V =
m
1
m
1
+m
2
u. What we want to
know is how the particles move after the collision. In other words, we need to
nd out what u
1
and u
2
are. For this, lets turn to the CM system.
The advantage of describing the motion in the CM system lies in the fact
that the system as a whole, represented by the center of mass, is not moving. In
other words, the total momentum of the system is zero, and thus, the momentum
49
CHAPTER 9. COLLISION THEORY 50
u
u1
V
u2
V
m1 m2

1
2
v1 v2
m1 m2
v1
v2
u
V
(a)
(b)
v1
v2
Figure 9.1:
of the two particles are always symmetric, that is, they are always of the same
magnitude in the opposite direction, before or after the collision.
Since the center of mass of the two particles is moving at velocity V with
respect to the LAB system, the CM system must be moving at V relative to
the LAB system. In the CM system, the two particles are initially moving with
velocities
v
1
= u V =
m
2
m
1
+m
2
u (9.2)
v
2
= 0 V =
m
1
m
1
+m
2
u (9.3)
As mentioned earlier, the total momentum in the CM system is P = m
1
v
1
+
m
2
v
2
= 0, which is consistent with the fact that the center of mass is at rest.
After the collision, the particles are moving at v
1
and v
2
. From conservation of
momentum, m
1
v
1
+m
2
v
2
= m
1
v
1
+m
2
v
2
, and conservation of kinetic energy
m
1
v
1
2
+m
2
v
2
2
= m
1
v
2
1
+m
2
v
2
2
, we get the relation
v
1
= v
1
=
m
2
m
1
+m
2
u (9.4)
v
2
= v
2
=
m
1
m
1
+m
2
u = V (9.5)
with v
1
and v
2
in the opposite directions. This is shown in Fig. 9.1(b). By the
same arguments that led us to Eq. (9.2) and Eq. (9.3), now u
1
and u
2
can be
determined.
u
1
= v
1
+V (9.6)
u
2
= v
2
+V (9.7)
The vector diagrams Fig. 9.1 represents this relation. Finally, lets nd out the
relation between the deection angles of particle 1,
1
, particle 2,
2
in the LAB
system and the deection angle of particle 1 and 2, in the CM system. Note
that the deection angle for two particles is the same in the CM system and
cannot exceed 2. See Fig. 9.1. Between
1
and the following relation holds:
tan
1
=
v
1
sin
V +v
1
cos
=
sin
(V/v
1
) + cos
=
sin
(m
1
/m
2
) + cos
. (9.8)
Here Eq. (9.4) and Eq. (9.5) are used to obtains the relation V/v
1
= m
1
/m
2
.
Similarly for
2
,
tan
2
=
v
2
sin
V v
2
cos
=
sin
(V/v
2
) cos
=
sin
1 cos
= cot(/2). (9.9)
Otherwise, from the geometry, one can easily show that
2
=
1
2
( ) (9.10)
which is an identical result.
9.2 Kinematics of Elastic Collisions
Lets focus on the energy relationship between the particles. In the LAB system,
the total energy of the particles, E
0
is conserved and is equal to
E
0
=
1
2
m
1
u
2
=
1
2
m
1
u
2
1
+
1
2
m
2
u
2
2
= E
1
+E
2
. (9.11)
The ratio between E
2
and E
0
is
E
2
E
0
=
1
2
m
2
u
2
2
1
2
m
1
u
2
=
m
2
u
2
2
m
1
u
2
. (9.12)
From the relation, (see Fig. 9.1)
v
2
= 2V sin(/2) (9.13)
we get
E
2
E
0
=
4m
1
m
2
(m
1
+m
2
)
2
sin
2
(/2) (9.14)
9.3 Inelastic Collisions
In the real world, not all collisions are elastic. In such a case, total kinetic
energy of the particles is not conserved, and energy is conserved through
Q+
1
2
m
1
u
2
1
+
1
2
m
2
u
2
2
=
1
2
m
1
v
2
1
+
1
2
m
2
v
2
2
(9.15)
where Q is called the Q-value and represents the energy loss or gain in the
collision.
Q = 0: Elastic collision, kinetic energy is conserved Q > 0: Exoergic col-
lision, kinetic energy is gained Q < 0: Endoergic collision, kinetic energy is
lost
An inelastic collision is an endoergic process, through which a part of kinetic
energy is converted to mass, heat, sound energy and etc.
There is a fairly simple relation between particle velocities in the direction
normal to the plane of contact that allows us to determine if a collision is elastic
or not. When the ratio known as the coecient of restitution
=
[v
2
v
1
[
[u
2
u
1
[
(9.16)
is equal to one, the collision is perfectly elastic. When = 0, the collision is
totally inelastic.
Any collision whether elastic or inelastic causes momentum of each particle
(but not of the two-particle system as a whole) to change. This change in
momentum is due to the impulsive forces F =
d
dt
(mv) acting on the particles
throughout the time period t = t
2
t
1
of the collision. We dene impulse P
as
_
t
2
t
1
Fdt P.
9.4 Scattering Cross Section
So far we have discussed kinematics of scattering, i.e. relationship between the
initial state before a collision and the nal state after it. The details of the
interaction between two colliding particles were not taken into account. We
now focus on that missing detail which species the kinematic relation. Despite
our tendency to think of a collision as two objects coming in direct contact and
bouncing o, such direct contact is not required for a collision or scattering
to take place. A force eld between two particles can cause the particles to
deect from their original trajectory and, in general, any such interactions can
be called scattering. In fact, even the collision through direct contact can be
understood with the force eld that only acts at a single distance of hard-core
repulsion. Here we impose one assumption that any force eld between these
colliding particles are vanishingly small at large distances, r .
Imagine, as shown in Fig. 9.3, the incident particle, m
1
, moving towards the
target particle, m
2
. Without any interacting force eld between the particles,
m
1
may get as close as the distance b away from m
2
but may not actually come
to contact with m
2
. If a repulsive interaction, such as Coulomb interaction
between the same-charge particles, is present, m
1
will be deected away from
the original path and the minimum distance between the two particles will be
greater than b. Nonetheless, this imaginary minimum distance, b, between the
two particles is what we call the impact parameter. The impact parameter has
a direct relation to the deection angle and our task is to nd that relationship.
In this section we will study the collisions due to central force eld, and it is
helpful to rst consider the collision in the CM system. When the internal forces
between the colliding particles is central, the force vector lies along the straight
line connecting the two particles. In the CM system, there is an advantage that
the trajectories of particles both before and after the collision are also on this
straight line in Fig. 9.1(b) where as in the LAB system, the trajectories of the
m
1
m
2
u
b
1
Figure 9.2:
particles are at some angle
1
+
2
and are not along the direction of the internal
force.
We can simplify the matter a bit further by reducing the two body problem
into a one body problem. Two interacting particles with mass m
1
and m
2
is
equivalent to a single particle with the reduced mass =
m
1
m
2
m
1
+m
2
under the
identical force eld.
Imagine we are shooting a beam of particles with mass m
1
is moving at a
velocity u in the LAB system towards the target with mass m
2
. The angular
momentum about the center of mass in the CM system is then
l = m
1
v
1
b
1
+m
2
v
2
b
2
(9.17)
= m
1
v
1
m
2
m
1
+m
2
b +m
2
v
2
m
1
m
1
+m
2
b = (v
1
+v
2
)b (9.18)
= ub (9.19)
and the total energy in the CM system is
E
0
= T
0
=
1
2
m
1
v
1
2
+
1
2
m
2
v
2
2
(9.20)
=
1
2
m
1
_
m
2
m
1
+m
2
u
_
2
+
1
2
m
2
_
m
1
m
1
+m
2
u
_
2
(9.21)
=
1
2
u
2
. (9.22)
We get the relation between l and E
0
as
l = b
_
2E
0
. (9.23)
Even if we try our best, the beam will not be a single line, but will be a
bundle that has some cross sectional area. In such a case, it is useful to dene
ux density, I, as the number of incident particles per unit area per time.
Once these incident particles come under the inuence of the force eld, they
will scatter in dierent directions. We can pick out an element of solid angle d
about a scattering angle and start counting the number of scattered particles,
dN, that fall into d
. We can then dene a dierential scattering cross

section, () as
()d
=
dN
I
(9.24)
() is identied as cross section because it has the dimensions of an area, as
shown in Fig. 9.4. Since the force eld is assumed to be central, the scattering
process is symmetric about the azimuthal axis, and therefore the solid angle
element is simply d
= 2 sin d. The number of particles scattering into

d
about is then dN = I () 2 sin d.

Figure 9.3: Scattering of an incident beam of particles by a center of force.
Figure from Classical Mechanics by Goldstein, Poole and Safko.
For a well-dened force eld, there should be one-to-one correspondence
between the impact parameter b and the scattering angle . In other words,
an incident particle ying in with a certain impact parameter b will always
scatter o at an angle . So the number of particles scattered into the solid
angle element d
about must be equal to the number of incoming particles

through a small range db about the impact parameter b. (See Fig. 9.4.)
I dA = I 2 b [db[ = I () 2 sin [d[ (9.25)
From this, we obtain the relation between the scattering cross section and the
impact parameter
() =
b
sin
db
d
. (9.26)
However, we need to know [db/d[ to actually calculate the scattering cross
section, and for this, we turn to something we learned in Chapter 7. We found
in the previous chapter that the change in angle for a particle of mass moving
in a central-force eld potential V (r) was given by
=
_
r
max
r
min
(l/r
2
)dr
_
2[E V (r) (l
2
/2r
2
)]
. (9.27)
In this case, the orbit is a particle at an innite distance, i.e. r
max
= ,
approaching the center of the force eld and then moving away to an innite
distance as shown in Fig. 9.4. Simple geometric calculation leads to = 2.
Combining Eq. (9.23) and (9.27) , we get
= 2
_

r
min
(b/r
2
)dr
_
1 (b
2
/r
2
) (V/E)
. (9.28)
with E = T
0
since at an innite distance, the potential energy V = 0 and thus
the total energy is equal to the original kinetic energy.
b
Scattering
center
Figure 9.4:
By inverting = (b), we can calculate b = b() and the dierential scatter-
ing cross section () can be expressed purely as a function of with Eq. (9.26).
However, the cross section we have calculated is in the CM system, and
most of the observations are made in the LAB system. We need to make a
transformation from the CM coordinate system to the LAB coordinate system.
Because the total number of particles scattered into a unit solid angle must be
the same in the LAB system as in the CM system, we have
()d
= (
1
)d (9.29)
()2 sin d = (
1
)2 sin
1
d
1
(9.30)
Here
1
and represent the same scattering angle in two dierent coordinate
system, i.e. the LAB system and the CM system, respectively. Similarly d
and d
represent the same element of solid angle in the two coordinate systems.
Thus (
1
) is the dierential cross sections for the scattering in the LAB system.
(
1
) = ()
sin
sin
1
d
d
1
= ()
d cos
d cos
1
(9.31)
The bar over indicates that the angle dependence or the functional form of
(
1
) is not identical to that of (). Eventually, one should be able to derive
d cos
d cos
1
=
d
d cos
1
_
_
m
1
m
2
_
_
sin
2
1
cos
1
_
m
2
m
1
_
2
sin
2
1
_
_
_
_
(9.32)
=
[ cos
1
_
1
2
sin
1
]
2
_
1
2
sin
1
(9.33)
where m
1
/m
2
. Also by inverting the Eq. (9.8), we can derive =
sin
1
( sin
1
) +
1
. Substituting for this relation in () for Eq. (9.31),
we get (
1
) purely as a function of
1
.
9.5 Rutherford Scattering Formula
We take a look at a very special case of the particle scattering we have discussed
so far. Hans Geiger and Enest Marsden in 1909 struck a gold foil with a beam
of -particles, i.e. He
2+
, under the direction of Ernest Rutherford. Although it
was not fully understood at the moment, the interaction between the -particles
and the gold foil is mostly due to Coulomb interaction between the -particles
and the positively charged nuclei of gold obeying the relation U =
k
r

q
1
q
2
4
0
r
with q
1
and q
2
the amounts of charge that the two interacting particles carry.
Substituting U = k/r in Eq. 9.63, we get
= 2
_

r
min
(b/r)dr
_
r
2
(k/E)r b
2
. (9.34)
Carrying out the integration in the right hand side with
k
2E
results in the
relation
cos =
/b
_
1 + (/b)
2
(9.35)
Rewriting Eq. 9.68 as tan = b/ and using the relation = 2, b is
expressed in terms of : b = k cot(/2). The scattering cross section is
() =

2
2
cot(/2)
sin sin
2
(/2)
=

2
2
1
sin
4
(/2)
=
k
2
(4E)
2
1
sin
4
(/2)
(9.36)
Chapter 10
Rigid Body Kinematics
In this chapter, we will turn our attention away from particles and focus on rigid
bodies. The rigid body is dened to be a system in which the distance between
any pair of composing particles is xed. For a rigid body, the main feature in
understanding its dynamics heavily relies on distinguishing translational and
rotational motions.
10.1 Rotation and Linear Velocity
In two dimensional space, rotation axis is always perpendicular to the plane
of motion, and angular velocity can be treated as a scalar. However, in three
dimensional space, the rotation axis can take arbitrary direction, say n and it is
useful to identify (right-handed) angular velocity as a vector along the rotation
axis,
= n (10.1)
We will leave proving that the angular velocity can be represented by a vector
as an exercise for the readers. The fact that the angular velocity is a vector
has a rather interesting implications, which is that the angular velocity can be
decomposed into two or more vector components:
=
1
+
2
. (10.2)
At rst glance, this might seem like a rather unintuitive and unnecessary prop-
erty of the angular velocity, but it will become extremely useful later.
With the angular velocity represented as a vector, the linear velocity of any
point, P, within the rigid body can be obtained by
v = (r b) (10.3)
where r is the position vector of P and b is the position vector of any xed
point, B, on the rotation axis. If the point B has translational velocity v
B
,
then the linear velocity of the point P should be rewritten as
v = v
B
+ (r b) (10.4)
57
CHAPTER 10. RIGID BODY KINEMATICS 58
Figure 10.1:
10.2 Inertia Tensor
Imagine a rigid body, composed of multiple particles, rotating about a xed
point O with angular velocity . The i-th particle within the rigid body would
have instantaneous velocity v
i
= r
i
about O. We will assume that there is
no translational motion here. The kinetic energy of the i-th particle is given by
T
i
=
1
2
m
i
v
2
i
=
1
2
m
( r
i
)
2
(10.5)
and the total kinetic energy is
T =
1
2
i
m
i
( r
i
)
2
=
1
2
i
m
i
_
_
_
__
r
2
i,
_
r
i,
_
_
_
r
i,
_
_
_
_
(10.6)
where the relation (A B)
2
= A
2
B
2
(A B)
2
was used to get the second
line from the rst line in the above equation. r
i
with = x, y, z denote three
Cartesian coordinate components, x
i
, y
i
and z
i
of r
i
.
Now, we can write
, so that
T =
1
2
m
i
_
r
2
i,
_
r
i,
r
i,
_
(10.7)
=
1
2
i
m
i
_
r
2
i,
r
i,
r
i,
_
(10.8)
and we can dene I
as
I
i
m
i
_
r
2
i,
r
i,
r
i,
_
(10.9)
so as to obtain the relation
T =
1
2
(10.10)
This is the general form for the rotational kinetic energy for a three dimensional
rigid body. The tensor I
is referred to as the moment of inertia tensor and

it represents the moment of inertia associated with the angular velocity in the
-th component that contributes to angular momentum in the -th component,
as shall be seen in the next section.
10.3 Angular momentum
Once again, in the body coordinate system where one or more points are xed,
the angular momentum of the body is simply
L =
i
r
i
p
i
(10.11)
where r
i
is the position of i-th particle relative to the xed point. In such a
coordinate system, the linear momentum of the i-th particle, p
i
can be expressed
as p
i
= m
i
r
i
and therefore
L =
i
m
i
r
i
( r
i
). (10.12)
Using the vector identity A(BA) = A
2
BA(A B), the -th component
of L is expressed as
L
i
m
i
_
_
r
2
i,
r
i,
r
i,
_
_
(10.13)
=
i
m
i
r
2
i,

r
i,
r
i,
_
(10.14)
=
i
m
i
_
r
2
i,
r
i,
x
i,
_
(10.15)
=
. (10.16)
or in matrix notation, L = I . Here, L and are three dimensional vectors
representing the angular momentum and angular velocity, respectively, and I is
a three-by-three matrix representation of the moment of inertia tensor.
From the Eq. (10.16), we can see that the angular momentum along the
r
-axis is a result of rotation along all three x, y and z-axes,

x
,
y
and
z
. In
other words, a general rotation given by an angular velocity in three dimensional
space is not necessarily colinear with the angular momentum. In describing a
rotational motion, the moment of inertia must be able to identify such non-
colinearity, which is done by utilizing tensor representation.
Eq. (10.16) has the following relation with the kinetic energy T,
1
2
=
1
2
= T (10.17)
Using the matrix form, this can also be expressed as,
T =
1
2
L =
1
2
I . (10.18)
x
y
z
L
Figure 10.2: Angular velocity and angular momentum representing a general
rotation of a dumbell.
10.4 Principal Axes of Inertia
Things are considerably simpler to describe rotation in two dimensions. Both
angular velocity and angular momentum are in the same direction, perpendic-
ular to the two dimensional surface and the moment of inertia is always scalar.
In three dimensional space, this is generally not the case. Only when certain
symmetry conditions are met, rotation of a rigid body can be described in a
manner similar to that of two dimensions.
Imagine a dumbbell connected by masses m
1
and m
2
at the ends of its
shaft as shown in Fig. 10.8. In general, the angular velocity and the angular
momentum L are not collinear, i.e. they are not along the same direction. In
other words, and L are not connected by a scalar number. There are however,
two distinct cases in which and L are collinear. One case is when the rotation
axis is along the shaft of the dumbbell, and the other is when the rotation axis
is perpendicular to the dumbbell. These particular set of axes are called the
principal axes of inertia. In general physics, rotation was always thought to be
taking place along these axes. Thus, the relation between the angular velocity
and angular momentum could be expressed as
L = I (10.19)
with I as a scalar number representing the moment of inertia for the rotation
along the principal axis.
The relation Eq. (10.19) can be used to nd the principal moment of axes
for a rigid body within the given coordinate system. When a body is rotating
about a principal axis, we set the angular momentum vector L = I to be equal
to L = I .
_
_
_
L
x
= I
x
= I
xx
x
+I
xy
y
+I
xz
z
L
y
= I
y
= I
yx
x
+I
yy
y
+I
yz
z
L
z
= I
z
= I
zx
x
+I
zy
y
+I
zz
z
(10.20)
We can rewrite the equations as follows
_
_
_
(I
xx
I)
x
+I
xy
y
+I
xz
z
= 0
I
yx
x
+ (I
yy
I)
y
+I
yz
z
= 0
I
zx
x
+I
zy
y
+ (I
zz
I)
z
= 0
(10.21)
For
s to have non-trivial solutions, the determinant of the coecients has to

be zero.

I
xx
I I
xy
I
xz
I
yx
I
yy
I I
yz
I
zx
I
zy
I
zz
I
= 0 (10.22)
The equation is third order equation with respect to I and yields three values
of I: I
A
, I
B
and I
C
. For each value of I
s ( = A, B, C), we can nd the ratio

between
x
,
y
and
z
by solving the equation
_
_
_
(I
xx
I
x
+I
xy
y
+I
xz
z
= 0
I
yx
x
+ (I
yy
I
y
+I
yz
z
= 0
I
zx
x
+I
zy
y
+ (I
zz
I)
z
= 0
(10.23)
This ratio denes the direction of
along which the moment of inertia has the

scalar value I
. In other words, the direction dened by s are the principal

axes of inertia. In linear algebraic terms, this is the eigenvalue and eigenvector
problem, where the principal axes represented by
are eigenvalues associated

to the principal moment of inertia, the eigenvectors.
Using the fact that the moment of inertia tensor is symmetric one can show
that the principal axes are orthogonal to one another, that is,
= 0 for
,= . Strictly speaking, if two or three of I
s have the same value,
s need
not be orthogonal to each other. Nonetheless, one can always nd orthogonal
set of
s that satises Eq. (10.23). We will leave it to the readers to prove this
point. (The proof falls into category of linear algebraic problem than a classical
mechanics problem.)
Because we now have three orthogonal axes as our princial axes, we can use
them as our bases of a coordinate system. In other words, we can represent
any vector as a linear combination of three vectors along e
1
, e
2
and e
3
where
e
1
|
A
, e
2
|
B
and e
3
|
C
.
An important feature of the principal axes is that if we choose them as our
basis for the coordinate system, the moment of inertia tensor becomes diagonal,
that is
I =
_
_
I
11
I
12
I
13
I
21
I
22
I
23
I
31
I
32
I
33
_
_
=
_
_
I
A
0 0
0 I
B
0
0 0 I
C
_
_
(10.24)
where the subscripts of I
ij
s come from the indices of e
i
. Therefore the rotation
presented within this coordinate system can be expressed through the following
relations
T =
1
2
i
I
ii
2
i
(10.25)
L
i
= I
ii
i
(10.26)
which are considerably simpler than Eq. (10.10) and (10.16). Because the co-
ordinate system is geometrically locked to the rigid body, such a coordinate
system is referred to as the body coordinate. Note that the body coorinate axes
e
i
s themselves are in motion, if the rigid body is in motion.
10.5 Moments of Inertia for Dierent Body Co-
ordinate Systems
There are cases when the moment of inertia tensor is given in a particular set
of body coordinate system, e.g. with the origin of the coordinate system at the
center of mass of a rigid body. However, if the body is rotating with the pivot
point of rotation not at the center of mass, it is helpful to understand how inertia
tensor transforms from the center-of-mass coordinate system to a dierent body
coordinate system. Lets denote three orthogonal axes of the center-of-mass
coordinate system with r
i
s. If we pick another body coordinate system with a
set of axes R
i
s that has same orientation as the center-of-mass coordinate and
is separated by distance a, the translational displacement between the two body
coordinate systems is then a, i.e. R = a +r.
If we substitute each component R
i
with a
i
+ x
i
, for inertia tensor in the
second body coordinate system, J
ij
, we get
J
ij
=
ij
k
X
2
,k
X
,i
X
,j
_
(10.27)
= I
ij
+M(a
2
ij
a
i
a
j
) (10.28)
where I
ij
is the original moment of inertia tensor dened at the center-of-mass
body coordinate system and M is the total mass of the rigid body. We will
leave the detailed calculation as an exercise for the readers.
The second term in the above equation, M(a
2
ij
a
i
a
j
) is simply moment
of inertia of a point particle sitting at a poition a with respect to the origin.
In other words, the moment of inertia tensor in a dierent coordinate system is
the sum of original moment of inertia tensor and the moment of inertia due to
a particle with mass M. It can be also shown from this that the orientation of
the body coordinates are not aected if the displacement vector a along one of
the three principal axes.
10.6 Eulers Equations
Now that we have established the methods for representing rotational dynamics,
we will set up equations of motion:
F =
dP
dt
(10.29)
N
CM
=
dL
CM
dt
(10.30)
where the rst equation is for the translational motion of the center of mass, and
the second equation for the rotational motion about the center of mass. If the
rigid body has a xed pivot O dierent from the center of mass, the equation
of motion can be reduced to a single torque equation
N =
dL
dt
(10.31)
We will rewrite this equation within the body coordinate system. With =
1
e
1
+
2
e
2
+
3
e
3
, the angular momentum is expressed as
L = I
A
1
e
1
+I
B
2
e
2
+I
C
3
e
3
(10.32)
and the equation of motion is then
N =
d
dt
_
I
A
1
e
1
+I
B
2
e
2
+I
C
3
e
3
_
= I
A

1
e
1
+I
B

2
e
2
+I
C

3
e
3
+I
A
1
e
1
+I
B
2
e
2
+I
C
3
e
3
(10.33)
Note that the coordinate axes are locked to the rigid body, and thus the moment
of inertia tensor is unaltered within this coordinate system. Instead, the coor-
dinate axes e
i
s are time dependent. Because the coordinate axes are changing
with time due to the rotation of the rigid body, the following relation holds,
from Eq. (10.3),
e
i
= e
i
(10.34)
By plugging =
1
e
1
+
2
e
2
+
3
e
3
into the above equation, Eq. (10.33) is
reduced to
N
1
= I
A

1
(I
B
I
C
)
2
3
N
2
= I
B

2
(I
C
I
A
)
3
1
N
3
= I
C

3
(I
A
I
B
)
1
2
which is known as the Eulers equations.
10.7 Free Body Rotation of a Rigid Body with
Axial Symmetry
A general motion of a rigid body is extremely complicated to solve, but there
are a few cases where the solution comes out in a rather straightforward fashion.
One of such problems is the rotation of an axisymmetric body with respect to
the center-of-mass in the absence of external torque. Here an axisymmetric body
(or a body with axial symmetry) means a body with a cylindrical symmetry, so
that I
A
= I
B
.
The angular velocity, , expressed in terms of the axial unit vector, e
3
, is
=
3
e
3
+e
3
e
3
(10.35)
The rst term represents the angular velocity component along the symmetric
axis, e
3
and the second term is the angular velocity component perpendicular
to the symmetric axis. Thus the angular momentum can be written as
L = I
A
e
3
e
3
+I
3
e
3
= constant (10.36)
where the rst term comes from the axial symmetry. Because the system has
cylindrical symmetry, we can choose any axis that is perpendicular to e
3
as a
principal axis, and the moment of inertia along such axis is equal to I
A
. Since
there is no external torque acting on the system, the angular momentum L must
be conserved.
Because the angular momentum is conserved, the following relation is also
true:
e
3
dL
dt
= I
C

3
= 0 (10.37)
which implies that the axial component of the angular velocity,
3
does not
change with time, even though the symmetric axis itself is not constant.
Taking the dot product between e
3
and L results in
e
3
L = I
C
3
= constant (10.38)
Thus the angle between the angular momentum and the symmetric axis is also
constant in time.
This time, we will take the cross product between e
3
and L and get
e
3
L = I
A
e
3
(e
3
e
3
) +I
C
3
e
3
e
3
= I
A
e
3
(10.39)
By rearranging the above equation, the following can be identied
e
3
=
L
I
A
e
3
= e
3
(10.40)
where L/I
A
. This equation states that the symmetry axis e
3
precesses
around the angular momentum axis L with constant angular speed L/I
A
.
We will rewrite the angular velocity as
= e
3
e
3
+
3
e
3
= e
3
_
L
I
A
e
3
_
+
e
3
L
I
C
e
3
=
L
I
A
+e
3
L
_
I
A
I
C
I
A
I
C
_
e
3
(10.41)
This can be decomposed as angular velocity of the symmetric axis e
3
about
the xed angular momentum axis L and the angular velocity of the rigid body
about the symmetric axis e
3
. In other words, the rigid body rotates about e
3
with an angular speed e
3
L
_
I
A
I
C
I
A
I
C
_
and the rotating body precesses about
the xed axis with an angular speed L/I
A
.
The same result can be derived by solving the Eulers equations from the
previous section. One can also show that three vectors L,
3
and e
3
are all on
the same plane, which is a good vector calculus exercise for the readers.
10.8 Precession of a Symmetric Top due to a
Weak Torque
Consider a symmetric top whose inertia tensor takes the form
I =
_
_
I
A
0 0
0 I
A
0
0 0 I
C
_
_
(10.42)
relative to the tops principal axes, e
1
, e
2
and e
3
. Here e
3
is the symmetry axis,
and e
1
and e
2
can be chosen to be any two orthogonal axes perpendicular to
e
3
. See Fig. ??.
CM
Mg
R
e
3
z
Figure 10.3: A weak torque acting on a spinning top.
If the rotation of the body is about the axis of symmetry, i.e. =
3
e
3
, the
angular momentum is simply,
L = I
C
= I
C
3
e
3
(10.43)
Torque due to gravity for such a top is
= RMg =

L = I
C
3
e
3
(10.44)
where R is the center of mass position with respect to the xed pivot point O
and g is the gravity along the negative z-axis, i.e. g = g z. The total mass of
the top is given by M. We can solve for e
3
,
e
3
=
MgR
I
3
3
z e
3
= e
3
(10.45)
where
=
MgR
I
3
3
z (10.46)
This again implies that the spinning top will precess with the angular speed
=
MgR
I
3
3
. An implicit condition here is that the precession frequency has
to be substantially smaller than
3
. Otherwise, the angular velocity of the
precession alters the total angular velocity of the spinning top suciently that
the angular velocity will no longer be along the symmetry axis e
3
. This violates
the initial assumption of our problem.
10.9 Steady Precession of a Symmetric Top un-
der a Uniform Torque
From what we have discussed up to this point, we can nd the general condition
for a steady precession of a spinning top under gravity. One can easily derive
the following equation of motion
d
dt
_
I
A
e
3
e
3
+I
C
3
e
3
_
= MgR(e
3
z) (10.47)
Steady precession is a precession of a top in which the angle, , between the
axial vector e
3
and the vertical z and the precession frequency are xed, i.e.
e
3
z = cos = constant and = constant. In such steady precession, the rate
of change of e
3
is given by
e
3
= z e
3
(10.48)
If such motion is possible, one should be able to rewrite the rst term in the
Eq. (10.47) as
d
dt
_
I
A
e
3
e
3
_
= I
A
d
dt
e
3
( z e
3
)
= I
A
d
dt
( z cos e
3
)
= I
A
2
cos (e
3
z) (10.49)
and the second term as
d
dt
(I
C
3
e
3
) = I
C
3
e
3
= I
C
3
(e
3
z) (10.50)
Using the relations given by Eq. (10.49) and (10.50), Eq. (10.47) can be simpli-
ed to
(I
A
cos
2
I
C
3
+MgR)(e
3
z) = 0 (10.51)
For the precession frequency to have a real value, the following condition has
to be met
(I
C
3
)
2
4I
A
MgRcos (10.52)
One can easily show that for the case (I
C
3
)
2
4I
A
MgRcos , two solutions
to the quadratic equation for ,
I
A
cos
2
I
C
3
+MgR = 0 (10.53)
becomes the two dierent precession frequencies obtained in the previous two
sections:
L
I
A
and
S
MgR
I
C
3
(10.54)
Readers can ponder upon the meaning of this statement.
10.10 Eulerian Angles
So far, we have learned how to describe rotation of rigid body in a few special
cases. Directly solving for the equation of motion or a rigid body is, in most
cases, not at all straightforward. Instead, one can utilize Lagrangian mechanics
which requires an appropriate choice of generalized coordinates. One widely
used generalized coordinates are the Eulerian angles.
To understand what Eulerian angles are, we have to rethink about the re-
lation between the xed coordinate system and the body coordinate system
which is geometrically locked to the rigid body of interest. Since both coordi-
nate systems are orthogonal three dimensional coordinate systems, there must
be a coordinate transformation that transforms the xed coordinates into the
body coordinates and vice versa. If the two coordinate systems share a common
origin, the coordinate transformation is a set of rotations.
There is always a single rotation about a particular axis that will transform
the xed coordinates into the body coordinates. However, this is not the sim-
plest method as it requires four parameters: three parameters to identify the
axis of rotation and the forth parameter to identify the amount of rotation, i.e.
the angle of rotation. Instead, a series of three rotations about known axes can
do the job just as well. There are dierent choices for such series of rotations,
but we will follow the method developed by Euler.
Consider the following series of rotations, which takes the x
i
system into the
x
i
system.
1. Rotation by counterclock-wise about x
3
-axis: the transformation leads
to
_
_
x
1
x
2
x
3
_
_
=
_
_
cos sin 0
sin cos 0
0 0 1
_
_
_
_
x
1
x
2
x
3
_
_
(10.55)
Because the rotation took place about the x
3
-axis, it is unaected by the
rotation and x
3
= x
3
.
2. Rotation by counterclock-wise about x
1
-axis:
_
_
x
1
x
2
x
3
_
_
=
_
_
1 0 0
0 cos sin
0 sin cos
_
_
_
_
x
1
x
2
x
3
_
_
(10.56)
This rotation will lead to the x
i
system with x
1
= x
1
.
3. Finally, rotation by counterclock-wise about x
3
(= x
3
)-axis:
_
_
x
1
x
2
x
3
_
_
=
_
_
cos sin 0
sin cos 0
0 0 1
_
_
_
_
x
1
x
2
x
3
_
_
(10.57)
The nal coordinate system is the x
i
system with x
3
= x
3
. Since these three
subsequent rotations will result in coordinate transformation from x
i
system to
x
i
, we can write the relation as
x = x
where
=
_
_
cos cos cos sin sin cos sin + cos cos sin sin sin
sin cos cos sin cos sin sin + cos cos cos cos sin
sin sin sin cos cos
_
_
As dicult as it is for one to see it, the beauty of this is that one can
transform between any two Cartesian coordinate systems that share a common
origin using this method with three parameters , and and these three
angles are known as the Eulerian angles.
Our goal was to nd appropriate coordinate transformation that yields e
1
e
2
e
3
system from x y z system. In this case z- and e
3
-axes play the role of x
3
and x
3
-
axes, respectively and is the angle between z- and e
3
-axes, i.e. cos = e
3
z.
We still need to identify x
1
-axis, which turns out to be the straight line formed
by the xy-plane and e
1
e
2
-plane. This is equivalent to the axis perpendicular to
the plane formed by z and e
3
vectors.
Therefore, one can get to e
i
system from the x y z system by 1) rotating the
x y z coordinate system about the z-axis until the x-axis matches x
= z e
3
.
We will denote the amount of rotated angle by : = cos
1
x ( z e
3
). 2)
Rotate the previously achieved coordinate system about x
by = cos
1
(e
3
z).
3) Finally, rotate the coordinate system about e
3
until the coordinate axes
are all aligned to e
1
e
2
e
3
system, that is x
becomes e
1
. The nal angle of
rotation is denoted by = cos
1
( z e
3
) e
1
. Note that all the angles can be
dened by the basis vectors of the two coordinate systems between which the
transformation is occurring.
10.11 Motion of a Symmetric Top with One Point
Fixed
If e
1
e
2
e
3
system represents the body coordinates of a rigid body under rotation,
then the basis vectors for this coordinate system e
1
, e
2
and e
3
must change with
time. However, this is not a problem per se, since any time dependence of the
coordinate axes can be dumped into the Eulerian angles. Therefore, the rigid
body dynamics simply becomes the matter of nding time dependence of (t),
(t) and (t) and they can be the perfect choice as the generalized coordinates.
This is especially useful considering how ineective Eulers equations given
in Section 10.6 are. The problem is that the equations themselves are set up
in the body coordinate system, and unless the solution is known or guessed
correctly in advance, there is no good way to identify the external torque of the
system.
To overcome this shortcoming, we will introduce Lagrangian mechanics. Let
us rst identify angular velocity in terms of Eulerian angles. Angular velocity
can be constructed with vectors having z, x
and e
3
as basis, =

+

+

.
We will denote
_
_
_
= angular speed about z
= angular speed about x
= z e
3
= angular speed about e

3
(10.58)
We can consider projection of each of these angular velocities along three body
coordinate axes, e
1
, e
2
and e
3
.
_
_
_
1
=

sin sin
2
=

sin cos
3
=

cos
(10.59)
_
_
_
1
=

cos
2
=

sin
3
= 0
(10.60)
_
_
_
1
= 0
2
= 0
3
=

(10.61)
From this, we can write the kinetic energy of the system
T =
1
2
I
ii
2
i
(10.62)
=
1
2
I
A
(
2
1
+
2
2
) +
1
2
I
C
2
3
(10.63)
=
1
2
I
A
(

2
sin
2
+

2
) +
1
2
I
C
(

cos +

)
2
(10.64)
If the center of mass of the rigid body is distance R away from the rotating
center in the body coordinate system, the potential energy of the system is given
by
V = MgRcos (10.65)
The Lagrangian L = T V is cyclical in and , therefore
p
=
L
= (I
A
sin
2
+I
C
cos
2
)

+I
C

cos = constant (10.66)
p
=
L
= I
C
(

+

cos ) = constant (10.67)
Eq. (10.67) yields
=
p
I
C

cos
I
C
(10.68)
and Eq. (10.68) combined with Eq. (10.66) gives us
=
p
cos
I
A
sin
2
(10.69)
We can then rewrite Eq. (10.68) as
=
p
I
C

(p
cos ) cos
I
A
sin
2
(10.70)
Note that the angular speeds

and

are now expressed in terms of a single
variable .
We now write the total energy of the system
E =
1
2
I
A
(
2
1
+
2
2
) +
1
2
I
C
2
3
+MgRcos (10.71)
which must be constant. Since p
= I
C
3
is constant,
1
2
I
C
2
3
must also be
constant. Therefore we can dene a dierent energy scale E
E
1
2
I
C
2
3
which is also constant.
E
=
1
2
I
A
(
2
1
+
2
2
) +MgRcos (10.72)
=
1
2
I
A
(

2
sin
2
+

2
) +MgRcos (10.73)
=
1
2
I
A

2
+
(p
cos )
2
2I
A
sin
2
+MgRcos (10.74)
=
1
2
I
A

2
+V () (10.75)
where V ()
(p
cos )
2
2I
A
sin
2
+ MgRcos is the eective potential energy in

-space. One can easily see that Eq. (10.75) is a one dimensional problem in -
space with rotational kinetic energy
1
2
I
A

2
and eective potential energy V ().
From this equation, we can calculate t()
t() =
_
d
_
(2/I
A
)[E
V ()]
(10.76)
and the nal result can be inverted to obtain (t). Because

and

were
expressed as functions of only in Eq. (10.69) and (10.70), (t) can be used to
nd (t) and (t). Without actually going into the full solution, which cannot
be analytically solved anyways, we can deduce qualitative motion by looking
into V () in Eq. (10.75). The eective potential is given by
V ()
(p
cos )
2
2I
A
sin
2
+MgRcos (10.77)
and one can easily see that the potential diverges at = 0 and due to the
sin
2
in the denominator of the rst term. If the total energy of the system
happens to coincide with the minimum of V (), the spinning top will precess
steadily with =
0
. Here
0
satises the condition V (
0
) = V
min
. One can nd
the precession speed by calculating

for =
0
. The result should be identical
to the solutions to the Eq. (10.53). This is a good exercise for readers.
In general, the angle will have a bound motion between some angles
1
and
2
. One can easily show that the angular velocity about the z-axis is non-
zero from Eq. (10.69). Therefore, the general motion of a spinning top includes
an oscillation of the tops symmetry axis between
1
and
2
superposed to the
precession around the z-axis. Such oscillatory precession is usually referred to
as nutation. Eq. (10.69) can be used to see if

changes sign at any point,
meaning whether the nutation includes a looping motion.
Chapter 11
Harmonic Oscillators
11.1 Simple Harmonic Oscillator
Let us consider a conservative potential well as shown in Fig. 11.1. The position
x
0
is the stable equilibrium position, i.e. a particle placed at x
0
will stay in
that position permanently and any perturbation away from x
0
will cause the
particle to return to x
0
thereby causing oscillation.
x
0
U
0
Figure 11.1:
The potential has to be at its minimum in terms of x.
U
x
0
= 0 (11.1)
where the subscript 0 denotes the dierentiation evaluated at the position x
0
.
We can set x
0
= 0 without losing generality. In other words, the position
is always measured with respect to the equilibrium position x
0
. Then, we can
express U as a Taylor series about x as follows
U = U
0
+
U
x
0
x +
1
2
2
U
x
2
0
x
2
+ (11.2)
71
CHAPTER 11. HARMONIC OSCILLATORS 72
From Eq. 11.1, the second term on the right hand side vanishes, and we can
always add or subtract constant from the potential energy such that U
0
= 0.
Thus, for suciently small x, we can always approximate the potential energy
of stable equilibrium as
U =
1
2
kx
2
(11.3)
where we dene k

2
U
x
2
0
.
Since force F and potential U has the relation F = U, stable equilibrium
yields the Hookes Law, F = kx, for one dimensional harmonic oscillators.
The equation of motion for a simple harmonic oscillator is given by equating
Hookes Law F = kx with Newtons second law F = m x.
kx = m x (11.4)
If we dene
2
0
k/m, then Eq. 11.4 becomes
x +
2
0
x = 0 (11.5)
The general solution for x is then x(t) = c
1
e
i
0
t
+ c
2
e
i
0
t
which can also be
expressed as
x(t) = A cos (
0
t ) (11.6)
A particle with mass m under the inuence of Hookes Law is subject to a
sinusoidal oscillation with its oscillation period
0
equals to
0
=
2
0
= 2
_
m
k
(11.7)
The variable
0
represents the angular frequency of the motion, which is related
to the frequency
0
by
0
= 1/
0
=
0
/2.
Let us now calculate the total energy E = T + U of the oscillator. From
Eq. 11.6, we can calculate x = A
0
sin (
0
t ) and
T =
1
2
m x
2
=
1
2
m
2
0
A
2
sin
2
(
0
t )
=
1
2
kA
2
sin
2
(
0
t ) (11.8)
U =
1
2
kx
2
=
1
2
kA
2
cos
2
(
0
t ) (11.9)
The total energy is then
E = T +U =
1
2
kA
2
[sin
2
(
0
t ) + cos
2
(
0
t )] =
1
2
kA
2
(11.10)
11.2 Harmonic Oscillations in Two Dimensions
We will now extend our focus to two dimensional oscillators. The equation of
motion that needs to be solved is
F = kr = mr (11.11)
which can be set up for two components
m x +kx = 0 (11.12)
m y +ky = 0 (11.13)
Once again, by dening
2
0
k/m, we get two solutions for each component
x(t) = Acos (
0
t ) (11.14)
y(t) = Bcos (
0
t ) (11.15)
If = , Eq. 11.15 can be rewritten as y(t) = Bcos (
0
t + ) and we
can extract information about time dependent oscillation amplitude from
_
x
A
_
2
+
_
y
B
_
2
= 1 + cos [2(
0
t ) +] cos (11.16)
1. = 0
_
x
A
_
2
+
_
y
B
_
2
= 1+cos [2(
0
t )] = 2 cos
2
(
0
t ) = 2
_
x
A
_
2
(11.17)
In this case, the oscillation is in fact one dimensional. This is an oscillation
along the straight line in the direction of A x +B y.
2. =
2
_
x
A
_
2
+
_
y
B
_
2
= 1 (11.18)
This is an equation describing an ellipse. In other words, the oscillator
follows an elliptic motion with two of its elliptical axes with length 2A
and 2B.
3. =
_
x
A
_
2
+
_
y
B
_
2
= 2 cos
2
(
0
t ) = 2
_
x
A
_
2
(11.19)
Once again, we get a one dimensional oscillation, but this time the oscil-
lation is along the direction of A x B y.
4. Other values of
For any other values of ,
_
x
A
_
2
+
_
y
B
_
2
becomes an elliptic motion as
shown in Fig. 4.
The oscillations along the x-axis and the y-axis have same frequency, thus the
oscillation ends up having a single loop.
However, a general two dimensional oscillation governed by the equation of
motion given below follows a more complicated path:
F
x
= k
x
x = m x (11.20)
F
y
= k
y
y = m y (11.21)
Then the solutions to these equations are
x(t) = Acos (
x
t ) (11.22)
y(t) = Bcos (
y
t ) (11.23)
Figure 11.2:
where
2
x
= k
x
/m and
2
y
= k
y
/m. In this case, because the oscillation frequen-
cies for oscillations in two dierent axes,
x
and
y
, are dierent, completing
one oscillation cycle along one axis is not always accompanied by the completion
of oscillation along the other axis. Thus, the motion does not necessarily form
a closed loop. The motion of the oscillator follows a closed loop only if

x
y
is
a rational number l/n. In such a case, the oscillation repeats itself with the
period
2l
x
=
2n
y
.
11.3 Damped Oscillations
Thus far, we have only considered free oscillations, that is oscillation without
dissipative or frictional forces, and thereby oscillating perpetually once set in
motion. Real physical systems, however, are always damped due to some form
of dissipation and we will take a look at the most simple case of such damping.
In our equation of motion given by Eq. ??, we will add frictional term that
depends on the velocity (or speed) of the motion, b x, and we have
m x +b x +kx = 0 (11.24)
We dene b/2m and
2
0
k/m and the equation can be rewritten as
x + 2 x +
2
0
x = 0 (11.25)
We assume that the solution to the equation takes the from x = e
rt
and obtain
the auxiliary equation
r
2
+ 2r +
2
0
= 0 (11.26)
By solving for r, we get the solution to the original dierential equation, Eq. 11.25,
x(t) = c
1
exp
_
_
+
_
2
0
_
t
_
+c
2
exp
_
_

_
2
0
_
t
_
= e
t
_
c
1
exp
_
_
2
0
t
_
+c
2
exp
_
2
0
t
__ (11.27)
1. Underdamped Oscillation: 0 <
2
<
2
0
The value of represents the strength of friction and this is the case where
the friction is relatively small and hence called underdamped oscillation.
In this case, we dene
2
1

2
0
2
and get
x(t) = e
t
(c
1
e
i
1
t
+c
2
e
i
1
t
) = Ae
t
cos (
1
t ) (11.28)
The solution can be viewed as an oscillation with its amplitude exponen-
tially decaying from Acos with the time constant 1/. This is the motion
we can see typically from a mass attached to a spring in the air.
Since the amplitude of the oscillation is not constant, strictly speaking, one
cannot dene a complete oscillation cycle. Therefore, angular frequency
of oscillation cannot be dened, either. Nonetheless, one can see that
the oscillation repeatedly goes through x = 0 position with the period of
1
= 2/
1
. We call
1
the angular frequency of the damped oscillator.
2. Critically Damped Oscillation:
2
=
2
0
Here,
2
2
0
= 0 and the solution to the equation of motion, Eq. 11.27,
becomes
x(t) = (c
1
+c
2
)e
t
= C
1
e
t
(11.29)
We only have one unknown constant C
1
, and the solution is overdeter-
mined with two initial conditions, one for position x(0) and the other for
velocity v(0) are given. In such a case, it is known that te
t
is also a
linearly independent solution to the original equation, and therefore we
get the solution of the form
x(t) = (c
1
+c
2
t)e
t
(11.30)
This is mostly a decaying function of time with the time constant almost
equal to 1/.
3. Overdamped Oscillation:
2
>
2
0
We dene
2
2
=
2
2
0
and the solution can be written as
x(t) = e
t
(c
1
e
2
t
+c
2
e
2
t
)
= c
1
e
(
2
)t
+c
2
e
(+
2
)t
(11.31)
Unless the initial conditions x(0) = c
2
and x(0) = c
2
( +
2
) are met,
the rst term on the right hand side of the above equation does not vanish.
Since the decay time constants for the two terms have the relation 1/(
2
) > 1/( +
2
), the rst term c
1
e
(
2
)t
becomes always dominant.
In other words, after time t > 1/( +
2
), the motion of the oscillator can
be approximated to x(t) c
1
e
(
2
)t
. Note that the decay rate for a
overdamped oscillator is slower than that of a critically damped oscillator.
This is because the friction due to large damping in an overdamped system
hinders the motion of the oscillator.
11.4 Sinusoidal Driving Forces
A damped oscillator always results in return to its equilibrium position regard-
less of the nature of damping. To create a steady state motion in a damped
harmonic oscillator, one has to apply external force. The equation of motion
under time dependent external force F
ext
(t) is
F = kx b x +F
ext
(t) = m x (11.32)
and after dening f(t) F
ext
(t)/m, we can rewrite the equation as
x + 2 x +
2
0
x = f(t) (11.33)
We will rst consider the case where the external driving force is a sinusoidal
function with frequency , i.e. F
ext
(t) = F
0
cos t. The equation of motion for
a system under such driving force is
x + 2 x +
2
0
x = a cos t (11.34)
where a F
0
/m. Since the right hand side of the equation contains sinusoidal
function with frequency , we will assume the particular solution to the equation
to have the from x
p
(t) = Acos(t ). Substituting x
p
(t) into the Eq. 11.34
results in
{a A[(
2
0
2
) cos + 2 sin ]} cos t
A[(
2
0
2
) sin + 2 cos ] sin t = 0
(11.35)
Since cos t and sin t are linearly independent, both coecients for cos t and
sin t must be equal to zero separately, i.e.
a A[(
2
0
2
) cos + 2 sin ] = 0 (11.36)
A[(
2
0
2
) sin + 2 cos ] = 0 (11.37)
From these equations, we get the relation
= tan
1
_
2
2
0
2
_
(11.38)
A =
a
(
2
0
2
) cos + 2 sin
=
a
_
(
2
0
2
)
2
+ 4
2
2
(11.39)
With A and expressed in terms of all the variables given in the Eq. 11.34,
x
p
(t) no longer contains unknown variables. Since the solution for force-free
damped oscillation obtained in the previous section, x
c
(t) (Eq. 11.27), satises
the x
c
+ 2 x
c
+
2
0
x
c
= 0 (this is called the complementary solution), the
general solution to the Eq. 11.34 is the sum of the complementary solution and
the particular solution
x(t) = x
c
(t) +x
p
(t)
= e
t
_
c
1
exp
_
_
2
0
t
_
+c
2
exp
_
2
0
t
__
+
a
_
(
2
0
2
)
2
+ 4
2
2
cos (t )
(11.40)
where is given by Eq. 11.38. Note, however, that x
c
(t) approaches zero after
a suciently long time t > 1/ and the solution becomes x(t) x
p
(t).
Both the amplitude and phase of the oscillation given by x
p
(t) are functions
of and . The maximum value for the oscillation amplitude is obtained when
the driving force frequency approaches
2
R
=
2
0
2
2
. This condition is found
by setting
dA
d
= 0. The frequency
R
is referred to as resonant frequency, that
is the oscillator is most responsive to the external driving force if the external
force frequency matches
R
.
The quality of the resonance is identied with the quality factor Q which
is dened to be Q
R
/2. In other words, the oscillator resonates better if
damping is small,
R
. If this condition is met,
R

0
and the quality
factor can also be expressed as Q
0
/ where represents the frequency
interval between the points on the amplitude resonance curve that are 1/
2 of
the maximum amplitude.
11.5 Principle of Superposition Fourier Series
Let us rewrite the Eq. 11.33 in the following form
_
d
2
dt
2
+ 2
d
dt
+
2
0
_
x(t) = f(t) (11.41)
and dene a linear dierential operator L
d
2
dt
2
+2
d
dt
+
2
0
. Since L is a linear
operator, it is distributive
L(x
1
+x
2
) = Lx
1
+Lx
2
(11.42)
If both x
1
and x
2
are solutions to the equations Lx
1
= f
1
(t) and Lx
2
= f
2
, we
can set a new equation
L(x
1
+x
2
) = f
1
(t) +f
2
(t) (11.43)
This can be generalized as
L
_
N
n=1
x
n
(t)
_
=
N
n=1
f
n
(t) = f(t) (11.44)
Thus, if the function f(t) can be expressed as a sum of sinusoidal functions,
f(t) =
N
n=1
a
n
cos (
n
t
n
) (11.45)
we can nd the solution to Eq. 11.44 from what we learned in the previous
section. Here the driving force contains a phase term
n
because individual
sinusoidal force may not be all in phase at t = 0. However,
n
only shifts
the phase (or timing) of the applied force and will shift the phase of harmonic
oscillators response to the force by the same amount. The steady-state solution
is thus
x(t) =
N
n=1
a
n
_
(
2
0
2
n
)
2
+ 4
2
n
2
cos (
n
t
n
n
) (11.46)
where
n
= tan
1
_
2
n
2
0
2
n
_
(11.47)
This is useful because any periodic function, i.e. a function that satises
f(t) = f(t +), can be expressed as a sum of sinusoidal functions
f(t) =
1
2
a
0
+
n=1
a
n
cos (nt
n
)
=
1
2
a
0
+
n=1
(b
n
cos nt +c
n
sin nt)
(11.48)
where a
n
,
n
, b
n
and c
n
have the following relationship
b
n
= a
n
cos
n
(11.49)
c
n
= a
n
sin
n
(11.50)
If the functional form for f(t) is known, the coecients b
n
and c
n
can be cal-
culated by performing the following integrals
b
n
=
2
_

0
f(t
) cos nt
dt
_
+/
/
f(t
) cos nt
dt
(11.51)
c
n
=
2
_

0
f(t
) sin nt
dt
_
+/
/
f(t
) sin nt
dt
(11.52)
11.6 The Response of Linear Oscillators to Im-
pulsive Forcing Functions
11.6.1 Response to a Step Function
Suppose an external force of a step function is acted upon an underdamped
harmonic oscillator sitting still at its equilibrium position
F(t) =
_
0, t < t
0
F
0
, t > t
0
(11.53)
At time t < t
0
, the motion is simply x(t) = 0 and for time t > t
0
, we have to
solve the equation of motion
x + 2 x +
2
0
x = a, t > t
0
(11.54)
where a F
0
/m. The particular solution is just a constant x
p
(t) = a/
2
0
.
The complementary solution, x
c
(t) for an underdamped oscillator is given by
Eq. 11.28 and the general solution can be expressed as
x(t) = e
(tt
0
)
[A
1
cos
1
(t t
0
) +A
2
sin
1
(t t
0
)] +
a
2
0
(11.55)
Applying the initial condition given for t = t
0
yields
A
1
=
a
2
0
, A
2
=
a
2
0
(11.56)
Figure 11.3:
Therefore, for t > t
0
, we have
x(t) =
a
2
0
_
1 e
(tt
0
)
_
cos
1
(t t
0
)

1
sin
1
(t t
0
)
__
(11.57)
This solution represents an under damped oscillation towards the new equilib-
rium position x =
a
2
0
from its original position x = 0.
11.6.2 Response to an Impulse Function
An impulse function can be treated as the dierence between two successive
step functions separated by a time = t
1
t
0
. In other words, an impulse
is superposition of a positive step of size a at t = t
0
followed by a negative
step of identical magnitude a at t = t
1
. The nal motion can be expressed as
superposition of the solutions to the equations of motion responding to each
step function,
x(t) =
a
2
0
_
1 e
(tt
0
)
_
cos
1
(t t
0
)

1
sin
1
(t t
0
)
__
2
0
_
1 e
(tt
0
)
_
cos
1
(t t
0
)

1
sin
1
(t t
0
)
__
=
a
2
0
e
(tt
0
)
_
e
cos
1
(t t
0
) cos
1
(t t
0
)
+

1
_
e
sin
1
(t t
0
) sin
1
(t t
0
)
_
_
, t > t
1
(11.58)
The solution represents an underdamped oscillation from x = 0 to x = a/
2
0
starting at t = t
0
and back to x = 0 starting at t = t
1
. By taking the limit
of approaching zero with holding a constant (otherwise, the solution just
becomes x = 0), we can nd the eect of an impulse.
x(t) =
a
1
e
(tt
0
)
sin
1
(t t
0
), t > t
1
t
0
(11.59)
11.6.3 Response to a General Forcing Function
Any forcing function can be dissected to a string of impulses of dierent strength
applied at dierent times. Then we set the equation of motion as
x + 2 x +
2
0
x =
N
n=
F
n
(t)
m
=
n
I
n
(t) (11.60)
where
I
n
(t) = I(t
n
, t
n+1
) (11.61)
=
_
a
n
(t
n
), t
n
< t < t
n+1
0, Otherwise
(11.62)
The solution for the nth impulse is, according to Eq. 11.59,
x
n
(t) =
a
n
(t
n
)
1
e
(tt
n
)
sin
1
(t t
n
), t > t
n
+ (11.63)
and the solution for all the impulses up to time t is
x(t) =
N
n=
a
n
(t
n
)
1
e
(tt
n
)
sin
1
(t t
n
) (11.64)
where Nth impulse is terminated at t
N+1
= t. If we allow the interval to
approach zero, the above summation becomes an integral:
x(t) =
_
t
a(t
1
e
(tt
)
sin
1
(t t
) dt
(11.65)
We dene
G(t, t
)
_
1
m
1
e
(tt
)
sin
1
(t t
), t > t
0, t < t
(11.66)
and using the relation F(t
) = ma(t
), we have
x(t) =
_
t
F(t
) G(t, t
) dt
(11.67)
The function G(t, t
) is known as the Greens function for the linear oscillator

equation. It is a function that reects the oscillators response at a moment t to
an impulse given at another moment t
for the given initial conditions, in our

case, an oscillator at rest in its equilibrium position. Because the Greens func-
tion relates the motion of the oscillator at a given moment t to the impulse given
in another moment t
, by performing the integral x(t) =

_
t
F(t
) G(t, t
) dt
,
we can see the eect of all the history of the applied force in the past up to the
present. Causality limits the integration to be performed for the past events
only, that is the range of integration runs from minus innity to the present
time t.
Chapter 12
Coupled Oscillations
12.1 Two Coupled Harmonic Oscillators
To ease into a more general and complicated coupled oscillators, we start this
chapter with two coupled harmonic oscillators. Two particles of mass m are
connected to three springs of spring constant
1
=
2
= at either end and
12
in between the particles. We can write equations of motion for two particles as
m x
1
+ ( +
12
)x
1
12
x
2
= 0 (12.1)
m x
2
+ ( +
12
)x
2
12
x
1
= 0. (12.2)
To solve these coupled equations, lets dene two variables
1
= x
1
+ x
2
and
2
= x
1
x
2
. By adding and subtracting above two equations, we get
m
1
+
1
= 0 (12.3)
m
2
+ ( + 2
12
)
2
= 0 (12.4)
Solution to these equations are
1
(t) = A
1
cos (
1
t
1
) (12.5)
2
(t) = A
2
cos (
2
t
2
) (12.6)
where
1
=
_
/m and
2
=
_
( + 2
12
)/m. From this, solutions for x
1
and
x
2
can be obtained,
x
1
(t) = C
1
cos (
1
t
1
) +C
2
cos (
2
t
2
) (12.7)
x
2
(t) = C
1
cos (
1
t
1
) C
2
cos (
2
t
2
). (12.8)
Here C
i
=
1
2
A
i
.
From Eq. 12.5 through Eq. 12.8, we can see that solutions to
i
(t) are dened
with single frequencies
i
s whereas the solutions to x
i
(t)s are superposition of
the two frequencies
1
and
2
. We call
i
s the normal coordinates, which is
dierent from spatial coordinate x
i
s, and two independent
i
s have two inde-
pendent modes of oscillation with two independent characteristic frequencies.
If we hold m
2
xed and nd the oscillation frequency of m
1
or vice versa, we
get
0
=
_
( +
12
)/M. Comparing this to
1
and
2
, we can see that one of
the two normal mode frequencies is higher and the other lower than the single
oscillator frequency
0
.
81
CHAPTER 12. COUPLED OSCILLATIONS 82
12.2 Weak Coupling
Suppose that
12
and rewrite
1
and
2
in terms of
0
.
1
=
_
+
12
M

12
M
=
_
+
12
M
_
1

12
+
12
(12.9)
2
=
_
+
12
M
+

12
M
=
_
+
12
M
_
1 +

12
+
12
(12.10)
Since
12
, we can dene

12
2(+
12
)
1 and
1
and
2
can be simplied
1
=
0
(1 ) (12.11)
2
=
0
(1 +) (12.12)
Now we hold m
2
xed and displace m
1
by a distance D initially. The initial
conditions are then x
1
(0) = D, x
2
(0) = 0, x
1
(0) = 0, x
2
(0) = 0. Impose
these initial conditions to Eq. 12.7 and 12.8. We then get C
1
= C
2
=
D
2
and
1
=
2
= 0. x
1
(t) and x
2
(t) becomes
x
1
(t) =
D
2
(cos
1
t + cos
2
t) (12.13)
= Dcos
_
1
+
2
2
t
_
cos
_
2
2
t
_
(12.14)
= (Dcos
0
t) cos
0
t (12.15)
x
2
(t) = (Dsin
0
t) sin
0
t (12.16)
(12.17)
12.3 General Problem of Coupled Oscillations
Let us consider a conservative system described in terms of a set of generalized
coordinates q
k
and the time t. A system with n degrees of freedom will have k
running from 1 to n. Since we are dealing with oscillations, there must be an
equilibrium position. Let q
k0
denote the equilibrium position in the generalized
coordinate q
k
. The potential has to be at its minimum in terms of the generalized
coordinate q
k
for the system to actually oscillate about q
k0
. In other words,
U
q
k
0
= 0 (12.18)
where the subscript 0 denotes the dierentiation evaluated at the position q
k0
.
For each of the generalized coordinates q
k
s, we can set q
k0
= 0 without losing
generality. In other words, the position in the generalized coordinates is always
measured with respect to the equilibrium position q
k0
. Then, we can express U
as a Taylor series about q
k0
s as follows
U(q
1
, q
2
, ..., q
n
) = U
0
+
k
U
q
k
0
q
k
+
1
2
j,k
2
U
q
j
q
k
0
q
j
q
k
+ (12.19)
From Eq. 12.20, the second term on the right hand side vanishes, and we can
always add or subtract constant from the potential energy such that U
0
= 0.
CHAPTER 12. COUPLED OSCILLATIONS 83
Thus,
U =
1
2
j,k
A
jk
q
j
q
k
(12.20)
where we dene A
jk

2
U
q
j
q
k
0
.
Weve shown earlier that if the equations connecting the generalized coor-
dinates and the rectangular coordinates do not explicitly contain the time, the
kinetic energy is a homogeneous quadratic function of the generalized velocities:
T =
1
2
j,k
m
jk
q
j
q
k
. (12.21)
where m
jk
=
i
x
,i
q
j
x
,i
q
k
=

2
T
q
j
q
k
.
Using Eq. 12.22 and Eq. 12.23, Lagranges equation can be reduced to
L
q
k
d
dt
L
q
k
=
U
q
k
+
d
dt
T
q
k
=
j
(A
jk
q
j
+m
jk
q
j
) = 0 (12.22)
To nd the solution to this equation, we rst assume the solution of the form
q
j
(t) = a
j
e
i(t)
. The equations of motion then becomes a set of n linear,
homogeneous, algebraic equations
j
(A
jk
2
m
jk
)a
j
= 0 (12.23)
For non-trivial solutions of a
j
to exist, the determinant of the coecients must
vanish:
det
A
jk
2
m
jk
= 0. (12.24)
This yields n roots for
2
. Because the frequency of oscillation is dened to be
a positive number, we have n roots for . Each of the n roots can be labelled
with
r
where r = 1, 2, ..., n. Much like the process of nding the principal axes
of inertia, n-dimensional vector a
r
can be found for each of
r
.
Because the principle of superposition applies for the dierential equation,
general solution for q
j
is a linear combination of the solutions for each of the n
values of r:
q
j
(t) =
r
a
jr
e
i(
r
t
r
)
(12.25)

CM-Kaist Lecture Note (2013)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

CM-Kaist Lecture Note (2013)

Uploaded by

Copyright:

Available Formats

Classical Mechanics

relative to the xed frame S. Then to an observer within the S

frame would see the motion of the puck also as force-free.

and a perpendicular component, r

(t) minimizes the functional J[x].

(t) can be expressed as

(t) +(x) (5.3)

where denotes each particle, i.e. = 1 through n. Instead of having to

. If we are to calculate the total

denotes the force acting on

lies on the straight line joining the two

. In such a case, the total amount of force acting on

, and by summing over we get the

is invoked in Eq. (9.6).

, the weak form of Newtons Third Law).

, in the system. Lets rewrite the position vector of a particle, r

is a position vector connecting the th particle and th particle, and

if the internal force is a central force.

is just an external torque exerted on the th particle, r

are derivable from potential

, respectively. We can dene the total potential energy

. Combining Eq. 9.21 and Eq. 9.22, we get W

. We can then dene a dierential scattering cross

= 2 sin d. The number of particles scattering into

about is then dN = I () 2 sin d.

about must be equal to the number of incoming particles

is referred to as the moment of inertia tensor and

-axis is a result of rotation along all three x, y and z-axes,

s to have non-trivial solutions, the determinant of the coecients has to

s ( = A, B, C), we can nd the ratio

along which the moment of inertia has the

. In other words, the direction dened by s are the principal

are eigenvalues associated

s have the same value,

= angular speed about z

= angular speed about x

= angular speed about e

+ MgRcos is the eective potential energy in

) is known as the Greens function for the linear oscillator

for the given initial conditions, in our

, by performing the integral x(t) =

You might also like