Classical Mechanics Block 2/4

UNIT-5: Central force problem
Central force, equations of motion and first integrals, Binet’s equation, classification of orbits,
Virial theorem, condition for closed orbits, Kepler problem of inverse square force.
5.0 Objectives
After studying this unit you will be able to
• Describe what are central forces.
• Explain how a two body problem having six degrees of freedom can be reduced to a one
body problem with just two or even one degree of freedom.
• Establish conservation of angular momentum in central force field.
• Obtain equation of motion and first integrals.
• Classify the orbits based on the energy of the particle
• Explain Binet’s equation and Virial theorem.
• Explain the Kepler’s problem and obtain the equation of orbits.
5.1 Introduction
Central force is one of the oldest and richest problems in classical mechanics. The Hooke’s law,
Newton’s law, Coloumb’s law are all involve central forces. In most of the situations wherein the
force is produced by a point object, the force turns out to be central. Out of the four basic forces,
the Electrostatic and gravitational forces are central in nature. Hence, he motion of electrons
around nucleus that leads to the diverse materials, their spectra, their physical and chemical
properties have their roots in central forces. Also the motion of planets, satellites, and all that
moves under the influence of gravitation are essentially motion in central force field. In this
regard, understanding the central forces is extremely important.
MPDSC 1.1 KSOU
5.2 Central force
Central forces are defined as the forces that are always directed either towards or away from a
fixed point.
Suppose we have a particle moving in such a way that the force acting on it is always directed
towards a point then we can write the expression for the force on the particle as F~(~ r) = f (~r )r̂
where f is a scalar function that determines the algebraic magnitude of the force. Depending on
the sign of the function f the central force can be attractive or repulsive in nature.
In general central forces doesn’t have to conservative. But here we are looking at conserva-
tive central forces. If the force is conservative than curl of the force will be zero. Because the
central force is always directed towards or away from a fixed point, if we chose that point as the
origin of the coordinate system used to describe the motion, the force will have only radial com-
ponent and no tangential component. This property makes the use of spherical polar coordinate
system to be highly suitable for the study of central forces.
~ × ~F = 0 =⇒ 1 ∂f 1 ∂f
∇ θ̂ − φ̂ = 0
r sin θ ∂φ r ∂θ
Since above equation is a vector equation both components of θ̂ and φ̂ has to be individually
zero. From this we can clearly see a conservative central force will only be a function of r (The
magnitude of ~r) therefore with all the conviction we can represent a general conservative central
force as
~F = f (r )r̂
We all know that a force field, that is a vector field can be better represented using a poten-
tial, that is a scalar function. This is more advantageous when the force field is conservative in
nature. The potential is a scalar function of the three spatial variables. In spherical polar coor-
dinate system, potential is a function of r, θ and φ. Hence a general potential function can be
represented as V = V (r, θ, φ).
Since the force is only a function of r, we can show that the potential will also be only a
function of r. As the curl of a conservative force field is zero, it can be represented as a gradient
of scalar function. Hence, we can write any conservative force as the negative gradient of the
potential function based on a theorem from vector calculus. The negative sign is included to
make the force to be always directed towards decreasing potential. Therefore,
72
MPDSC 1.1 KSOU
∂V 1 ∂V 1 ∂V
− f (r )r̂ = r̂ + θ̂ + φ̂
∂r r ∂θ r sin θ ∂φ
In the above equation, the left hand side does not have θ̂ and φ̂ components. Hence, by
equating the corresponding terms on either side we can write
∂V ∂V
=0 =0
∂θ ∂φ
We can conclude that the potential function does not depend on θ or φ. In other words, we
can see that the potential is only a function of r.
5.2.1 Two body problem
Now let us take two bodies which are interacting with each other through central force. When
the force of interaction between the two bodies is always acts along the line joining, then the
force between them is said to be central force.
Figure 5.11: Two body problem with central force
We shall assume that the force on the particle 2 is given by the following equation, and will
be along the line joining the two particles.
r2 | − ~
F2 = f (|~ r2 − ~
r1 )(~ r1 )
There are six degrees of freedom for this particle and we can write down the Lagrangian of
the system as
73
MPDSC 1.1 KSOU
1 1
L= r˙12 + m2 ~
m1 ~ r˙22 − V (r2 − r1 )
2 2
Now let us change the coordinate system in which R gives position vector of centre of mass.
~R = m1 ~r1 + m2 ~r2
m1 + m2
r2 − ~
Let ~r = ~ r1 be the relative position vector of the particles. We can express ~r1 and ~r2 in
terms of ~R and ~r as,
m2~r m1~r
~r1 = ~R − ~r2 = ~R +
m1 + m2 m1 + m2
In terms of the newly introduced coordinates, the Lagrangian can be written as,
1 m2~r˙ 1 m1~r˙
L= m1 (~R˙ − ) + m2 (~R˙ + ) − V (r )
2 m1 + m2 2 m1 + m2
With a little bit of simplification by using some basic algebra which you can check for your-
self we can rewrite the above equation as,
1 1 m1 m2 2
L= (m1 + m2 ) Ṙ2 + ṙ − V (r )
2 2 m1 + m2
You can see ~R doesn’t appear in the Lagrangian and hence ~R is cyclic and can be ignored.
This implies that the corresponding canonical momentum is conserved. In other words, as ~R is
cyclic, ~Ṙ will be a constant.
Hence, that the centre of mass is moving with a uniform velocity. Now if we use a coordinate
system attached to center of mass, that we call center of mass frame, the problem will be much
more simplified. This reduces the number of degrees of freedom of the system. Hence the
process is also called reduction. Then the reduced Lagrangian will be
1 ˙2
L= µ~r − V (r )
2
m1 m2
In the above expression, µ = is called the reduced mass.
m1 + m2
Note that the Lagrangian is similar to the Lagrangian of a particle in central force field. The
reduction process has reduced a two body problem into a one body problem.
Now we try to find out the first integrals of motion and see what are the constants of motion.
74
MPDSC 1.1 KSOU
We will get a lot of information about the system much before solving the problem.
5.3 Conservation of angular momentum
When a body moves in central force, because of the spherical symmetry of the potential and
forces, the total angular momentum, ~l = ~r × ~p gets conserved. It therefore follows that ~r is
always perpendicular to the fixed direction of ~l in the space. This can be true only if ~r always
lies in a plane whose normal is parallel to ~l. Then we can use the plane polar coordinate system
instead of spherical polar coordinate system that will simplify the problem even further. But,
before we do that, let us establish the fact that the angular momentum is conserved in central
force field.
Consider the Lagrangian and express it in terms of cartesian coordinates as
1
L= µ( ẋ2 + ẏ2 + ż2 ) − V (r )
2
p
We know r = x2 + y2 + z2 and also we know the derivatives
∂r x ∂r y ∂r z
= = =
∂x r ∂y r ∂z r
Using the above derivatives and solving the Lagrangian we get the equation of motion as
∂V x ∂V y ∂V z
µ ẍ = − µÿ = − µz̈ = −
∂r r ∂r r ∂r r
We have,
~l = ~r × ~p = µ~r × ~v
From the above expression we can write the components of angular momentum as
lx = µ(yż − zẏ) ly = µ(z ẋ − x ż) lz = µ( x ẏ − y ẋ )

=⇒
dlx dly dlz
= µ(yz̈ − zÿ) = µ(z ẍ − x z̈) = µ( x ÿ − y ẍ )
dt dt dt
Using the equations of motion we obtained earlier in the above equations we can write,
75
MPDSC 1.1 KSOU
dlx ∂V z ∂V y dly ∂V x ∂V z dlz ∂V y ∂V x

= −y +z = −z + x( = −x +y
dt ∂r r ∂r r dt ∂r r ∂r r dt ∂r r ∂r r
It is evident from the above set of equations that
dlx dly dly

= 0;
dt dt dt
=⇒
lx = constant; ly = constant; lz = constnat
Hence we have established that the each component of angular momentum is a constant
(The constants have different value for each component). Hence the angular momentum is a
constant of motion for a particle moving in central force field. We shall prove this using another
method in the next section.
5.4 Equations of motion and first integrals
Since we found out the angular momentum is a constant of motion, we can restrict our degrees
of freedom from three to two. We have understood an important aspect of the central force
problem that is the central force is a planar problem because the angular momentum vector of
the particle moving in central force field is a constant. From six degrees of freedom we have
brought down to two degrees of freedom. Hence to describe the motion of a particle in central
force field we can use just two generalized coordinates that may be chosen as r and θ from plane
polar coordinate system. Then the Lagrangian of the system can be written as
1
L= µ(ṙ2 + r2 θ̇ 2 ) − V (r )
2
Note that in the Lagrangian the coordinate θ is absent. Hence, the corresponding conjugate
momentum will be a constant of motion. Hence,
∂L
= µr2 θ̇ = constant
∂θ̇
The above equation that describes the conservation of angular momentum is known as first
integral of motion.
Angular momentum being constant can help us prove the Kepler’s second law or show that
76
MPDSC 1.1 KSOU
areal velocity is constant. So even though Kepler’s laws are written for a planetary motions,
it can be very much true for all central force problems. The area swept in the orbit at equal
intervals of time is always equal
dA 1
= r (rdθ ) = 0
dt 2
If you are dealing with a very small arc in an ellipse or a circle,we can approximate the
1
segment into a triangle and the area of the triangle can be written as (r2 dθ ). Since l = µr2 θ̇, is
2
a constant of motion, we can write
d d 1 2 1 2
(µr2 θ̇ ) = 0 =⇒ ( r θ̇ ) = 0 =⇒ r θ̇ = constant
dt dt 2 2
Thus, Areal velocity is a constant.
Since Lagrangian is independent of time, according to Nöether’s theorem, because of the
symmetries of time invariance, Energy of the system will be a constant of the motion. Hence,
1
E= µ(ṙ2 + r2 θ̇ 2 ) + V (r ) = constant
2
We can modify the above expression as
1 2 1
µṙ = E − V (r ) − µr2 θ̇ 2
2 2
l
We know l = µr2 θ̇ is a constant of motion, Hence we can use the relation θ̇ = in the
µr2
above expression
2
l2

1 1 l 1
E = µṙ2 + µr2 ) + V (r ) = µṙ2 + + V (r )
2 2 µr2 2 2µr2
=⇒
1 2 l2 2
µṙ = E − V (r ) − θ̇
2 2µr2
1
Dividing the entire equation by µ and taking the square root we will get
2
s
dr 2 l2
= ( E − V (r ) − )
dt µ 2µr2
77
MPDSC 1.1 KSOU
dr
dt = s
2 l2
( E − V (r ) − )
µ 2µr2
integrating the above equation under proper limits
Z r
dr
t − t0 = s
r0 2 l2
( E − V (r ) − )
µ 2µr2
Above integral provides an equation of motion that with known energy, angular momentum
and reduced mass of the system one can determine t as a function of r that can be inverted to
get r = r (t).
In parallel, we find the equation describing θ as,
Z t
dθ l ldt
= 2 =⇒ θ − θ0 =
dt µr t0 µr (t)2
So once we have the expression r = r (t), we can solve this integral and find the exact position
of the particle as a function of time. The above two equations together form the equations of
motion of the particle or the system.
5.5 Classification of orbits
Now we will try to reduce the problem into a one dimensional problem, that is since θ is cyclic
l
and θ̇ = we can substitute this in the Lagrangian that gives us,
mr2
1
L= µ(ṙ2 + r2 θ̇ 2 ) − V (r )
2
=⇒
1 2 l2 1 2
L= µṙ − (V (r ) − ) = µṙ − Ve f f (r )
2 2µr2 2
As the Lagrangian is valid for any particle in central force field we can replace the reduced
mass, just by the mass of the particle in central force field and write the general Lagrangian as
1 2
L= µṙ − Ve f f (r )
2
78
MPDSC 1.1 KSOU
Note that we have reduced the problem to just one dimension. Also note that the potential
has changed which is different from the original potential and we will call new one as effective
potential.
Using the effective potential we can find the force by taking the negative of the gradient of
the effective potential which became possible because our force was conservative
dVe f f l2
fe f f = − = f+ 3
dr mr
The first term is the central force term and the second is the centrifugal term
Figure 5.12: The effective potential of central force field.
Since Classical mechanics is deterministic, by knowing the four initial conditions that is by
knowing initial positions and initial velocities, we can determine the motion of the system.
We can write r˙0 and θ̇ in terms of Energy and angular momentum.
s
2 l2 l
r˙0 = (E − V − ) and θ˙0 =
m 2mr02 mr02
One important piece of information is that θ doesn’t change the sign in the middle of the
motion. That is an object moving in the clockwise direction, will not change its direction and
79
MPDSC 1.1 KSOU
l
rotate in the counter clockwise direction. This is from the fact that θ˙0 = . Since l and m are
mr02
constants and doesn’t change and hence θ cannot change as well.
1 2
For a given value of l, the shape of the orbit will be fixed. We have, E − Ve f f α mṙ .
2
Suppose the energy is fixed and E − Ve f f is positive, the kinetic energy will be positive. This
is an allowed situation of the orbit. If we deviate from the orbit, the effective potential will be
different and usually more than its value in the region of the orbit. Hence, outside the orbit
on both sides of the curve, E < Ve f f . With this case, the energy of the system is less than
the potential. The particle cannot access those regions or in other words these are classically
forbidden regions. Since the curve in the figure is for a harmonic oscillator potential, the particle
will be oscillating between r1 and r2 . The shape of the orbit can be seen in the figure below. Since
the motion is oscillatory there will be a sinusoidal wave which can be drawn between r1 and r2
and the circular nature represents the closed nature of its orbit that is it comes back to its original
position after a complete cycle of motion.
Figure 5.13: The nature of the bound orbits in central force field.
5.6 Differential equation for the orbit or Binet’s equation
We have the the equations of motion in central force field as
l2 l
mr̈ − = f (r ) and θ̇ =
mr3 mr2
In principle the idea is to solve the these differential equation and get the complete solution
r (t) and θ (t). Once we know the complete solution we get the orbital equation as well that is
r in terms of θ. We shall try to find r (θ ) without solving the differential equations representing
80
MPDSC 1.1 KSOU
equation of motion. Since we need a solution in terms of θ we will transform the differential
equation in terms of θ.
d dθ d l d
= =
dt dt dθ mr2 dθ
=⇒
d2 r d d l d l dr
2
= ( r) = 2
( 2 )
dt dt dt mr dθ mr dθ
1
Let us use a substitution r = in the above equation. With this consideration we will have,
u
dr dr du 1 du
= =− 2
dθ du dθ u dθ
=⇒
1 dr du
2
=−
r dθ dθ
d2 r l 2 u2 d2 u
= −
dt2 m2 dθ 2
Substituting all these in the equation of motion we get,
l 2 u2 d2 u l2 3 1
− 2
− 3
u = f( )
m dθ mr u
l 2 u2
Dividing the entire equation by − we get, a modified differential equation as
m
d2 u m 1
2
+ u = − 2 2 f( )
dθ l u u
This equation is the differential equation of orbit. This equation is also known as Binet’s
equation. This equation remains same under the transformation of θ → −θ. However, for this
equation to be invariant, the first derivatives under this transformation must be equal at θ0 . This
actually happens at certain points called the turning points.
du du
i.e. = at θ0
dθ d(−θ )
Further, we can obtain the relation between r and θ using the first integrals that we have
mr2
already obtained by using the relation dt = dθ in the first integral expressions as
l
dr
dt = r
2 l2
(E − V − )
=⇒ m 2mr2
81
MPDSC 1.1 KSOU
dr
dθ = r
2 l2
(E − V − )
m 2mr2
By integrating this equation
dr
Z r
θ − θ0 = r r2
r0 2mE 2mV 1
( 2 − 2 − 2)
l l r
We will rewrite the above equation in terms of u
Z u
du
θ − θ0 = r
u0 2mE 2mV
( − 2 − u2 )
l2 l
This represents the integral equation of the orbit of the particle in central force field. Because
the above integral often cannot be expressed in terms of well-known functions. In fact only cer-
tain types of forces have been investigated and the most important are the power-law functions
of r with potential as,
V = ar n+1
5.7 Virial theorem
Another property of central force motion can be derived as a special case of a general theorem
valid for a large variety of systems. This is known as Virial theorem.
Consider a general system of mass points with position vectors ri and applied forces Fi . From
Newton’s laws of motion we have,
~ṗi = ~Fi
let us define a quantity
G= ∑ ~ṗi ·~ri
i
The total time derivative of this quantity is
82
MPDSC 1.1 KSOU
dG
dt
= ∑~ṙi · ~pi + ∑ ~ṗi · ri
i i
The first term can be written as
∑~ṙi · ~pi = ∑ mi~ṙ~i˙ṙi = ∑ mi v2i = 2T

i i i
While the second term becomes
∑ ~ṗi ·~ri = ∑ ~Fi ·~ri

i i
Therefore the equation reduces to
d
dt ∑
~pi ·~ri = 2T + ∑ ~Fi ·~ri
i i
The time average over time τ is obtained by integrating both sides with respect to t from 0
to τ and dividing by τ.
1 τ dG
Z
lim ¯ + Fi¯ri
dt = 2T
τ 0 dt
or
1
2T + Fi ri = [ G (τ ) − G (0)]
τ
If the motion is periodic, that is all coordinates repeat after a certain time and is τ is chosen
to be the period, then the right hand side vanishes.
1
T = − Fi ri
2
If the forces are derivable from a potential then the theorem becomes,
1
T = − ∇Vri
2
and for a single particle moving under a central force it reduces to
1 ∂V
T= r
2 ∂r
83
MPDSC 1.1 KSOU
If V is a power law function of r
V = ar n+1
∂V
r = ( n + 1)V
∂r
and therefore
n+1
T̄ = V̄
2
for attractive inverse square law n=-2 and therefore
1
T=− V
2
5.8 Condition for closed orbits
We have a beautiful theorem relating the power law of the forces with the nature of orbits to de-
cide on what condition the orbits will be closed. This theorem is called the Bertrand’s theorem.
Before talking about the Bertrand’s theorem, Let us know what are Bounded and Unbounded
orbits.
When the energy of a particle or a system is positive in the central force field, then the system
will eventually escape from the potential and tend to move towards infinity. Hence gradually
the system will not be under the influence of a central force field. When this happens as t → ∞
the system will be out of the field and may have a straight line path if it does not come under
any other force fields. Such path of the system or orbits are called Unbounded orbits.
If the energy of the particle or the system is negative, then the particle will never be able
to escape the potential of the central force. Hence the particle will be continuously moving in
the force field only and that implies they are bounded to the force field. The path traced by the
system or the corresponding orbits are called bounded orbit.
All the unbounded orbits are open, but the bounded orbits may be closed or may not be
closed. All bounded orbits are not closed. Closed orbits are those orbits which are retraced
by the system. For example lets take a simple pendulum which is oscillating under a potential
with some damping effects. The particle will not go back to its original amplitude because of
the damping. This is a bounded system having an orbit that is not closed.
84
MPDSC 1.1 KSOU
In addition, the force must be attractive to have bounded orbits because, an attractive po-
tential is negative and can make the total energy negative when the magnitude of potential en-
ergy becomes more than that of kinetic energy thereby leading to the condition of closed orbits.
Bound orbits are not possible in repulsive potential as the total energy will never be negative
under repulsive force field.
With the above stated conditions or the closed orbits, for any given l, this will occur if the
equivalent potential has and extremum, that is either minimum or maximum at some distance
ro and if the energy E is equal to Ve f f (ro ). The requirement that Ve f f have an extremum is
equivalent to vanishing f at r0 .
l2
f (r0 ) = −
mr03
Which says that the force must be attractive for circular (in general closed) orbits to be pos-
sible in addition the energy must be
l2
E = V (r0 ) +
2mr02
The stability of the circular orbit is defined by the sign of the second derivative of the poten-
tial at the radius of the circle.
∂2 V 0 ∂f 3l 2
= − + greaterthan0
∂r2 ∂r mr04
∂f 3 f (r0 )
<−
∂r r0
dln f
< −3
dlnr
If the force behaves like a power law of r
k
f =−
r n +1
then the stability condition becomes
( n + 1) k 3k
− n +2
< − n +2
r0 r0
or
85
MPDSC 1.1 KSOU
n<2
1
A power law attractive potential varying more slowly than is thus capable of stable circu-
r2
lar orbits for all values of r0 .
5.9 Kepler problem of inverse square force
The inverse square law is the most important of all central force laws and it deserves detailed
treatment
k k
f =− and V=
r2 r
We have the differential equation for the orbit is
1
d2 u −m f ( )
+u = u = mk
dθ 2 2
l u 2 l2
mk
Changing the variable to y = u − ,the differential equation becomes
l2
d2 y
+y = 0
dθ 2
which has the solution

y = Bcos(θ − θ 0 )
B and θ 0 being two constants of integration,in terms of r the solution
1 mk
= 2 (1 + ecos(θ − θ 0 ))
r l
We write the solution in the form of θ
du
Z
θ =θ− r
2mE 2mku
− u2
l2 l2
86
MPDSC 1.1 KSOU
This indefinite integral has a standard form
dx 1 β + 2γx
Z
= −√ arccos − √
−γ
p
α + βx + γx2 q
where
q = β2 − 4αx
to apply this we must set

2mE
α=
l2
2mk
β=
l2
γ = −1
and the discriminant q is therefore
2mk 2 2El 2
q=( ) ( 1 + )
l2 mk2
with these substitutions the equation of θ becomes
l2u
−1
0
θ = θ − arccos rmk
2El 2
1+
mk2
1
finally for solving for u = the equation of the orbit becomes
r
s
1 mk 2El 2
= 2 (1 + 1+ cos(θ − θ 0 ))
r l mk2
It is to be noted that only three out of four constants appear in the equation of orbits. Thus we
use conservation of angular momentum to get the last constant.
mr2 dθ = ldt
Now the general equation of the conic with one focus at the origin is
1
= C (1 + ecos(θ − θ 0 ))
r
87
MPDSC 1.1 KSOU
Where e is the eccentricity of the conic section. By comparing with the previous equation the
eccentricity can be written as s
2El 2
e 1+
mk2
The nature of the orbit depends on the magnitude of eccentricity e. If e > 1, the energy will
be positive and hence the orbit will be a hyperbola. If e = 1, the energy will be zero. In this case
the path of the particle will be a parabola. If e < 1, then the energy will be negative and the orbit
mk2
of the particle will be an ellipse. If e = 0, the energy will be negative and equal to - 2 . Under
2l
this condition, the orbit will be a circle.
From Virial theorem we know that
V V
E = T+V = − +V =
2 2
Hence,
k
E=
2r0
but from the statement of equilibrium between the central force and the effective force we can
write
k l2
− = −
r02 mr03
or
l2
r0 =
mk
with this formula for the orbital radius, the energy equation becomes
mk
E
2l 2
In the case of the elliptical orbit it can be shown that the major axis depends solely on the energy.
The semi major axis is one half the sum of the two apsidal distances r1 and r2 . By definition the
radial velocity is zero at these points and the conservation of energy implies that the apsidal
distances are the roots of the equation.
l2 k
E=− 2
+ =0
2mr r
88
MPDSC 1.1 KSOU
or
k l2
r2 + r− =0
E 2mE
Now the coefficients of the linear term in a quadratic equation is the negative of the sum of the
roots and hence the semi major axis is given by
r1 + r2 k
a =−
2 2E
In terms of the semi-major axis the eccentricity of the ellipse is given by

s
l2
e= 1−
mka
or
l2
= a (1 − e2 )
mk
in terms of which the elliptical orbit equation can be written as
a (1 − e2 )
r=
1 + ecos(θ − θ 0 )
it follows that the two apsidal distances (which occur when θ − θ 0 is 0 and π respectively) are
equal to a(1 − e) and a(1 + e) which are expected as these are the properties of the ellipse.
5.10 Check your progress
Check your progress by answering the questions below.
1. What is a central force?
2. How many degrees of freedom are there for a general two body problem and to how many
number of degrees of freedom they can be reduced if the central forces are involved?
3. Briefly explain why angular momentum is conserved in central force field.
4. Classify different types of orbits in central force field.
5. When the central force involves a power law, what is the condition for closed orbits?
89
MPDSC 1.1 KSOU
5.11 Keywords
• Central force
• First integrals
• Orbits
• Binet’s equation
• Virial theorem
• Kepler problem
5.12 Worked examples
1. The orbit of a particle of mass m moving in a central force is given by r = kθ 2 , where k is a

constant. Find the law of force
Answer: In central force problem, the equation of orbit is given by
d2

1 1 m
+ = − 2 r 2 F (r )
dθ 2 r r l
1
Given that r = kθ 2 or = d f rac1kθ 2 . Then,
r

d 1 d 1 2
= = −
dθ r dθ kθ 2 kθ 3
d2

1 6 6k
= 4 = 2
dθ 2 r kθ r
Then substituting this in the equation of orbit we get,
6k 1 m
2
+ = − 2 r 2 F (r )
r r l
Then,
l2

6k 1
F (r ) = − 4
+ 3
m r r
90
MPDSC 1.1 KSOU
8200 km
2. A satellite in an elliptical orbit around the earth has the equation = 1 + 0.2 cos θ.
r
Find (i) the values of semi major and semi minor axes; (ii) the period of the satellite; and
(iii) the altitude of perigee and apogee. Radius of earth 6380 km, mass of the earth 5.97 ×
1024 kg, gravitational constant G = 6.67 × 10−11 Nm2 kg−2 .
Answer: We know the equation of ellipse
1/C 1
= 1 + e cos θ with = a (1 − e2 )
r C
Now comparing the given equation with the standard equation we can write
a(1 − e2 ) = 8200 km and e = 0.2
8200 km
(i) The Semi-major axis a = = 8541.7 km
0.96
√
The semi-minor axis b = a 1 − e2 = 8369.1 km
2πa3/2
(ii) Period of the satellite T = √ = 2 hrs 11 min 2 s
GMe
(iii) The rmin = a(1 − e) = 6833.4 km and rmax = a(1 + e) = 10250.0 km
Then the altitude of perigee = rmin − R E = 453.4 km
And the altitude of apogee = rmax − R E = 3870.0 km

k
3. A particle moves in a circular orbit in a force field F (r ) = −. Suddenly k becomes k/2
r2
without change in velocity of the particle. Show that the orbit becomes a parabolic.
Answer: In elliptical orbits, we have the total energy E = −k/2a. In the case of zero
eccentricity elliptical orbit reduces to a circle, the semi major axis a equals to semi major
axis b and is just the radius of the circle r. For circular orbit, the total energy E potential
energy V and kinetic energy T are given by
k k k
E=− ; V=− ; T=
2r r 2r
When k becomes k/2 with there is no change in velocity, the kinetic energy remains same
but the potential energy becomes half of the initial value
k k
V=− ; T=
2r 2r
91
MPDSC 1.1 KSOU
Then the total energy will be
E = T+V = 0
When the total energy becomes zero, the eccentricity of the orbit becomes 1 and hence the
orbit will become parabola.
5.13 Questions for self study
1. Explain what is central force?
2. Describe two body problem and explain how the number of degrees of freedom is reduced
from six to two in two body problem.
3. Explain conservation of angular momentum in central force field.
4. Obtain the first integrals and equations of motion of a particle in central force field.
5. Describe how orbits of a particle in central force field are classified.
6. Explain Binet’s equation.
7. State and prove Virial theorem.
8. Deduce the condition for closed orbits when the central force follows a power law be-
haviour.
9. Discuss the Kepler’s problem of inverse square force.
5.14 Answers to check your progress
1. A force which is always directed towards or away from a single point in force field is called
central force. Or when the force of interaction between two particles is always along the
line joining the two particles, then the force is said to be central force.
2. There are six degrees of freedom for a general two body problem and they can be reduced
to one degree freedom if the central forces are involved?
92
MPDSC 1.1 KSOU
3. The angular momentum of a particle in central force field is always a constant. In the
Lagrangian of the particle in central force field φ is not present that means it is a cyclic
coordinate. Hence, the corresponding conjugate momentum is a constant. Because φ is
the angle corresponding conjugate momentum will be angular momentum.
4. There are two kinds of orbits for a particle in central force field they are bound orbits
(closed orbits most of the time) and unbound orbits (open orbits).
1
5. When the central force involves a power law like f α − , There can be closed orbits
r n +1
only if n¡2
5.15 References
• Herbert Goldstein, Charles Poole, John Safko, (2010) Classical Mechanics. Pearson Educa-
tion.
• G. Aruldhas, (2008) Classical Mechanics. Prentice Hall of India pvt. Ltd.
• R. G. Takwale and P. S. Puranik (2006) Introduction to Classical Mechanics. Tata McGraw-

Hill Publishing Company Ltd.
• N. C. Rana and P. S. Joag (2005) Classical Mechanics. Tata McGraw-Hill Publishing Com-
pany Ltd.
93
UNIT-6: Scattering theory
Scattering in a central force field, differential and total scattering cross section, impact parameter,
lab frame and center of mass frame, Rutherford scattering.
6.0 Objectives
• Explain scattering cross section and impact factor.
• Obtain an expression for the scattering cross section for a hardsphere scattering.
• Explain the scattering due to central force field and obtain the expression for the scattering
cross section.
• Describe the importance of center of mass frame and explain the transformation between
the lab frame and center of mass frame.
6.1 Introduction
Scattering is one of the fundamental phenomena that is widely observed in nature. The indirect
day light, the colors of sky and sea, pressure, temperature, electrical resistance and many other
well known properties based on scattering. Scattering can be defined as the change in direction
of motion of a particle due to the interaction with a field. In other words, scattering deals with
the unbounded motion of particles influenced by a force field.
6.2 Scattering cross section and impact factor
Consider a uniform beam of particles of same mass and energy and incident upon a centre of
force. It will be assumed that the force falls off to zero for every large distances. The number
of particles crossing a unit area placed normal to the bean in unit time is the intensity I. This is
also called flux density. When a particle approaches a center of force, it will either be attracted
or repelled and its orbit will deviate from the incident path. After passing the center of force,
MPDSC 1.1 KSOU
the force acting on the particle will diminish, so that the orbit once again approaches a straight
line. In general, the final direction of motion is not the same as the incident direction and the
particle is said to be scattered.
s + ds s θ θ − dθ
O
Figure 6.14: Scattering of particles.
We will take the a small region on the detector which has an element of solid angle dΩ that
makes an angle θ with the scatterer. Let I be the number of particles per unit time per unit area
or in other words which is the incident intensity. The differential cross section σ(Ω) is defined as
the ratio of number of particles scattered per unit solid angle into a region of solid angle element
to the incident number of particles. Mathematically,
Number of particles scattered in dΩ per unit time/dΩ

σ(Ω) =
Incident intensity
We can notice that the cross section does not depend on the angle φ because of the symmetry
of the potential. Hence we will simplify the formula, by considering the the annular region with
corresponding solid angle dΩ is given by
dΩ = 2π sin θdθ
Where θ is the scattering angle.

Since we have already argued the independence of the differential cross section with φ, we
can rewrite the previous equation of differential cross section as a function of θ alone. Let us
define a new parameter called the impact parameter. It is the perpendicular distance between
the path of the projectile and the centre of the potential field created by the scatterer. The impact
parameter and scattering angle are connected as the angle of scattering is a function of impact
parameter along with the force field.
If the impact parameter is small then the scattering angle will be large and vice-versa. We
95
MPDSC 1.1 KSOU
Figure 6.15: Impact parameter and scattering angle.
already know that for a single particle in a central force motion, E and l are constants of motion.
In a typical scattering experiment, the energy of the particle is fixed because as mentioned earlier
we are only allowing particles with certain kinetic energies to reach the scatterer. Also at the
initial points, the potential of scatterer has no role because of very large distance.
Therefore we can write
r
2E
v= and l = m~v ×~r
m
But we have already defined the perpendicular component of the radius vector as the impact
parameter (s) so,
l = mvs
Now let us take an annular region which is at a distance ds above the the trajectory of a single
particle. In other words, consider a region at a distance s + ds from the axis of the scatterer.
The number of particles crossing the annular ring is given by area of the annular ring multi-
plied by the intensity which is equal to
dN = 2πsdsI = σ (Ω)2π sin θdθ I
We can rewrite the expression for differential cross section as

2πsdsI s ds
σ(θ ) = =
2π sin θdθ I sin θ dθ
ds
Absolute value of is taken so as to remove the negative sign in the differential cross
dθ
ds
section. Since can be negative because as the impact parameter is small, the scattering angle
dθ
96
MPDSC 1.1 KSOU
will be more. Hence the derivative of such a inverse relationship will have negative sign.
6.3 Scattering by hardsphere
ψ
θ
ψ
s R
Figure 6.16: hard sphere scattering.
Let us consider a hard ball at which we are throwing small balls and it will be undergoing
elastic scattering. In this case the potential is given by V (r ) = ∞ when r < R and V (r ) = 0
when r greater than R. Let us find the scattering cross section which is given by
s ds
σ= | |
sinθ dθ
From the figure 6.16 s = R sin ψ and θ = π − 2ψ. Therefore,
π θ θ
s = Rsin( − ) = Rcos
2 2 2
θ
s ds Rcos ds
σ= | |= 2| |
sinθ dθ sinθ dθ
ds R sinθ
=−
dθ 2 2
substituting in σ(θ ) equation ,we get
θ θ
Rcos Rsin 2 2
σ(θ ) = 2 2 = R sinθ = R
sinθ 4 sinθ 4
From the above equation we can say that the scattering cross section is uniformly distributed as
it only depends on R. We can find the total scattering cross section by integrating over the entire
volume,
97
MPDSC 1.1 KSOU
R2
Z π
σ(θ )2πsinθdθ = 2π [−cosθ ]0 π = πR2
0 4
6.4 Scattering by central force and Rutherford scattering
Figure 6.17: Scattering in central force field.
Consider a particle being scattered by central force field. When a particle with a positive
total energy approaches the center of the force field, the motion of the particles gets influenced
by the force. As a result of this, the particle takes an hyperbolic path. Interestingly the normal
drawn to the hyperbolic orbit from the scatterer that is center of the central force field forms the
shortest distance of the particle from the point. Let rc be this distance of closest approach of the
beam to the scatterer.
Let the initial conditions are as the radius vector goes from infinity to rm , θ will change from
π to π − φ. Note that θ and Θ are different. θ is the angle measured at the point of closest
approach. Whereas Θ is the scattering angle.
Let us express the angular momentum, one of the constants of motion in terms of impact
parameter s as
√
l = mvs = 2mEs
I am going to borrow the equation of orbits from the previous chapter which is given by
Z um
du
θ − θ0 = − φ = − r
0 2mE 2mV
− 2 − u2
l2 l
1 1
In the above expression u = and um = . Expressing l in terms of the impact parameter
r rm
98
MPDSC 1.1 KSOU
s, the resultant expression for Θ(s) is
Z um
sdu
Θ = π−2 r
0 V
1− − s2 u2
E
When the potential is known we can solve the integral then our scattering problem is solved.
The solution to the problem is similar to that of the keplers problem.
In case we consider scattering of charged particle due to another point charge, the central
force that is responsible for scattering will be electrostatic force. This is similar to Rutherford’s
alpha scattering experiment in which an alpha particle of positive charge Ze is being scattered
by a nucleus of positive charge Z 0 e. Then the force will be
ZZ 0 e2
f =
r2
The energy E is greater than zero because the potential is greater than zero. Hence the orbit
is hyperbola with the eccentricity e given by,
s r
2El 2 2Es 2
e= 1+ = 1+( )
m( ZZ 0 e2 )2 ZZ 0 e2
if θ 0 is chosen to be π and θ = 0 then the orbit equation becomes,
1 mZZ 0 e2
= (ecosθ − 1)
r l2
The direction of the incoming asymptote ψ is then determined by the condition r → ∞
1
cosψ =
e
or
Θ 1
sin =
2 e
Therefore,
Θ 2Es
cot =
2 ZZ 0 e2
The relationship between the impact parameter and scattering angle will be
99
MPDSC 1.1 KSOU
ZZ 0 e2 Θ
s= cot
2E 2
Using some simple rearrangement we can get,
1 ZZ 0 e2 2 Θ
σ(θ ) = ( ) cosec4
4 2E 2
this is the expression for the scattering cross section in Rutherford alpha scattering experi-
ment.
6.5 Laboratory and center of mass frames
Till now we were concerned with the one-body problem of the scattering of the particle by the
fixed centre of force. In practice the scattering always involves two bodies. For example, in the
case of the Rutherford scattering, we have the α particle and the atomic nucleus. The second
particle is not fixed but it recoils.
This is not a problem since we have already found that you can reduce a two body problem
into a one body problem. However, the scattering angle which was measured in the laboratory
θ is the angle between the final and initial direction of the scattered particle. On the other hand
the angle Θ calculated from the equivalent one body problem is the angle between the final and
initial directions of the relative vector between the two particles. The two angles will be equal
only if the second particle remain stationery but that only happens initially. Thus the equivalent
one body problem does not furnish the scattering angle as measured in the laboratory frame.
~r
θ Θ Θ
Figure 6.18: (a)Laboratory and (b)Center of mass frame of reference.
100
MPDSC 1.1 KSOU
The relationship between the scattering angle Θ and θ can be determined by examining
how the scattering take place in the centre of mass frame. In such a system, the total linear
momentum is zero that is the two particles move with equal and opposite momenta. Before
the collision they will be moving towards each other and after the collision they will be moving
away from each other. The angle between the initial and final directions of the relative vector
is Θ. r1 and v1 are the position and velocity after scattering of the incident particle 1 in the
laboratory system. r10 and v10 are the position and velocity after scattering of particle 1 in the
center of mass system. Let R and v are the position and velocity of the center of mass in the lab
system.
At any instant
r1 = R + r 0 1 and v1 = V + v10
After scattering, v1 and v10 makes an angle θ and Θ respectively with V. From conservation
of linear momentum,
µ
(m1 + m2 )V = m1 v0 =⇒ V= v0
m2
v1 sin θ = v10 sin Θ and v1 cosθ = v10 cosΘ + V
The ratio of the two equations gives us the relation between θ andΘ.
sin Θ
tan θ =
cos Θ + ρ
Where ρ is defined by
µv0
ρ=
m2 v10
From the law of cosines we can get
v21 v102 + V 2 + 2v10 V cos Θ
we will eliminate v1 and V is expressed in terms of v0
101
MPDSC 1.1 KSOU
cos Θ + ρ
cos θ = p
1 + 2ρ cos Θ + ρ2
Both these relations still involve the ratio of speeds through ρ. By definition of center of mass
the speed of particle in the center of mass frame v10 is connected with the relative speed v by
µ
v10 = v
m1
Hence ρ can also be written as
m1 v0
ρ=
m2 v
When the collision is elastic, the kinetic energy is conserved the ρ is
m1
ρ=
m2
If the collision is inelastic, the total kinetic energy of the two particles in altered, since the
kinetic energy of the center of mass motion must be constant. By conservation of linear momen-
tum, the energy change can be expressed as
µv2 µv2
= 0 +Q
2 2
The ratio of relative speeds before and after collision can be written as
s
v m1 + m2 Q
= 1+
v0 m2 E
where E is the energy of the incoming particle. Thus for the inelastic scattering ρ becomes,
m1
ρ= r
m + m2 Q
m2 1+ 1
m2 E
Even though θ and Θ are different, in an experiment the number of particles scattered into a
given element of solid angle must be same whether we measure the event in terms of θ and Θ.
2πIσ(Θ) sin Θ|dΘ| = 2πIσ0 (θ ) sin θ |dθ |
102
MPDSC 1.1 KSOU
or
d(cos Θ)
σ0 (θ ) = σ(Θ)
d(cos θ )
Where σ0 is the differential scattering cross section expressed in terms of scattering angle in
the lab frame. By finding the derivative of cos θ we can get
(1 + 2ρ cos Θ + ρ2 )3/2
σ0 (θ ) = σ(Θ)
+ρ cos Θ
When the masses of the two particles are same in elastic scattering then
r
1 + cos Θ Θ
cos θ = = cos
2 2
Θ
θ=
2
Thus with equal masses, scattering angles cannot be greater than 90◦ and the scattering cross
section is given by
σ0 (θ ) = 4 cos θσ(Θ)
1. Define scattering cross section.
2. What is impact factor?
3. What will be the usual path of the particle being scattered by a central force field.
4. What is a center of mass frame?
6.7 Keywords
• Scattering
• Cross section
• Impact factor
• Rutherford’s scattering
103
MPDSC 1.1 KSOU
6.8 Worked examples
1. Consider a scattering of particles by a rigid sphere of radius R and calculate the differential
and total scattering cross section.
Answer: Since the sphere is rigid, the potential outside the sphere is zero and that inside
is infinity. A particle with impact factor b > R will not get scattered. If b < R due to the
law of conservation of momentum and energy a particle incident at an angle α with the
normal to the surface of the sphere will be scattered off on the outside of the normal at the
same angle α
We can write
b
sin α = and φ = π − 2α
R
π−φ φ
α= or sin α = cos( )
2 2
Then the impact factor
φ
b = R cos
2
The expression for differential scattering cross section will be
b db R2
σ(φ) = =
sin φ dφ 4
The total scattering cross section would be
Z Z π
σT = σ (Ω)dΩ = 2π σ(φ) sin φ dφ
4π 0
R2
σT = 2π [1 − cos φ]0π = 4π 2
R
2. Determine the scattering cross section of incident α particles each having 10 keV energy
when they get scattered in Rutherford alpha scattering experiment at right angles to the
incident beam
104
MPDSC 1.1 KSOU
Answer: We know the expression for the scattering cross section in Rutherford’s alpha
scattering experiment as
1 ZZ 0 e2 2 Θ
σ(θ ) = ( ) cosec4
4 2E 2
We know Z = 4, Z 0 = 79, E = 10keV and Θ = 90o
Then 2
4 × 79 × e2

1
σ θ) = ( cosec4 (45o )
4 2 × 10 × 103 × e
Then
σ(90o ) = 6.39078 × 10−42
3. In an experiment, a particle of mass 10 g moving with a velocity 10 ms−1 collides with

another particle of mass 50 g at rest. If the particle of mass 10 g gets deviated by 30o with
respect to its initial direction and start to move with a velocity 8 ms−1 , determine the recoil
velocity of the scatterer.
Answer: We know from the conservation of linear momentum, the total initial momentum
must be equal to total final momentum.
With respect to the horizontal direction,
m1 v1 − m1 v2 cos θ
m1 v1 = m1 v2 cos θ + m2 v x =⇒ vx =
m2
The horizontal component of recoil velocity will be
10 × 10 − 10 × 8 × cos 30
vx = = 0.614359ms−1
50
With respect to the vertical direction,
10 × 8 × sin 30o
0 = m1 v2 sin θ + m2 vy =⇒ vy = −
m2
vy = 0.8ms−1
Then the recoil velocity of the scatterer is (0.614359, 0.8)ms−1
105
MPDSC 1.1 KSOU
1. Explain the scattering phenomena and define scattering cross section and impact factor.
2. Describe the scattering of particles by a hard sphere.
3. Explain the phenomena of scattering due to a repulsive central force.
4. Obtain the expression for scattering cross section in Rutherford α scattering experiment.
5. What are laboratory and center of mass frames of references? Explain their importance.
6. Obtain the relation between the scattering cross section in laboratory frame and center of
mass frame.
1. Scattering cross section is defined as the ratio of number of particles getting scattered into
an small element of solid angle per unit time per unit solid angle to the incident intensity.
2. Impact parameter is the perpendicular distance between the path of the projectile and the
center of the potential field created by the scatterer.
3. Parabola.
4. A frame of reference attached to the center of mass of the system (scatterer and projectile
together) is called center of mass frame.
6.11 References
tion.

pany Ltd.
106
UNIT-7: Hamiltonian formulation
Legendre transformations and the Hamilton’s equations of motion, cyclic coordinates and con-
servation theorems, Hamilton’s equations from variational principle, applications of Hamilto-
nian formulation.
7.0 Objectives
• Describe Legendre transformations and take the Legendre transformation for some func-
tions
• Define Hamiltonian of a system as the Legendre transformation of the Lagrangian and

obtain the Hamilton’s equations of motion.
• Define cyclic coordinates and explain its importance.
• Explain the Routh’s procedure
• Derive the Hamilton’s equations of motion from the variational principle.
• Describe the usage of Hamiltonian formulation to find the equations of motion of any
system.
7.1 Introduction
Hamiltonian mechanics being equivalent with the Lagrangian mechanics provides an equiv-
alent analytical approach to find the equations of motion of physical systems. Similar to La-
grangian formulation, Hamiltonian formulation is also set up using generalized coordinates.
Instead of generalized velocities, conjugate momenta are used as another set of variables in
Hamiltonian formulation. An advantage of Hamiltonian formulation is that it gives the equa-
tions of motion as first order differential equations as you would see unlike Lagrangian formu-
lation that provides second order differential equations. Because, the generalized coordinates
MPDSC 1.1 KSOU
and conjugated momenta are used as variables in Hamiltonian formulation, it will be suitable to
use for the studies of waves and their interaction with matter. This can not be carried out using
Lagrangian. Further, the Hamiltonian formulation have found ways into quantum mechanics
also.
7.2 Legendre transformation
Before getting into Legendre transformation first let us define a convex function. If a function is
defined over a real space so that the second derivative is positive at all points of the space, then
such functions are called convex functions.
d2 f
f : I ∈ R → R is convex if >0
dx2
Similarly we can also define concave functions over a real space such that the second deriva-
tive is always less than zero.
d2 f
f : I ∈ R → R is convex if <0
dx2
Consider a smooth convex function. Let us define a variable s = f 0 ( x ). Now let us define
the Legendre transformation as a transformation from x to s as,
F (s) = sx − f ( x )
In the above transformation, we have to remember that x is a function of s. Hence, to be

precise, we must write the Legendre transformation equation as,
F (s) = sx (s) − f [ x (s)]
For a given function f ( x ), its derivative must be found and the variable x and hence f ( x )
must be expressed in terms of f 0 ( x ). Then they can be substituted in the above transformation
equation to obtain F (s). Let us see an example
Let us obtain Legendre transformation of f ( x ) = a + bx + cx2 .
First we must find the derivative of the function, and define that as a new variable.
s−b
s = f 0 ( x ) = b + 2cx =⇒ x=
2c
108
MPDSC 1.1 KSOU
Then the Legendre transformation is defined as,
s−b s−b
F (s) = sx (s) − f [ x (s) = s − f( )
2c 2c
s−b s − b ( s − b )2
=s −a−b −
2c 2c (2c)2
s−b s−b
= [s − b − c ]−a
2c 2c
s − b 2c − c
= (s − b) − a
2c 2c
( s − b )2
= −a
4c
This the the Legendre transformation of f ( x ) = a + bx + cx2 .
Properties of Legendre transformation
1. If f is a convex function then F is also a convex function
Proof:
df
F0 = = x = ( f 0 ) −1 ( s )
ds
F (s) = xs − f ( x )
then
dF dx d f dx
F0 = = x+s −
ds ds dx ds
d f dx d f dx
x+ − = x = ( f 0 ) −1 ( x )
dx ds dx ds
To prove that F(s) is also convex we have to take the second derivative
d 0 −1 1
F”(s) = (f ) = greaterthan0
ds f ”( x )
Because the original function f(x) was greater than zero
2. Legendre transformation are involuntary
This means, if LT ( f ) = F then LT ( F ) = f

dF
proof: Let LT ( F ) = G = G (q) with q = .
ds
G (q) = qs − F (s(q)) = qs − (qs − f (q)) = f (q)
109
MPDSC 1.1 KSOU
Henc, G = f .
df dF
3. If s = , then d f = sdx then x = implies dF = xds
dx ds
4. There is at-most one minimum f min = − F (0) and vice versa Fmin = − f (0)
If you have more than one variable, then the Legendre transform can be computed by taking
the partial derivatives of the function with respect to the all the variables and treating them as a
new set of variables and carrying out the transformation to their space.
For example, consider a function of two variables f = f ( x, y).
Let us define two new variables as
∂f ∂f
u= and v=
∂x ∂y
Then the Legendre transform of f ( x, y) is defined as
LT ( f (u, v)) = F (u, v) = xu + yv − f ( x, y)
let us do an example problem to understand this better. Consider a two variable function
f = f ( x, y) = ax2 + 2bxy + cy2 .
Let us define the two variables as
∂f ∂f
u= = 2ax + 2by and v= = 2bx + 2cy
∂x ∂y
For writing the Legendre transformation, we have to express x and y in terms of u and v.
Hence using simple algebra we can write
bu − av bu − cv
y= and x=
2(b2 − ac) 2(b2 − ac)
Then the Legendre transform F (u, v) can be written as
u[bu − av] v[bu − cv] a[bu − av]2

F (u, v) = ux + vy − f ( x, y) = + −
2(b2 − ac) 2(b2 − ac) 4(b2 − ac)2
2b[bu − av][bu − cv] c[bu − cv]2

+ +
4(b2 − ac)2 4(b2 − ac)2
1
F (u, v) = [−cu2 + 2buv − av2 ]
4( b2 − ac)
110
MPDSC 1.1 KSOU
I am going to give you an important example ,there are a lot of places Legendre transforma-
tions are useful like in thermodynamics but here we are going to see how we can transform the
Lagrangian into Hamiltonian and vice-versa
7.3 Hamilton’s Equations of motion
Hamiltonian to start with is defined as the Legendre transform of Lagrangian. We know that the
Lagrangian is a function of generalized coordinates and generalized velocities. Also, the deriva-
tive of Lagrangian with respect to the generalized velocity is defined as conjugate or canonical
momentum.
∂L
L = L(q, q̇, t) and p=
∂q̇
Hamiltonian is constructed by taking the Legendre transform of Lagrangian by eliminating
the generalized velocity replaced in terms of canonical momentum as,
H (q, p, t) = pq̇ − L = pq̇(q, p, t) − L(q, q̇(q, p, t), t)
Once we have the Lagrangian of the system, we shall find the canonical momentum and
then take the Legendre transform to the Lagrangian to obtain the Hamiltonian. While doing
so, note that the generalized velocities are expressed as functions of generalized coordinates
and canonical momenta. Unlike the Lagrangian formulation, the generalized coordinates and
velocities are not independent.
To obtain the Hamiltonian equations of motion, let us take the partial derivatives of the
above function.

∂H ∂q̇ ∂L ∂L ∂q̇
=p − +
∂q ∂q ∂q ∂q̇ ∂q
∂H ∂q̇ ∂L ∂L ∂q̇
=p − −
∂q ∂q ∂q ∂q̇ ∂q
The last term arises because the Lagrangian is an implicit function of q through q̇ . Addition-
∂L
ally, we know = p. Using this in above equation we get,
∂q̇
∂H ∂q̇ ∂L ∂q̇ ∂L
=p − −p =−
∂q ∂q ∂q ∂q ∂q
111
MPDSC 1.1 KSOU
From Lagrangian equations of motion, we know that

d ∂L ∂L ∂L d ∂L
− =0 =⇒ =
dt ∂q̇ ∂q ∂q dt ∂q̇
Then,

∂H d ∂L dp ∂H
=− =− = ṗ =⇒ ṗ = −
∂q dt ∂q̇ dt ∂q
Now differentiating the Hamiltonian with respect to p
∂H ∂q̇ ∂L ∂q̇ ∂q̇ ∂q̇

= q̇ + p − = q̇ + p − p
∂p ∂p ∂q̇ ∂p ∂p ∂p
∂H
= q̇
∂p
Now differentiate the Hamiltonian with respect to t
∂H ∂q̇ ∂L ∂q̇ ∂L
=p − −
∂t ∂t ∂q̇ ∂t ∂t
Therefore
∂H ∂L
=−
∂t ∂t
The important two equations obtained in above discussion,
δH δH
q̇ = and ṗ = −
δp δq
These two equations are the Hamiltonian equations of motion.
You can see that both q̇ and ṗ are on equal footing as both are first order equations. There
is a nice symmetry. Because of the structure of the equation these are also known as canonical
equations.
7.4 Cyclic Coordinates
In Lagrangian formulation, cyclic coordinates are those which do not appear on the Lagrangian
d
of the system. If L is independent of some q j then, ( ṗ j ) = 0. This implies p j is constant of
dt
∂H ∂H ∂L
motion. In Hamiltonian formalism, we have q˙j = and ṗ j = − = .
∂p j ∂q j ∂q j
112
MPDSC 1.1 KSOU
∂L ∂H
Hence, if q j is cyclic in Lagrangian, that means, = 0, then = 0. If a particular coor-
∂q j ∂q j
dinate is cyclic in Lagrangian then it will also be cyclic in the Hamiltonian. Nöether’s theorem
is a consequence of the Hamilton’s equations that says any conserved quantity is because of a
symmetry law. The Hamiltonian is a function of q, p and t. Now the total derivative of H with
respect to time is
dH ∂H ∂H ∂H
= q̇ + ṗ +
dt ∂q ∂p ∂t
We have,
∂H ∂H
= − ṗ and = q̇
∂q ∂p
Therefore,
dH ∂H ∂L
= =−
dt ∂t ∂t
If L is independent of t that implies H is also independent of t which means Hamiltonian is
also conserved quantity. This is a simple consequence similar to the case of the Lagrangian. We
shall see soon that the Hamiltonian is the energy function. Hence when either Lagrangian or
Hamiltonian are independent of time, then the energy of the system becomes conserved.
7.5 Routh’s Procedure
So What have we done when we moved from n second order differential equations to 2n first
order differential equation, did we achieve any computational ease? We definitely know that
going from Newtonian to Lagrangian made our life easier. but strangely, going from Lagrangian
to Hamiltonian haven’t reduced the time for computation. But this transformation reduces the
computation time when there is cyclic coordinates.
Let us take a system of n degrees of freedom in which q j one of the coordinates is cyclic.
∂L
Then, p j = = constant.
∂q˙j
Does this help us in Lagrangian, the answer is, it does not, because even though p j is a
constant, it does not mean q˙j is a constant. And p j is not contained in the Lagrangian. But in the
Hamiltonian p j is present. So we have reduced the problem from n degrees of freedom to n − 1
degrees of freedom. This procedure will reduce the degrees of freedom by the number of cyclic
113
MPDSC 1.1 KSOU
coordinates.
Let us take a coordinate system with n in which some coordinates are cyclic. Let us assume
the coordinates qs+1 to qn are cyclic.
We define another function called, Routhian denoted by R = R(q1 , q2 , , , qs , q˙1 , q˙2 , , , q˙s , ps+1 , , , pn )
The difference between the Routhian and Hamiltonian is that in the case of Routhian we
take the Legendre transform of Lagrangian replacing all generalized velocity terms of the cyclic
coordinates. However, in Hamiltonian, we take the Legendre transform of Lagrangian with
respect all the generalized velocities.
n
R= ∑ pi q̇i − L
i = s +1
Hence, in Routhian there are only s second order differential equations. The equations of
motion in terms of Routhian are given similar to that of Lagrangian equations of motion as,

d ∂R ∂R ∂R
− =0 where, i = 1, 2, 3, , , , s and, q̇i = for i = s + 1, , , n
dt ∂q̇i ∂qi ∂pi
7.6 Derivation of Hamilton’s equation from the Variational principle
We have already shown that Lagrange’s equation as a consequence of the variational principle.
The variational method has been often proved to be preferable method of deriving Lagrange’s
equation. It can also be applicable to types of system not usually comprised within the scope of
mechanics. So it would be advantageous if we can find a variational principle that can lead us
to Hamilton’s equation of motion.
We have the Hamilton’s principle of least action as
Z t2
δI = δ L dt = 0
t1
Now if you remember carefully this is formulated over the configuration space. So our first
task is to change the space from the configuration space to a phase space. In Hamiltonian, both
q and p are independent of each other and hence we can’t just write a simple transformation
from q to p. But the transformation relationship has to be a function of both q, p and their time
derivatives
114
MPDSC 1.1 KSOU
Z t2
δI = δ pi q̇i − H (q, p, t) dt = 0
t1
As a variational principle in phase space, this equation is sometimes referred to as Modified

Hamilton’s principle. The modified Hamilton’s principle is exactly of the form of the variational
problem in the Lagrange’s case. The above equation then becomes,
Z t2
∂H ∂H
t1
∑ pi δq̇i + q̇i δpi −
∂qi
δqi − δp
∂pi i
dt = 0
i
Consider the first term,
Z t2 Z t2 Z t2
d d
pi δq̇i dt = pi δ qi dt = pi δq dt
t1 t1 dt t1 dt i
Now we shall integrate the first term in the above equation by parts,
Z t1
| pi δqi |tt21 − ṗi δqi dt
t1
The integrated term vanishes at the end points t1 and t2 and therefore,
Z t2 Z t1
pi δq̇i dt = − ṗi δqi dt
t1 t1
Substituting the above result in the Hamilton’s principle we get,
Z t2
∂H ∂H
t1
∑ − ṗi +
∂qi
δqi + q̇i −
∂pi
δpi dt = 0
i
Since the modified Hamilton’s principle is a variational principle in phase space, the δq’s
and δp’s are arbitrary and independent. Hence, the coefficients of δq’s and δp’s in the above
equation must vanish individually. Hence we get,
∂H ∂H
ṗi + =0 =⇒ ṗi = −
∂qi ∂qi
∂H ∂H
q̇i − =0 =⇒ q̇i =
∂pi ∂pi
These are the Hamiltonian equations of motion.
115
MPDSC 1.1 KSOU
7.7 Application of Hamiltonian formulation
7.7.1 A particle in Cartesian coordinate system
Consider a particle of mass m moving under the influence of a potential V = V ( x, y, z). The
Lagrangian of the particle will be
1 2 1
L = T−V = mv − V = m( ẋ2 + ẏ2 + ż2 ) − V ( x, y, z)
2 2
The canonical momenta would b e
∂L ∂L ∂L
px = = m ẋ py = = mẏ pz = = mż
∂ ẋ ∂ẏ ∂ż
=⇒
px py pz
ẋ = ẏ = ż =
m m m
Constructing the Hamiltonian using the Legendre transform,

1
H= ∑ pi q̇i L = px ẋ + py ẏ + pz ż − 2
m( ẋ2 + ẏ2 + ż2 ) − V ( x, y, z)
i
py p x 2 p y 2 p z 2

px pz 1
H = px + py + pz − m + + + V ( x, y, z)
m m m 2 m m m
1 2
H= p x + p2y + p2z + V ( x, y, z)
2m
Note that the Hamiltonian is the total energy function. Hence we can also write the Hamil-
tonian as H = T + V. Let us wait before we generalize let us look into some more examples.
The Hamiltonian equations of motion would be
∂H ∂H
ṗi = − and q̇i =
∂qi ∂pi
The first equation would be
∂V ∂V ∂V
ṗ x = − ; ṗy = − ; ṗz = −
∂x ∂y ∂z
=⇒
116
MPDSC 1.1 KSOU
~ṗ = −∇V
The above equation is nothing but the Newton’s second law.

The second equation would be
∂V ∂V ∂V
ẋ = ; ẏ = ; ż =
∂p x ∂py ∂pz
=⇒
px py pz
ẋ = ẏ = ż =
m m m
7.7.2 A charged particle in an electric field produced by another point charge
Consider a particle of charge Q in an electric field produced by another charge Q0 . Let the
electric field intensity be represented by ~E and the potential be V.
~E = 1 Q0 1 Q0
and V=
4πeo r2 4πeo r
Note that the force field is central force. Hence, we can use the plane polar coordinate system
in which the kinetic energy expression can be taken as
1
T= m(ṙ2 + r2 θ̇ 2 )
2
The potential energy of the charged particle will be
1 QQ0
U=
4πeo r
Then the Lagrangian of the particle will be
1 1 QQ0
L = T−U = m(ṙ2 + r2 θ̇ 2 ) −
2 4πeo r
QQ0
Let = K. Then the Lagrangian will be
4πeo
1 K
L= m(ṙ2 + r2 θ̇ 2 ) −
2 r
117
MPDSC 1.1 KSOU
To construct the Hamiltonian, first we need to determine the canonical momenta. Then
express the generalized velocities in terms of generalized coordinates and conjugate momenta.
∂L pr
pr = = mṙ =⇒ ṙ =
∂ṙ m
Similarly
∂L pθ
pθ = = mr2 θ̇ =⇒ θ̇ =
∂θ̇ mr2
Constructing Hamiltonian as Legendre transform of the Lagrangian as,

1 K
H = pr ṙ + pθ θ̇ − L = pr ṙ + pθ θ̇ − m(ṙ2 + r2 θ̇ 2 ) −
2 r
p p m p 2 mr2 p 2 K
r θ r θ
H = pr + pθ 2
− − +
m mr 2 m 2 mr2 r
p2r p2θ K
H= + 2
+
2m 2mr r
Again note that the Hamiltonian is same as the total energy function. Now let us find the
equations of motion of the particle,
∂H ∂H
ṗi = − and q̇i =
∂qi ∂pi
With respect to r,
∂H ∂H
ṗr = − and ṙ =
∂r ∂pr
=⇒
p2r K pr
ṗr = 3
− 2 and ṙ =
mr r m
Similarly, with respect to θ we can find the equations of motion. But before we proceed, note
that θ is not present in the Hamiltonian. In other words, θ is cyclic coordinate. Thus, we can
directly write the equations of motion with respect to θ as,
pθ
pθ = constant and θ̇ =
mr2
118
MPDSC 1.1 KSOU
1. What is Hamiltonian of a system?
2. Define Hamiltonian as Legendre transform of Lagrangian.
3. What is the physical significance of Hamiltonian of a system?
4. What are cyclic coordinates?
5. Mention any two advantages of Hamiltonian formulation.
7.9 Keywords
• Hamiltonian
• Legendre transform
• Cyclic coordinates
• Hamilton’s equation of motion
1. An object of mass m is suspended by a massless spring of spring constant k. The suspen-

sion point is pulled upwards with a constant acceleration ao . Find the Hamiltonian of the
system and Hamilton’s equation of motion.
Answer: Let the vertical be the z-axis. As the acceleration due to gravity is downwards,
taking the net acceleration as ( g − ao ).
1 2
Potential energy V = kz + m( g − ao )z
2
1 2
Kinetic energy T = mż
2
Then the Hamiltonian of the system will be H = T + V
1 2 1 2
H= mż + kz + m( g − ao )z
2 2
119
MPDSC 1.1 KSOU
But the variables in Hamiltonian must be coordinate and momenta, not velocities. Replac-
ing the velocity term appropriately by momentum, we get the Hamiltonian as
p2z 1
H= + kz2 + m( g − ao )z
2m 2
The Hamilton’s equation of motion are
∂H pz
ż = =
∂pz m
∂H
p˙z = − = −kz − m( g − ao )
∂z
We can combine the above two equations and write a single equation of motion as
1 1
z̈ = p˙z = [−kz − m( g − ao )]
m m
z̈ = −kz − m( g − ao )
2. Obtain the Hamiltonian of a charged particle in an electromagnetic field.
Answer: The potential enegy of a charged particle in an electric field is given by
~ · ~v) = qφ − q( A x ẋ + Ay ẏ + Az ż)
U = qφ − q( A
Similarly the kinetic energy is
1 2 m
T= mv = ( ẋ2 + ẏ2 + ż2 )
2 2
Then the Lagrangian of the particle will be
m 2
L = T−V = ( ẋ + ẏ2 + ż2 ) − qφ + q( A x ẋ + Ay ẏ + Az ż)
2
120
MPDSC 1.1 KSOU
Now let us find the conjugate momenta of x, y and z respectively.
∂L p x − qA x
px = = m ẋ + qA x =⇒ ẋ =
∂ ẋ m
∂L py − qAy
py = = mẏ + qAy =⇒ ẏ =
∂ẏ m
∂L pz − qAz
pz = = mż + qAz =⇒ ż =
∂ż m
The Hamiltonian is defined as the Legendre transform of Lagrangian
H= ∑ pi q̇i − L = px ẋ + py ẏ + pz ż − L
i
m
2 2 2
H = p x ẋ + py ẏ + pz ż − ( ẋ + ẏ + ż ) − qφ + q( A x ẋ + Ay ẏ + Az ż)
2
m
H = p x ẋ + py ẏ + pz ż − ( ẋ2 + ẏ2 + ż2 ) + qφ − q( A x ẋ + Ay ẏ + Az ż)
2
p x − qA x py − qAy pz − qAz
H = px + py + pz
m m m
m p x − qA x 2 m py − qAy 2 m pz − qAz 2

− − − + qφ
2 m 2 m 2 m
p x − qA x py − qAy pz − qAz
−q( A x + Ay + Az )
m m m
1 1 2 1
H= ( p x − qA x )2 + py − qAy + ( pz − qAz )2 + qφ
2m 2m 2m
1 ~ |2 + qφ
H= |~p − q A
2m
This is the Hamiltonian of a charged particle in electromagnetic fields.
3. Obtain Hamilton’s equations for a simple pendulum.
Answer: Consider a simple pendulum of length l. The angle subtended by the string with
respect the vertical is used as generalized coordinate.
The expression for kinetic energy will be
1 2 2
T= ml θ̇
2
The expression for the potential energy will be
121
MPDSC 1.1 KSOU
V = mgl (1 − cos θ )
Then the Lagrangian of the system will be
1 2 2
L = T−V = ml θ̇ − mgl (1 − cos θ )
2
∂L pθ
pθ = = ml 2 θ̇ =⇒ θ̇ =
∂θ̇ ml 2
Then the Hamiltonian of the system will be H = pθ θ̇ − L
pθ 1 2 p θ 2
H = pθ − ml + mgl (1 − cos θ )
ml 2 2 ml 2
1 2 2
H= ml θ̇ + mgl (1 − cos θ )
2
Then the Hamiltonian equation of motion will be
∂H p
θ̇ = = θ2
∂pθ ml
∂H
p˙θ = − = −mgl sin θ
∂θ
Now we can combine the above two equations to obtain the solution
g sin θ g
θ̈ = − =⇒ θ̈ + sin θ = 0
l l
This is the equation of motion of the pendulum
1. Describe Legendre transformations and compute the Legendre transform of the function
f ( x ) = ax2 − bx + c
2. Obtain the Hamiltonian equations of motion through Legendre transformation of Lagrangian.
3. What are cyclic coordinates and explain their importance.
122
MPDSC 1.1 KSOU
4. Write a note on Routh’s principle.
5. Derive the Hamilton’s equation of motion using variational principle.
6. Deduce the equations of motion of a linear harmonic oscillator using Hamiltonian formu-
lation.
1. Hamiltonian is defined as Legendre transform of Lagrangian. Hamiltonian is also used

like Lagrangian to determine equations of motion of systems.
2. H f = ∑i pi q̇i − L.
3. For conservative systems, Hamiltonian is the total energy function.
4. Coordinates that are absent in the Hamiltonian are called cyclic coordinates
5. Hamiltonian formulation involves first order differential equations that are easy to solve.
In case of the presence of cyclic coordinates, the equations of motion can be directly written
in case of Hamiltonian.
7.13 References
tion.

pany Ltd.
123
UNIT-8: Canonical transformations
Point transformations, general canonical transformation, invariants of canonical transforma-

tions, generating functions, applications of canonical transformations.
8.0 Objectives
• Describe the importance of canonical transformations.
• Define point transformations and obtain the condition for a point transformation to be
canonical transformation.
• Describe the generating function and mention different types of generating function.
• Explain symplectic approach to canonical transformations
• Describe infinitesimal canonical transformations.
• Write a note on canonical invariants.
• Use the canonical transformations to find the equations of motion of a simple harmonic
oscillator.
8.1 Introduction
When applied in a straight forward manner, the Hamiltonian does not give any decrease in the
difficulty of solving any given problem. One ends up with the same differential equation to be
solved as provided by the Lagrangian procedure. The advantages of Hamilton’s formulation
lie not in the computational case but rather in the equal status of the coordinates and momenta.
There is one type of problem for which the solution of the Hamilton’s equation is trivial.
Let us consider a system of n degrees of freedom and then using Hamilton’s equations of
motion we can write that
MPDSC 1.1 KSOU
∂H ∂H
q̇i = and ṗi = − Where, i = 1, 2, , , , n
∂pi ∂qi
If all the coordinates become cyclic then ṗi = 0 for all values of i. This implies we can write
pi = αi . Since the generalised momentum is conserved, we can write it as equal to some constant
α. Then we can rewrite the Hamiltonian as a function of αi . That is H = H (α1 , α2 , , , αn ).
∂H
=⇒ q̇i = = ωi
∂pi
By integrating with respect to time we can get qi (t).
qi (t) = ωi t + qi0
This can be done only if all the coordinates are cyclic. When this is not the general case, if
there are any transformations which helps to change to new coordinates in which every variable
is cyclic, then the problems become trivial and can be solved as above. This is the spirit of
canonical transformations.
8.2 Point transformation
Consider Lagrangian of the system L(q, q̇, t). We want to change to a new coordinate system in
which the Lagrangian is written as L̃(s, ṡ, t). One of the important features of the Lagrangian is
that it does not change its form when changing from one coordinate to the other and therefore,
L̃(s, ṡ, t) = L(q(s, t), q̇(s, ṡ, t), t)
In order to study the motion of a particle in a plane, one can either use a Cartesian or plane
polar coordinate. Both are perfectly valid. However, one is more suitable than the other in
various cases. For example in the central force problem neither x nor y are cyclic. Whereas, θ is
cyclic. So using plane polar coordinates is more advantageous. So we must first derive a specific
procedure for transforming from one set of variable to some other that may be more suitable.
These transformations are called point transformations as they map point to point. A simple
transformation could be as follows.
Qi = Qi (q, t)
125
MPDSC 1.1 KSOU
These transformations one-to-one map a point of phase space to another point of trans-
formed new phase space.
8.3 Canonical transformations
We learned about point transformations defined as Qi = Qi (q, t) in the previous section. But
such a transformation will not work here because in Hamiltonian both q and p are independent
variables. The concept of transformations should be widened to include the simultaneous trans-
formation of q as well as p. There is another restriction that we have to keep in mind that in the
old Hamiltonian, q and p were both canonical variables, so even in the New Hamiltonian, it has
to be maintained that they remain canonical.
So we will introduce a function K ( Q, P, t) such that the equation of motion in the new set are
∂K ∂K
Q̇i = and Ṗi = −
∂Pi ∂Qi
If Q and P are canonical coordinates, they have to satisfy the modified Hamilton’s principle
just like the old canonical variables satisfied the principle.
Z t2 Z t2
δ lim( Pi Q̇i − K )dt = 0 as well as δ lim( pi q̇i − H )dt = 0
t1 t1
Just because both the above equations are simultaneously equal to zero, doesn’t mean the
integrands are equal. But they must be connected by the relationships of the following form.
dF
λ( pi q̇i − H ) = Pi Q̇i − K +
dt
Where F is a function of the phase space coordinates with continuous second derivatives
and λ is a constant which is known as the scale of the transformation. λ increases or decreases
dF
the size. When λ=1 we will get ( pi q̇i − H ) = Pi Q̇i − K + and it is called simply as canonical
dt
transformation. If there is a scaling factor involved, that is when λ 6= 1 then it is called extended
canonical transformations. If there is no explicit dependence on time, it is known as Restricted
canonical transformation.
The function F acts a bridge between the old set of coordinates and the new set and is called
as the generating function.
126
MPDSC 1.1 KSOU
8.4 Generating functions
There are four different types of generating functions classified based on the independent vari-
ables involved in the generating functions that generate the transformation. In general, any
linear combination of these type of generating functions will also be a generating function that
results in a valid canonical transformations.
1. Let us assume the generating function is a function of old generalized coordinates and
new generalized coordinates.
F = F1 (q, Q, t)
then
dF1
( pi q̇i − H ) = Pi Q̇i − K +
dt
∂F1 ∂F1 ∂F
Pi Q̇i − K + + q̇i + 1 Q̇i
∂t ∂qi ∂Qi
Since the old and new coordinates are separately independent, the above equation is valid
only if the coefficients of q̇i and Q̇i individually become equal to zero. Then, we get,
∂F1 ∂F1 ∂F1

pi = ; Pi = − ; K= H+
∂qi ∂Qi ∂t
These equations provide the transformation equations from the old canonical variables to
new set of canonical variables.
It sometime so happen that, it is not suitable to describe the canonical transformation by

the generating function of the type F( q, Q, t), then we extend the generating function by
considering other combinations of variables.
2. Let the second type of generating function be the function of old coordinates and new
momenta.
F = F2 (q, P, t) − Qi Pi
Then,
127
MPDSC 1.1 KSOU
dF2
( pi q̇i − H ) = Pi Q̇i − K +
dt
∂F1 ∂F1 ∂F
( pi q̇i − H ) = Pi Q̇i − K + + q̇i + 1 Ṗi − Pi Q̇i − Qi Ṗi
∂t ∂qi ∂Pi
Therefore the transformation equations become,
∂F2 ∂F2 ∂F2

pi = ; Qi = ; K= H+
∂qi ∂Pi ∂t
3. Let the generating function be the function of old momenta and new coordinates. Then,
F = F3 ( p, Q, t) + qi pi
dF3
( pi q̇i − H ) = Pi Q̇i − K +
dt
by the same process used in the previous case we will get the transformation equations as,
∂F3 ∂F3 ∂F3

qi = − ; Pi = − ; K= H+
∂pi ∂Qi ∂t
4. The fourth kind of generating function is assumed to be a function of old momenta and
new momenta. Then,
F4 ( p, P, t)
F = qi pi − Qi Pi + F4 ( p, P, t)
With the similar procedure, we can write the transformation equations as,
∂F4 ∂F4 ∂F4

qi = − ; Qi = − ; K= H+
∂pi ∂Pi ∂t
It is tempting to look at all the four generating function and see that they are Legendre
transformations of each other. You may think that we can get all the four generating functions
from one another. But that is not possible. Further, some kind of generating functions are more
suitable than the others in some cases.
128
MPDSC 1.1 KSOU
8.5 Symplectic approach to Canonical Transformation
Another method of treating canonical transformations, can be derived in terms of the matrix or
symplectic formulation of Hamilton’s equation by introducing a restricted canonical transfor-
mation. Let,
Qi = Qi (q, p) Pi = Pi (q, p)
The time derivative of Qi is
∂Qi ∂Qi ∂Qi ∂H ∂Qi ∂H

Qi = q˙j + ṗ j = −
∂q j ∂p j ∂q j ∂p j ∂p j ∂q j
The differentiation of the Hamiltonian H (q, p, t) with respect to P is,
∂H ∂H ∂p j ∂H ∂q j
= +
∂Pi ∂p j ∂Pi ∂q j ∂Pi
Comparing the above equations we get
∂H
Q̇i =
∂Pi
That is the transformation is canonical only if
∂Qi ∂p j
( )q,p = ( )Q,P
∂q j ∂Pi
∂Qi −∂q j
( )q,p = ( )Q,P
∂p j ∂Pi
Since finding the generating function is difficult, a new approach is necessary. Let us take a
system with one degree of freedom. The Hamilton’s equation can be written as,
∂H ∂H
q̇ = and ṗ = −
∂p ∂q
We will write this in a matrix form. with,

 
q
η= 
p
129
MPDSC 1.1 KSOU
and  
q̇
η̇ =  
ṗ
 
∂
 ∂q 
∇η =  ∂ 

∂p
Now the Hamilton’s equation will be like
 
  ∂H
q̇  ∂p 
η̇ =   = 
 ∂H 

ṗ −
∂q
 
 ∂
0 1  ∂q 
=  H
 ∂ 
−1 0
∂p
 
0 1
J= 
−1 0
∴ η̇ = J ∇η H
If we take a two dimensional case, the canonical coordinates are (q1 , q2 , p1 , p2 ) then
 
q
 1
 q2 
 
η= 
 p1 
 
 
p2
 
q˙1
 
 q˙2 
 
η̇ =  
 p˙1 
 
 
p˙2
130
MPDSC 1.1 KSOU
= 
  ∂
0 0 1 0  ∂q1 
 
  ∂ 
 0 0 0 1  ∂q2 
  
η=  ∂ H
 
 −1 0 0 0 


   ∂p1 
−1
 
0 0 0  ∂ 
∂p2
∴ η̇ = J ∇η H
The structure of J matrix is something which we can easily figure out

 
0n × n In×n
J= 
− In×n 0n × n
2n×2n
η̇ = J ∇η H
above equations are known as the Hamilton’s equation in symplectic form.
Properties of J matrix
1. J 2 = − I2n×2n
2. J T = J −1 = − J
3. det(J)=+1
If M is a 2n ×2n such that MJM T =J, then M is called as symplectic matrix.

Now let us take the restricted canonical transformation,
   
q Q
 1  1
 ,   , 
   
   
η= ,  and ζ= , 
   
   
 
 , 
 
 , 
   
pn Pn
Then,
d ∂ζ dηi
dt
ζ= ∑ ∂ηij dt
j
∂ζ i
is the Jacobian which is denoted by M. Now let us take the gradient as,
∂η j
131
MPDSC 1.1 KSOU
∂ ∂ζ j ∂
(∇η )i =
∂ηi
= ∑ lim
j ∂ηi ∂ζ j
= ∑ lim = M ji (∇zeta ) j
j
We can write the Hamilton’s equation will become
ζ̇ = Mη̇ = MJ M T ∇ζ H
if the transformation is canonical

ζ̇ = J ∇ζ H
So if we compare the two previous equation we will get transformation is canonical if and only
if the Jacobian matrix is symplectic. Hence this is a tool to verify whether a given set of trans-
formation is canonical or not.
1. Let us take the identity transformation that is Qi = qi and Pi = pi

   
q Q
 1  1
 ,   , 
   
   
η= ,  and ζ= , 
   
   
 
 , 
 
 , 
   
pn Pn
∂ζ i
=⇒ ζ = η The Jacobian matrix Mij = = δij The Jacobian matrix is the identity matrix
∂η j
in this case and identity Matrices are symplectic. Therefore the above transformation is
canonical.
2. Consider another set of transformation Q = qcosα − psinα and P = qsinα + pcosα
 
∂Q ∂Q
 ∂q ∂p 
M=  ∂P ∂P 

∂q ∂p
 
cosα −sinα
M= 
sinα cosα
   
cosα − sinα 0 1 cosα sinα
∴ MJ M T =    
sinα cosα −1 0 −sinα cosα
132
MPDSC 1.1 KSOU
 
0 1
= =J
−1 0
Since the above J matrix is symplectic, The given set of transformation is canonical.
8.6 Infinitesimal Canonical transformation
We are transforming from (q, p) → ( Q, P) in which Qi = Qi (q, p, t) and Pi = Pi (q, p, t). We

are going to make that this transformation is infinitesimal in which ζ = η + δη. We will take a
generating function F2 (q, p, t) = ∑ limi qi Pi + eG (q, P, t). Where e is an infinitesimal number. By
using the generating function we will find Qi and pi ,
∂F2 ∂G
Qi = = qi + e
∂Pi ∂Pi
∂F2 ∂G
pi = = Pi + e
∂qi ∂qi
∂G ∂G
∴ Qi − qi = δqi = e =
∂Pi ∂pi
P is related to p by a linear equation and they are equal in the zeroth order Similarly,
∂G
Pi − pi = δpi = −e
∂qi
∴ δη = eJ ∇η G
The Jacoian of the matrix of this transformation is
∂ζ i
Mij = =⇒ M = I + eJ (∇l ∇k G )
∂η j
where ∇l ∇k G is the Laplacian

M T = I − eJ (∇l ∇k G )
since I and (∇l ∇k G ) are both symmetric matrices.

By applying the symplectic condition we can write,
MJ M T = ( I + eJ (∇l ∇k G )) J ( I − eJ (∇l ∇k G ))
= J + O + O ( e2 ) = J
133
MPDSC 1.1 KSOU
Hence the infinitesimal transformations are symplectic and therefore they are canonical. You
can try this for finite transformations also but since finite transformations can be made from an
infinite number of infinitesimal transformation, the jacoian of the finite transformation can be
written in terms of product of jacobians of infinitesimal transformation and therefore the finite
transformations are also symplectic.
8.7 Canonical Invariants
When we change from one set of coordinates to another sets but some variables remain invari-
ant. Such as length when there is an orthogonal transformation. Let us take two functions of
canonical variable u(q, p) and v(q, p).
The Poisson bracket is defined as
∂u ∂v ∂u ∂v
[u, v]q,p = ∑( − )
i
∂qi ∂pi ∂pi ∂qi
The three fundamental Poisson brackets are as follows,
∂qi ∂q j ∂q ∂q j
[qi , q j ]q,p = ∑( − i )=0
k
∂qk ∂pk ∂pk ∂qk
since q and p are independent variables, above identity becomes zero. Similarly we can show
[ pi , p j ] = 0
However, interestingly,
∂qi ∂p j ∂q ∂p j
[qi , p j ]q,p = ∑( − i ) = δjk
k
∂qk ∂pk ∂pk ∂qk
because the second term is always zero, for the first term to be non zero, i must be equal to k
and j must be equal to k. This is represented by the kronecker delta function.
In Quantum mechnaics, the Poisson brackets are replaced by the commutator brackets. We
can write the Poisson bracket in a symplectic form as,
∂u T ∂u
[u, v]q,p = ( ) J( )
∂η ∂η
134
MPDSC 1.1 KSOU
where  
∂u
 ∂qi 
 
 , 
 
∂u  
( ) = ∇η u =  , 
 
∂η  

 , 

 
 ∂u 
∂pn
∂u ∂v ∂u ∂v
= ∑( − )
i
∂qi ∂pi ∂pi ∂qi
When we transform the old set to a new set, they are connected by the Jacobian matrix. (q, p) →
( Q, P)
∇η = M T ∇ζ
where ζ → ( Q, P) and η → (q, p) and M is the Jacobian matrix
[u, v]η = (∇η u)T J (∇η u)
(∇ζ u)T MJ M T (∇η u)
for this transformation to be canonical the Jacobian has to be symplectic MJ M T = J which gives
us
(∇ζ u)T J (∇η u) = [u, v]ζ
Any Poisson bracket of any arbitrary coordinate will be exactly same in any new coordinate
system Poisson brackets are invariant under canonical transformation.
In corollary, Poisson brackets which preserves the form are canonical in nature. The other
invariance is the phase space volume element. The transformation from (q, p) → ( Q, P). The
elementary volume in the old set is (dq1 dq2 dqn ) which should be equal to | M |(dQ1 dQ2 dQn ).
If the transformation is canonical, then the Jacobian matrix is symplectic and hence | M| = +1
which means the volume element is invariant.
135
MPDSC 1.1 KSOU
8.8 Applications of canonical transformations
As an example, let us consider a canonical transformation that can be used to solve the problem
of the simple harmonic oscillator in one dimension. If the force constant is k, the Hamiltonian
for this problem can be written as
p2 kq2
H= +
2m 2
Designating the ratio k/m by ω 2 , the Hamiltonian can be written as
1 2
H= p + m2 ω 2 q2
2m
This form of the Hamiltonian, as the sum of two squares suggests a transformation in which
the Hamiltonian is cyclic in the new coordinate. If we could find a canonical transformation of
the form,
f ( p)
p = f ( p) cos Q and q= sin Q
mω
Then the Hamiltonian as a function of Q and P would be simply
f 2 ( P) 2 2
f 2 ( P)
K=H= cos Q + sin Q =
2m 2m
In the above transformed Hamiltonian, Q is cyclic. The problem is to find the form of the yet
unspecified function f ( P) that makes the transformation canonical . If we use the generating
function of the first kind given by
mωq2
F1 = cot Q
2
We can then provide the corresponding transformation equations as,
∂F1 ∂F1 mωq2

p= = mωq cot Q and P=− =
∂q ∂Q 2 sin2 Q
Then solving for q and p from the above equations, we can write,
r
2P p
q= sin Q and p= 2pmω cos Q
mω
From the above set of equations, we can write
136
MPDSC 1.1 KSOU
√
f ( P) = 2mωP
Then the New Hamiltonian will become,
H = ωP
Since the Hamiltonian is cyclic in Q, the conjugate momentum P is a constant. It is seen from
the equation that P is in fact equal to the constant energy divided by ω.
E
P=
ω
The equation of motion for Q reduces to the simple form,
∂H
Q̇ = =ω =⇒ Q = ωt + α
∂P
Where α is a constant of integration fixed by the initial conditions. Then the equations of
motion for q and p will become,
√
r
2E
q= sin(ωt + α) and p= 2mE cos(ωt + α)
mω 2
1. What is the intention of canonical transformation.
2. Define the condition for canonical transformation.
3. How many different types of generating functions are there?
4. What are canonical invariants?
8.10 Keywords
• Canonical transformations
• Canonical invariance
137
MPDSC 1.1 KSOU
• Generating functions
• Symplectic approach
1. Show that the following transformation is canonical
2qe−α sin p
p p
Q= 2qeα cos p and P=
Where α is a constant.
Answer: The transformation is said to be canonical ifPdQ − pdq is a perfect differential.
we have
p
Q= 2qeα cos p
!
p cos p
∴ dQ = eα − 2q sin pdp + p dq
2q
Then,
" !#
hp i cos p
2qe−α sin p
p
PdQ − pdq = eα − 2q sin pdp + p dq − pdq
2q
PdQ − pdq = −2q sin2 pdp + sin p cos ppdq − pdq

2 sin 2p
PdQ − pdq = −2q sin pdp + − p dq
2

d sin 2p
−p = cos 2p − 1 = −2 sin2 p
dp 2

sin 2p sin 2p
PdQ − pdq = qd −p + − p dq
2 2
138
MPDSC 1.1 KSOU

sin 2p
PdQ − pdq = d q −p
2
Since the right hand side is a perfect differential, the canonical transformation is canonical.
2. Show that the transformation
mωq2
p = mωq cot Q and P=
2 sin2 Q
is canonical, and obtain the generator of the transformation.
Answer: From the given data,
p
cot Q =
mωq
Differentiating
sin2 Q

2 1 dp pdq pdq dp
−cosec QdQ = − 2 =⇒ dQ = −
mω q q mω q2 q
mωq2 sin2 Q

pdq dp
PdQ − pdq = − − pdq
2 sin2 Q mω q2 q
q2

pdq dp
PdQ − pdq = − − pdq
2 q2 q
1 pq
PdQ − pdq = − ( pdq + qdp) = −d( )
2 2
Thus PdQ − pdq is a total differential, the given transformation is canonical.
The generating function will be the integral of this total differential. Hence the generating
function will be
1 1 1
F= pq = qmωq cot Q = mωq2 cot Q
2 2 2
This is the generating function that is a function of old and new coordinates
139
MPDSC 1.1 KSOU
3. Show that the following transformation is canonical
√ √
q= 2P sin Q and p= 2P cos Q
Answer:
√
Consider q = 2P sin Q
Differentiating the above equation,
√ sin Q
dq = 2P cos QdQ + √ dP
2P
Now consider,
√ √
sin Q

PdQ − pdq = PdQ − 2P cos Q 2P cos QdQ + √ dP
2P
PdQ − pdq = PdQ − 2P cos2 QdQ − cos Q sin QdP

sin 2Q sin 2Q
PdQ − pdq = PdQ − 2P cos2 QdQ − dP = (1 − 2 cos2 Q) PdQ + dP
2 2

sin 2Q P sin 2Q
PdQ − pdq = − cos 2QPdQ − dP = −d
2 2
This is a total differential. Hence the given transformations are canonical.
1. What are canonical transformations? Explain their importance.
2. Deduce the conditions for a point transformation to be canonical.
3. Explain the four different types of generating functions with the corresponding transfor-
mation equations involved.
4. Write a note on symplectic approach to canonical transformations.
5. Describe infinitesimal canonical transformations.
6. Write a note on canonical invariants.
140
MPDSC 1.1 KSOU
1. The intention of canonical transformation is to increase the number of cyclic coordinates

and if possible to make all the coordinates cyclic.
2. A transformation from (q, p, t) to ( Q, P, t) is said to be canonical if it satisfies the condition

dF
λ( pi q̇i − H ) = Pi Q̇i − K + .
dt
3. There are four different types of generating functions
4. Physical quantities that do not change under a canonical transformations are called canon-
ical invariants.
8.14 References
tion.

pany Ltd.
141

Classical Mechanics Block 2/4

Uploaded by

Copyright:

Available Formats

Classical Mechanics Block 2/4

Uploaded by

Document Information

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

Classical Mechanics Block 2/4

Uploaded by

Copyright:

Available Formats

UNIT-5: Central force problem

After studying this unit you will be able to

• Describe what are central forces.

• Establish conservation of angular momentum in central force field.

• Obtain equation of motion and first integrals.

• Classify the orbits based on the energy of the particle

• Explain Binet’s equation and Virial theorem.

• Explain the Kepler’s problem and obtain the equation of orbits.

5.2 Central force

5.2.1 Two body problem

Figure 5.11: Two body problem with central force

5.3 Conservation of angular momentum

lx = µ(yż − zẏ) ly = µ(z ẋ − x ż) lz = µ( x ẏ − y ẋ )

dlx  ∂V z   ∂V y  dly  ∂V x  ∂V z  dlz  ∂V y   ∂V x 

It is evident from the above set of equations that

dlx dly dly

lx = constant; ly = constant; lz = constnat

5.4 Equations of motion and first integrals

integrating the above equation under proper limits

5.5 Classification of orbits

Figure 5.12: The effective potential of central force field.

5.6 Differential equation for the orbit or Binet’s equation

We have the the equations of motion in central force field as

5.7 Virial theorem

let us define a quantity

The total time derivative of this quantity is

The first term can be written as

∑~ṙi · ~pi = ∑ mi~ṙ~i˙ṙi = ∑ mi v2i = 2T

While the second term becomes

∑ ~ṗi ·~ri = ∑ ~Fi ·~ri

Therefore the equation reduces to

If V is a power law function of r

5.8 Condition for closed orbits

5.9 Kepler problem of inverse square force

which has the solution

B and θ 0 being two constants of integration,in terms of r the solution

We write the solution in the form of θ

This indefinite integral has a standard form

to apply this we must set

and the discriminant q is therefore

with these substitutions the equation of θ becomes

In terms of the semi-major axis the eccentricity of the ellipse is given by

5.10 Check your progress

Check your progress by answering the questions below.

1. What is a central force?

3. Briefly explain why angular momentum is conserved in central force field.

4. Classify different types of orbits in central force field.

5.12 Worked examples

1. The orbit of a particle of mass m moving in a central force is given by r = kθ 2 , where k is a

Answer: In central force problem, the equation of orbit is given by

Then substituting this in the equation of orbit we get,

Answer: We know the equation of ellipse

a(1 − e2 ) = 8200 km and e = 0.2

Then the altitude of perigee = rmin − R E = 453.4 km

And the altitude of apogee = rmax − R E = 3870.0 km

Then the total energy will be

5.13 Questions for self study

1. Explain what is central force?

3. Explain conservation of angular momentum in central force field.

5. Describe how orbits of a particle in central force field are classified.

dlx ∂V z ∂V y dly ∂V x ∂V z dlz ∂V y ∂V x