Hamiltonian Mechanics

ide
Sk Jahiruddin*

gu *Assistant Professor
Sister Nibedita Govt. College, Kolkata
ics
Author was the topper of IIT Bombay M.Sc Physics 2009-2011 batch
He ranked 007 in IIT JAM 2009 and 008 (JRF) in CSIR NET June 2011

He has been teaching CSIR NET aspirants since 2012

ys
Ph

1
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Contents

1 Phase space 4

ide
2 Hamiltonian 5

2.1 Hamiltonian of particle in EM field . . . . . . 6

2.2

2.3
gu
Does all Lagrangian have a proper Hamiltonian?

Some solved problems using Hamilton’s equa-

tion of motion . . . . . . . . . . . . . . . . . .
6

8
ics
3 Drawing Phase space 16
ys

3.1 Fixed points and separatrix . . . . . . . . . . 23

4 Canonical Transformation 29
Ph

4.1 Basic ideas . . . . . . . . . . . . . . . . . . . . 29

4.2 Generating functions . . . . . . . . . . . . . . 31

4.3 Four basic types of generating functions . . . 33

[email protected] 2 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

4.4 Canonical transformation and cyclic coordinates 37

5 Poission Bracket 38

ide
6 Equation of motion of a dynamical variable 48

7 Infinitesimal Canonical Transformation 50

7.1 gu
Generators of linear momentum, angular mo-
mentum and time . . . . . . . . . . . . . . . . 52
ics
7.2 integration of the equation of motion . . . . . 53

8 Phase space volume and Liouville’s Theorem 55

ys

9 Hamilton Jacobi and Action Angle 58

Ph

9.1 Hamilton Jacobi equation . . . . . . . . . . . 58

9.2 Separation of variable and H-J method . . . . 64

9.3 Action-Angle Variables . . . . . . . . . . . . . 67

[email protected] 3 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

We see that Lagrange’s equations for n coordinates leads

to total n 2nd-order differential equations. And to solve those
we need 2n initial conditions ( qi (t = 0) q̇i (t = 0)) also. Now
the Qs is can we do with 1st-order differential equations? Yes,

ide
we can, but for that we’ll need 2n equations.

∂L (qj , q̇j , t)
We take qi and the conjugate momenta, pi ≡
∂ q̇i

1
gu
as independent variables.

Phase space
ics
Consider coordinates and momenta as independent. Now the
state of the system is given by
ys

(q1 , . . . , qn , p1 , . . . , pn ). Consider just a particle first. At a

particular time the particle has a definite coordinate and mo-
mentum. Consider that coordinate and momentum as a state.
Ph

Draw coordinate in x axis and momentum in y axis. That is

the phase space of the particle. Every state represent a point
in the 2 dimensional phase space. Now if the system have n
independent coordinate and so n no of momentum, then ev-
ery state represent a point in the 2n -dimensional phase space.
Keep in mind that we are switching the independent variables

[email protected] 4 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

(qi , q̇i , t) → (qi , pi , t) We will draw many phase spaces later.

First define the Hamiltonian.

ide
2 Hamiltonian

H(q, p, t) = q̇i pi − L(q, q̇, t) (2.1)

gu
Hamiltonian is same as the ”energy function” defined earlier,
just written in terms of q and p instead of q and q̇.
ics
You can see that the total derivative of Hamiltonian would
be
∂L
dH = q̇i dpi − ṗi dqi − dt
∂t
ys

This must be equivalent to

∂H ∂H ∂H
dH = dpi + dqi + dt
∂pi ∂qi ∂t
Ph

Hence we find
∂H ∂H ∂H ∂L
= q̇i = −ṗi and =− (2.2)
∂pi ∂qi ∂t ∂t

These equations are called Hamilton’s equations.

[email protected] 5 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

2.1 Hamiltonian of particle in EM field

For a particle in an EM field we know the Lagrangian is

ide
1
L = mv 2 − qφ + qA · v
2

When writing as a sum of coordinates we get

gu L=
m 2
ẋ − qφ + qAi ẋi
2 i
=⇒pi = mẋi + qAi
mẋ2i
ics
=⇒ = H = (mẋi + qAi ) ẋi − L = + qφ
2

Now we see that the conjugate momentum is pi = mẋi +

qAi .
ys

Writing H in terms of conjugate momentum we get

(pi − qAi )2
H (xi , pi ) = + qφ (2.3)
Ph

2m

2.2 Does all Lagrangian have a proper Hamil-

tonian?

We see through an example

[email protected] 6 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Example 2.1: The Lagrangian of a system by a single

generalized coordinate q is
1
L = q̇ sin2 q

ide
2
Its Hamiltonian is [TIFR 2017]
(a) not defined (b) zero (c) −q̇ sin2 q (d)
q̇ p − 21 sin2 q

gu
Solution: We see that

p=
∂L sin2 q
∂ q̇
=
2
; pq̇ = q̇
sin2 q
2
ics
hence we see
H = pq̇ − L = 0
Now the question is can the Hamiltonian of a physical system
ys

be ZERO? Ans is NO! To explore more let’s try to calculate

the Lagrangian equation of motion of the given Lagrangian.
Ph

∂L sin2 q ∂L 2q̇
p= = ; = sin q cos q
∂ q̇ 2 ∂q 2
   
d ∂L d ∂L ∂L
= q̇ sin q cos q; − =0
dt ∂ q̇ dt ∂ q̇ ∂q
So we did not get a valid Lagrangian equation of motion
of the particle.

[email protected] 7 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Hence we conclude that the Hamiltonian is NOT DE-

FINED. And we see that there may not always be a valid
Hamiltonian for a given Lagrangian.

ide
2.3 Some solved problems using Hamilton’s
equation of motion
Hamilton’s canonical equations of motion

gu q˙i =
∂H
∂pi
; ṗi = −
∂H
∂qi

Example 2.2: Write down the Hamiltonian and Hamil-

(2.4)
ics
ton’s equation of motion for central inverse square force
F (r) = rk2 .
ys

Solution: Central force is azimuthaly symmetric. Its La-

grangian is
1  2 2 2
 k
L = µ ṙ + r θ̇ +
Ph

2 r
To calculate the Hamiltonian we start with calculating the
momentum.
∂L ∂L
pr = = µṙ and pθ = = µr2 θ̇
∂ ṙ ∂ θ̇
So the Hamiltonian is
1  2 2 2
 k
H = pr ṙ + pθ θ̇ − L = µ ṙ + r θ̇ −
2 r
[email protected] 8 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Replace the velocities with momentum

pr pθ
ṙ = and θ̇ = 2
µ µr

ide
p2r p2θ k
H= + −
2µ 2µr2 r
Hamilton’s equation’s of motions are - Equation-2.4

∂H pr ∂H pθ

and
gu ṙ =
∂pr
=
m

p˙r = −
and θ̇ =

∂H
∂r
p2θ k
= 3− 2
µr r
∂pθ
= 2
µr
ics
∂H
p˙θ = − =0
∂θ
Example 2.3: Find Hamilton’s equation of motion and
ys

solve the coordinate and momentum for time dependent

Hamiltonian given-

p2
H(p, q, t) = − mtq
Ph

2m

p
q̇ = and ṗ = mt
m
Second equation gives
1
p(t) = mt2 + p0
2
[email protected] 9 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Substitute the value of p(t) into first equation

1 p0
q̇ = t2 +
2 m
integrating we get

ide
1 p0 t
q = t3 +
6 m
value of p0 depends on the initial condition.
Example 2.4: Problem: Find the Hamiltonian for an-
harmonic oscillator with the given lagrangian

gu 1 1
L = ẋ2 − ωx2 − αx3 + βxẋ2
2 2
ics
Solution:
∂L
p= = ẋ + 2βxẋ
∂ ẋ
p
ẋ =
(1 + 2βx)
ys

H = pẋ − L
substitute all ẋ with p, x, β
p2 1
− ω 2 x2 + αx3
Ph

H=
2(1 + 2βx) 2

Example 2.5: Problem:Describe the motion of a par-

ticle with Hamiltonian
 2
1 2 1 2 2 1 2 1 2 2
H = p + ω0 x + λ p + ω0 x
2 2 2 2

[email protected] 10 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Solution: Hamilton’s canonical equations of motion

∂H ∂H
ẋ = ; ṗ = −
∂p ∂x

ide
 
1 2 1 2 2
ẋ = p + 2λ p + ω0 x p
2 2
 
2 1 2 1 2 2
ṗ = −ω0 x − 2λ p + ω0 x ω0 x
2 2
Take
gu 1
E0 = ω02 a2 ; ω = (1 + 2λE0 ) ω0
2
Take derivative again →substituite ẋ and ṗ and get harmonic
ics
oscillator solution. (Calculation is little tough, but you need
to do it )

x = a cos (ωt + φ) p = −ω0 a sin (ωt + φ)

ys

Example 2.6: Find the equations of motion for light

traveling inside a medium using Hamilton’s equation.
Hamiltonian is given
Ph

c|~p|
H=
n (~p, ~r)
where n is the refractive index of the medium. Deter-
mine the trajectory for the case when n(r) = ax.

[email protected] 11 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Solution: Using Hamilton’s equations of motion

p~ c|p| ∂n c|p|
ṙ = c − 2 ; ṗ = ∇n
n|p| n ∂~p n2

ide
when n = ax, then
 
y
x = c1 cosh + c2
c1

gu
Example 2.7: Problem: Describe the motion of
a charged particle in uniform magnetic field B in ẑ
direction by solving the Hamilton’s equations of motion.
ics
Solution: We know the Hamiltonian for a charged particle
in uniform magnetic field is.
ys

1
H= (p − qA)2 + qφ
2m
For magnetic field in ẑ direction we can take vector potential
Ph

A either ŷBx or x̂By.

So our Hamiltonian can be written [taking A = ŷBx

p2x + p2y 1
H= + (py − qBx)2
2m 2m
H does not depend on y and z. So using Hamilton’s equation
of motionp˙y = p˙z = 0 From here we get py and pz = constant.

[email protected] 12 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Now we write the Hamiltonian in different form by taking

eB p
ω = mc , x0 = eBy .

p2x 1 2 2 p2z
H= = mω (x − x0 ) +

ide
2m 2 2m
So the solution is for x and px
px
ẋ = ; p˙x = −mω 2 (x − x0 )
m

gu ẍ =
p˙x
m
= −ω 2 (x − x0 )

p¨x = −mω 2 ẋ = −ω 2 px
ics
Hence the solution is

x = x0 + a cos(ωt + φ); px = −mωa sin(ωt + φ)

∂H 1
ys

ẏ = = (py − eBx) = ωa cos(ωt + φ)

∂py m
pz
ż =
m
Ph

pz
y = a sin(ωt + φ) = y0 ; z = t + z0
m
The particle is moving along a spiral with axis parallel to z axis
(i.e the axis of applied magnetic field). Note: This problem is
very important quantum mechanically. Try to solve the same
using quantum mechanics.

[email protected] 13 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Example 2.8: The Lagrangian for a system can be

written as
ẏ p
L = aẋ2 + b + cẋẏ + f y 2 ẋż + g ẏ − k x2 + y 2

ide
x
Where a,b,c,f,g, k are constants. What is the Hamilto-
nian? What quantities are conserved?

First we calculate the momentum

gu px =
∂L
∂ ẋ
∂L
= 2aẋ + cẏ + f y 2 ż

b
ics
py = = + cẋ + 2g
∂ ẏ x
∂L
pz = = f y 2 ẋ
∂ ż
Hamiltonian is
ys

X
H= pi q̇i − L
i
H = ẋpx + ẏpy + żpz − L
Ph

Substitute the values of px , py , pz

p
H = aẋ2 + cẋẏ + f y 2 ẋż + g ẏ + k x2 + y 2

Now comes the trick. You need to substitute the values of

velocities by momentum. You can do it by so many ways. My
trick is write ẋ, ẏ, and ż in terms of p˙x , p˙y , p˙z , and substitute

[email protected] 14 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

in the Hamiltonian expression. You can use your own trick.

There is a short way also.

ap2z
 
px pz 1 b cpz
H= − 2 4+ py − − 2

ide
f y2 f y 2 x fy
p
+ k x2 + y 2

pz is conserved since ∂L
∂z = 0. Energy is conserved since the
Hamiltonian is time independent.

gu
Example 2.9: Problem: A dynamical system has the
Lagrangian

q˙2 2
ics
L = q˙1 2 + 2 + k1 q12 + k2q˙1 q˙2
a + bq1
Where a, b, k1 , k2 are constants. Find the equation of
motion in Hamiltonian.
ys

Solution:
∂L
p1 = = q˙1 + k q˙2
∂ q˙1
Ph

∂L 2q˙2
p2 = = + k q˙1
∂ q˙2 a + bq12

H = q˙1 p1 + q˙2 p2 − L
q˙2 2
= q˙1 2 + 2 − k1 q12 + k2q˙1 q˙2
a + bq1

[email protected] 15 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Now the calculation is tough. First you need to calculate

q˙1 and q˙2 in terms of p˙1 and p˙2 . Then you substitute q˙1 and q˙2
in terms of p˙1 and p˙2 in the Hamiltonian. Do only if you have
patience and time!

ide
3 Drawing Phase space
We can simply say A diagram which consists all information

diagrams. gu
about the system. Phase space is basically constant energy

Two phase space drawing of a same system can not inter-

ics
sect each other.
Example 3.10: Constant velocity Consider a parti-
cle that moves with constant velocity in 1−D. since there
is no force, the potential energy is constant (which we
ys

will take to be zero), so the Hamiltonian is H = p2 /2m.

How will the phase space look like ?
Ph

You can quickly show that Hamilton’s equations are q̇ =

p/m and ṗ = 0. The second of these is the statement that
p doesn’t change, as expected. So the possible ”curves” in
phase space are horizontal lines, as shown in Figure.

[email protected] 16 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

ide
gu
The actual line the particle is on is determined by the ini-
tial coordinates (q0 , p0 ) . Note that the lines associated with
negative p are traced out leftward (that is, in the direction of
ics
decreasing q) as should be the case. Note also that although
all the lines look basically the same on the page, the parti-
cle traverses the ones with larger |p| more quickly, due to the
above q̇ = p/m equation; the larger the |p|, the greater the
ys

rate at which q changes, as expected.

Example 3.11: Falling balls: Hamiltonian for a ver-
tically thrown ball is H = p2 /2m − mgq. (With positive
Ph

q defined to be downward.) Draw the phase space.

You can quickly show that Hamilton’s equations are q̇ =

p/m and ṗ = mg. Eliminating p gives the familiar equation
of motion, q̈ = g. Integrating twice then gives the standard
result, q(t) = q0 + v0 t + gt2 /2.

[email protected] 17 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

The phase space plot will follow the general trajectory

p2 − q = E = Constant

ide
The drawing of the parabola will be like the figure below.
When q is zero, p should be minimum and when q increases
p increases too, to keep p2 − q fixed.
How can we find the direction of the arrows? Well we have

gu
taken positive q defined to be downward. So when the ball
moves towards the ground from some height. It’s q increases
and as velocity increase so as momentum also increases. Then
the arrow will be like the figure.
ics
ys
Ph

Now what happened when the ball drops and go u again

with a full elastic collision? Well at the point of collision q
does not change. Only the momentum flips sign. And then
q starts decreasing to zero and momentum starts increasing

[email protected] 18 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

starting with negative value. This is shown in left side of the

next figure.
What happens when the collision is inelastic? Same dia-
gram, the only change is the maximum value of momentum

ide
in the second collision is less than the first collision as there
are some energy is lost during collision. This is shown in right
side of the diagram

gu
ics
ys

Example 3.12: Falling balls: Again: Now if we take

Ph

positive q upwards then the Hamiltonian is

H = p2 /2m + mgq; q≥0

Simply the trajectory would be

p2 + q = E = constant

[email protected] 19 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

So the trajectory will be a parabola. Just the radius of cur-

vature will be in the opposite direction. See the next figure.

ide
gu
ics
Example 3.13: Harmonic oscillator The Hamilto-
nian for a harmonic oscillator is
p2 kq 2
ys

H= +
2m 2
What will be the phase space?
Ph

You can quickly show that Hamilton’s equations are q̇ =

p/m and ṗ = −kq. To solve for q in terms of p, you could
solve for q in terms of t.We have
p2 kq 2 p20 kq02
+ = +
2m 2 2m 2
This is the equation for an ellipse in the q −p plane. The (q, p)

[email protected] 20 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

coordinates of the particle keep going around and around a

particular ellipse as time goes by, as shown in the figure.

ide
gu
ics
ys
Ph

[email protected] 21 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Example 3.14: A particle moves in one dimension

under a potential V (x) = α|x| with some non-zero
total energy. Which one of the following best describes

ide
the particle trajectory in the phase space? [GATE
2018]

gu
ics
ys
Ph

The Phase space trajectory will follow the equation

p2 /2m + α|q| = E = Constant

What does this mod sign do in the phase space? You see
that at q = 0, the potential changes sign. So the force is
discontinuous there.

[email protected] 22 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

ide
So the forces flip direction at q = 0, hence as the force is

gu
the derivative of momentum, the phase space trajectory will
display the discontinuity of the derivative of the momentum
at position q = 0. So the ans will be A.
ics
3.1 Fixed points and separatrix
Now we will learn something more to plot phase space diagram
ys

properly.
Fixed points are the points where

ṗ = q̇ = 0
Ph

If a solution of Hamiltonian possible which passes through

the fixed point then the phase space diagram either directly
emerge from fixed point or directly converge to fixed point.
No line can cross the fixed points.
The lines of phase diagram which passes directly over the
fixed point are called the separatrix. No phase space curve

[email protected] 23 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

crosses the separatrix.

And if solution not possible through fixed point then the
phase space diagram circulate around fixed point. Now we
will see an example.

ide
Example 3.15: The Hamiltonian of particle of mass m
p2 αq 2
is given by H = − . Which one of the following
2m 2
figures describes the motion of the particle in phase
space?

gu [GATE 2014]
ics
ys
Ph

To have an idea about the diagram we need to consider

[email protected] 24 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

only
H = p2 − q 2
. Hamilton’s equation says

ide
∂H ∂H
ṗ = − = 2q; q̇ = = 2p
∂q ∂p
To find the fixed points we just make ṗ = q̇ = 0. So the
fixed points become

gu p=q=0
First we see that whether H = 0 solutions are possible or
not. If we make H = 0 then
ics
H=0 =⇒ p2 − q 2 = 0 =⇒ p = ±q

These p = ±q lines are the solutions which will passes through

p = q = 0. So we got straight lines solutions through p = q =
ys

0 which will be called the separatrix.

Now we will see the direction of force to check the direction
of the arrows in the separatrix.
Ph

∂V
V (q) = −q 2 =⇒ F (q) = = 2q
∂q
Now when q > 0 force is +ive. So the force will increase the
displacement in positive direction. And when q < 0, the force
is −ive, so the force will try to increase the displacement in
negative direction.

[email protected] 25 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Now consider the first quadrant q > 0, p > 0. Here the

velocity is already in the positive q direction and the force is
increasing with q. So obviously both the q and p increases.
So the direction of the arrow will be outwards.

ide
Next, consider the fourth quadrant q > 0, p < 0. The co-
ordinate is positive and the velocity is negative, means the
particle moving towards the origin. Now the force is posi-
tive as F = 2q, so the force try to move the particle towards

gu
positive q but as the particle velocity already is in the −q di-
rection, the the particle continues to move towards the origin
but because of the force towards q direction the velocity de-
ics
creases and eventually the particle moves towards the origin
with less and less velocity. hence the direction of the arrow
will be inwards.
Similarly you can check the direction of the second and
ys

third quadrants. One more thing you need to remember that

the contiguous lines besides the separatrix will follow the same
directions. Hence just evaluating the direction of separatrix
Ph

is enough.

[email protected] 26 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Example 3.16: Which of the following figures is a

schematic representation of the phase space trajectories
(i.e contour of constant energy) of a particle moving in

ide
a one dimensional potential
1 1
V (x) = − x2 + x4
2 4
[NET June 2015]

gu
ics
ys
Ph

The Hamiltonian is
p2 1 1
H= − x2 + x4
2m 2 4

[email protected] 27 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

First question is where are the fixed points?

∂H ∂H
ṗ = − = x − x3 ; q̇ = − = p/m
∂q ∂p

ide
Now to make
ṗ = q̇ = 0
we need
x − x3 = 0; and p/m =

gu
Hence the fixed points are

x = 0, 1, −1; and p = 0
ics
Hence we get three fixed points (0, 0), (1, 0), (−1, 0) Now we
will check the H = 0 solution.
p2 1 2 1 4
H=0= − q + q
ys

2 2 4
indeed have a solution which passes through (0, 0). So, among
the fixed points only (0, 0) point implies H = 0. So the (0, 0)
Ph

points will have lines emerges out and converges in.

Now for the other two fixed points, there will be no lines
which emerges out or converge in. So, there will only be closed
loops encirculating the fixed points.
It is not always easy to determine whether a fixed point is
stable (circular / elliptic) or unstable (hyperbolic). One very

[email protected] 28 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

useful shortcut is for constant energy systems if

d2 H 

= −ive
dx2 x0 ,y0

ide
(x0 , y0 ) is a fixed point, then that fixed point is repulsive (hy-
d2 H 
perbolic). And if = +ive, then the fixed point is
dx2 x0 ,y0
stable.

gu
Now you have learned enough of the phase space trajectory.
Attempt the exercises in the problem set.

4 Canonical Transformation
ics
4.1 Basic ideas
Lagrangian → Hamiltonian formalism meant moving from the
ys

configuration space to the phase space, i.e

(q1 , . . . , qn ) =⇒ (q1 , . . . , qn , p1 , . . . , pn )
Ph

Lagrangian formalism is invariant under coordinate transfor-

mations such as Qi = Qi (q1 , . . . , qn , t). Now the Qs is, Is
Hamiltonian formalism invariant under similar but more gen-
eral transformations?

Qi = Qi (q1 , . . . , qn , p1 , . . . , pn , t) ;

[email protected] 29 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Pi = Pi (q1 , . . . , qn , p1 , . . . , pn , t)
Not all these types of general transformations are dynam-
ically possible. But we have certain types of transformations

ide
which are called canonical transformations. Now what is so
special about canonical transformation?
Canonical Transformation is a special kind of transforma-
tion of coordinate and momentum

gu Qi = Qi (q1 , . . . , qn , p1 , . . . , pn , t) ;

Pi = Pi (q1 , . . . , qn , p1 , . . . , pn , t)
ics
that satisfy Hamilton’s equation of motion, i.e after trans-
forming the coordinate and momentum the transformed Hamil-
tonian K(Q, P, t), sometimes called Kamiltonian satisfies the
Hamilton’s equations of motion with respect to the new trans-
ys

formed coordinate and momentum Q, P

∂H ∂H ∂K ∂K
q̇i = ṗi = − =⇒ Q̇i = Ṗi = − (4.1)
dpi dqi dPi dQi
Ph

Hamilton’s least action principle requires

Z t2
δ (pi q̇i − H(q, p, t)) dt = 0
t1
Z t2  
and δ P Q̇i − K(Q, P, t) dt = 0 (4.2)
t1

[email protected] 30 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Now we have two ways such that the Hamilton’s least ac-
tion principle remains intact. One is just to multiply the old
Hamiltonian by a constant number which is called a scale /
contact transformation. The other way is to add total time

ide
derivative of a function of coordinate, momentum and time
which is the actual canonical transformation.

Pi Q̇i − K = λ (pi q̇i − H) ; scale transformation

gudF
Pi Q̇i − K +
dt
Combining we get
= pi q̇i − H; canonical transformation
ics
dF
Pi Q̇i − K + = λ (pi q̇i − H)
dt (4.3)
extended canonical transformation

4.2 Generating functions

ys

Now this quantity F is very important in canonical transfor-

mation is we see from
Ph

dF
Pi Q̇i − K + = pi q̇i − H
dt
To preserve Hamilton’s principle we must have
Z t2  Z t2  
 dF
δ Pi Q̇i − K dt = δ pi q̇i − H − dt
t1 t1 dt (4.4)
= −δ[F ]tt21 =0

[email protected] 31 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Satisfied if δp = δq = δP = δQ = 0 at t1 and t2 .
We see that F can be any function of pi , qi , Pi , Qi and t. F
is called the generating function or generator of the transfor-
mation. The transformation depends on F .

ide
Now we see some special types of generating functions and
evaluate what they actually do.
Example 4.17: Try a generating function: F = qi Pi −
Qi Pi .

gu
As we know
dF
Pi Q̇i − K +
dt
= pi q̇i − H
ics
We see that the canonical transformation generated by F is
dF
Pi Q̇i − K + = −K + (qi − Qi ) Ṗi + Pi q̇i = pi q̇i − H
dt
ys

Hence we have

Q i = qi ; P i = pi ; K=H
Ph

This is the identity transformation.

Example 4.18: Try this generating function:

F = fi (q1 , . . . , qn , t) Pi − Qi Pi

where fi are arbitrary functions of q1 . . . qn and t

[email protected] 32 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

See
dF ∂fi ∂fi
Pi Q̇i − K + = −K + (fi − Qi ) Ṗi + Pi q̇j + Pi
dt ∂qj ∂t

ide
= pi q̇i − H

So we get by comparing the two sides of the equation

∂fj ∂fi
Qi = fi (q1 , . . . , qn , t) ; pi = Pj ; K=H+ Pi
∂qi ∂t

gu
Using this type of generating function we can transform the
coordinate to any function of generalized coordinates.
ics
One thing you must know at this point- The generating
function is not unique for a particular type of canonical trans-
formation. There are many possible F for each canonical
transformation.
ys

4.3 Four basic types of generating functions

Ph

There are four basic types of generators

F1 (q, Q, t), F2 (q, P, t), F3 (p, Q, t), F4 (p, P, t)

[email protected] 33 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

ide
gu
There is no reason to consider any of these 4 definitions to
ics
be more fundamental than the others.
We arbitrarily chose the first form (which happens to be
the Lagrangian form) to write the generating functions in the
table.
ys

Example 4.19: Suppose you have the Hamiltonian

p2 1
H= + 2
Ph

2 2q
Now you need find a canonical transformation and the
co-responding generating function F2 (q, P, t). You are
given P = pq. Find F and Q. Construct the new Hamil-
tonian and then solve Hamilton’s equation of motion to
get P and Q as function of time.

[email protected] 34 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Solution: As for the second type of the generating func-

tions
∂F2 ∂F2
= p; =Q
∂q ∂P

ide
We are given
P
P = pq =⇒ p =
q
∂F2 P
=p=

gu
Integrating we get
∂q q

F2 = P log q
Now differentiate F2 with respect to P to get
ics
Q = log q ⇒ q = eQ

Now the new Hamiltonian is

ys

p2 1 P 2 + 1 −2Q
H= + 2= e =E Constant
2 2q 2
Now
∂H
Ph

= P 2 + 1 e−2Q = 2E


Ṗ = −
∂Q
Integrating
P = 2Et + C
We also get
r r
Q P2 + 1 C2 + 1
q=e = = 2Et2 + 2Ct +
2E 2E
[email protected] 35 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

r
C2 + 1
Q = log 2Et2 + 2Ct +
2E

Example 4.20: If a generating function of type 1 is

ide
given by
Q
F1 = −
q
Find the new canonical variables Q and P

gu
Solution: We see

p=
∂F1
∂q
Q
= 2
q
and P =−
∂F1
∂Q
=
1
q
ics
Writing the new coordinates as function of the old ones gives
1
Q = pq 2 and P =
q
ys

Example 4.21: Prove that the following transforma-

tion is canonical and find a suitable generating function
for it.
Ph

   2 
p q p
Q = ln and P = − +1
q 2 q

Solution: We know that for type 1 generating function

∂F1
= p = qeQ
∂q

[email protected] 36 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Which gives us

q2 Q
Z
Q
F1 = qe dq + g(Q) = e + g(Q)
2

ide
And
q 2 Q dg
 2   2 
∂F1 q p q
P =− =− e − =− +1 =− + 1 eQ
∂Q 2 dQ 2 q 2
⇒ g(Q) = eQ

gu
So the generating function is given by

F1 =
 2
q

+ 1 eQ
ics
2

4.4 Canonical transformation and cyclic co-

ordinates
ys

Let us imagine that we find coordinates qi that are all cyclic.

Then
ṗi = 0,
Ph

so
p i = αi
are all constant. If H is conserved, then:

H = H (α1 , . . . , αn )

[email protected] 37 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

is also constant in time. In such a case the remaining equa-

tions of motion:
∂H
q̇i = = ωi (α) ⇒ qi = ωi t + δi

ide
∂αi
All coordinates are linear in time and the motion becomes very
simple. We might imagine searching for a variable transfor-
mation to make as many coordinates as possible cyclic. The

gu
search of such types of coordinates is guided by the canonical
transformation as we can not transform the coordinates and
momentums arbitrarily and make the new coordinates cyclic.
ics
5 Poission Bracket
Poisson bracket of two dynamical variables u(p, q, t) and v(p.q.t)
ys

is defined by
X  ∂u ∂v ∂v ∂u

{u, v}q,p = − (5.1)
i
∂q i ∂p i ∂q i ∂p i
Ph

It is easy to check that the following fundamental Poisson

bracket relations are satisfied:

{qi , qj }q,p = {pi , pj }q,p = 0 and {qi , pj }q,p = δij (5.2)

There are a few other properties of note. These include:

[email protected] 38 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

{u, u} = 0
{u, v} = −{v, u}

ide
{au + bv, w} = a{u, w} + b{v, w}
{uv, w} = u{v, w} + {u, w}v
{u, {v, w}} + {v, {w, u}} + {w, {u, v}} = 0
df

gu {f (u), v} =
du
df
{u, v}

{u, f (v)} = {u, v}

dv
(5.3)
ics
Now we need to learn one very important Theorem:
A transformation Qj = Qj (q, p, t) and Pj = Pj (q, p, t) is
canonical if and only if:
ys

{Qi , Qj }q,p = {Pi , Pj }q,p = 0 and {Qi , Pj }q,p = δij (5.4)

Poission bracket of any two dynamical variable is invarian

Ph

under a canonical transformation, i.e

[A, B]q,p = [A, B]Q,P (5.5)

[email protected] 39 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Example 5.22: Consider the following transformation

p
Q= ; P = −q 2
2q

ide
(a)Is the transformation canonical?
(b) Determine the new Hamiltonian.
(c) Write down the Hamilton’s equations for both
Hamiltonians and solve for the variables.

gu
(d) Check that the solutions set of both old and new
Hamiltonian are same.
ics
Solution: Calculate Poisson Bracket for the new coordi-
nates with respect to old coordinates
∂Q ∂P ∂Q ∂P
[Q, P ]q,p = − =1
∂q ∂p ∂p ∂q
ys

So the transformation is canonical.

(b)
√ √
q = −P and p = 2Q −P
Ph

Substituting the values of p and q in the old Hamiltonian to

get the new Hamiltonian as function of new coordinates.
1
H 0 (Q, P ) = − 1 + 4Q2


2
(c) Hamilton’s equations in old coordinates
q̇ = p; ṗ = −q

[email protected] 40 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Hamilton’s equation in new coordinates

∂H 0 1
= − 1 + 4Q2


Q̇ =
∂P 2
∂H 0

ide
Ṗ = − = 4P Q
∂Q
We solve for old coordinates

q̈ = ṗ = −q; → q = A sin t + B cos t

gu p̈ = −q̇ = p; → p = C sin t + D cos t

At t= 0; q = q0 and p = p0 . Hence the solution

q = q0 cos t + p0 sin t and p(t) = −q0 sin t + p0 cos t

ics
For new coordinates the equations are not very easy to solve.
However we can just substitute the solutions in old coordi-
nates to new coordinates with the transformation given in
ys

the problem.
p −q0 sin t + p0 cos t
Q(t) = =
2q 2 (q0 cos t + p0 sin t)
Ph

P (t) = −q 2 = − (q0 cos t + p0 sin t)2

Example 5.23: Show that the time independent trans-

formation given below is canonical and obtain a suitable
generating function of type F3 (p, Q)

P = p + q 2 + pq 2 ; Q = tan−1 q

[email protected] 41 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Solution:
∂Q ∂P ∂Q ∂P 1 + q2
[Q, P ]q,p = − = =1
∂q ∂p ∂p ∂q 1 + q2

ide
So the transformation is canonical.
It is given that we need to find the generating function of the
type F3 (p, Q). Hence we need to write q and P in terms of Q
and p.

So
gu
q = tan Q; P = p(1 + q 2 ) + q 2 = p sec2 Q + tan2 Q

q = tan Q = −
∂F3
ics
∂p
∂F3
P = p sec2 Q + tan2 Q = −
∂Q
From first equation F3 = −p tan Q + f1 (Q) and from 2nd
ys

equation F3 = −p tan Q − tan Q + Q + f2 (p).

So the generating function is

F3 = −p tan Q − tan Q + Q
Ph

otherwise we can write at first

∂F3 ∂F3
dF3 = dp + dQ
∂p ∂Q
= − tan Qdp − (p sec2 Q + tan2 Q)dQ

F3 = −p tan Q − tan Q + Q

[email protected] 42 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Example 5.24: Find the conditions that need to be sat-

isfied by the real constants α, β, γ, δ so that the trans-
formation is canonical.

ide
Q = αpq γ ; P = βq δ

Solution:
∂Q ∂P ∂Q ∂P
[Q, P ]q,p = − = αpγq γ−1 × 0 − αq γ βδq δ−1

gu ∂q ∂p
= αβδq γ+δ−1
∂p ∂q

If the Poisson Bracket is to be 1 then αβδ = −1 and γ+δ−1 =

ics
0
ys
Ph

[email protected] 43 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Example 5.25: A particle that has coordinate-

dependent mass mq 2 is attached to a spring that has
spring constant k = mg. The particle’s Hamiltonian is

ide
p2 1
H= 2
+ mgq 2
2mq 2
(a) Derive the HAmilton’s equation of motion.
(b) Show that the coordinate transformation

gu 1
z = q2
2
and pz =
p
q
is a canonical transformation. Derive the Hamilton equa-
ics
tions of motion for z and pz and find the general solutions
for z(t) and pz (t) in terms of intial values z(0) and pz (0).
(c) Let us define a new coordinate P which is equals to
ys

new Hamiltonian.
p2z
P =H= + mgz
2m
Ph

Find another coordinate Q which is canonical to P. (Q

and P that make up a pair of canonical coordinates.)
Derive the Hamilton’s equation for Q and P, and solve
Q(t) and P(t) in terms of initial values of Q and P.

[email protected] 44 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Solution: (a)

∂H p ∂H p2
q̇ = = ; ṗ = − = − mgq
∂p mq 2 ∂q mq 3

ide
(b)
∂z ∂pz ∂z ∂pz 1
[z, pz ]q,p = − =q −0=1
∂q ∂p ∂p ∂q q
So the transformation is canonical.

gu
Let us try to find a second type generating function F2 (q, pz ).
We have studied the convention of Q and P as the transformed
coordinates. Here the new variables are Q = z and P = pz .
ics
∂F2 1
= p = qpz −→ F2 = q 2 pz + f (pz )
∂q 2
and
∂F2 1
z= = q 2 + f 0 (pz )
∂pz 2
ys

As the partial derivative ∂F ∂pz matches with the definition of z

2

(z = 21 q 2 ), we can take f 0 (pz ) = 0, then. So

Ph

1
F2 (q, pz ) = q 2 pz
2
Well, now you see that the Hamiltonian becomes in terms
of new variables
p2
H = z + mgz
2m

[email protected] 45 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

This hamiltonian is exactly same hamiltonian as a particle in

a gravitational field. So we get
pz
ż = and p˙z = −mg

ide
m
So the solution
1 pz
z = gt2 + 0 t + z0
2 m

gu
(c) Given
P =H=
pz = −mgt + pz0

p2z
2m
+ mgz
ics
We need to find a Q, such that the transformation is canonical.
Lets try to find a generating function F2 (z, P ).
∂F2 p
= pz = 2m(P − mgz)
ys

∂z
Hence
√
2 2m
Ph

F2 (z, P ) = − (P − mgz)3/2 + f (P )
3 mg
√
2 2
=− (P − mgz)3/2 + f (P )
3g m
r
∂F2 1 2
Q= =− (P − mgz)1/2 + f 0 (P )
∂P g m

[email protected] 46 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

p2z
Now use P = 2m + mgz to get
2
pz 0 pz
Q=− +f ( + mgz)
mg 2m

ide
pz
If we choose f 0 (P ) = 0.We get Q = − mg . We could not have
0
chosen f (P ) = 0. But as we see
pz p2z
{Q, P } = {− , + mgz} = 1
mg 2m

gu
So P and Q are canonical variables. If we have chosen f 0 (P ) 6=
0, then the Q would have been more complex. And as our
task is to choose just another Q which is canonical to P , why
ics
should we go for more complex Q when we have already have
got simpler Q. So our Q is just
pz
Q=−
mg
ys

Now we have H = P so
∂H ∂H
Ṗ = − = 0 and Q̇ = =1
∂Q ∂P
Ph

Hence the solution is

P = P0 and Q = t + Q0

Easy, isn’t it?? In this problem you have seen that an other-
wise complex problem can be made simple by use of canonical
transformation.

[email protected] 47 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

6 Equation of motion of a dynami-

cal variable

ide
let us consider a dynamical variable (function of q, p, t ) u =
u(q, p, t) Then we have by application of partial derivative
∂u ∂u ∂u
u̇ = q̇i + ṗi +
∂qi ∂pi ∂t

gu
Using Hamilton’s equation of motion we get

u̇ =
∂u ∂H
−
∂u ∂H ∂u
∂qi ∂pi ∂pi ∂qi
+
∂t
In terms of Poission Bracket we write the above equation
ics
more concisely as
∂u
u̇ = {u, H} + (6.1)
∂t
ˆ If u = qi then
ys

∂qi ∂H
q̇i = {qi , H} + =
∂t ∂pi
ˆ If u = pi , then
Ph

∂pi ∂H
ṗi = {pi , H} + =−
∂t ∂qi
These two equations are Hamilton’s equation of motion.
ˆ Now if u = H then
∂H ∂H
Ḣ = {H, H} + =
∂t ∂t
[email protected] 48 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

ˆ Next consider what we can say if a quantity u is conserved.

Here:
∂u
u̇ = 0 = {u, H} +
∂t

ide
So we can say if
∂u
= 0 and {u, H} = 0 ⇒ u is conserved
∂t
Another fact, is that if u and v are conserved then so is {u, v}.

gu
This could potentially provide a way to compute a new con-
stant of motion. To prove it, first consider the special case
where:
ics
∂u ∂v
= =0 =⇒ {u, H} = {v, H} = 0
∂t ∂t
then using the Jacobi identity we have:
ys

{H, {u, v}} = −{u, {v, H}}−{v, {H, u}} = −{u, 0}−{v, 0} = 0

For the most general case we have

Ph

du ∂u ∂u
= {u, H} + =0 ⇒ {u, H} = −
dt ∂t ∂t
And
dv ∂v ∂v
= {v, H} + =0 ⇒ {v, H} = −
dt ∂t ∂t

[email protected] 49 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

So we get
{{u, v}, H} = {u, {v, H}} + {v, {H, u}}
   
∂v ∂u
= − u, + v,

ide
∂t ∂t
∂
= − {u, v}
∂t
Now the equation 6.1 for tha variable {u, v} is
d ∂

gu
Hence we get
dt
{u, v} = {{u, v}, H} + {u, v}

d
dt
∂t

{u, v} = 0
ics
7 Infinitesimal Canonical Transfor-
mation
ys

Let us now consider the generating function:

F2 (q, P, t) = qi Pi + G(q, P, t)
Ph

where F2 = qi Pi is an identity transformation, and ||  1

is infinitesimal. The function G(q, P, t) is known as the gen-
erating function of an infinitesimal canonical transformation.
Using the properties of an F2 generating function we have:
∂F2 ∂G ∂G
pj = = Pj +  ⇒ δpj = Pj − pj = − (7.1)
∂qj ∂qj ∂qj
[email protected] 50 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

giving the infinitesimal transformation in the momentum. Sim-

ilarly
∂F2 ∂G
Qj = = qj +  (7.2)
∂Pj ∂Pj

ide
since Pj = pj +O() and  and  is infinitesimal we can replace
∂G(q, P, t)/∂Pj = ∂G(q, p, t)/∂pj + O(). Therefore we have:
∂G ∂G
+ O 2


Q j = qj +  ⇒ δqj = Qj − qj =  (7.3)
∂pj ∂pj

gu
where now we note that we can consider G = G(q, p, t), a
function of q and p, to this order.
We can see
ics
 
∂qi ∂G ∂qi ∂G ∂G
ε [qi , G] = ε − =ε = δqi (7.4)
∂qj ∂pj ∂pj ∂qj ∂pi
and
ys

 
∂pi ∂G ∂pi ∂G ∂G
ε [pi , G] = ε − = −ε = δpi (7.5)
∂qj ∂pj ∂pj ∂qj ∂qi

For an arbitrary function u(q, p, t), the ICT does

Ph

ICT ∂u ∂u ∂u
u −→ u + δu = u + δqi + δpi + δt
∂qi ∂pi ∂t
∂u ∂G ∂u ∂G ∂u
=u+ ε − ε + δt
∂qi ∂pi ∂pi ∂qi ∂t
∂u
= u + ε[u, G] + δt
∂t
[email protected] 51 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

That is
∂u
δt
δu = ε[u, G] + (7.6)
∂t
Hence we get the equation of motion of a dynamical vari-

ide
able again
∂u du ∂u
δu = δt[u, H] + δt → = [u, H] + (7.7)
∂t dt ∂t

7.1
gu
Generators of linear momentum, angu-
lar momentum and time
Now look at the infinitesimal canonical transformation again.
ics
Just see if G = pi then δpi = 0 and δqj = δij , which is why
momentum is the generator of spatial translations.

Now what happens if G is the z component of the angular

ys

momentum??
First writing as if there is only one particle.
Ph

G = Lz = (xpy − ypx )

then the infinitesimal change corresponds to a rotation

δx = [x, G] = −y , δy = [y, G] = x , δz = [z, G] = 0

[email protected] 52 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

and
δpx = [px , G] = −py 
δpy = [py , G] = px 
δpz = [pz , G] = 0

ide
Now for which infinitesimal change of  the x changes by
−y? and y changes by x? Obviously it is the δθ,the change
of angle in the xy plane. As we know for a rotation in xy
plane

gu
δx = −yδθ , δy = xδθ , δz = 0
So we see that  = δθ. Hence we find that if G = angular
momentum in z direction then the infinitesimal change  = δθ
ics
Which is why angular momentum is the generator of rotations.
Now consider the very important example: if G = H then
∂H δp
δp = − = −(−ṗ) = 
∂q δt
ys

Hence we see  = δt
So if Hamiltonian is the generator then  is the time trans-
lation
Ph

7.2 integration of the equation of motion

We can “integrate” ICT to get finite CT. How do we integrate
δu = ε[u, G]? First, let’s rewrite it as
du
du = dα[u, G] → = [u, G]
dα
[email protected] 53 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

We want the solution u(α) as a function of α, with the initial

condition u(0) = u0
Taylor expansion u(α) about α = 0 gives

ide
α2 d2 u  α3 d3 u 
  
du 
u(α) = u0 + α + + + ···
dα 0 2! dα2 0 3! dα3 0

du 
α dα 0
is [u, G]0 .

gu du
Now since dα = [u, G] is true for any u, we can say
[, G]. Apply this operator repeatedly
d2 u
dα2
=
d
dα
[u, G] = [[u, G], G]
d
dα =
ics
dj u
→ j = [[· · · [u, G], G], · · · , G]
dα
Going back to the Taylor expansion,
α2 d2 u  α3 d3 u 
  
du 
ys

u(α) = u0 + α + + + ···
dα  2! dα2 
0 0 3! dα3  0
2 3
α α
= u0 + α[u, G]0 + [[u, G], G]0 + [[[u, G], G], G]0 + · · ·
2! 3!
Ph

(7.8)
Now we have a formal solution – But does it work?
Example 7.26: An object is falling under gravity, so
the Hamiltonian is
p2
H= + mgz
2m

[email protected] 54 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Here the infinetisimal change dα = dt so use α = t in

equation (7.8). Integrate the time ICT.

t2 t3
z(t) = z0 + t[z, H]0 + [[z, H], H]0 + [[[z, H], H], H]0 + · · ·

ide
2! 3!
You see
p
[z, H] = , [[z, H], H] = −g, [[[z, H], H], H] = 0
m
So you get
gu z(t) = z0 +
p0
m
g
t − t2
2
ics
8 Phase space volume and Liouville’s
Theorem
ys

There are two views in dynamics while considering the canon-

ical transformation and the phase space.
Static view: CT moves a point in one phase space to a point
in another phase space
Ph

Dynamic view: CT moves a point in one phase space to

another point in the same space.
We will consider the static view as of now. If you consider
a set of points, CT moves a volume to anther volume as shown
in the figure

[email protected] 55 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

ide
gu
Now calculate the jacobian of the Canonical Transforma-
tion in 1-D

dQdP = |M|dqdp where M =

"
∂Q/∂q ∂Q/∂p
#
ics
∂P/∂q ∂P/∂p

∂Q ∂P ∂P ∂Q
|M| = − = [Q, P ] = 1 → dQdP = dqdp
∂q ∂p ∂q ∂p
ys

i.e., volume in 1-dim. phase space is invariant. This is true

for n-dimensions. So we conclude, Volume in Phase Space
is invariant in canonical transformation.
Ph

Suppose we have a harmonic oscillator and we transform p

and q by using a canonical transformation

√
r
2P
q= sin Q; p = 2P mω cos Q
mω
See that
1
p2 + m2 ω 2 q 2 = ωP


K=H=
2m
[email protected] 56 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

E ∂K
P = const = ; Q̇ = = ω; Q = ωt + α
ω ∂P
The phase space representation for one period of the har-
monic oscillator in both the set of coordinates are represented

ide
in the following figure.

gu
ics
The area remains invariant. One cycle draws the same area
2πE
in both spaces.
ω
Liouville’s Theorem says that The phase volume occu-
ys

pied by a group of particles (ensemble in stat. mech.) is

conserved. Thus the density in phase space remains constant
with time.
Ph

This theorem form some of theoretical basis of the 2nd law

of thermodynamics.
The theorem holds true when there are large enough num-
ber of particles so that the distribution may be considered
continuous.

[email protected] 57 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

9 Hamilton Jacobi and Action An-

gle

ide
9.1 Hamilton Jacobi equation
Let’s search for new canonical variables that are all cyclic,
such that:
Q̇i = Ṗi = 0 ⇒ (Q, P ) (9.1)

gu
As the new variables are constants then the new Hamiltonian
K can be independent of (Q, P ) then:
∂K ∂K
ics
Q̇i = =0 and Ṗi = − =0 (9.2)
∂Pi ∂Qi
We could look for a constant K, but it it is simplest to simply
look for K = 0. Using a generating function F = F2 (q, P, t),
ys

then we need
∂F2
K = H(q, p, t) + =0 (9.3)
∂t
Because pi = ∂F
∂qi , then we can rewrite this condition as
2
Ph

 
∂F2 ∂F2 ∂F2
H q1 , . . . , q n , ,..., ,t + =0 (9.4)
∂q1 ∂qn ∂t

which is the time dependent Hamilton-Jacobi equation (hence-

forth abbreviated as the H-J equation). This is a 1 st order
partial differential equation in n + 1 variables (q1 , . . . , qn , t)

[email protected] 58 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

for F2 . The solution for F2 has n + 1 independent constants of

integration, One of these constants is trivial (F2 → F2 +C for
a pure constant C), so we’ll ignore this one. Hence, suppose
the solution is

ide
F2 ≡ S = S (q1 , . . . , qn , α1 , . . . αn , t) (9.5)

where S is called Hamilton’s principal function and each αi

is an independent constant. We can pick our new momenta

gu
to be the constants of integration Pi = αi for i ∈ {1, . . . , n}
(so that Ṗi = 0), thus specifying F2 = F2 (q, P, t) as desired.
Then, using again the property of an F2 generating function
ics
( and K = 0), we have that the new constant variables are:

∂S(q, α, t)
Pi ≡ α i and Qi ≡ βi = (9.6)
∂αi
ys

We introduce the notation βi to emphasize that these are

constants. From these results we can obtain a solution for the
motion as follows. From the invertibility of our transforma-
Ph

tions we have:

∂S
βi (q, α, t) = ⇒ qi = qi (α, β, t)
∂αi
∂S (9.7)
pi (q, α, t) = ⇒ pi = pi (q, α, t)
∂qi
= pi (q(α, β, t), α, t) = pi (α, β, t)

[email protected] 59 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Now, What is S? As S = F2 so S = S(q, P, t)

∂S ∂S ∂S
Ṡ = q̇i + Ṗi +
∂qi ∂Pi ∂t

ide
we also know that:
∂S ∂S
= pi , Ṗi = 0 and = −H (9.8)
∂qi ∂t
The above two equations say
Z
Ṡ = pi q̇i − H = L ⇒

gu S= Ldt

Thus S is the classical action which is an indefinite integral

over time of the Lagrangian (so it is no coincidence that the
same symbol is used).
(9.9)
ics
We should also know that the choice of constants αi is
somewhat arbitrary, as any other independent choice γi =
ys

γi (α) is equally good. Thus, when solving the H-J equation,

we introduce the constants αi in whatever way is most conve-
nient.
Ph

The H-J equation is also the semiclassical limit of the quan-

tum mechanical Schrödinger equation (00 th order term in the
WKB approximation . To see this consider the Schrödinger
equation)
~2 ∂ 2
 
∂ψ
i~ = Hψ = − + V (q) ψ (9.10)
∂t 2m ∂q 2
[email protected] 60 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

with the wavefunction ψ = exp(iS/~). At this point we are

just making a change of variable, without loss of generality,
and S(q, t) is complex. Plugging it in, and canceling an ex-
ponential after taking the derivative, we find

ide
 2
∂S i~ ∂ 2 S 1 ∂S
− =− + + V (q) (9.11)
∂t 2m ∂q 2 2m ∂q
This equation is just another way of writing the Schrödinger

∂S
gu
equation, to solve for a complex S instead of ψ. If we now
take ~ → 0 then we find that the imaginary term goes away
and what is left is
1 ∂S
 2
∂S

∂S

ics
0= + + V (q) = + H q, (9.12)
∂t 2m ∂q ∂t ∂q
which is the Hamilton-Jacobi equation for S with a standard
p2 /2m kinetic term in H
ys

Example 9.27: Example : Let us consider the har-

monic oscillator
Ph

1
p2 + (mωq)2 = E


H=
2m

Here we will look for one constant P = α and one constant

Q = β. The H-J equation says
 2 !
1 ∂S ∂S
+ (mωq)2 + =0
2m ∂q ∂t

[email protected] 61 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

In solving this, we note that the dependence of S on q and t

is separable

S(q, α, t) = W (q, α) + g(α, t)

ide
so we get
 2 !
1 ∂W ∂g
+ (mωq)2 =− =α
2m ∂q ∂t

gu
since the left side is independent of t and the right hand
side is independent of q, then the result must be equal to a
separation constant α that is independent of q and t. We will
choose our new P = α . Now we have
ics
∂g
= −α ⇒ g = −αt
∂t
where we have avoided the addition of a further additive con-
stant (since our convention was to always drop an additive
ys

constant when determining S ). To identify what α is note

that
∂S ∂g
H=− =− =α
Ph

∂t ∂t
which corresponds to the constant energy,

α=E

The other equation we have to solve is

 2 !
1 ∂W
+ (mωq)2 = α
2m ∂q

[email protected] 62 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

so rearranging and integrating this gives the indefinite integral

Z p
W =± 2mα − (mωq)2 dq

ide
which we will leave unintegrated until we must do so. The
full solution is then given by:
Z p
S = −αt ± 2mα − (mωq)2 dq

gu
With this result for Hamilton’s Principal function in hand we
can now solve for the equations of motion. The equations of
motion come from (we now do the integral, after taking the
ics
partial derivative):
Z
∂S dq
β= = −t ± m p ⇒t+β
∂α 2mα − (mωq)2
r !
2
ys

1 mω
= ± arcsin q
ω 2α

integrating we get
Ph

r
2α
q=± sin(ω(t + β))
mω 2
so β is related to the phase. Next we consider p and use this
result to obtain:
∂S p √
p= = ± 2mα − (mωq)2 = ± 2mα cos(ω(t + β))
∂q
[email protected] 63 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

So we get our familiar solution of p and q. To complete the

story, the constants α and β must be connected with the initial
conditions q0 and p0 at time t = 0. By squaring p and q and
using t = 0 we get

ide
2mα = p20 + m2 ω 2 q02

Now dividing p and q at t = 0 we get

gu tan β = mω
q0
p0
The choice q0 = 0 and hence β = 0 corresponds to starting the
motion with the oscillator at its equilibrium position q = 0 .
ics
9.2 Separation of variable and H-J method
Separation of variables is the main technique to solve the H-J
ys

equation. In particular, for a time independent H where

∂H
Ḣ = =0 (9.13)
Ph

∂t
we can always separate time by taking:

S(q, α, t) = W (q, α) − α1 t (9.14)

where α1 has been chosen as the separation constant, then

plugging this into the time dependent H-J equation yields

[email protected] 64 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

(just as in our Harmonic Oscillator example):

 
∂W ∂W
H q1 , . . . , q n , ,..., = α1
∂q1 ∂qn

ide
This result is referred to as the time independent Hamilton-
Jacobi equation. since Ḣ = 0 , H is conserved, and equal to a
constant α1 . If H = E then this constant is energy, α1 = E.
The function W is called Hamilton’s characteristic function.

method. gu
Now let’s try to solve the central force Kepler problem by H-J

Example 9.28: As an extended example, let us con-

ics
sider the Kepler problem of two masses m1 and m2 in a
central potential (with the CM coordinate R = 0). The
Lagrangian is:
ys

1 1 1 1
L = mṙ2 − V (r) where ≡ +
2 m m1 m2
and here m is the reduced mass.
Ph

Any V (r) conserves L = r × p, so the motion of r and p is

in a plane perpendicular to L. The coordinates in the plane
can be taken as (r, ψ), so:
m 2 2 2

L= ṙ + r ψ̇ − V (r)
2

[email protected] 65 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

ψ is a cyclic coordinate

pψ = mr2 ψ̇ is a constant

ide
Now check that we can write the hamiltonian as
!
2
1 p ψ
H= p2r + 2 + V (r) = α1 = E (9.15)
2m r

As ψ is cyclic, then pψ ≡ αψ is constant. Using:

gu W = W1 (r) + αψ ψ

then the time independent H-J equation is:

ics
2 !
2
α

1 ∂W1 ψ
+ 2 + V (r) = α1
2m ∂r r

After simplification
ys

s
∂W1 αψ2
= 2m (α1 − V (r)) −
∂r r2
Integration gives
Ph

Z r
α2
W = αψ ψ + 2m (α1 − V (r)) − 2v dr
r
Transformation equations are
Z
∂W dr
t + β1 = =m q
∂α1 αψ2
2m (α1 − V (r)) − r 2

[email protected] 66 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

and
Z
∂W dr
β2 = = ψ − αψ q
∂αψ 2
αw
r2 2m (α1 − V (r)) − r2

ide
9.3 Action-Angle Variables
For many problems, we may not be able to solve analyti-
cally for the exact motion or for orbital equations, but we can

gu
still characterize the motion. For periodic systems we can
find the frequency and time period by exploiting action-angle
variables.
ics
For periodic motion, we replace P = α1 by the action vari-
able I
J = pdq

The coordinate conjugate to J is the angle variable

ys

∂W
ω=
∂J
Ph

After some derivatons and arguments we get ν(J) is the

frequecy of the periodic motion.
∂H(J)
ν=
∂J

[email protected] 67 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Example 9.29: Consider a general Hamiltonian

p2
H= + α|x|ν
2m

ide
Find the time period of a particle in this Hamiltonian
using action angle variable.

Solution: As the potential is symmetric then the magni-

gu
tudes of the turning points are same

xturning = ±
 1/ν
E
α
ics
√ Z x2 √
Z x2 p
J= 2m E− αν dx =2 2m (E − αxν )dx
x1 0

√ x2
r Z
α
ys

= 2 2mE 1 − xν dx
0 E
Now we have E = αxν2 . Substituite
α ν
Ph

z≡ x
E
so  1/ν  1/ν
zE E 1 1 −1
x= ; dx = z ν dz
α α ν
Then  1/ν Z 1
√ E 1 1 √
J = 2 2mE z ν −1 1 − zdz
α ν 0
[email protected] 68 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

 1/ν
√ E 1 Γ(1/ν)Γ(3/2)
= 2 2mE
ν Γ ν1 + 23


α
 1/ν
√
√ E Γ ν1 + 1 12 π
= 2 2mE

ide
Γ ν1 + 23


α
 1/ν
√ Γ ν1 + 1


E
= 2πmE
Γ ν1 + 23


α
1 3


1 1 J Γ +
E ν +2 = √ α1/ν ν1 2

gu k=√
1
2πm

Γ +
Γ ν +1
Put the constants inside another constant
1
α1/ν ν1 2

3


ics
2πm Γ ν +1

and write
1 1
E ν + 2 = kJ
ys

Taking derivative with respect to J

1 1 ∂E
E ν + 2 −1 =k
∂J
Ph

Then the frequency

∂E 1 1
ν= = k/E ν − 2
∂J
Then the Time period
1 1 1
T = = E ν − 2 /k
ν
[email protected] 69 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Hence the time period proportional to

1 1
T ∝ E ν −2

ide
Example 9.30: Let us consider the Harmonic oscillator
problem again
1
p2 + m2 ω 2 q 2 = E = α = constant


H=
2m
Now J is
gu J=
I
pdq =
I p
2mα − m2 ω 2 q 2 dq
ics
where α is the constant total energy and ω 2 = k/m.
r
2α
If we substitute q = sin θ Then the integral reduces
mω 2
to
ys

2α 2π
Z
J= cos2 θdθ
ω 0
Limits should be over one complete cycle of q.
Ph

The integral after integration

2πα
J=
ω
Solving for α
Jω
α≡H=
2π

[email protected] 70 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

The frequency of oscillation is then

r
∂H ω 1 k
=v= =
∂J 2π 2π m

ide
Example 9.31: A particle of mass m moves in one di-
mension subject to the potential
k
U (x) =
sin2 (x/a)

gu
(a) Obtain an integral expression for Hamilton’s charac-
teristic function.
(b) Under what conditions may action-angle variables
ics
be used?
(c) Assuming that action-angle variables are permis-
sible, determine the frequency of oscillation by the
ys

action-angle method.
(d) Check your result for the oscillation frequency in
the limit of small oscillations.
Ph

Solution:
(a) As the total energy is conserved.
p2 k
H= + =E= constant
2m sin2 (x/a)
The motion is therefore between the turning points. Turn-

[email protected] 71 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

ing points are those where total energy becomes equal with
potential energy
p
x− (E) = nπa + a sin−1 k/E

ide
p
x+ (E) = (n + 1)πa − a sin−1 k/E
where n is any integer. The Hamilton Jacobi equation is then
 2
1 dW k

gu 2m dx
Which We may then write

√
+
sin2 (x/a)

Z x
=Q

s
k
ics
W (x, E) = 2m dx0 E−
x− (E) sin2 (x0 /a)

The lower limit may be left as unspecified; this only changes

the result by a constant.
ys

(b) We need that the motion is bounded. In our case,

x− (E) ≤ x ≤ x+ (E)
(c) The action is
Ph

s
√ I
k
J = 2m dx0 E −
sin2 (x0 /a)

This integration is tough. However if we substituite

r  
k 1
cos(x/a) = 1 − cos u
E 2
[email protected] 72 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Then J becomes
√ Z 2π
1 − cos u
J = a 2mE du E+k
0 E−k − cos u

ide
This integral can be solved by method of contour integration
by using techniques of complex analysis. However the stan-
dard integral table may be used
Z 2π
du 1 1

gu I=
0 2π b − cos u
=√
b2 − 1
Using this integral we got, with b = (E + k)/(E − k),
ics
2
√ √ √ √

J
J= 2ma( E − k) , E= √ + k
2ma
Note that the minimum energy is Emin = k. The oscillation
frequency is given by
ys

r r
∂E J 2k 2E
ν(J) = = 2
+ 2
=
∂J ma ma ma2
Ph

(d) With U (x) = k/ sin2 (x/a) we have

2k cos(x/a)
U 0 (x) = − ·
a sin3 (x/a)
2k sin4 (x/a) + 3 sin2 (x/a) cos2 (x/a)
U 00 (x) = 2 ·
a sin6 (x/a)

[email protected] 73 Physicsguide
©Sk Jahiruddin, 2020 Hamiltonian Mechanics

Setting U 0 (x∗ ) = 0 we obtain x∗ = n + 12 πa, where n is

an integer . At any of these equilibria, U 00 (x∗ ) = 2k/a2 .

Therefore, the frequency of small oscillations is

ide
r r
U 00 (x∗ ) 2k
ωs.o. = =
m ma2
which agrees the result of (c)

gu
ics
ys
Ph

[email protected] 74 Physicsguide