Classical Mechanics II

Undergraduate School - B2

Year 2015-2016

Lecturer: Nguyen Thi Hong Van

Institute of Physics
10 Dao Tan, Ba Dinh, Ha Noi

For internal use only

Introduction

The lecture notes are written for internal use only.

The target audience for these lecture notes is students in the B2 of the Undergrad-
uate school in USTH.
These lecture notes contain the basic concepts of classical mechanics. They cover
most of the major topic areas, but are far from comprehensive in scope. Students
are still encouraged to consult recommended text books for in depth explanations.

2
Recommended text books and
resources

1. David Tong, Classical Dynamics, University of Cambridge Part II Mathe-

matical Tripos.

2. Tai L. Chow, Mathematical Methods for Physicists: A concise introduction,

Cambridge University Press.

3. H. Goldstein, C. Poole and J. Safko, Classical Mechanics.

3
Course Syllabus

• Time Commitment: 36 hours of lectures

• Pre-requisites: Students are required to have attended elementary courses

on calculus and differential equations; vector and tensor analysis; classical
mechanics I.

• Cource Contents:
Class Contents Number of hours Note
1 Mathematical Backgrounds (part 1) 3 Lecture
2 Mathematical Backgrounds (part 2) 3 Lecture
Reminder of Newtonian Mechanics (part 1)
3 Reminder of Newtonian Mechanics (part 2) 3 Lecture
4 The Lagrangian Formalism 3 Lecture
5 Noether’s Theorem and Symmetries 3 Lecture
6 Applications of the Lagrangian Formalism (part 1) 3 Exercise
7 Applications of the Lagrangian Formalism (part 2) 3 Exercise
8 Mid-term test 3
9 The Hamiltonian Formalism (part 1) 3 Lecture
10 The Hamiltonian Formalism (part 2) 3 Lecture
11 Poisson brackets and a relation between CM and QM 3 Lecture
12 Canonical Transformation 3 Lecture
13 Applications of the Hamiltonian Formalism 3 Exercise

4
Contents

1 Mathematical backgrounds 8
1.1 Unit vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.1.1 Unit vectors in Cartesian coordinates . . . . . . . . . . . 8
1.1.2 Unit vectors in cylindrical coordinates . . . . . . . . . . . 8
1.1.3 Unit vectors in spherical coordinates . . . . . . . . . . . . 9
1.1.4 General unit vectors and vector algebra . . . . . . . . . . 9
1.2 Vector differentiation of a scalar field and the gradient . . . . . . . 11
1.3 Conservative vector field . . . . . . . . . . . . . . . . . . . . . . 12
1.4 The vector differential operator and its action . . . . . . . . . . . 12
1.4.1 The divergence of a vector . . . . . . . . . . . . . . . . . 12
1.4.2 Laplacian operator ∇2 . . . . . . . . . . . . . . . . . . . 13
1.4.3 The curl of a vector . . . . . . . . . . . . . . . . . . . . . 13
1.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 Reminder of Newtonian mechanics 15

2.1 Newtonian mechanics for a single particle . . . . . . . . . . . . . 15
2.1.1 Angular momentum and Torque . . . . . . . . . . . . . . 16
2.1.2 Conservation of momentum and angular momentum . . . 16
2.1.3 Energy and energy conservation . . . . . . . . . . . . . . 16
2.1.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Newtonian mechanics for a system of many particles . . . . . . . 18
2.2.1 Momentum and angular momentum . . . . . . . . . . . . 19
2.2.2 Energy and energy conservation of a system of many par-
ticles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3 Lagrangian formalism 21
3.1 The principle of least action . . . . . . . . . . . . . . . . . . . . 21
3.1.1 Configuration space . . . . . . . . . . . . . . . . . . . . 21
3.1.2 Lagrangian . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.2 Lagrange’s equation under changing coordinate systems . . . . . 23

5
Classical Mechanics II

3.3 Lagrange’s equation in rotating coordinate systems and rotational

motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.4 Lagrange’s equation in constraint and generalised coordinates and
oscillatory motions . . . . . . . . . . . . . . . . . . . . . . . . . 24

4 Noether’s Theorem and Symmetries 27

4.1 Noether’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.2 Homogeneity of space and momentum conservation . . . . . . . . 28
4.3 Isotropy of space and angular momentum conservation . . . . . . 29
4.4 Homogeneity of time and energy conservation . . . . . . . . . . . 29

5 Applications of Lagrangian formalism 30

5.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
5.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.6 Bead on a Rotating Hoop . . . . . . . . . . . . . . . . . . . . . . 31
5.7 Double Pendulum . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.8 Two Body Problem . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.9 Charged particles moving in electromagnetic fields . . . . . . . . 33

6 The Hamiltonian formalism 34

6.1 Hamilton’s equations . . . . . . . . . . . . . . . . . . . . . . . . 34
6.1.1 The Legendre transform . . . . . . . . . . . . . . . . . . 34
6.1.2 Hamilton’s equations . . . . . . . . . . . . . . . . . . . . 35
6.1.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . 36
6.2 Conservation laws in the Hamiltonian formalism . . . . . . . . . 38
6.3 Hamilton’s equations from the principle of least action . . . . . . 39
6.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

7 Poisson Brackets and a relation between classical mechanics and quan-

tum mechanics 40
7.1 Poisson brackets . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
7.2 Angular momentum and Runge-Lenz vector . . . . . . . . . . . . 42
7.3 Magnetic monopoles . . . . . . . . . . . . . . . . . . . . . . . . 43
7.4 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

8 Canonical transformations 45
8.1 The equation of canonical transformation . . . . . . . . . . . . . 45
8.2 Infinitesimal canonical transformations . . . . . . . . . . . . . . . 48

6
Classical Mechanics II

8.3 Noether’s theorem in Hamiltonian formalism . . . . . . . . . . . 49

8.4 The Hamilton-Jacobi equation . . . . . . . . . . . . . . . . . . . 49
8.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

7
Chapter 1

Mathematical backgrounds

1.1 Unit vectors

Unit vectors are usually used to form a vector space. In a normed vector space, a
unit vector has a length of 1.
Unit vector v̂ of a vector v is defined as a vector of length 1 and having the
same direction as vector v:
v
v̂ = , (1.1)
kvk
where kvk is the length of vector v. Here vectors are denoted by the bold font. A
unit vector is thus sometimes called a “direction vector”.
Unit vectors are often chosen to form the basis of a vector space. Every vector
in the space may be written as a linear combination of unit vectors.

1.1.1 Unit vectors in Cartesian coordinates

In a three dimensional Cartesian coordinate system, the unit vectors x̂, ŷ and ẑ,
respectively, in the direction of the x, y, and z axes are:
     
1 0 0
x̂ =  0 ; ŷ =
  1 ; ẑ =
  0 . (1.2)
0 0 1
It is easily to see that these vectors are orthogonal. These vector are also called
Cartesian basic vectors.

1.1.2 Unit vectors in cylindrical coordinates

The three orthogonal unit vectors in cylindrical coordinates that describe cylindri-
cal symmetry are denoted as ρ̂, φ̂ and ẑ and related with Cartesian basic vectors

8
Classical Mechanics II

as followed: 
 ρ̂ = cos(ϕ) · x̂ + sin(ϕ) · ŷ
φ̂ = − sin(ϕ) · x̂ + cos(ϕ) · ŷ (1.3)
ẑ = ẑ


We see that ρ̂ and ϕ̂ depend on ϕ and are not constant in direction. Hence when
differentiating or integrating in cylindrical coordinates, these unit vectors them-
selves must also be operated on:

∂ ρ̂
 ∂ ϕ = − sin(ϕ) · x̂ + cos(ϕ) · ŷ


∂ ϕ̂
∂ϕ = − cos(ϕ) · x̂ − sin(ϕ) · ŷ (1.4)
 ∂ ẑ
=0


∂ϕ

1.1.3 Unit vectors in spherical coordinates

The three orthogonal unit vectors in spherical coordinates that describe spherical
symmetry are denoted as r̂ (the direction in which the radial distance from the
origin increases), ϕ̂ ( the direction in which the angle in the x − y plane counter-
clockwise from the positive x − axis is increasing) and θ̂ (the direction in which
the angle from the positive z axis is increasing). These unit vectors are related to
Cartesian unit vectors as follows:

 r̂ = sin(θ ) cos(ϕ) · x̂ + sin(θ ) sin(ϕ) · ŷ + cos(θ ) · ẑ
θ̂ = cos(θ ) cos(φ ) · x̂ + cos(θ ) sin(ϕ) · ŷ − sin(θ ) · ẑ (1.5)
ϕ̂ = − sin(ϕ) · x̂ + cos(ϕ) · ŷ


Similarly with cylindrical unit vectors, spherical unit vectors are functions of θ
and ϕ and are not constant in direction so when differentiating or integrating in
spherical coordinates, these unit vectors themselves must also be operated on.

1.1.4 General unit vectors and vector algebra

An extended space which has n − dimensions is formed by a basic vectors sys-
tem containing n unit vectors ê1 , ê2 , . . . , ên . In a normed vector space, these unit
vectors satisfy:
êi · ê j = δi j , i, j = 1, . . . , n, (1.6)
and
n
êi × ê j = ∑ εi jk êk ≡ εi jk êk , i, j, k = 1, . . . n, (1.7)
i, j=1
where δi j is the Kronecker delta function:

0, if i 6= j
δi j = (1.8)
1, if i = j

9
Classical Mechanics II

And εi jk is the Levi-Civita symbol which is a permutation matrix satisfying:


 +1, if (i, j, k) is an even permutation of (1, 2, 3)
εi jk = −1, if (i, j, k) is a odd permutation of (1, 2, 3) (1.9)
0, otherwise, (2 or more indices are equal)


So we have
εi jk = εki j = ε jki = −ε jik = −εik j = −εk ji
and useful relations between the Levi-Civita symbol and Kronecker delta function:
3
∑ εmnk .εi jk = δmi .δn j − δm j .δni
k=1
3
∑ εm jk .εn jk = 2δmn (1.10)
j,k=1
3
∑ εi jk .εi jk = 6
i, j,k=1

Once a basic vectors are defined, an arbitrary vector A in the n-dimensional

space can be written in terms of unit vectors as followed:
n
A = ∑ Ai .êi ≡ Ai êi , (1.11)
i=1

where Ai , i = 1 . . . n is a component/projection of vector A along the direction of

vector êi .
We can define a unit vector  along vector A as follows:

A = kAk · Â = Ai êi
Ai
⇒ Â = · êi
kAk
In the following we list some important properties of vector algebra (example
in three dimensional vector space):
• Equality of vectors:

A = B ⇔ (A1 , A2 , A3 ) = (B1 , B2 , B3 )

 A1 = B1
⇔ A2 = B2
A3 = B3


Equal vectors are parallel and have the same length, but don’t necessarily
have the same position.

10
Classical Mechanics II

• Vector addition
A + B = (A1 , A2 , A3 ) + (B1 , B2 , B3 )
(1.12)
= (A1 + A2 , A2 + B2 , A3 + B3 )

• Multiplication by a scalar
λ .A = λ .(A1 , A2 , A3 ) = (λ .A1 , λ .A2 , λ .A3 )

• Scalar product of two vectors

A · B = kAk.kBk cos(A, dB)

(1.13)
= A1 B1 + A2 B2 + A3 B3 = Ai Bi
– Scalar product is commutative: A · B = B · A
– If A · B = 0, (A 6= 0, B 6= 0) ⇒ A and B are perpendicular.
• Vector (or cross or outer) product
C = A×B
kCk = A · B. sin(A,
d B)
⇒ A × B = −B × A → not commutative
If A and B are parallel ⇒ A × B = 0
In vector components we have:
A × B = Ai B j .εi jk êk
A × B|k = εi jk Ai B j

1.2 Vector differentiation of a scalar field and the

gradient
In a three-dimensional space, consider the scalar field Φ(x1 , x2 , x3 ). With a change
of dr = (dx1 , dx2 , dx3 ), the field Φ(x1 , x2 , x3 ) will change an amount of dΦ deter-
mined as follows:
∂φ ∂φ ∂φ
dφ = dx1 + dx2 + dx3 .
∂ x1 ∂ x2 ∂ x3
dφ can be expressed as a scalar product of two vectors:
∂φ ∂φ ∂φ
dφ = dx1 + dx2 + dx3 ≡ (∇φ ) · dr,
∂ x1 ∂ x2 ∂ x3
where ∇φ is called the gradient of φ and is often written as gradφ .

11
Classical Mechanics II

1.3 Conservative vector field

• Definition: a vector field is said to be conservative if the line integral of the
vector along any closed path vanishes.
• Example: F is a conservative vector field ⇒
I
F · ds = 0

• A necessary and sufficient condition for F to be conservative is F = −∇Φ,

where φ is a scalar field.
Z b Z b
F · ds = − ∇φ · ds = φ (a) − φ (b)
a a
I I
⇒ F · ds = − ∇φ = 0

1.4 The vector differential operator and its action

∂ ∂ ∂ ∂
∇ = ê1 + ê2 + ê3 ≡ êi
∂ x1 ∂ x2 ∂ x3 ∂ xi
This gradient operator acts on a scalar field as follow:
∂φ ∂φ ∂φ ∂φ
∇φ = ê1 + ê2 + ê3 ≡ êi .
∂ x1 ∂ x2 ∂ x3 ∂ xi

1.4.1 The divergence of a vector

In a three dimensional space, a vector field is written in terms of unit vectors:

V(x1 , x2 , x3 ) = V1 ê1 +V2 ê2 +V3 ê3 ≡ Vi êi

Divergence of V is defined as
   
∂ ∂ ∂ ∂

∇ · V = ê1 + ê2 + ê3 · (V1 ê1 +V2 ê2 +V3 ê3 ) ≡ êi · V j ê j
∂ x1 ∂ x2 ∂ x3 ∂ xi
 
∂V j
⇒ ∇·V = (êi · ê j )
∂ xi
From (1.6) we have
   
∂V j ∂Vi ∂V1 ∂V2 ∂V3
∇·V = δi j = ≡ + +
∂ xi ∂ xi ∂ x1 ∂ x2 ∂ x3

12
Classical Mechanics II

Note that ∇ · V 6= V · ∇:
∂ ∂ ∂ ∂
V · ∇ = Vi = V1 +V2 +V3 .
∂ xi ∂ x1 ∂ x2 ∂ x3

1.4.2 Laplacian operator ∇2

∂2 ∂2 ∂2
∇2 = ∇ · ∇ = + +
∂ x12 ∂ x22 ∂ x32
The Laplacian ∇2 ≡ ∆ is a scalar differential operator.
Laplace’s equation:
∇2 φ = 0

1.4.3 The curl of a vector

∂Vk
curlV = ∇ × V = εi jk êi
∂xj
Stoke’s theorem: Z I
(∇ × A) · da = A · dl
S Γ

1.5 Exercises
Given differentiable vector fields functions A and B and differentiable scalar field
functions f and g of position (x1 , x2 , x3 ), prove the following relations:

1. ∇ ( f g) = f ∇ g + g∇
∇f,

2. ∇ · ( f A) = f ∇ · A + ∇ f · A,

3. ∇ × ( f A) = f ∇ × A + ∇ f × A,

4. ∇ × (∇
∇ f ) = 0,

5. ∇ · (∇
∇ × A) = 0,

6. ∇ · (A × B) = (∇
∇ × A) · B − (∇
∇ × B) × A,

7. ∇ × (A × B) = (B · ∇)A − B(∇ · A) + A(∇

∇ · B) − (A · ∇ )A,

8. ∇ × (∇ ∇ · A) − ∇ 2 A,
∇ × A) = ∇ (∇

9. ∇ (A · B) = A × (∇
∇ × B) + B × (∇
∇ × A) + (A · ∇ )B + (B · ∇ )A,

13
Classical Mechanics II

10. (A · ∇ )r = A,

11. ∇ · r = 3,

12. ∇ × r = 0,

13. ∇ · (r−3 r) = 0,

14. dF = (dr · ∇ )F + ∂∂tF dt (F is a differentiable vector field quantity),

15. dϕ = dr · ∇ ϕ + ∂∂tϕ dt

14
Chapter 2

Reminder of Newtonian mechanics

2.1 Newtonian mechanics for a single particle

A particle is defined as a point-like mass having infinitesimal size such as an
electron, a tennis ball or a planet....
Motion of a particle is described by the second law of Newton:

F(r, ṙ) = ṗ, (2.1)

where:

• F(r, ṙ) is the force acting on the particle. In general this force depends on
the position r and the velocity ṙ of the particle.

• ṗ = dp/dt is the first derivative of the momentum p = mṙ of the particle.

• Both F and p are 3-vectors which we denote by the bold font.

• If ṁ ≡ dm/dt = 0, (2.1) becomes F(r, ṙ) = ma, where a = r̈ is the accel-

eration of the particle.

Equation (2.1) is applied in an inertial frame which is a frame in which a free

particle with ṁ = 0 travels in a straight line (with a constant velocity):

r = r0 + v.t (2.2)
This is the main content of first law of Newton stating an existent of such an inertial
frame.
An inertial frame is not unique. There are an infinite number of inertial frame.
Let S be an inertial frame. Then there are 10 linearly independent transformations
S → S0 such that S0 is also an inertial frame:

15
Classical Mechanics II

• 3 rotations: r0 = Or, where O is 3 × 3 orthorgonal matrix,

• 3 translations: r = r + c, where c is a constant vector,
• 3 boosts: r0 = r + ut, where u is a constant velocity,
• 1 time translation: t 0 = t + c, where c is a constant real number.
These are space and time transformations which make the Galilean group.
Under these transformations, Newton’s laws are invariant.

2.1.1 Angular momentum and Torque

The angular mometum of a particle and the torque acting on it are defined as
follows:
L = r × p, (2.3)

τ = r×F (2.4)
From (2.3) and (2.4) we have
τ = L̇ (2.5)

2.1.2 Conservation of momentum and angular momentum

From (2.1) and (2.5) we see that:
• If F = 0 → dp/dt = 0 or momentum of the particle is conserved.
• If τ = 0 → dL/dt = 0 or angular momentum of the particle is conserved.

2.1.3 Energy and energy conservation

• Kinetic energy
1
T = mṙ.ṙ (2.6)
2
Derivative of T with respect to time
dT
= F.ṙ (2.7)
dt
The change in kinetic energy when the particle travel from position r1 at
time t1 to position r2 at time t2 is:
Z t2 Z t2 Z r2
dT
T (r2 ) − T (r1 ) = dt = F.ṙdt = F.dr (2.8)
t1 dt t1 r1

16
Classical Mechanics II

(2.8) implies that the change in kinetic energy is equal to the work done by the
force acting on the particle. This is the content of the ”work-kinetic energy ”
theorem.
In a special case of conservative force field in which the force depends only on
position r rather than velocity ṙ and the work done is dependent of the path taken.
Thus, for a closed path, the work done vanishes
I
F.dr = 0 ⇔ ∇ × F = 0. (2.9)
We thus can write F in this form:
F = −∇
∇V (r), (2.10)
where V (r) can be understood as the potential. Systems having a potential of this
form can be gravitational, electrostatic and interatomic force-fields.
Substituting (2.10) into (2.8) we have:
Zt2
T (r2 ) − T (r1 ) = − ∇V (r).dr = −(V (r2 ) −V (r1 )), (2.11)
t1
⇒ T (r1 ) + T (r2 ) = V (r1 ) +V (r2 ) ≡ E, (2.12)
where E is the energy of the particle and we see that it is conserved when the
particle is moving in conservative force-fields.

2.1.4 Exercises
• The simple harmonic oscillator: a system moving in one-dimension with
a force proportional to the distance x to the origin: F(x) = −kx, (k is a con-
stant). This force arises from the potential V = kx2 /2. Examine which quan-
tities between momentum, angular momentum and energy are/is conserved?
(ans: momentum is not conserved because F 6= 0, angular momentum is not
defined in one-dimension; energy is conserved).
• Particle moving under gravity: Consider a particle of mass m moving in
3-dimensions under the gravitational pull of a much larger particle of mass
M. The force is F = −G.M.m.r̂/r2 which arises from the potential V =
−G.M.m/r. Are momentum, angular momentum and energy are conserved?
(Ans: momentum is not conserved because F 6= 0; angular momentum is
conserved because F ∼ r; energy is conserved).
• The damped simple harmonic oscillator: We now include a friction term
so that F = −kx − γ ẋ. Examine the conservation of momentum, angular
momentum, and energy? (Ans: the force is not conservative so energy is not
conserved but lost until the partice stops moving).

17
Classical Mechanics II

2.2 Newtonian mechanics for a system of many par-

ticles
Consider a system containing N particles. Apply the second law of Newton for
particle ith having mass mi and position ri :
Ftot
i = ṗi , (2.13)
where Ftot th
i is a vector sum of all forces acting on particle i and ṗi = dpi /dt is
the first derivative of momentum of particle ith with respect to time.
We can write Ftoti in this form:
Ftot
i = ∑ fi j + Fi , (2.14)
j6=i

where fi j is the internal force which particle jth acts on particle ith and Fi is the
sum of all external forces acting on particle ith .
Summing over (2.14) for all particles we have:
N
∑ Ftot
i = ∑ fi j + ∑ Fi = ∑ (fi j + f ji ) + ∑ Fi (2.15)
i=1 i, j, j6=i i i< j i

According to the the third law of Newton: fi j = −f ji , the first term in (2.15)
vanishes. We denote:
• Total of all external forces acting on the system of particles:
N
Fext = ∑ Fi
i=1

• Total mass of the system of particles:

N
M = ∑ mi
i=1

• Center of mass of the system of particles:

∑ mi ri
R= i
∑ mi
i

Then the second law of Newton for the whole of system can be written as
Fext = M R̈. (2.16)
(2.16) has the same form as the second law of Newton for a point particle. This
formula implies that the centre of mass of a system of particles acts just as if all
the mass were concentrated there.

18
Classical Mechanics II

2.2.1 Momentum and angular momentum

Momentum of a system of particles is defined to be
N
P = ∑ pi = M Ṙ,
i=1

We can write the second law of Newton (2.16) for the center of mass as

Fext = P̈.

So if there is no external force acting on the system of particles, Fext = 0, momen-

tum of the system is conserved.
Angular momentum of a system of particles is defined to be

L = ∑ Li = ∑(ri × pi )
i i

Time derivative of angular momentum:

L̇ = ∑(ri × ṗi )
i
= ∑ ri × ( ∑ fi j + Fi )
i j6=i
= ∑ ri × fi j + ∑ ri × Fi
i, j, j6=i i
= ∑ (ri − r j ) × fi j + ∑ ri × Fi
i< j i

If fi j ∝ (ri − r j ) (this means fi j is parallel to the line joining to two particles) we

have
(ri − r j ) × fi j = 0
We define total external torque as

τ ext = ∑ ri × Fi ,
i

then we have
L̇ = τ ext .
So if the total external torque τ ext = 0 the angular momentum of the system of
particles is conserved.

19
Classical Mechanics II

2.2.2 Energy and energy conservation of a system of many par-

ticles
The total kinetic energy of a system of particles:
1
T = ∑ mi ṙ2i (2.17)
i 2

We write ri in this form:

ri = R + r0i ,
where r0i is the distance from the centre of mass to the particle i. Then we have
1 1
T = M Ṙ2 + ∑ mi (ṙ0i )2 (2.18)
2 2 i

(2.18) tells that the kinetic energy splits up into the kinetic energy of the centre
of mass, together with an internal energy describing how the system is moving
around its centre of mass.
Similarly to the case of single particle we have the following:
• The change in total kinetic energy:
Z
T (t2 ) − T (t1 ) = ∑ Fi .dri + ∑ fi j .dri
i i, j,i6= j

• Energy is conserved when both following conditions are satisfied:

– conservative external forces: Fi = −∇

∇iVi (r1 , r2 , ..., rN ),
– conservative internal forces: fi j = −∇
∇iVi j (r1 , r2 , ..., rN ),

where ∇ i = ∂ /∂ ri
In order to get fi j = −f ji and fi j ∝ (ri − r j ) we can require the internal potential to
depend only on the distance between particle ith and particle jth :

Vi j (r1 , r2 , ..., rN ) = Vi j (|ri − r j |)

For the external potential, we assume

Vi (r1 , ..., rN ) = Vi (ri )

so that the external force acting on particle i doesn’t depend on position of other
particles. Hence, similarly to the case of single particle we can prove (exercise)
T +V = const

20
Chapter 3

Lagrangian formalism

3.1 The principle of least action

3.1.1 Configuration space
So far we have used coordinate ri to describe position of particle ith , (i = 1, 2, ..., N)
in a system containing N particles. This coordinate is a vector in three-dimensional
space. So degrees of freedom of the system is 3N.
Now we rewrite the coordinate of the system as xA , (A = 1, 2, ..., 3N) in 3N −
dimensional space known as configuration space C. Each point in C specifies a
configuration of the system such as the position of all particles.
In the configuration space, the second law of Newton can be written as:

∂V
ṗA = − , (3.1)
∂ xA
where pA = mẋA .

3.1.2 Lagrangian
We define Lagrangian L(xA , ẋA ), a function of position xA and velocity ẋA , as
follow:

L(xA , ẋA ) = T (ẋA ) −V (xA ), (3.2)

where
1
T (ẋA ) = mA (ẋA )2
2∑A

is kinetic energy and V (xA ) is the potential energy of the system.

21
Classical Mechanics II

To describe the principle of least action, we consider all smooth paths xA (t) in
C with fixed end points:
xA (ti ) = xinitial
A
; xA (t f ) = xAfinal
To each path, we assign a number called the action S defined as
Z tf
S[xA (t)] = L(xA (t), ẋA (t))dt. (3.3)
ti
The principle of least action is “The actual path taken by the system is an
extremum of S.”
Proof: Consider varying a given path slightly, so
xA (t) → xA (t) + δ xA (t) (3.4)
Z tf    tf
∂L d ∂L ∂L
⇒ δS = A
− A
δ xA dt + A δ xA (3.5)
ti ∂x dt ∂ ẋ ∂ ẋ ti
Since δ xA (ti ) = 0 = δ xA (t f ), the final term vanishes. In order to have δ S = 0
for all variation δ xA (t) we require
 
∂L d ∂L
− = 0, with A = 1, 2, ..., 3N (3.6)
∂ xA dt ∂ ẋA
These are known as Lagranges equations (or sometimes as the Euler-Lagrange
equations).
To finish the proof, we need only show that Lagranges equations are equivalent
to Newtons ⇒ Exercise!!.
• All the fundamental laws of physics can be written in terms of an action
principle. This includes electromagnetism, general relativity, the standard
model of particle physics, and attempts to go beyond the known laws of
physics such as string theory. For example, (nearly) everything we know
about the universe is captured in the Lagrangian
 
√ 1 µν
L = g R − F Fµν + ψ̄ 6 Dψ (3.7)
2
where the terms carry the names of Einstein, Maxwell (or Yang and Mills)
and Dirac respectively, and describe gravity, the forces of nature (electro-
magnetism and the nuclear forces) and the dynamics of particles like elec-
trons and quarks.
• There are two very important reasons for working with Lagranges equa-
tions rather than Newtons. The first is that Lagranges equations hold in any
coordinate system, while Newtons are restricted to an inertial frame. The
second is the ease with which we can deal with constraints in the Lagrangian
system.

22
Classical Mechanics II

3.2 Lagrange’s equation under changing coordinate

systems
We consider a changing of the coordinate system xA , A = 1, 2, ..., 3N to a coor-
dinate system qa , a = 1, 2, ..., 3N. Remember that these coordinates change with
time t. We have
qa = qa (x1 , x2 , ..., x3N ,t) (3.8)
so
dqa ∂ qa A ∂ qa
q̇a ≡
= A ẋ + (3.9)
dt ∂x ∂t
Here we use the “summation convention” in which repeated indices are summed
over. Inversely we also have

xA = xA (q1 , q2 , ..., q3N ,t) (3.10)

so
dxA ∂ xA ∂ xA
ẋA ≡ = q̇a + (3.11)
dt ∂ qa ∂t
The change of Lagrange L due to the change of coordinate system:
 2 A
∂ L ∂ xA ∂ L ∂ 2 xA

∂L ∂ x
= + q̇b + (3.12)
∂ qa ∂ xA ∂ qa ∂ ẋA ∂ qa ∂ qb ∂t∂ qa

∂L ∂ L ∂ ẋA
= (3.13)
∂ q̇a ∂ ẋA ∂ q̇a
we can prove (???)
∂ ẋA ∂ xA
=
∂ q̇a ∂ qa
Finally we obtain:

∂ L ∂ xA
     
d ∂L ∂L d ∂L
− = − A (3.14)
dt ∂ q̇a ∂ qa dt ∂ ẋA ∂ x ∂ qa

Equation (3.14) implies that if the choice of coordinates is invertible (i.e: det(∂ xA /∂ qa ) 6=
0), the Lagrange’s equation is invariant with the change between coordinates xA
and qa . So the form of Lagrange’s equations holds in any coordinate system. This
is in contrast to Newton’s equations which are only valid in an inertial frame.

23
Classical Mechanics II

3.3 Lagrange’s equation in rotating coordinate sys-

tems and rotational motion
Motion of a free particle with mass m moving in the coordinate system r = (x, y, z)
is described by the Lagrangian
1
L = mṙ2
2
Now we consider the motion of this particle in a coordinate system r’ =
(x0 , y0 , z0 ) which is obtained by rotating r about the z axis with an angular velocity
ω = (0, 0, ω). The relation between (x, y, z) and (x0 , y0 , z0 ) is expressed by
 0
 x = x. cos(ωt) + y. sin(ωt)
y0 = y. cos(ωt) − x. sin(ωt) (3.15)
 0
z =z
In the rotating frame with coordinate system (x0 , y0 , z0 ) the Lagrangian has this
form:
1 1
L = m[(ẋ0 − ωy0 )2 + (ẏ0 + ωx0 )2 + (ż0 )2 ] = m(ṙ0 + ω × r0 )2
2 2
Equations of motion for the particle in the rotating frame can be derived by
using Euler-Lagrange’s equation:
∂L
0
= m(ṙ0 × ω − ω × (ωω × r0 ))
 ∂ r
d ∂L
= m(r̈0 + ω × ṙ0 )
dt ∂ ṙ0
So  
d ∂L ∂L
− = m(r̈0 + ω × (ω
ω × r0 ) + 2ω
ω × ṙ0 ) = 0
dt ∂ ṙ0 ∂ r0
• Centrifugal force: −mω ω × r0 )
ω × (ω
ω × ṙ0
• Coriolis force: −2mω

3.4 Lagrange’s equation in constraint and generalised

coordinates and oscillatory motions
Another advantage of the Lagrangian formulation is to deal with “constraint forces”
in Newtonian mechanics as constraints and generalized coordinates. The follow-
ing is some examples using this advantage.

24
Classical Mechanics II

The pendulum with Newtonian mechanics

A pendulum of a mass m connected to a fix point by a string of length `. The
motion of the pendulum can be described using Cartesian coordinate (x, y). The
constraints in this case can be expressed as

x = `. sin θ
(3.16)
y = `. cos θ
and
x2 + y2 = `2
Using Newtonian mechanics, the tension T which the rod acts on the mass m
satisfies: 
mẍ = −T x/`
(3.17)
mÿ = mg − Ty/`
Applying the constraint condition we have
θ̈ = −(g/`) sin θ
T = m`(θ̇ )2 + mg cos θ
Now we will deal with constraints more rigorously and we can avoid the step
of calculating constraint forces by using Lagrangian formulation. Before going to
this, we introduce a method of holonomic constraints.

Holonomic Constraints
“Holonomic constraints” are relationships between the coordinates of the form
fα (xA ,t) = 0, α = 1, 2, ..., (3N − n)
Holonomic constraints can be solved in terms of n “generalised coordinates”
qi , i = 1, 2, ..., n. So we have
xA = xA (q1 , q2 , ..., qn )
The system is said to have n degrees of freedom.
We now introduce (3N − n) new variables λα , called “Lagrange multipliers”
and define a new Lagrangian:

L0 = L(xA , ẋA ) + λα fα (xA ,t)

and λα is treated as new coordinates. Lagrange’s equation applied for λα gives
back the constraints:
∂ L0
= fα (xA ,t) = 0
∂ λα

25
Classical Mechanics II

and equations for xA are:

 
d ∂L ∂L ∂ fα
− = λα
dt ∂ ẋA ∂ xA ∂ xA
The LHS is the equation of motion for the unconstrained system. The RHS is
the manifestation of the constraint forces in the system. We can now solve these
equations as we did in the Newtonian formulation.

The pendulum with Lagrangian formulation

The Lagrangian for the pendulum is given by
1 1
L0 = m(ẋ2 + ẏ2 ) + mgy + λ (x2 + y2 − `2 )
2 2
Applying Lagrange’s equation for x and y we obtain:

mẍ = λ x
(3.18)
ÿ = mg + λ y
while the equation of motion for λ gives the constraint
x2 + y2 − `2 = 0.
Comparing to the Newtonian approach we have the relation between the La-
grange multiplier λ and the tension T
λ = −T /`
Theorem: For constrained systems, we may derive the equations
of motion directly in generalised coordinates qi :
L[qi , q̇i ,t] = L[xA (qi ,t), ẋA (qi , q̇i ,t)]
Example:
1
L = m(ẋ2 + ẏ2 ) + mgy
2
Changing the coordinates gives
1
L = m`2 θ̇ 2 + mg` cos θ
2
Lagrange’s equation for θ gives
 
d ∂L ∂L
− = m`2 θ̈ + mg` sin θ = 0
dt ∂ θ̇ ∂θ

26
Chapter 4

Noether’s Theorem and Symmetries

We consider functions F(qi , q̇i ,t), i = 1, 2, ..., n of coordinates, where qi satisfies

Euler-Lagrange equation. Their time derivatives and time t is called a “constant
of motion” (or a conserved quantity) if the total time derivative vanishes:
dF ∂F ∂F ∂F
= q̇ j + q̈ j + =0
dt ∂qj ∂ q̇ j ∂t
• Example 1: If L does not depend explicitly on time t (∂ L/∂t = 0), the
quantity defined as follows is conserved:
∂L
H = q̇ j −L
∂ q̇ j
When H is a function of qi and pi , it is known as “Hamiltonian”. It is
usually identified with the total energy of the system.
Proof ????
• Example 2: If ∂ L/∂ q j = 0 for some q j , then p j = ∂ L/∂ q̇ j is conserved. In
this case, q j is said to be “ignorable” (or “cyclic”).

4.1 Noether’s theorem

Consider a one-parameter family of maps
qi (t) → Qi (s,t), s ∈ R
such that Qi (0,t) = qi (t). This transformation is said to be a “continuous symme-
try” of the Lagrangian L if
∂
L[Qi (s,t), Q̇i (s,t),t] = 0
∂s

27
Classical Mechanics II

Noether’s theorem states that for each such continuous symmetry there exists a
conserved quantity.
Proof:
We have  
∂L d ∂ L ∂ Qi
=
∂ s s=0 dt ∂ q̇i ∂ s s=0
Obviously we see that if
∂L
∂s s=0
then this quantity is conserved:

∂ L ∂ Qi
∂ q̇i ∂ s s=0

4.2 Homogeneity of space and momentum conser-

vation
Consider a closed system of N particles with Lagrangian:
1
L= mi ṙ2 −V (|ri − r j |).
2∑
This Lagrangian is invariant under the translation ri → ri + s.n, where n is an
arbitrary vector and s is an arbitrary real number. This implies that

L(ri , ṙi ,t) = L(ri + sn, ṙi ,t).

This is the statement that space is homogeneous and a translation of the sys-
tem by sn does nothing to the equations of motion. Note that these translations
are elements of Galilean group. From Noether’s theorem, conserved quantities
associated with translations can be determined as:
∂L
∑ ∂ ṙi .n = ∑ pi.n.
i=1 i

This quantity is recognized as the total linear momentum of the system in the
direction n. Since n is arbitrary so we can conclude that the total momentum of
the system ∑i pi is conserved.

28
Classical Mechanics II

4.3 Isotropy of space and angular momentum con-

servation
The isotropy of space is the statement that a closed system, described by the La-
grangian as mentioned above is invariant under the rotations around an axis n̂. To
find out conserved quantities in this case, we will work with the infinitesimal form
of the rotations

ri → ri + δ ri
= ri + α n̂ × ri

where α is considered infinitesimal. The Lagrangian of the system is invariant of

this transformation:

L(ri , ṙi ) = L(ri + α n̂ × ri , ṙi + α n̂ × ṙi ) (4.1)

According to Noether’s theorem, the conserved quantity associated with this sym-
metry is
∂L
.(n̂ × ri ) = n̂.(ri × pi ) = n̂.L (4.2)
∂ ṙi
where L is total angular momentum of the system.

4.4 Homogeneity of time and energy conservation

Homogeneity of time implies that the Lagrangian is invariant under the transfor-
mation t → t + s, where s is an arbitrary real number. This means ∂ L/∂t = 0. We
already know that in this case the conserved quantity is

∂L
H = q̇i − L.
∂ q̇i
In the system we are considering, this quantity is total energy.

29
Chapter 5

Applications of Lagrangian
formalism

5.1
If L is a Lagrangian for a system of n degrees of freedom satisfying Lagrange’s
equations, show by direct substitution that
dF(q1 , q1 , ..., qn ,t)
L0 = L +
dt
also satisfies Lagrange’s equations, where F is any arbitrary, but differentiable,
function of its arguments.

5.2
Obtain the Lagrangian equations of motion for a spherical pendulum, i.e. a mass
point suspended by a rigid weightless rod.

5.3
A particle of mass m moves in one dimension such that it has the Lagrangian

m2 ẋ4
L= + mẋ2V (x) −V 2 (x),
12
where V is some differentiable nature of the system on the basis of this equation.
Find the equation of motion for x(t) and describe the physical nature of the system
on the basic of this equation.

30
Classical Mechanics II

5.4
Obtain the equation of motion for a particle falling vertically under the influence
of gravity when frictional force obtainable from a dissipation function kv2 /2 are
present. Integrate the equation to obtain the velocity as a function of time and
show that the maximum possible velocity for fall from rest is v = mg/k.

5.5
The Lagrangian for a particle in an electromagnetic field is
q
L = T − qΦ(r(t)) + A(r(t)).v(t).
c
The electromagnetic field is invariant under a gauge transformation of the scalar
and vector potential given by

A → A + ∇Ψ(r,t),
1 ∂Ψ
Φ → Φ− ,
c ∂t
where Ψ is arbitrary (but differentiable). What effect does this gauge transforma-
tion have on the Lagrangian of a particle moving in the electromagnetic field? Is
the motion affected?

5.6 Bead on a Rotating Hoop

The hoop is of radius a and rotates with frequency ω as shown in the figure on the
right. The bead, of mass m, is threaded on the hoop and moves without friction.
Determine the following:

31
Classical Mechanics II

• Find the relation between Cartesian co-

ordinates and generalized coordinates de-
scribing the motin of the bead.

• Lagrangian of the bead in terms of Carte-

sian coordinates (x, y).

• Lagrangian of the bead in terms of gener-

alized coordinate Ψ.

• Find positions of the bead so that it can re-

main stationary on the hoop.

5.7 Double Pendulum

A double pendulum is drawn in the right figure, consisting of two particles of
mass m1 and m2 , connected by light rods of length l1 and l2 .

• Find Lagrangian of the system in terms of

Cartesian coordinates (x, y).

• Find Lagrangian of the system in terms of

generalized coordinates (θ1 , θ2 )

• Find equations of motion of the system in

generalized coordinates.

5.8 Two Body Problem

Consider two objects of mass m1 and m2 interacting through a central force. In
this case, the potential V (|r12 |) depends on the length of separation r12 = r1 − r2 .

32
Classical Mechanics II

• Write lagrangian of the system in terms of separation r12 and the center of
mass R = (m1 r1 + m2 r2 )/(m1 + m2 )

• Find conserved quantities in this case.

• Find equation of motion and then the effective potential of the system.

5.9 Charged particles moving in electromagnetic fields

A particle of charge e moving in an electromagnetic field with electric field E and
magnetic field B that can be written in terms of a vector potential A(r,t) and a
scalar potential φ (r,t) as follows:

B = ∇×A
1 ∂A
E = −∇φ −
c ∂t
where c is the speed of light. The Lagrangian describing motion of the particle is
 
1 2 1
L = mṙ − e φ − ṙ.A
2 c

• Find the momentum conjugated to r of the particle?

• Using Euler-Lagrange’s equation to find the Lorentz Force acting on the

particle in terms of E and B?

33
Chapter 6

The Hamiltonian formalism

6.1 Hamilton’s equations

In the Lagrangian formalism, a system of particles is described by the Lagrangian
L(qi , q̇i ,t) which is a function of generalized coordinates qi (i = 1, . . . , n). Equa-
tions of motion are given by Lagrange’s equations:
 
d ∂L ∂L
− = 0, (6.1)
dt ∂ q̇i ∂ qi

where i = 1, . . . , n. We have totally n second order differential equations with 2n

unknown parameters. We thus need 2n initial conditions to solve those equations
including n initial conditions for qi (qi (t = 0)) and n initial conditions for q̇i (q̇i (t =
0)). The basic idea of Hamiltons approach is to try and place qi and q̇i on a more
symmetric footing. Technically, we introduce n generalized momenta pi defined
as follows:
∂L
pi = , i = 1, . . . , n (6.2)
∂ q̇i
In Hamiltonian formalism, n generalized momenta pi are used instead of n gener-
alized coordinates q̇i . Thus to determine the system of particles, the n pair {qi , pi }
are used to define a point in the 2n−dimensional phase space.

6.1.1 The Legendre transform

We want to find a function of qi and pi (but not q̇i ) on the phase space that con-
tains the same information as the Lagrangian L(qi , q̇i ,t) to determine the unique
evolution of qi and pi . The Legendre transform is a mathematical trick to do this.

34
Classical Mechanics II

Assume we have an arbitrary function f (x, y) of two variables x and y. Its total
derivative is
∂f ∂f
df = dx + dy. (6.3)
∂x ∂y
Using f (x, y) we can define a new function which depends on three variables x, y
and u as follows:
g(x, y, u) = u x − f (x, y), (6.4)
and the total derivative of g can be calculated as

∂f ∂f
dg = u dx + x du − dx − dy. (6.5)
∂x ∂y
If u is a specific function of x and y:

∂f
u(x, y) = , (6.6)
∂x
then (6.5) becomes:
∂f
dg = x du − dy. (6.7)
∂y
We can invert (6.6) to get x as a function of u and y which is denoted as x(u, y).
So we can obtain g as a function of only u and y as follows:

g(u, y) = u x(u, y) − f (x(u, y), y). (6.8)

(6.8) is the Legendre transform. It helps to transform one function f (x, y) to an-
other function g(u, y) where u = ∂ f /∂ x. In this transformation, we have not lost
any information. Vice versa, we can always recover f (x, y) from g(u, y) by

∂g ∂g ∂f
= x(u, y) , = , (6.9)
∂u y ∂y u ∂y

then
∂g
f= u−g (6.10)
∂u
will take us back to the original function.

6.1.2 Hamilton’s equations

Assuming that a system of particles is described by a Lagrangian L(qi , q̇i ,t), i =
1, . . . , n depending on the coordinates qi , theirs derivatives q̇i and time. We can

35
Classical Mechanics II

define the Hamiltonian to be the Legendre transform of the Lagrangian L(qi , q̇i ,t)
with respect to the variables q̇i :
n
H(qi , pi ,t) = ∑ pi q̇i − L(qi , q̇i ,t), (6.11)
i=1

where q̇i is eliminated from the right hand side in favor of pi by this relation:
∂L
pi = ≡ pi (qi , q̇i ,t). (6.12)
∂ q̇i
q̇i = q̇i (qi , pi ,t) can be get back by inverting (6.12).
Consider the variation of H(qi , pi ,t):
 
∂L ∂L ∂L
dH = (q̇i d pi + pi d q̇i ) − dqi + d q̇i + dt
∂ qi ∂ q̇i ∂t
∂L ∂L
= q̇i d pi − dqi − dt. (6.13)
∂ qi ∂t
On the other hand, dH can be written as
∂H ∂H ∂H
dH = dqi + d pi + dt. (6.14)
∂ qi ∂ pi ∂t
Equating terms of (6.13) and (6.14) we have
∂H
ṗi = −
∂ qi
∂H
q̇i =
∂ pi
∂L ∂H
− = (6.15)
∂t ∂t
(6.15) are Hamilton’s equations. We see that by transforming the Lagrangian
formalism to the Hamiltonian formalism we replace n second order differential
equations for qi to 2n first order differential equations for pi and qi . Technically,
for solving equations, this is not helpful but we will see, conceptually, this is very
helpful.

6.1.3 Examples
A particle in a potential
A particle moving in a potential in 3-dimensional space. The Lagrangian is simply
1
L = mṙ2 −V (r) (6.16)
2
36
Classical Mechanics II

We calculate the momentum by taking the derivative with respect ṙ:

∂L
p= = mṙ, (6.17)
∂ ṙ
in this case we see that p defined in (6.17) coincides with what we usually call
momentum (but in general it is not).
The Hamilton is given by
1 2
H = p · ṙ − L = p +V (r) (6.18)
2m
Hamilton’s equations are:
∂H 1
ṙ = = p (6.19)
∂p m
∂H
ṗ = − = −∇ ∇V (6.20)
∂r
(6.20) is the definition of momentum in terms of velocity and (6.20) is the second
law of Newton which we knew before.

A particle in an electromagnetic field

We have known that the Lagrangian for a charge particle moving in an electro-
magnetic field is  
1 2 1
L = mṙ − e φ − ṙ · A . (6.21)
2 c
From (6.21) we can compute the momentum conjugate to the position as follows:
∂L e
p= = mṙ + A, (6.22)
∂ ṙ c
in this case we see that, because of the last term in (6.22), the momentum conju-
gate to the position of the charge particle is different from the “normal” momen-
tum which we usually call.
Inverting (6.22) we have
1 e 
ṙ = p− A . (6.23)
m c
From (6.21) and (6.23) we can compute Hamiltonian as follows
H(p, r) = p · ṙ − L
 
1  e  1  e 2 e  e 
= p· p− A − p − A − eφ + p− A ·A
m c 2m c cm c
1  e 2 
= p − A + eφ . (6.24)
2m c

37
Classical Mechanics II

Now we can derive Hamilton’s equations:

∂H
ṙ =
∂p
1 e 
= p− A (6.25)
m c
ṗ = −∂ H/∂ r can be expressed in terms of components as:

∂H
ṗi = −
∂ ri
∂φ e  e  ∂Aj
= −e + pj − Aj (6.26)
∂ ri cm c ∂ ri
?? Exercise: Derive the Lorentz force which is has the same form as derived
using the Lagrangian formalism !

6.2 Conservation laws in the Hamiltonian formal-

ism
• If the Hamiltonian H of the system does not depend on time explicitly
(∂ H/∂t = 0), then H itself is a constant of motion.
We have:
dH ∂H ∂H ∂H
= q̇i + ṗi +
dt ∂ qi ∂ pi ∂t
∂H
= − ṗi q̇i + q̇i ṗi +
∂t
∂H
= (6.27)
∂t

• If an ignorable coordinate q doesn’t appear in the Lagrangian then, by con-

struction, it also doesn’t appear in the Hamiltonian. The conjugate momen-
tum p is then conserved:

∂H
ṗ = = 0. (6.28)
∂q

38
Classical Mechanics II

6.3 Hamilton’s equations from the principle of least

action
As we know, in the Lagrangian formalism, Lagrange’s equations can be derived
by the principle of least action by requiring δ S = 0 for all paths with fixed end
points so that δ qi (t1 ) = 0 = δ qi (t2 ), where the action is defined as follows:
Z t2
S= L(qi , q̇i ,t)dt. (6.29)
t1

In the Hamiltonian formalism, we define the action:

!
Z t2 n
S=
t1
∑ piq̇i − H dt, (6.30)
i=1

where, q̇i = q̇i (qi , pi ).

Now we consider varying qi and pi independently. And remember that in the
Hamiltonian formalism, we treated qi and pi on the equal footing. We have:
Z t2  
∂H ∂H
δS = q̇i δ pi + pi δ q̇i − δ pi − δ qi dt
t1 ∂ pi ∂ qi
Z t2     
∂H ∂H
= q̇i − δ pi + − ṗi − δ qi dt + (pi δ qi )|tt21 (6.31)
t1 ∂ pi ∂ qi

Analogous to the case of Lagrangian formalism, we require that the end points of
qi are fixed:
δ qi (t1 ) = 0 = δ qi (t2 ). (6.32)
δ pi can be free at the end points t = t1 and t = t2 . So we see that, in this context,
qi and pi are not symmetric in the Hamiltonian formalism as we expect.
Thus, from (6.31), if we look for extrema δ S = 0 for all δ qi and δ pi we get
Hamilton’s equations:

∂H
q̇i =
∂ pi
∂H
ṗi = − (6.33)
∂ qi

6.4 Exercises

39
Chapter 7

Poisson Brackets and a relation

between classical mechanics and
quantum mechanics

7.1 Poisson brackets

In this section, we’ll present algebraic description of classical dynamics which
makes it look almost identical to quantum mechanics.
Let f (q, p) and g(q, p) be two functions on phase space. Then the Poisson
bracket is defined to be
∂ f ∂g ∂ f ∂g
{ f , g} = − (7.1)
∂ qi ∂ pi ∂ pi ∂ qi
In the following, we list some of properties of the Poisson bracket:

• Anti-commutative:
{ f , g} = −{g, f }

• Linearity:
{α f + β g, h} = α{ f , h} + β {g, h}

• Leibniz rule:
{ f g, h} = f {g, h} + { f , h}g

• Jacobi identity:

{ f , {g, h}} + {g, {h, f }} + {h, { f , g}} = 0

40
Classical Mechanics II

We see that the Poisson bracket {, } satisfy the same algebraic structure as matrix
commutator [, ], and the differentiation operator d. This is related to Heisenberg’s
and Schrödinger’s point of view on quantum mechanics respectively.
The relationship between classical mechanics and quantum mechanics is also
shown via these expressions:

{qi , q j } = 0
{pi , p j } = 0
{qi , p j } = δi j (7.2)

?? Exercise: Using Hamilton’s equations, we can proof this useful formula:

df ∂f
= { f , H} + (7.3)
dt ∂t
An interesting consequence of (7.3) is that if we can find a function I(p, q)
(doesn’t depend on time explicitly) which satisfy

{I, H} = 0 (7.4)

then I(p, q) is a constant of motion. I and H are said to be Poisson commute. Here
are some examples of this consequence:

• It is easily to see that H “commutes” with itself : {H, H} = 0 so when H

doesn’t depend explicitly on time, H is a constant of motion as we have
known.

• If qi is ignorable then its momentum conjugate pi will commute with H

{pi , H} = 0

and as we knew before that in this case pi is conserved.

Another important property of Poisson bracket is that if I and J are constant

of motion then, from the Jacobi identity, we have

{{I, J}, H} = {I, {J, H}} + {{I, H}, J} = 0.

This means that {I, J} is also a constant of motion. We say that the constants
of motion form a closed algebra under the Poisson bracket.

41
Classical Mechanics II

7.2 Angular momentum and Runge-Lenz vector

Angular momentum
Angular momentum of a particle is defined as
L = r × p, (7.5)
which is in component form:
L1 = r2 p3 − r3 p2 , L2 = r3 p1 − r1 p3 , L3 = r1 p2 − r2 p1 . (7.6)
Using relation at (7.2) we can derive (?? exercise !):
{L1 , L2 } = L3 . (7.7)
So if L1 and L2 are conserved, L3 will also conserved finally the whole vector
L is conserved. We can prove also:
{L2 , Li } = 0 (7.8)
where L2 = L12 + L22 + L32 . This should all be looking familiar from quantum me-
chanics. L1 , L2 , L3 form a closed algebra (which is SU(2)) under the Poisson
bracket.

Runge-Lenz vector
The Runge-Lenz vector was proposed by Hermann, Bernoulli, Laplace and Pauli.
It is defined as
1
A = p × L − r̂, (7.9)
m
where r̂ = r/krk. This vector satisfies these relations:
A·L = 0 (7.10)
and
2 p2 1
 
{Li , A j } = εi jk Ak , {Ai , A j } = − − εi jk Lk (7.11)
m 2m r
If the Hamiltonian has a fork
p2 1
H= − (7.12)
2m r
then we have
2H
{Ai , A j } = − εi jk Lk (7.13)
m
and
{H, A} = 0. (7.14)
We see that the Hamiltonian with −1/r potential has another constant of motion
A.

42
Classical Mechanics II

7.3 Magnetic monopoles

A particle of charge e in a magnetic field B = ∇ × A is described by the Hamilto-
nian: 2 m
1  e
H= p − A(r) = ṙ2 , (7.15)
2m c 2
where ṙ = ṙ(r, p) (is a function of r and p).
Let’s compute the Poisson bracket structure for this system (?? Exercise):
e
{mṙi , mṙ j } = εi jk Bk , {mṙi , r j } = −δi j (7.16)
c
Now we will use (7.16) to describe a postulated object known as a magnetic
monopole. In the fact, all magnets ever discovered are dipoles which have both a
north and south pole. Chop the magnet in two, and each piece also has a north and
a south pole. This fact can be explained using Maxwell’s equations as follows.
When we use the gauge potential A to formulate the electromagnetism, then the
magnetic field can be defined as B = ∇ ×A, we obtain one of Maxwell’s equations

∇ · B = 0. (7.17)

(7.17) states that any flux that enters a region must also leave. This means that
there can be no magnetic monopole.
A monopole would have a radial magnetic field:
r
B=g . (7.18)
r3
It’s easy to point out that (7.18) does not satisfy (7.17). This means magnetic
monopoles are forbidden by Maxwell’s equations.
Now we can consider the motion of an electron in the background of a monopole.
We don’t have a gauge potential A to set up the Lagrangian. But we can use the
Poisson brackets (7.16) which contain only magnetic field.
Consider the generalized angular momentum
ge
J = mr × ṙ − r̂ (7.19)
c
where r̂ = r/r. Using (7.16) we can show that the Hamiltonian given at (7.15) and
J satisfy
{H, J} = 0. (7.20)
This implies that J is a constant of motion. Since J is conserved, we can look
at r̂ · J = −eg/c to learn that the motion of an electron in the background of a
magnetic monopole lies on a cone of angle cos θ = eg/cJ pointing from the vector
J.

43
Classical Mechanics II

7.4 Exercises

44
Chapter 8

Canonical transformations

8.1 The equation of canonical transformation

In the Hamiltonian formalism, we want to have a symmetry between q and p.
However, the Hamilton’s equations as shown in (6.33) is not really symmetric.
We thus introduce a way to write the Hamilton’s equations so that they look more
symmetric. Define the 2n vector

x = (q1 , . . . , qn , p1 , . . . , pn )T (8.1)

and the 2n × 2n matrix J:

 
0n×n 1n×n
M= (8.2)
−1n×n 0n×n

With this notation, the Hamilton’s equations are written in the form:

∂H
ẋ = M . (8.3)
∂x
So far we know in the Lagrangian formalism, the Lagrange’s equations are in-
variant under a change of the coordinate system. In the Hamiltonian formalism
we will do the same and let’s see whether the Hamilton’s equations are invariant
under a change of the coordinate system.
Assuming we have this transformation of the coordinate system defined at
(8.1):
xi → yi (x), i = 1, . . . , 2n. (8.4)

45
Classical Mechanics II

We have:
∂ yi
ẏi = ẋ j (8.5)
∂xj
∂ yi ∂ H ∂ yl
= M jk
∂xj ∂ yl ∂ xk
∂H
= (Ji j M jk Jlk )
∂ yl
Collating all indices we have

∂H
ẏ = (JMJ T ) (8.6)
∂y
where
∂ yi
Ji j =
∂xj
is the Jacobian of the transformation of coordinates. We see that Hamilton’s equa-
tions are invariant under any transformation of coordinates whose Jacobian M
satisfies:
JMJ T = M (8.7)
or
∂ yi ∂ yl
M jk = Mil (8.8)
∂xj ∂ xk
The Jacobian M that satisfies this condition is said to be symplectic and a trans-
formation of coordinates with a symplectic Jacobian is said to be a canonical
transformation.
We have this theorem relating canonical transformation with a symplectic Ja-
cobian:

Theorem:
The Poisson bracket is invariant under canonical transformations. Conversely,
any transformation which preserves the Poisson bracket structure so that

{Qi , Q j } = 0 = {Pi , Pj } and {Qi , Pj } = δi j (8.9)

is canonical.

46
Classical Mechanics II

Proof
• Firstly, we show that the Poisson bracket is invariant under canonical trans-
formations. Consider two functions:
∂ f ∂g ∂ f ∂g
{ f , g} = − (8.10)
∂ qi ∂ pi ∂ pi ∂ qi
∂f ∂g
= Mi j
∂ xi ∂xj

Now we consider a transformation x → y(x), then:

∂f ∂f
= Jki (8.11)
∂ xi ∂ yk
Subtituting (8.11) into (8.11) we have

∂f ∂g
{ f , g} = Jki Mi j Jl j (8.12)
∂ yk ∂ yl
If the transformation (8.11) is canonical where (8.7) is hold then

∂f ∂g
{ f , g} = Mkl . (8.13)
∂ yk ∂ yl
This emplies that Poisson brackets are invariant under a canonical transfor-
mation.

• Conversely, we will write the Jacobian using notation (qi , pi ) (in the old
co-ordinates) and (Qi (q, p), Pi (q, p)) as:
 
∂ Qi /∂ q j ∂ Qi /∂ p j
Ji j = (8.14)
∂ Pi /∂ q j ∂ Pi /∂ p j

If we compute JMJ T in componets we will get

 
T {Qi , Q j } {Qi , Pj }
(JMJ )i j = (8.15)
{Pi , Q j } {Pi , Pj }

so whenever the Poisson bracket structure is preserved, the transformation

is canonical.

47
Classical Mechanics II

8.2 Infinitesimal canonical transformations

We consider the following transformations:
qi → Qi = qi + αFi (q, p) (8.16)
pi → Pi = pi + αEi (q, p)
where α is infinitesimal small. The Jacobian of this transformation is computed
to be
 
δi j + α∂ Fi /∂ q j α∂ Fi /∂ p j
Ji j = . (8.17)
α∂ Ei /∂ q j δi j + α∂ Ei /∂ p j
The condition of the transformation to be canonical (JMJ T = M) giving
∂ Fi ∂ Ei
=− (8.18)
∂qj ∂ pj
This condition is satisfied if
∂G ∂G
Fi = and Ei = − (8.19)
∂ pi ∂ qi
for some function G(q, p). We say that G generates the transformation.
In this point of view, we can interpret the canonical transformation in a differ-
ent way. Consider these transformations:
qi → Qi (q, p; α) (8.20)
pi → Pi (q, p; α).
Suppose that these transformations are canonical for all α ∈ R and have the prop-
erty that
Qi (q, p; α = 0) = qi and Pi (q, p; α = 0) = pi . (8.21)
Then we can say that the transformations take us from one point in the phase
space (qi , pi ) ≡ (Qi (q, p; α = 0), Pi (q, p; α = 0)) to another point in the same
phase space (Qi (q, p; α), Pi (q, p; α)) by varying the parameter α. Each way of
varying α, we trace out lines in phase space. From (8.17) and (8.19) we see that
the tangent vectors to these lines are given by
dqi ∂G d pi ∂G
= and =− . (8.22)
dα ∂ pi dα ∂ qi
If G is the Hamiltonian H and α is the time t, (8.22) becomes Hamilton’s equa-
tions.
We can conclude from what we have found is that there is a link between one-
parameter family of canonical transformation and Hamiltonian flow and a link
between the Hamiltonian and time.

48
Classical Mechanics II

8.3 Noether’s theorem in Hamiltonian formalism

So far we have learnt that in the Lagrangian formalism, there is a connection
between symmetries and conservation quantities. In the Hamiltonian formalism
we will also see there is a similarity.
Consider an infinitesimal canonical transformation generated by G. We have
∂H ∂H
δH = δ qi + δ pi (8.23)
∂ qi ∂ pi
∂H ∂G ∂H ∂G
= α −α + O(α 2 )
∂ qi ∂ pi ∂ pi ∂ qi
= α{H, G}
The generator G is called a symmetry of the Hamiltonian if δ H = 0. This means:
{G, H} = 0 (8.24)
On the other hand,
dG
= {G, H} (8.25)
dt
So we can conclude that if G is a symmetry then G is conserved. Vice versa, if
we have a conserved quantity G we can always use this to generate a canonical
transformation which is a symmetry.

8.4 The Hamilton-Jacobi equation

We know that in the Lagrangian formalism, the action is defined as:
Z T
S= L(qi , q̇i ,t)dt (8.26)
0

This integration is evaluated for all paths q(t) with fixed end points
f inal
qi (0) = qinitial
i , qi (T ) = qi . (8.27)
Then the true path taken is an extremum of the action or δ S = 0.
Now we consider the action evaluated only along the true path qclassical
i and
define f inal
initial,qi ,T
W (qi ) = S[qclassical
i ]. (8.28)
From the Lagrangian formalism we know how to get the variation of the action
S caused by varying qi (t):
Z T     T
∂L d ∂L ∂L
δS = dt − δ qi (t) + δ qi (t) (8.29)
0 ∂ qi dt ∂ q̇i ∂ q̇i 0

49
Classical Mechanics II

If we evaluate this on the classical path (qi (t) satisfying Euler-Lagrange’s equa-
tion) and qi (t) is fixed at the initial point (δ qi (0) = 0) then we have
∂L
δS = δ qi (t) (8.30)
∂ q̇i t=T
f inal
On the other hand, from (8.28) we have (if we consider only a variation of qi )
∂W f inal
δ S = δW = f inal
δ qi (8.31)
∂ qi
So we get
∂W ∂L f inal
f inal
= = pi (8.32)
∂ qi ∂ q̇i t=T

If we consider a variation T → T + δ T we have

dW ∂W ∂W f inal ∂W f inal f inal
= + f inal q̇i = + pi q̇i (8.33)
dT ∂T ∂q ∂T
i

On the other hand we have

dW dS
= =L (8.34)
dT dT
so
∂W 
f inal f inal f inal f inal

f inal f inal
= − pi q̇i − L(qi , q̇i , T ) = −H(qi , pi , T ). (8.35)
∂T
f inal
Now if we replace qi → qi and T → t we have
∂W ∂W
= pi and = −H(qi , pi ,t) (8.36)
∂ qi ∂t
thus
∂W
= −H(qi , ∂W /∂ qi ,t). (8.37)
∂t
This is the Hamilton-Jacobi equation!

8.5 Exercises
Exercise 1
• A system moving in one-dimensional space described by the Hamiltonian:
p2 1
H= − 2. (8.38)
2 2q

50
Classical Mechanics II

Show that this quantity is a constant of motion:

pq
D= − Ht (8.39)
2

• For a motion in a plane with the Hamiltonian

H = kpkn − ar−n (8.40)

where p is the vector of the momenta conjugate to the Cartesian coordinates,

show that there is a constant of motion
p·r
D= − Ht (8.41)
n

Exercise 2
Show directly that the transformation
 
1
Q = log sin p , P = q cot p (8.42)
p

is canonical.

Exercise 3
Show directly for a system of one degree of freedom that the transformation

αq2 p2
 
αq
Q = arctan , P= 1+ 2 2 (8.43)
p 2 α q

is canonical, where α is an arbitrary constant of suitable dimensions.

Exercise 4
Solve the problem of the motion of a point projectile in a vertical plane, using
the Hamiltonian-Jacobi method. Find both the equation of the trajectory and the
dependence of the coordinates on time, assuming the projectile is fired off at time
t = 0 from the origin with the velocity v0 , making an angle α with the horizontal.

51