Unit-4

Hamilton’s Equations
Of
Motion
Hamiltonian Formulation

Newtonian → Lagrangian → Hamiltonian

 Describe same physics and produce same results

 Difference is in the viewpoints
 Flexibility of coordinate transformation

Hamiltonian formalism linked to the development of

 Hamilton-Jacobi theory
 Classical perturbation theory
 Quantum mechanics
 Statistical mechanics
 Lagrangian → Hamiltonian
Hamiltonian formulation  closer to Lagrangian formulation

For most systems of interest:

L = T −V (difference of kinetic and potential energies)

Lagrange’s equations for n coordinates

2nd order differential
d  ∂L  ∂L
  − =0 i = 1, …..n equations, n variables
dt  ∂q&i  ∂qi
 n equations

Lagrangian, L  n generalised coordinates (q1, q2 ,...., qn )

n generalised velocities (q&1, q&2 ,...., q&n )
and time t

L = L(qi , q&i , t ) Function of

2n+1 variables

Can we do with 1st-order differential equations?
 Yes, but you’ll need 2n equations
 We keep qi and replace q&i with something similar
 We take the conjugate momenta

Consider conservative system → F = −∇V
 V  independent of velocities
 Constraints  independent of time
⇒ L not an explicit function of t

Consider system of mass points under influence of F
L = T −V
∂L ∂T ∂V
= − =
∂
∑
∂x&i ∂x&i ∂x&i ∂x&i 2
( )
1 m x& 2 = m x& = p
i i i i ix

∂L
Using generalised velocity q&i ⇒
= pi
∂q&i
Configuration Space

We considered (q1 , q2 ,...., qn ) as a point in an n-dim. space

 Called configuration space

 Motion of the system →

A curve in the config space

When we take variations,

we consider qi and q&i as
independent variables

 i.e., we have 2n independent variables in n-dim. space

 Isn’t it more natural to consider the motion in 2n-dim space?
Phase Space

Consider coordinates and momenta as independent

 State of the system is given by (q1 ,...., qn , p1 ,...., pn )

 Consider it a point in the 2n-dimensional phase space

 We are switching the

independent variables

(qi , q&i , t ) → (qi , pi , t )

Hamiltonian
Lagrangian describing a system where angular momentum is
conserved, does not depend on time explicitly, i.e.

dL
=0
dt
we can express the dynamics in terms of the 2n + 1
variables qi, pi, and t.

The Lagrangian: L = L(qi , q&i , t )

Therefore, differentiating w.r.t. time:
dL ∂L ∂L
= ∑ q&i + ∑ q&&i (1)
dt i ∂qi i ∂q&i
 ∂L d  ∂L 
Lagrange’s equation: =  
∂qi dt  ∂q&i 

dL d  ∂L  ∂L
So (1) → = ∑   q&i + ∑ q&&i (2)
dt i dt  ∂q&i  i ∂q&i
Equation (2) is simply the result of differentiation of a product of two
functions,
dL d  ∂L 
therefore = ∑  q&i 
dt i dt  ∂q&i 

d  ∂L  dL
∑ dt  q&i ∂q&  − dt = 0
i  i

d   ∂L  
or  ∑ q&i  − L  = 0 (3)
dt  i  ∂q&i  
Integrating equation (3):

 ∂L 
∑ q&i − L = H (constant) (4)
i  ∂q &i 
∂L
The generalised momenta pi are defined as: pi =
∂q&i
So (4) → ∑ {q&i pi } − L = H
i
Therefore, H (qi , pi , t ) = ∑ q&i pi − L(qi , q&i , t ) (5)
i

Transformation of formulation of mechanics (Independent

coordinates are generalised coordinates and generalised velocities )
→ a formulation in which the independent coordinates are generalised
coordinates and generalised momenta.
Hamilton’s Equations

So the Hamiltonian is: H = H (qi , pi , t ) (6)

The total derivative:

∂H ∂H ∂H
dH = ∑ dqi + ∑ dpi + dt (7)
i ∂qi i ∂pi ∂t

Also from equation (5):

∂L ∂L ∂L
dH = ∑ pi dq&i + ∑ q&i dpi − ∑ dqi − ∑ dq&i − dt (8)
i i i ∂qi i ∂q&i ∂t
∂L d  ∂L  ∂L

because: pi = ⇒   = p& i =
∂q&i dt  ∂q&i  ∂qi
 ∂L
dH = ∑ q&i dpi − ∑ p& i dqi − dt (9)
i i ∂t

Equating (7) and (9):
∂L ∂H ∂H ∂H
∑ q&i dpi − ∑ p& i dqi − ∂t dt = ∑ ∂q dqi + ∑ ∂p dpi + ∂t dt
i i i i i i

Comparing coefficients, we get 2n + 1 relations:

∂H
q&i =
∂pi Canonical equations (10)
∂H of Hamilton
− p& i = (conforming to well established
∂qi rule or procedure)

∂L ∂H
and − = (11)
∂t ∂t
 ∂H ∂H

We find = q&i = − p& i and ∂H = − ∂L
∂pi ∂qi ∂t ∂t

2n equations replacing the n Lagrange’s equations

1st order differential instead of 2nd order

“Symmetry” between q and p is apparent

There is nothing new – We just rearranged equations

First equation links momentum to velocity

This relation is “given” in Newtonian formulation

Second equation is equivalent to Newton’s/Lagrange’s

equations of motion
How to apply Hamiltonian formulation?

1. Choose the generalized coordinates qi and construct the

Lagrangian L ( qi , q&i , t )
∂L
2. Obtain canonical momenta using the relations pi = (qi , q&i , t )
∂q&i
to express q& as a function of ( qi , pi , t ) .

3. Construct the Hamiltonian H as a function of ( qi , pi , t ) .

4. Derive Hamilton’s equations of motion.
Examples

Free particle in one-dimension
Lagrangian: L(q, q& , t ) = 12 mq& 2
Momentum conjugate to q is:
∂L
p= = mq&
∂q&
p
or q& =
m
Transformation equation: H ( qi , pi , t ) = ∑ q&i pi − L( qi , q&i , t )
i
the corresponding Hamiltonian:

H (q, p ) = pq& − L
mq& 2 p 2 mp 2 p 2
= pq& − = − 2
=
2 m 2m 2m
 p2
So we can write: H ( q, p ) =
2m
L and H assume same values  both equal to the kinetic energy
But they are different functions  depend on different variables

Carrying out partial differentiation → the equations of motion are:

∂H p
q& = =
∂p m

∂H
p& = − =0
∂q
So p is a constant of motion.

Particle under Hooke’s law force F = –kx

L = T −V
∂L
L= 1 mx& 2 − 12 kx 2 p= = mx&
2 ∂x&
p 2 p 2 kx 2
Hamiltonian H = px& − L = − +
m 2m 2
p 2 kx 2
H= +
2m 2
Hamilton’s equations:

∂H p ∂H
x& = = p& = − = − kx
∂p m ∂x
Conservation Theorems
and Physical Significance of Hamiltonian

 Some problems involve cyclic coordinates  Hamiltonian procedure

is adapted to treatment of such problems

Cyclic coordinates
A coordinate q j which does not appear in the Lagrangian.
Then Lagrange’s equations → its conjugate momentum p j is
constant.

∂L ∂H
Then p& j = =− =0
∂q j ∂q j
⇒ A cyclic coordinate will also be absent from H
 Conversely  if a generalised coordinate does not occur in H,
then its conjugate momentum is conserved.

Let cyclic coordinate  qn

then L = L(qi .....qn−1 , q&i .....q& n , t )
still, the problem to be solved is of n degrees of freedom.

Hamiltonian formulation  a cyclic coordinate can also be labelled as

ignorable coordinate
Let the conjugate momentum pn is some constant α
H = H ( qi .....qn−1 , pi ..... pn−1 , α , t )
Now problem involves n – 1 coordinates  can be solved ignoring the
cyclic coordinate
 As a cyclic coordinate in L is also cyclic in H ⇒ conservation
theorems also apply in Hamiltonian formulation

 Just like in case of Lagrangian  Physical significance of

Hamiltonian also lies in the fact  if H is not an explicit function
of t then H is a constant of motion

If Hamiltonian is: H = H (qi , pi , t )

dH  ∂H ∂H  ∂H
Then time derivative: = ∑  q&i + p& i  +
dt  ∂qi ∂pi  ∂t
∂H ∂H
as q&i = and p& i = −
∂pi ∂qi

dH  ∂H ∂H ∂H ∂H  ∂H
then = ∑  −  +
dt i  ∂qi ∂pi ∂pi ∂qi  ∂t
 ⇒ dH ∂H ∂L
= =− ⇒ H is a constant of motion
dt ∂t ∂t
⇒ Hamiltonian is a conserved quantity if it does not
explicitly contain time

 In many problems, canonical momenta are obvious from physical

considerations and Hamiltonian is also equal to total energy.

The Hamiltonian: H = q&p − L = mv 2 − L

In a conservative system: L = T −V
so H = 2T − T + V
or H = T +V
Example

Motion of a particle in 2-dim. in a central force field.

The total energy is then: H = T + V (r )

where, V(r) is central potential, and K.E. is:
T = 12 mv 2 = 12 m( r& 2 + r 2θ& 2 )
Hamiltonian formulation → r,θ converted to conjugate momenta pr,pθ

So we have:
pr = mvr = mr& (linear momentum)
pθ = mrvθ = mr 2θ& (angular momentum)
In terms of momenta and generalised velocities:

pr pθ
r& = and θ= 2
&
m mr
  p 2
r 2 2
pθ 
therefore: 1 
T = 2 m 2 + 2 4 
r
m m r 
pr2 pθ2
= +
2m 2mr 2
2 2
p
Hamiltonian: H = r +
pθ θ is cyclic
2
+ V (r )
2m 2mr
pθ = constant=l
Hamilton’s equations:

∂H pr ∂H pθ2 ∂V
r& = = p& r = − = 3−
∂pr m ∂r mr ∂r

Cyclic variable drops off by itself

Derivation of Hamilton’s Equations from
Variational Principle

Variational principle  Hamilton’s principle →

useful in derivation of Lagrange’s equations

t2
Hamilton’s principle δI = δ ∫t Ldt = 0 (12)
1

Lagrangian formulation  qi and q&i independent variables in

configuration apace

Hamiltonian formulation  q and p independent coordinates in
phase space

express Hamilton’s principle in terms of q and p

Hamiltonian H (qi , pi , t ) = ∑ q&i pi − L(qi , q&i , t )

i

L(qi , q&i , t ) = ∑ q&i pi − H (qi , pi , t ) (13)

i

Substituting in Hamilton’s Principle:

δI = δ ∫tt12  ∑ q&i pi − H (qi , pi , t ) dt

Modified Hamilton’s
(14)
principle  i 
Review  δ - variation process:

Configuration space formed by n generalised coordinates

The initial and final configurations of the system, at times t1 and t2

 each represented by a point

“variation of the integral” ⇒ variation in the value of integral
as we change the path traversed by the system between two
end points.

Variation occurring in constant time ⇒ end point times are fixed

⇒ “Travel time” between two configurations is same for all paths

The variation of integral about actual path is zero ≡ integral has an
extremum for this path

Integral I  function of parameter α → δ is identified with
change dα ∂
δ → dα (a)
∂α
Hamilton’s Equations

t2  
Modified Hamilton’s
δI = δ ∫t  ∑ q&i pi − H (qi , pi , t ) dt (14)
Principle: 1
 i 
In terms of α:
∂I ∂ t2  
δI = dα = dα ∫ 
 ∑ q&i pi − H (qi , pi , t ) dt = 0
∂α ∂α t1  i 
Since, the end-point times are not varied ⇒ they are not functions of α

∂  
 ∑ q&i pi − H (qi , pi , t ) dt = 0
t2
dα ∫
t1 ∂α  i 

t2  ∂q&i ∂pi ∂H ∂qi ∂H ∂pi 

dα ∫ ∑  ∂α pi + q&i ∂α − ∂q ∂α − ∂p ∂α  dt = 0 (15)
t1
i  i i 
First term can be written as:

t2 ∂q&i t2 d ∂qi
∫t pi dt = ∫ pi dt
1 ∂α t1 dt ∂α
t (16)
∂qi 2 t2 ∂qi
= pi − ∫ p& i dt
∂α t1 t1 ∂α

= 0 Q ∂qi ∂α vanishes at the end-points

Equation (15) takes the form:

t2  ∂pi ∂qi ∂H ∂qi ∂H ∂pi 

dα ∫ ∑  q&i ∂α − p& i ∂α − ∂q ∂α − ∂p ∂α  dt = 0 (17)
t1
i  i i 
 t2  ∂pi ∂qi ∂H ∂qi ∂H ∂pi 
dα ∫ ∑  q&i ∂α − p& i ∂α − ∂q ∂α − ∂p ∂α  dt = 0 (17)
t1
i  i i 
Using correspondence in equation (a):
∂qi ∂pi
dα = δqi and dα = δpi
∂α ∂α
t2  ∂H ∂H 
∫t ∑  q&iδpi − p& iδqi − ∂q δqi − ∂p δpi  dt = 0
1
i  i i 
t2   ∂H   ∂H  
or ∫t ∑  δpi q&i − ∂p  − δqi  p& i + ∂q   dt = 0 (18)
1
i   i  i 
As variations δpi and δqi are independent, integral can only vanish if
coefficients vanish separately.
∂H ∂H Required Hamilton’s
therefore: q&i = and p& i = − (20)
∂pi ∂qi Equations
Applications of Hamilton’s Equations
Example-1:
Use Hamiltonian method to find equations of motion of a particle
constrained to move on the surface of a cylinder defined by x2 + y2 =
R2. The particle is subjected to a force directed toward the origin and
proportional to the distance of the particle from the origin: F = -kr

Solution: z

The potential corresponding to force is:

V = − ∫ F(r )dr R

= − ∫ − krdr r
y

V = 12 kr 2 θ
So
x

as r 2 = x2 + y 2 + z 2
 V = 12 k ( x 2 + y 2 + z 2 )
= 12 k ( R 2 + z 2 )
If the particle traces a path s in x,y plane, then:
ds 2 = dx 2 + dy 2 + dz 2 (i)
In cylindrical coordinates: x = R cosθ
y = R sin θ
Find dx, dy, dz, substitute in (i):

therefore ( ds ) 2 = (dR) 2 + R 2 (dθ ) 2 + (dz ) 2

Dividing by ( dt ) 2 :

2  dθ 
2 2 2 2
 ds   dR   dz 
We get: =
    + R    +
 dt   dt   dt   dt 
 or v 2 = R& 2 + R 2θ& 2 + z& 2
As R is constant: v 2 = R 2θ& 2 + z& 2
Therefore the K.E. is: T = 12 m( R 2θ& 2 + z& 2 )

Now we can write the Lagrangian as:

L = T − V = 12 m( R 2θ& 2 + z& 2 ) − 12 k ( R 2 + z 2 )

Generalised coordinates are θ and z → generalised momenta then:

∂L
pθ = = mR 2θ&
∂θ& (ii)

∂L
and pz = = mz& (iii)
∂z&
conservative system  equations of transformation between rectangular
and cylindrical coordinates do not involve time explicitly,
⇒ Hamiltonian is just total energy, expressed in terms of θ, pθ , pz, z
(θ is cyclic)

so H ( z , pθ , p z ) = T + V = 12 mR 2θ& 2 + 12 mz& 2 + 12 kR 2 + 12 kz 2
pθ2 p z2 1 2 1 2
Using (ii) and (iii): H= 2
+ + kR + kz
2mR 2m 2 2
∂H
now p& θ = − =0 (iv)
∂θ
∂H
p& z = − = − kz (v)
∂z
 ∂H p
θ& = = θ2
∂pθ mR (vi)

∂H p z
z& = = (vii)
∂p z m
Equations (iv) and (vi) give:
pθ = mR 2θ& = Constant

⇒ angular momentum about the z-axis is constant of motion

From equations (v) and (vi): m&z& = − kz

or &z& + ω02 z = 0
k
where ω02 =
m
Therefore, motion in z-direction is simple harmonic motion.
Example-2:
Use the Hamiltonian method to find the equations of motion for a
spherical pendulum of mass m and length b .
Solution: z

The generalized coordinates are θ and φ . x

Using spherical polar coordinates:

x = b sin θ cos φ θ
b
y
y = b sin θ sin φ
φ
z = b cosθ
K.E. is given by: T = 12 m( x& 2 + y& 2 + z& 2 ) mg

and P.E. : V = − mgz

Perform simple calculations to transform the K.E. and P.E. equations

using spherical coordinates.
The results are: T = 12 m(b 2θ& 2 + b 2φ& 2 sin 2 θ )
V = − mgb cosθ
The Lagrangian is: L = T −V
therefore L = 1 m(b 2θ& 2 + b 2φ& 2 sin 2 θ ) + mgb cosθ
2
The generalised momenta are:
∂L
pθ = = mb 2θ& (i)
∂θ&
∂L
pφ = = mb 2φ& sin 2 θ (ii)
∂φ&
Since the system is conservative and that the transformation from
Cartesian to spherical coordinates does not involve time, we have:
H = T +V
H = 12 mb 2θ& 2 + 12 mb 2φ& 2 sin 2 θ − mgb cosθ (iii)
Using (i) and (ii) in (iii):
2
pθ2 pφ
H= + − mgb cosθ
2mb 2mb sin θ
2 2 2

∂H p
Finally, the equations of motion are: θ& = = θ2
∂pθ mb
∂H pφ
φ=
& =
∂pφ mb 2 sin 2 θ
∂H pφ2 cosθ
p& θ = − = − mgb sin θ
∂θ mb sin θ
2 3

∂H
p& φ = − =0
∂φ
Because φ is cyclic (or ignorable), the generalized momentum pφ about
the symmetry axis is a constant of motion. pφ is actually the component
of the angular momentum along the z-axis .
Example-3:

Consider a particle of mass m moving freely in a conservative force field,

whose potential function is V. Find the Hamiltonian function and show
that the canonical equations of motion reduce to Newton’s equations. Use
rectangular coordinates.
Solution:
For a particle moving freely in a conservative field:
K.E.: T = 12 m( x& 2 + y& 2 + z& 2 )
P.E.: V = V ( x, y , z )
The Lagrangian is: L = T −V
= 12 m( x& 2 + y& 2 + z& 2 ) − V ( x, y, z )
Generalised momenta are then:

∂L ∂L ∂L
px = = mx& , py = = my& , pz = = mz&
∂x& ∂y& ∂z&
The Hamiltonian can be written as: H = T +V
H = 12 m( x& 2 + y& 2 + z& 2 ) + V ( x, y, z )
p x2 p 2y
p z2
= + + + V ( x, y , z )
2m 2m 2m
Canonical equations of motion are:
∂H ∂V ∂H ∂V ∂H ∂V
p& x = − =− , p& y = − =− , p& z = − =−
∂x ∂x ∂y ∂y ∂z ∂z
however, F = −∇V
∂V ∂V ∂V
therefore: Fx = − , Fy = − , Fz = −
∂x ∂y ∂z
so p& x = Fx , p& y = Fy , p& z = Fz
which are simply Newton’s equations of motion.
Example-4:
Determine the Hamiltonian and Hamilton’s equations of motion for a
simple pendulum of mass m and length l.
Solution: x

The K.E. of the system is: T = 12 mv 2

θ
and P.E.: V = − mgy l

Here the velocity v is related to angular velocity ω

v = lω = lθ&
and y = l cosθ y

Therefore, we write the lagrangian:

L = T − V = 12 mv 2 + mgy
= 12 ml 2θ& 2 + mgl cosθ
∂L pθ
now pθ = = ml 2θ& → θ= 2
& (i)
∂θ& ml
Now the Hamiltonian can be written as: H = pθ θ& − L
H = ml 2θ& 2 − 12 ml 2θ& 2 − mgl cosθ
= 12 ml 2θ& 2 − mgl cosθ
Substituting from (i): pθ2
H= 2
− mgl cosθ
2ml
So the equations of motion are:
∂H pθ
θ=
& = 2
∂pθ ml
∂H
p& θ = − = − mgl sin θ ⇒ ml 2θ&& = − mgl sin θ
∂θ
g
or θ = − sin θ
&&
l
For small θ : sin θ ≈ θ g
so θ =− θ
&&
l