You are on page 1of 22

Analytical Dynamics - Graduate Center CUNY - Fall 2008

Professor Dmitry Garanin

Hamiltonian mechanis
December 10, 2008

Preface

Both Newtonian and Lagrangian formalisms operate with systems of second-order differential equations for
time-dependent coordinates or generalized coordinates, qi = . . .. For a system with N degrees of freedom,
N such equations can be reformulated as systems of 2N first-order differential equations if one considers
velocities or generalized velocities vi = qi as additional dynamical variables. This system of equations has
the form qi = vi , v i = . . . that is non-symmetric with respect to qi and vi .
Hamiltonian formalism uses qi and pi as dynamical variables, where pi are generalized momenta defined
by
L
pi =
.
(1)
qi
The resulting 2N Hamiltonian equations of motion for qi and pi have an elegant symmetric form that is
the reason for calling them canonical equations. Although for most of mechanical problems Hamiltonian
formalism is of no practical advantage, it is worth studying because of the similarity between its mathematical
structure and that of the quantum mechanics. In fact, a significant part of quantum mechanics using the
matrix and operator algebra grew out of the Hamiltonian mechanics. The latter is invoked in constructing
new field theories. Hamiltonian formalism finds application in statistical physics, too.

Hamiltonian function and equations

Hamiltonian equations can be obtained from the Lagrange equations that can be written in the form
pi =

L
qi

(2)

using a Legendre transformation. The differential of the Lagrange function has the form

X  L
L
dL =
dqi +
dqi
qi
qi

(3)

that with the help of Eqs. (1) and (2) can be rewritten as
X
dL =
(pi dqi + pi dqi ) .

(4)

Let us intoduce the Hamiltonian function, or the Hamiltonian, defined by


X
H=
pi qi L.

(5)

With the help of the above its differential becomes


X
dH =
(qi dpi pi dqi ) .

(6)

One can see that the natural variables of H are q and p and
qi =

H
,
pi

pi =
1

H
qi

(7)

that are the Hamiltonian equations of motion for q and p.


As the Lagrange function is bilinear in qi ,
L = T (q, q)
U (q),

T =

1X
aij (q)qi qj ,
2

(8)

ij

the first term in Eq. (5) is the double kinetic energy,


X

pi qi =

X L
qi = 2T.
qi

(9)

Now one can see that H is the full energy,

H = T + U.

(10)

In most cases the kinetic energy is written in a natural way in terms of qi , so that expressing it in terms
of pi may require work. For unconstrained particles there is the linear relation pi = mvi and reexpressing
H in terms of pi is not a problem. In particular, for a system of particles one obtains
H=

X p2
i
+ U.
2mi

(11)

On the other hands, for systems with constraints, such as the double pendulum, different generalized velocities qi couple since aij (a)ij in Eq. (8) is a nondiagonal matrix. Eq. (8) can be written in the vectorized
form
1
T = q T a q,

(12)
2
where q = (q1 , q2 , . . .) is a column and the transposed q T is a line. Further, from Eqs. (1) and (8) one
obtains
p = a q
(13)
and thus
q = a1 p,

q T = a1 p

T

= pT a1

T

= pT a1 ,

(14)

where we have taken into account that a1 , as well as a, is a symmetric matrix. Plugging this result into
Eq. (12) one obtains
1
H = pT a1 p + U.
(15)
2
Although inverting the kinetic-energy matrix a is possible, at least numerically, it makes the Hamiltonian formalism less appropriate. For this reason it is never used for systems with constraints relevant in engineering,
exactly where the Lagrangian approach shows its strength.
Let us, as an illustration, find the Hamiltonian of a particle in the spherical coordinates. The kinetic
energy has the form

2 
 m 2  2 
mv 2
m 2
2
2

T =
=
v + v + v =
r + r + r sin
.
(16)
2
2 r
2
Thus the generalized momenta (1) are given by
pr =
p =
p =

T
= mr
r
T
= mr 2

T

= mr 2 sin2 .

2

(17)

and ,
yields
Solving these equations for r,
,
r =

p
=
,
mr 2

pr
,
m

p
.
mr 2 sin2

(18)

Inserting the results into Eq. (16) one obtains


1
H=T +U =
2m

p2
p2
p2r + 2 + 2 2
r
r sin

+ U (r, , ).

(19)

One can see that Eqs. (18) are Hamilton equations


r =

H
,
pr

H
=
,
p

H
=
.
p

(20)

Hamiltonian equations from the least-action principle

As the Lagrange equations follow from the least-action principle and the Hamiltonian equations can be derived from the Lagrange equations, one can obtain the Hamiltonian equations from the least-action principle
directly. The calculation starts from the action written with the help of Eq. (5) in the form
! Z
!
Z t2
Z t2
t2 X
X
S=
dtL(q, q,
t) =
dt
pi qi H =
pi dqi Hdt .
(21)
t1

t1

t1

Here, in contrast to the Lagrange formalism, one considers both pi and qi as independent dynamical variables
and makes a variation
qi qi + qi ,
pi pi + pi .
(22)
The corresponding variation of the action has the form

X Z t2 
H
H
S =
pi dqi + pi dqi
qi dt
pi dt .
qi
pi
t1

(23)

Integrating the second term by parts and using that qi = 0 at the begining and the end, one can rewrite
this as



X Z t2  
H
H
S =
pi dqi
dt qi dpi
dt .
(24)
pi
qi
t1
i

Since for the actual trajectories S = 0 for arbitrary and independent qi and pi , one concludes that both
expressions in round brackets are zero that leads to the Hamilton equations (7).

Poisson brackets and conservation

Time derivative of any quantity f = f (q, p, t) has the form




df
f
f X f
=
+
qi +
pi .
dt
t
qi
pi

(25)

With the help of Eqs. (7) it can be rewritten as


df
f
=
+ {f, H} ,
dt
t

(26)

where {f, H} is a Poisson bracket defined by



X  f g
f g
{f, g}

.
qi pi pi qi

(27)

One can see that if f does not depend on time explicitly and its Poisson bracket with the Hamiltonian is
zero, f is an integral of motion. In particular, obviously {H, H} = 0, so that in the absence of explicit time
dependence the energy is an integral of motion, H = E = const. Note the similarity between Eq. (26)
and the quantum
mechanical
equation of motion for an operator in the Heisenberg representation where the
h
i

commutator f, H replaces the Poisson bracket.


Poisson brackets satisfy obvious relations such as
{f, g1 + g2 } = {f, g1 } + {f, g2 } ,

{g, f } = {f, g} ,

(28)

etc. There is also the nontrivial Jacoby identity


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

(29)

that we quote without proof here. (To prove this you need a large piece of paper and a hot cup of coffee.
Expand out all 24 terms and watch them cancel one by one David Tong.) It follows then that if both f
and g are integrals of motion, then {f, g} is also an integral of motion. For the proof we take h = H, then
the Jacoby identity takes the form

f, {g, H}
+ {H, {f, g}} + g, {H, f }
= {H, {f, g}} = 0,
(30)

| {z }
| {z }

0
0

thus {f, g} = const. In principle, this could be used to obtain new integrals of motion from already known
old integrals of motion. However, in most cases the new integrals of motion {f, g} are trivial, such as
functions of the old integrals of motion.
The straightforwardly obtainable Poisson brackets
{qi , qj } = 0,

{pi , qj } = ij ,

{qi , pj } = ij

{pi , pj } = 0

(31)

are called fundamental Poisson brackets.

4.1

Example: Particles motion in spherical coordinates

As an example of nontrivial integrals of motion detectable with the help of Poisson brackets consider a
particle moving in the potential
b()
U (r, , ) = a(r) + 2
(32)
r
in the spherical coordinates. In this case with the help of Eq. (19) one obtains the Hamiltonian function in
the form
"
!#
p2
1
1
2
2
H=
p + 2ma(r) + 2 p +
+ 2mb()
.
(33)
2m r
r
sin2
One can show that the combination in the round brackets is an integral of motion,
p2 +

p2
sin2

+ 2mb() = = const.
4

(34)

It is sufficient to calculate the Poissson bracket


H
H H H
H
H
= { , H} =

r
p
p
r

p
p

p
p
r
r

|{z}
|{z}
|{z}
|{z}
0
0
0
0
1
=
{ , } = 0.
(35)
2mr 2
Also p is an integral of motion since is a cyclic variable, H being independent of . The Hamiltonian
equation for has the form
p
H
=
=
(36)
p
mr 2 sin2
and it cannot be solved without knowing r(t) and (t). Finding the motion of r is simplified by the energy
conservation, H = E = const. From Eq. (33) one obtains
p

pr = mr = 2m [E Ueff (r)],
Ueff (r) a(r) +
(37)
2mr 2

that can be integrated to implicitly find r(t). The motion of can be most conveniently obtained from the
integral of motion (34)
s
p2
p = mr 2 = 2mb()
.
(38)
sin2
This equation can be integrated as follows
Z
Z t
d 0
dt0
q
=
.
(39)
mr 2 (t0 )
2mb( 0 ) p2 / sin2 0

Note that to work out the integral in the rhs, one at first has to find r(t) from Eq. (37). After r(t) and (t)
have been found, one can integrate Eq. (36). The situation allowing to obtain the solution of this problem
in quadratures is called separation of variables.

4.2

Poisson brackets and commutators

Let us finally work out the correspondence between the Poisson brackets and commutators. In quantum
mechanics
d
qi = qi ,
pi = i~ ,
[
qi , pj ] = i~ ij
(40)
dqi
and operator functions like f (
q, p) that correspond to classical functions f (q, p) are interpreted as Taylor
series with full symmetrization over permutations of terms. The commutator [f, g] can be calculated, for
instance, by expanding both f and g in Taylor series in q and p, commuting term by term, and then summing
the series back. The final result of this procedure can be obtained much easier by formally substituting

 X

X  f
f
g
g
[f, g]
qi +
pi ,
qj +
pj
(41)
qi
pi
qj
pj
i

and then commuting the momentum and coordinate operators considering partial derivatives as numbers.
With the help of Eq. (40) this yields the relation
[f, g] = i~ {f, g} .

(42)

In this formula the Poisson bracket should be symmetrized over permutations, qf pg (qf pg + pg qf ) /2
etc., since q and p are operators. If ~ is formally considered as small, as is sometimes done in the analysis
of the semiclassical case, the symmetrization is irrelevant. This is because each commutation introduces a
factor ~, so that changing the order of terms changes the result by terms starting with ~2 .
5

Canonical transformations

As any system of differential equations, Hamiltonian equations (7) allow change of variables
Qi = Qi (q, p, t) ,

Pi = Pi (q, p, t) ,

(43)

where q, p are old variables and Q, P are new variables. The transformation of variables above is called
canonical if the transformed Hamiltonian equations also have a Hamiltonian (canonical) form
H0
Q i =
,
Pi

H
Pi =
,
Qi

(44)

where, in principle, H0 can differ from H. Sometimes one can find a canonical transformation that results in
a simple Hamiltonian function allowing for an easy solution. In particular, if H0 does not depend on Qi , the
variable Qi is called cyclic. The second of the above equations then yields Pi = 0 and Pi = const. Now, in
the case of one degree of freedom, the rhs of the first equation above is an integral of motion, Q = = const.
This equation can be easily integrated, Q = t + const.

5.1

Least-action principle and generating functions

Possible relations between old and new variables can be obtained in the elegant and powerful form from
the least-action principle. Validity of both Eqs. (7) and (44) implies that the action S in terms of the old
variables, Eq. (21), is equivalent to the action in terms of the new variables. It is sufficient to require that
the integrands of the two actions differ by a full differential of some function F. Then variations of both
actions coincide, so that S = 0 for the old variables entails S = 0 for the new variables. Then validity of
Eq. (7) entails validity of Eq. (44). The difference between the two action integrands having the form
X
X

pi dqi
Pi dQi + H0 H dt = dF
(45)
i

suggests that F = F (q, Q, t) and

pi =

F
,
qi

Pi =

F
,
Qi

H0 H =

F
.
t

(46)

These formulas establish relations between the new and old variables. The function F (q, Q, t) is the generating function of the canonical transformation. In the absence of explicit time dependence of the canonical
transformation, that is the main case, one has H0 = H.
Canonical transformation specified by Eq. (46) can be put in other equvalent forms with the help of
different Legendre transformations. For instance, the differential of the function
X
=F+
Qi Pi
(47)
i

with the help of Eq. (45) takes the form


X
X

d =
pi dqi +
Qi dPi + H0 H dt.
i

(48)

thus = (q, P, t) and

,
Qi =
,
H0 H =
.
(49)
qi
Pi
t
P
P
P
One can introduce other types of generating functions such as F i qi pi , F i qi pi + i Qi Pi , and
many functions in which the Legendre transformation is done with respect to some select canonical pairs
qi pi or Qi Pi with particular values of i. In all cases generating functions contain old and new variables for
each degree of freedom i. It should be stressed that Legendre transformations lead to other expressions
for the same canonical transformation specified by the generating function F rather than to new canonical
transformations.
pi =

5.2

Trivial examples of canonical transformations

The transformation defined by the generating function


(q, P ) =

qi Pi

(50)

has the form


pi =

= Pi ,
qi

Qi =

= qi
Pi

(51)

or
Qi = q i ,

Pi = pi .

This is the identity transformation.


The transformation with
F (q, Q) =

q i Qi

(52)

(53)

has the form


pi =

F
= Qi ,
qi

Pi =

F
= qi
Qi

(54)

or
Qi = p i ,

Pi = qi .

(55)

This transformation interchanges generalized coordinates and momenta. The above example shows that
there is no essential difference between generalized coordinates and momenta in the Hamiltonian formalism.
One cannot say that generalized momenta are related to velocities while generalized coordinates not.
It should be noted that for the above two transformations that are obviously canonical an attempt to make
the Legendre transformation (47) does not work. For instance, one cannot find the primary form F (q, Q)
of the transformation, the secondary form (q, P ) of which is given by Eq. (50). Since only the nonexistent
F (q, Q) follows from the least-action principle, (q, P ) loses its relation to the general formalism. It looks
like there are more canonical transformations than those following from the least-action principle.

5.3

Harmonic oscillator via canonical transformation

Another example of canonical transformations is the harmonic oscillator having the Hamiltonian
H=


kq 2
1
p2
+
=
p 2 + m2 2 q 2 ,
2m
2
2m

(56)

p
where = k/m is the oscillators eigenfrequency. Quadratic dependence on q and p suggests to use a
transformation of the type
f (P )
q=
sin Q,
p = f (P ) cos Q
(57)
m
that leads to the transformed Hamiltonian
H=

 f 2 (P )
f 2 (P )
cos2 Q + sin2 Q =
2m
2m

(58)

that is independent of Q. If the transformation is canonical, one obtains a simple solution with P = const.
The form of the required canonical transformation can be guessed to be
F (q, Q) =
From this one obtains
p=

F
= mq cot Q,
q

mq 2
cot Q.
2
P =
7

F
mq 2 1
=
.
Q
2 sin2 Q

(59)

(60)

Resolving the second of these equations for q and then substituting the result into the first equation one
obtains
r

2P
(61)
q=
sin Q,
p = 2P m cos Q
m
that indeed has a form of Eq. (57) with

f (P ) = 2P m.
(62)
Now the new Hamiltonian has the form
H = P.

(63)

The generalized momentum

E
H
=
(64)

is conserved. The interpretation of P is action over the period of motion divided by 2, see Eq. (120). The
equation of motion for the cyclic variable Q is
P =

H
Q =
=
P

(65)

Q = t + 0 ,

(66)

the solution of this equation reads


where 0 = const. One can see that Q is the oscillators phase angle. Inserting the above results into Eq.
(61) one finally obtains the solution
r

2E
q=
sin
(t
+

)
,
p
=
2mE cos (t + 0 ) .
(67)
0
m2

5.4

Symplectic formalism for canonical transformations

Some of canonical transformations do not depend on time. For this group of transformation the symplectic or
matrix-vector formalism is of a great help since it allows to formulate criteria for a particular transformation
to be canonical in a compact and elegant form. Let us introduce the dynamical vector
x = {x1 , x2 , . . . , x2N } = {q1 , q2 , . . . , qN , p1 , p2 , . . . , pN }

(68)

that allows to formulate the Hamiltonian equations (7) in the vector or component forms
x = J

H
,
x

x i = Jij

H
,
xj

(69)

where summation over repeated indices is assumed and the matrix J is given by
Jij = i+N,j iN,j .

(70)

Let us now perform a canonical transformation (43) without time dependence and introduce the dynamical
vector
y = {y1 , y2 , . . . , y2N } = {Q1 , Q2 , . . . , QN , P1 , P2 , . . . , PN } .
(71)
The equation of motion for y follows from Eq. (69)
yi =

yi
yi
H
yi
yl H
x j =
Jjk
=
Jjk
.
xj
xj
xk
xj
xk yl

(72)

The condition that the resulting equation is Hamiltonian and thus the transformation is canonical in the
vector and component forms reads
yi
yl
Jjk
= Jil ,
xj
xk

M J MT = J,
8

(73)

where Mij yi /xj is the Jacobian matrix of the transformation.


With the help of Eq. (70) one can rewrite Eq. (73) as
yi yl
yi yl

= i+N,l iN,l
xj xj+N
xj xjN

(74)

This can be rewritten in the form



X  yi yl
yi yl

= i+N,l iN,l
qj pj
pj qj

(75)

that has particular cases

X  Qi Pl
j

Qi Pl

qj pj
pj qj

=1

(76)

etc. Collecting all these results, one obtains that the fundamental Poisson brackets of the new variables
with respect to the old ones have the form
{Qi , Qj } = 0,

{Pi , Qj } = ij ,

{Qi , Pj } = ij

{Pi , Pj } = 0

(77)

and zero in other cases. This is the final form of the criterion for checking whether a particular transfromation
is canonical. Its form is similar to the trivial Eq. (31).
One can see that Eq. (73) gives the expression for the fundamental Poisson brackets in the symplectic
formalism in its lhs. Eq. (73), as well as the more explicit Eq. (77) state that the criterion of canonicity is
the invariance of the fundamental Poisson brackets,
{A, B}x = {A, B}y ,

(78)

where A, B = Qi , Pi and subscripts x and y mean that the Poisson brackets are calculated with respect to
the old (x) and new (y) sets of dynamic variables. The lhs of this equation is Eq. (77) and the rhs is trivial
and similar to Eq. (31).
One can prove the inverse and more general statement: If the transformation is canonical, Poisson brackets
of any two variables A and B are invariant with respect to the transformation, Eq. (78). The proof uses
Eq. (73):
A
B
A yk
yl B
A
B
{A, B}x =
Jij
=
Jij
=
Jkl
= {A, B}y .
(79)
xi xj
yk xi
xj yl
yl
yl

5.5

Harmonic oscillator again

In Sec. 5.3 the problem of a harmonic oscillator was solved by guessing the generating function of the
canonical transformation, Eq. (59). To bypass the guesswork, one can find the function f (P ) in Eq. (57)
from the canonicity criterion given by Eq. (77). In particular, one should have
q p
q p

Q P
P Q

=
=

f (P )
df (P )
1 df (P )
cos Q
cos Q +
sin Q f (P ) sin Q
m
dP
m dP
f (P ) df (P )
1 df 2 (P )
=
= 1.
m dP
2m dP

(80)

One of the solutions of this differential equation is given by Eq. (62). Since one is looking for a one particular
canonical transformation of the simplest form, one does not have to care for integration constants. Other
Poisson brackets such as {Q, Q} are invariant trivially.

5.6

Relation between the generator and symplectic formalisms

It is of interest to perform the direct canonicity check provided by Eq. (77) for canonical transformations
via generating functions obtained from the least-action principle in Sec. 5.1. The resulting equivalence
between the two seemingly unrelated approaches is referred to as Caratheodory theorem. For simplicity, we
will produce the proof for one degree of freedom. Using the second of relations (46), one can write


Q P
Q P
Q 2 F (q, Q)
Q 2 F (q, Q)
{Q, P }x =

=
+
.
(81)
q p
p q
q
Qp q
p
Qq p

Here differentiation of F over q and p should be done taking into account F = F (q, Q(q, p)):


Q 2 F (q, Q) Q Q 2 F (q, Q) Q 2 F (q, Q) Q
Q p
{Q, P }x =
+
+
=
= 1.
q
2Q
p
p Qq
p
Q2
q
p Q

(82)

This proves canonicity. In the remaining term we used the first of relations (46) and simplified this term as
Q(q, p(q, Q)) p(q, Q)
Q
=
= 1.
p(q, Q)
Q
Q

(83)

For more than one degree of freedom the proof seems to be much more involved.

6
6.1

Action as function of coordinates and the Hamilton-Jacoby equation


General formulation with a simple example

The action S given by Eq. (21) was used to derive the Lagrangian and Hamiltonian equations of motion
from the least-action principle, S = 0. This condition singles out the real physical trajectory from all
other competing trajectories. Here we will consider the action for the real trajectory as the function of the
upper-limit variables t2 t and q2 q. The expression for S in the Hamiltonian form, the last term in Eq.
(21), shows that as qi and t change in the course of motion, the action acquires corresponding increments,
so that infinitesimally one has
X
dS =
pi dqi Hdt.
(84)
i

This implies that S = S(q, t) and

S
= pi ,
qi

S
= H.
t

(85)

Let us illustrate this on a free particle in one dimension. The Lagrangian has the form L = T = mv 2 /2,
and the trajectory is described by q = vt. Thus the action reads
Z t
mv 2
mq 2
S=
dt L =
t + const =
+ const.
(86)
2
2t
Now one can check that
S
q
S
t

mq
= mv = p
t
mq 2
mv 2
p2
= 2 =
=
= H,
2t
2
2m
=

(87)

as it should be.
If the problem does not explicitly depend on time, the energy is conserved, H = const = E. Then the
second of equations (85) can be integrated that yields
S (q, t) = S0 (q) Et.
10

(88)

Here S0 (q) is the so-called short or abbreviated Raction Rthat can be obtained by integrating the first of
t
t
equations (85). Alternatively, using Eq. (21) with t12 and H E, one obtains Eq. (88) with
X Z qi
S0 (q) =
pi dqi .
(89)
i

Equations (85) can be used to set up the famous Hamilton-Jacoby equation that together with canonical
transformations is an efficient tool for finding analytical solutions of mechanical problems. The HamiltonJacoby equation


S
S
+ H q,
,t = 0
(90)
t
q
is a nonlinear first-order partial differential equation (PDE) for the function S (q, t) . As usual, q and S/q
in the arguments stand for the whole sets of qi and S/qi = pi . For practical purposes it is sufficient to
find just some solution of the Hamilton-Jacoby equation rather than its most general solution.
In particular, for the free particle considered above, Eq. (90) becomes
 2
S
1
S
= 0.
(91)
+
t
2m q
The solution of this equation can be searched for in the form
S (q, t) = S0 (q) + S 0 (t) .

(92)

(93)

Then Eq. (91) becomes


S 0
1
+
t
2m

S0
q

2

= 0.

To satisfy this equation, both terms should be constants compensating each other, so that
S 0
= const = E,
t

S0
= const = p
q

(94)

and the constants satisfy

p2
,
p = 2mE.
2m
Integrating Eq. (94) one obtains S (q, t) in the two alternative forms
E=

S (q, t) = pq

p2
t = 2mEq Et,
2m

(95)

(96)

up to an irrelevant constant. One can check that this result coincides with Eq. (86):
S (q, t) = pq

 mq 2 t
p2
mq
mq 2
t=
q
=
.
2m
t
t
2m
2t

(97)

S
=H+
= H H = 0.
t
t

(98)

Whereas the solution of the Hamilton-Jacoby equation for one degree of freedom such as Eq. (96) depends
on one constant, the so-called complete integral of the Hamilton-Jacoby equation for N degrees of freedom
depends on N constants that we call Pi . On the top of it, one can always add an irrelevant constant to S (q, t)
that has been suppressed in Eq. (96). The complete integral yields the solution for the systems dynamics
if one uses it as the generating function of a canonical transformation in terms of the old coordinates and
new momenta, (q, P, t) of Eq. (47). The new momenta are the constants Pi in the complete integral. The
new Hamiltonian H0 given by Eq. (49) vanishes according to the Hamilton-Jacoby equation:
H0 = H +

11

Thus the Hamiltonian equations for the new dynamic variables Qi and Pi become trivial:
Q i = 0
Pi = 0

Qi = const

Pi = const.

(99)

Time dependences of the old variables qi and pi can be obtained from the first two equations (49). At first
qi are found from the equations
S
Qi =
.
(100)
Pi
Then pi can be found by the formulas
pi =

S
.
qi

(101)

Some literature uses i and i as new momenta and coordinates within the Hamilton-Jacoby formalism,
i Pi and i Qi .
Let us illustrate finding the dynamical solution on the free particle. Let us choose the constant p in the
first expression in Eq. (96) as the new momentum P,
S (q, P, t) = P q

P2
t.
2m

(102)

Equation (100) takes the form


S
P
=q t
P
m

(103)

P
P
t + Q = t + const.
m
m

(104)

S
= P = const.
q

(105)

Q=
that yields the solution
q=
The momentum is defined by Eq. (101):
p=

Thus Eq. (104) reproduces the well-known solution q = (p/m) t + const for a free particle.
Alternatively one can use the second expression in Eq. (96) and choose the energy E as the conserved
new momentum P . Now instead of Eq. (103) one obtains
r
S
m
Q=
=
qt
(106)
E
2E
that yields the familiar solution
q=

2E
t+
m

2E
Q=
m

2E
t + const.
m

(107)

The old momentum is given by the familiar formula


p=

S
= 2mE.
q

(108)

This alternative solution using the energy rather than the momentum as the conserved new momentum
is preferred because it survives for systems with nontrivial potential energy where the momentum is not
conserved. As an example we will consider the harmonic oscillator in the next section.

12

6.2

Harmonic oscillator by the Hamilton-Jacoby method

With the harmonic-oscillator Hamiltonian of Eq. (56) the Hamilton-Jacoby equation (90) becomes
 2
1
S
S
m2 q 2
+
.
+ U (q) = 0,
U (q) =
t
2m q
2
The action has the form of Eq. (88) with S0 satisfying the equation


1
S0 2
+ U (q) = E.
2m q

(109)

(110)

Resolving this equation and integrating, one obtains


Z q
p
S (q, E, t) =
dq 0 2m [E U (q 0 )] Et.

(111)

For the harmonic oscillator one can calculate the integral analytically as follows
Z q r
Z
m
1 q
d
q0
1
p
dq 0

t
=
t = arcsin q t,
Q=
2
02
2E m q

1 q02

(113)

Using this as the generating function (q, P, t) with P = E, one obtains the implicit formula for q(t)
Z q
r
S
m
0
=
t.
(112)
Q=
dq
E
2 [E U (q 0 )]

where

m2
q.
2E
Inverting Eq. (113) one obtains the well-known solution
q

q = sin (t + Q) = sin (t + 0 )
or
q=
After that one finds p as

2E
sin (t + 0 ) .
m 2

p
S
= 2m [E U (q)]
q
that for the harmonic oscillator yields the well-known expression
p

p = 2mE 1 q2 = 2mE cos (t + 0 ) .


p=

(114)

(115)

(116)

(117)

(118)

Finally we calculate the short action S0 , the integral term in Eq. (111). The result has the form
 p

E
S0 (q) = I q 1 q2 + arcsin q ,
I= .
(119)

Here I is the harmonic-oscillator form of the so-called action variable that will be used below. S0 (q) is a

multivalued function of q, taking into account different branches of the . . . and arcsin (. . .) functions, see
Fig. 1. As q changes from 0 to 1 that is the quarter of the oscillation period, the expression in brackets
changes from 0 to /2. The same happens each quarter period, so that the change over the period makes
up 2. Thus the short action over the period is given by
(Period)

S0

= 2I.

(120)

Unlike the short action S0 , the full action S does not increase with time on average.
R Since L = T U and
both kinetic and potential energies oscillate having the same average values, S = dtL oscillates without
growing.
13

S0(q) (2E/)1 1.2


1.0
0.8
0.6
0.4
0.2
1.0

0.5

~
q

0.5

1.0

0.2

Figure 1: Short action of the harmonic oscillator

6.3

Separation of variables

The method of solving the Hamilton-Jacoby equation applied to the harmonic oscillator in the preceding
section works for any system with one degree of freedom, described by the potential energy U (q). This
method can be generalized for systems with N degrees of freedom than allow separation of variables. In
such systems one or more canonical pairs (qi , pi ) enter the Hamiltonian H as combinations that do not
contain other dynamical variables. In the case of one such pair the Hamilton-Jacoby equation has the form


 
S
S
S
+ H qi6=1 ,
, F1 q1 ,
, t = 0.
(121)
t
qi6=1
q1
The solution can be searched in the form of the sum
(1)

S = S (N 1) (qi6=1 , t) + S0 (q1 ) .

(122)

With this Ansatz Eq. (121) becomes


S (N 1)
S (N 1)
+ H qi6=1 ,
, F1
t
qi6=1

(1)

S
q1 , 0
q1

! !
,t

= 0.

(123)

Since this equation has to be valid for any value of the coordinate q1 , the condition F1 = const = 1 should
be fulfilled. Thus Eq. (123) splits up into two equations
!
(1)
S0
F1 q1 ,
= 1
q1
!
S (N 1)
S (N 1)
+ H qi6=1 ,
, 1 , t
= 0.
(124)
t
qi6=1
This is why this situation is called separation of variables. The first equation here is an ordinary differential
equation that allows for a solution in quadratures. The second equation is a Hamilton-Jacoby equation with
N 1 degrees of freedom.
If the problem is time independent, the one can search the solution in the form
(N 1)

S = Et + S0

(1)

(qi6=1 ) + S0 (q1 ) .

14

(125)

This results in simpler equations


(1)

F1

S
q1 , 0
q1
(N 1)

S
H qi6=1 , 0
, 1
qi6=1

= 1

= E.

(126)

A particular case of separation of variables is the case of a cyclic variable q1 that does not enter the
(1)
(1)
Hamiltonian. Then F1 q1 , S0 /q1 reduces to S0 /q1 , so that the first equation (124) can be easily
integrated and Eq. (122) becomes
(127)
S = S (N 1) (qi6=1 , t) + 1 q1 .
Time also is a cyclic variable, and Et in Eq. (125) is similar to 1 q1 in the above equation.
If there is another separating variable q2 , the second equation (124) can be further simplified in terms of
(2)
2 , S0 , and the remainder action S (N 2) . If all N variables separate, the procedure results in the complete
integral of the Hamilton-Jacoby equation of the completely additive form
S = S (0) (t)+

N
X
i

(i)

S0 (qi , ) ,

(128)

whereas S (0) satisfies the equation


S (0)
+ H (1 , 2 , . . . , N , t) = 0
t

(129)

that also can be solved in quadratures. Each term of Eq. (128) depends on one or more constants i . For
time-independent problems one obtains
S (0) (t) = Et,

E = H (1 , 2 , . . . , N ) .

(130)

The complete integral above can be used now to define the systems dynamics as was explained in Sec. 6.1,
see Eqs. (100) and (101). Let us write these equations again:
i =

S
,
i

i = const.

(131)

(i)

Note the equivalence i Pi and i Qi . As, in general, S0 (qi , ) depends not only on its own constant
i but on other constants as well, these equations for different i are coupled. For simpler problems such
(i)
(i)
as the three-dimensional harmonic oscillator, S0 (qi , ) = S0 (qi , i ) , and different equations above are
uncoupled.

6.4

Example: Particles motion in spherical coordinates

As an example consider a generalization of the Kepler problem in spherical coordinates having the Hamiltonian (19). Here the variables separate for
U (r, , ) = a(r) +

b()
c()
+ 2 2 .
r2
r sin

(132)

In particular, for c() = 0 that is of the most interest, the solution has been found in Sec. 4.1. Now we
solve the same problem by the Hamilton-Jacoby formalism. The Hamilton-Jacoby equation has the form
" 
 2
 2 #
S
1
S
1
S 2
1
S
+
+ a(r) +
+ 2mb() +
= 0.
(133)
2
2
t
2m r
2mr

sin
15

The variable is cyclic. The action S is of the additive form


S = Et + p + Sr (r) + S ().

(134)

For the actions one obtains equations





p2
S 2
+ 2mb() +

sin2

2

Sr
+ 2ma(r) + 2
r
r

= = const
= 2mE = const.

(135)

Note that S depends on the constants p and , whereas Sr depends on and E. Integration of these
equations yields
s
Z r r
Z
p2

0
0
0
0
S = Et + p +
dr 2m [E a(r )] 02 +
d 2mb( )
(136)
r
sin2 0
that depends on the three constants E, p , and . Differentiating S with respect to these constants and
equating the results to other constants, one obtains equations of motion in the implicit form that include
six constants, as it should be. These equations are as follows:
Z r
S
m
= t +
dr 0 p
r =
E
2m [E a(r 0 )] /r 02
Z
Z r
S
1
1
1
=
=
d0 q

dr 0 p

2 2m [E a(r 0 )] /r 02 r 02
2 2mb( 0 ) p2 / sin2 0
Z
p
S
1
=
=
d 0 q
(137)
2 0.
p
2mb(0 ) p2 / sin2 0 sin

The first of these equations after integration and solving for r yields r(t). Then the second equation after
integration and solving for yields (t), expressed via r(t). Finally, the third equation yields the solution
(t), expressed via (t) that is in turn expressed via r(t). The whole procedure is much more complicated
than direct numerical solution of Newtonian, Lagrangian, or Hamiltonian equations of motion.

7
7.1

Angle-action variables and slow evolution of mechanical systems


Definition of the angle-action variables

Consider a system with one degree of freedom with a time independent Hamiltonian
H=

p2
+ U (q).
2m

(138)

The Hamiltonian is conserved, H = E = const. Assume that the potential energy U (q) is such that the
system performs a periodic motion between the turning points q . The short action
Z q
Z qp
0
S0 (q) =
pdq =
2m [E U (q 0 )]dq 0
(139)
depends on the energy E as a parameter. Let us introduce the so-called action variable
I
1
I=
pdq
2
16

(140)

that is proportional to the action over the period of motion, c. f. Eq. (120). One can see that I depends on E
only. Thus one can consider S0 of Eq. (139) as parametrically dependent on I and use the function S0 (q, I)
as a generating function of a time-independent canonical transformation, I being the new momentum. The
relation between the old and new dynamic variables is given by Eq. (49) in the form
=

S0 (q, I)
,
I

p=

S0 (q, I)
.
q

(141)

Since the Hamiltonian H = E(I) does not depend on , the latter is a cyclic variable. The Hamilton
equations have the form
H
I =
=0

H
dE
=
=
= ,
I
dI

(142)

where = 2/T is the frequency of motion, T = T (E) being the period of motion. The proof is as follows
I p
I r
I
I
d 1
1
1
dq
1
T
1
dI
m
=
2m [E U (q)]dq =
dq =
=
dt =
= . (143)
dE
dE 2
2
2 [E U (q)]
2
q
2
2

Note the difference between the relation = dE/dI and = E/I in Eq. (119) that is valid only for a
harmonic oscillator. The choice of the action variable I above anticipates the nice result = that allows
to interpret as the phase or the angle of the periodic motion that linearly grows with time, (t) = t + 0 .
Time dependence of q can be found from Eq. (141).
For the harmonic oscillator, S0 (q) and the relation between E and I are given by Eqs. (119) and (120). It
can be obtained by the integration of dE/dI = in the second equation (142) with = const. The relation
between the (q, p) and (, I) variables that follows from Eq. (141) reads
r

2I
q=
sin ,
p = 2mI cos .
(144)
m
This is obviously a form of Eq. (67) or Eqs. (116) and (118). One can obtain a convenient presentation of
S0 in terms of


Z q
Z
Z
dq(0 ) 0
1
2 0
0
pdq 0 =
p(0 )
d
=
2I
cos

d
=
I

+
sin
(2)
.
(145)
S0 =
d0
2
Together with the first equation (144) this formula gives a parametric presentation of S0 (q) shown in Fig.
1.
Equation (140) can be written in the form
Z Z
1
I=
dpdq,
(146)
2
where the double integral is the area circumscribed by the closed orbit in the phase plane (q, p) . To the
contrary, the trajectory of the system in the phase plane (, I) is just a straight line. One can say that
transformation to the angle-action variables straightens the trajectory. Of course, the action-angle formalism
is merely a variant of the Hamilton-Jacoby method of solving mechanical problems. In the latter, the full
action S is used as the generating function of a canonical transformation rather than the short action S0 here.
The Hamilton-Jacoby method is even more radical because after the canonical transformation trajectories
reduce to points.

17

7.2

Integrable and non-integrable systems

If the motion of a system with N degrees of freedom can be determined by the Hamilton-Jacoby method,
e. g., in the case of separation of variables, one can also find a transformation to the angle-action variables
{i , Ii } , i = 1, 2, . . . , N. Then the transformed Hamiltonian of the system depends only on the N constants
Ii , H = H(I1 , I2 , . . . , IN ), whereas all angles linearly increase with time, i (t) = i t + 0i . Thus, again, the
trajectory of the system is a straight line in a 2N -dimensional space. The corresponding motion in the real
phase space {qi , pi } is a multiperiodic motion with in general incommensurate periods. Topologically this
is a motion on multidimensional tori that are known as invariant tori.
The systems that allow a complete transformation to angle-action variables are called integrable systems.
Unfortunately, this definition does not provide an explicit integrability check for a particular system. Obviously systems with separating variables are integrable. Also one can transform a system with separating
variables to a non-separating form by some wild canonical transformation. The resulting system will be
integrable, too, although non-separable. It will be very difficult to integrate this system without knowing
the canonical transformation. One can give up and erroneously conclude that this system is non-integrable.
Trajectories of integrable systems corresponding to different initial conditions are straight lines that do
not cross and depend smoothly on the initial conditions. To the contrast, for nonintegrable systems (that
are the majority of mechanical systems) angle-action variables cannot be found and trajectories cannot be
straightened. This usually leads to an apparently irregular behavior known as dynamical chaos.
One apparent difference between integrable and non-integrable systems is that the former have many
integrals of motion Ii , one for each separable degree of freedom i. These integrals of motion, expressed
through the natural variables {qi , pi } , impose limitations on the regions in the phase space accessible to
the system. This makes the motion of the system regular. To the contrary, non-integrable systems do not
have integrals of motion depending on a small subset of dynamical variables. Thus much more phase space
becomes accessible to them.

7.3

Time-dependent systems

Let us now consider a system with one degree of freedom and the Hamiltonian depending on time via a
parameter (t), i.e., H = H(q, p, (t)). If the time dependence of is slow, the energy E is a slow function
of time as well. The formalism of angle-action variables allows one to separate this slow dynamics from the
fast (or regular) orbiting dynamics of the system. One can define the function S0 by the same Eq. (139)
formally considering as a parameter depending on the current time. This function S0 (q, I, (t)) is not
the short action since the latter would take into account the time dependence of in the integration over
the time-dependent trajectory in Eq. (139). Similarly to Sec. (7.1), we use S0 (q, I, (t)) as the generating
function of a canonical transformation. Since this canonical transformation is time dependent, it changes
the Hamiltonian,
S0

H0 = H +
= E (I, ) + ,
(147)
t
where

S0 (q, I, )

(148)
.

q,I
In the function one has to eliminate q with the help of the transformation formulas
=

S0 (q, I, )
,
I

p=

S0 (q, I, )
,
q

(149)

after which = (, I, ) . The Hamilton equations are modified by the term with :
H0

I =
=

0
H

=
= (I, ) +
,
I
I
18

(150)

where (I, ) = E (I, ) /I. One can see that I and hence E are no longer integrals of motion because of
the time dependence of .
Let us consider, as an illustration, a harmonic oscillator with time-dependent frequency, the Hamiltonian
being given by Eq. (56) with = (t). The short action is given by Eq. (119), where the dependence on
enters via q defined by
r
m
q
q
(151)
2I
that follows from Eqs. (114) and (119). Thus for in Eq. (148) with the help of Eq. (115) one obtains


p
S0 q
q
I
I
= 2I 1 q2
= sin cos =
sin (2) .
(152)
=


q I q,I
2

Now Eqs. (150) take the form

I = I cos (2)

= + sin (2) .
2

(153)

Note that the second of these equations is autonomic.

7.4

Adiabatic invariants

The function of Eq. (148) is a periodic function of with a zero average. Although the short action S0
increases acquiring the increment 2I every period of motion, Eq. (120), this increment does not depend
on and makes no contribution into . An example is for the harmonic oscillator given by Eq. (152).
Periodicity of can be proven rigorously if one uses the Legendre-transformed generating function
S0 (q, , (t)) = S0 (q, I, (t)) I

(154)

instead of S0 (q, I, (t)) for the canonical transformation. Then one obtains the same equations (147) and
(150), whereas is now given by


S0 (q, , )
S0 (q, I, )

(155)
=
.

q,
q,I

One can see that this is the same as the above. However, S0 (q, , ) does not grow on average with ,
because the increment 2I of S0 (q, , ) over the period of motion is exactly compensated for by the term
I 2 in Eq. (154). Thus it is obvious that is periodic with a zero average.
In the first equation (150) the coefficient / is also periodic with a zero average. If the change of
over the period of oscillations T is small,


T


(156)

 1,

the change of I becomes very small upon integration on time as the oscillations in the rhs average out. Thus
I is the so-called adiabatic invariant of motion. To the contrary, the energy E is not an adiabatic invariant.
For instance, if the frequency of a harmonic oscillator is slowly changing, one has
I=

E
= const,

E .

(157)

Adiabatic invariants also emerge in the motion of a charged particle in a weakly non-uniform magnetic field
and in quantum mechanics. The Bohr-Sommerfeld quasiclassical quantization condition has the form
I
1
I=
pdq = n~,
(158)
2
19

I
0.010

0.005

0.000

0.005
0.010
60

40

20

20

40

Figure 2: Time dependence of the change of the action variable I of a harmonic oscillator with slowly
changing frequency. Since I is an adiabatic invariant, I is small at all times. (See parameters for the
numerical calculation in the text.)
where n is an integer. If parameters of the system are changed slowly, I does not change, and so does
not n. Were I not an adiabatic invariant, it would change continuously with the parameters of the system.
However, n cannot change continuously. Thus we conclude that only adiabatic invariants are suitable to
impose quantization while going from the classical to quantum mechanics.
Change of I becomes especially small if slowly changes over the time interval (, ) . In this case the
integral
Z

I =
dt
(159)

can be usually transformed by shifting the integration contour into the complex plane to suppress oscillations
of the integrand. Then the dominant contribution to I comes from the singularity of the integrand closest
the greater is the negative exponential
to the real axis and I becomes exponentially small. The smaller is ,
in I.
As a numerical example one can consider a harmonic oscillator with the energy function

m 2
x + 2 (t)x2 ,
(160)
E=
2
(t) changing from () to () according to

(t) = 0 +

1 + tanh (t)
2

(161)

with > 0. The change of (t) is slow if  0 . Let us set = 0 /10. Also we set the oscillator
mass m = 2. For 0 = = 1 one has () = 1 and () = 2. For the initial state x() = 1
and x()

= 0 the initial oscillator energy is E() = 1. The inital value of the action variable is
I() = E()/() = 1. Numerical integration of the equation of motion x
+ 2 (t)x = 0 yields the
results for I(t) shown in Fig. 2. One can see that the change I(t) I(t) I() is small at all times,
that is, I is indeed an adiabatic invariant. Whereas the asymptotic change I() is so small that it cannot
be seen in the plot, I(t) at t around zero, where the change of (t) occurs, is noticeable. Although I()
is exponentially small, I(t) for a general t is not.

7.5

Parametric resonance via angle-action variables

The formalism of angle-action variables can be used for a more insightful solution of the parametric-resonance
problem considered above. We have to rename the unperturbed frequency of the oscillator as 0 since
20

we need to denote the pumping frequency. The time-dependent frequency of the oscillator can be written
as
0 (t) = 0 [1 + cos(t)],
 1.
(162)
The second autonomic equation (153) has the explicit form
= 0 [1 + cos(t)]

1
cos(t)
sin(2)
.
2
1 + cos(t)

(163)

As we shall see, parametric resonance occurs if is close to 2 0 . Thus we use


= 2 0 + 

(164)

with a small resonance detuning . Solution of the equation for can be searched for in the form


1
t + f (t),
(t) = 0 +
2
2
where f (t) is a slow phase. f (t) satisfies the equation
(2 0 + ) sin [(2 0 + ) t]
f(t) =  + 2 0 cos [(2 0 + ) t] sin [(2 0 + ) t + f (t)]
.
1 + cos [(2 0 + ) t]

(165)

(166)

This is still an exact equation. Taking into account that f (t) is a slow function of time and  1, one
can drop the term with in the denominator and fast oscillating terms that average to zero. Using
2 sin [(2 0 + ) t + f (t)] sin [(2 0 + ) t] = cos f (t) cos [2 (2 0 + ) t + f (t)]

(167)

and dropping the term , one obtains the slow equation


f(t) =  0 cos f (t).

(168)

This equation can be written in the potential form


dUeff
f =
,
df

Ueff (f ) = f + 0 sin f,

(169)

a tilted washboard potential. For small resonance detuning || < 0 , the potential Ueff has local maxima
and minima, so that the phase f (t) relaxes down to a constant value that satisfies
cos f =
thus
sin f =


0



0

(170)
2

(171)

This result for sin f corresponds to the minima of Ueff (f ), whereas the solution with the sign (+) in front
of the square root corresponds to the maxima of Ueff (f ). The latter is unstable and it should be discarded.
The result above means that the oscillator locks into the frequency /2, as also follows from the solution
in natural variables. In the case || > 0 the potential Ueff (f ) is monotonic, and the f (t) performs slow
nonlinear motion with a variable rate without stopping anywhere.
To see the parametric instability that develops for || < 0 , let us now consider the first of the angleaction equations (153). This equation can be written as
d ln I
= 2 0 cos [(2 0 + ) t + f (t)] sin [(2 0 + ) t] .
dt
Reducing the product of trigonometric functions as
21

(172)

2 cos [(2 0 + ) t + f (t)] sin [(2 0 + ) t] = sin f (t) + sin [2 (2 0 + ) t + f (t)]

(173)

and dropping the fast oscillating term, one obtains


d ln I
= 0 sin f (t).
dt
After f (t) approaches a constant given by Eq. (171), this equation becomes
d ln I
1p
= 2,
=
( 0 ) 2 2 ,
dt
2
where is the parametric resonance exponent. Solution of Eq. (175) has the form
I = I0 e2t ,

(174)

(175)

(176)

where I0 is the initial values of I. One can see the exponential divergence of I and thus the oscillators energy
E in the region of parametric resonance. On the other hand, for large detuning, || > 0 , the exponent
is imaginary and I oscillates without growing.
As we have seen, the analytical solution in the angle-action variables is more elegant than the straightforward Newtonian solution and it provides more insight. Numerical solution can be done with both formalisms
to the same effect. However, including damping and nonlinearities in the Newtonian formalism is straightforward, whereas in the angle-action formalism it requires a significant work.

22

You might also like