Professional Documents
Culture Documents
Mechanics
Brian Tu
Contents
1 Introduction 2
2 Preliminaries 2
3 Poisson Bracket 3
3.1 Characterization of Canonical Transforms . . . . . . . . . . . . . . . . . 3
3.2 Application to Integrals of Motion . . . . . . . . . . . . . . . . . . . . . . 5
5 Generating Functions 6
1
1 Introduction
2 Preliminaries
and hence,
y = (J JJ T )H,
where (J)ij = yi /xj is the Jacobian of y. The form of Hamiltons equation is preserved
if
J JJ T = J. (1)
Hence a transformation is canonical if its Jacobian satisfies (1).
Incidentally, such a matrix J is called symplectic.
2
3 Poisson Bracket
We saw above that one way to characterize a canonical transformation is by checking (1).
Another way of characterizing these transformations is by using an operation called the
Poisson bracket.
Definition 3.1. A Poisson bracket is a binary operation on two functions f (q, p),
g(q, p) defined as
n
X f g f g
{f, g} = .
i=1
qi pi pi qi
Proof. For the first part, let f (q, p) and g(q, p) be any two functions. We have
n
X f g f g
{f, g} =
i=1
qi pi pi qi
f g
= J .
x x
If we have a transformation y(x) with Jacobian J , then we have
f f
= J
x y
and thus
f g
{f, g} = J JJ T
x y
f g
= J .
x y
3
This shows the invariance of the Poisson bracket. For the other direction, note that we
can write the Jacobian as
Q Q
q p
J = P P .
q p
Computing, we have
T {Q, Q} {Q, P }
J JJ =
{P, Q} {P, P }
where
({Q, P })ij = {Qi , Pj },
and the other entries are defined similarly. Thus we have
J JJ T = J
and the transformation is canonical.
4
3.2 Application to Integrals of Motion
The Poisson bracket also neatly characterizes the integrals of motion. A function f is an
integral of motion iff {f, H} = 0 [4]. To see this, note that
n X n
X f H f H f f df
{f, H} = = qi + pi = ,
i=1
qi pi pi qi i=1
qi pi dt
and
Q(0) = q P (0) = p.
If is very small, we can write
Qi () = qi + dqi , Pi () = pi + dpi .
We wish to know what relationships dq and dp must satisfy for the transformation to be
canonical. The Jacobian is
In + dq/q dq/p
J =
dp/q In + dp/p
5
which must satisfy J JJ T = J. This gives us
dq dp
= . (7)
q p
If we could find a function F such that
F F
= dq and = dp, (8)
p q
then (7) would be satisfied [5].
Note that the equations in (8) look exactly like Hamiltons equations. Indeed, if we let
be the time t, the transformation induced by the Hamiltonian H matches the differential
equations for the flow of the system. Hence the advance of time in a Hamiltonian system
is a canonical transformation.
Now that we have established that time evolution of a Hamiltonian system is a canonical
transformation, we can give a very quick proof of Liouvilles Theorem.
Theorem 4.1 (Liouvilles Theorem). The phase flow of a Hamiltonian system is volume
preserving.
Proof. Let t be fixed. Let the evolution of the Hamiltonian system after time t be a
canonical transformation taking q to Q(q, p) and p to P (q, p). It suffices to show that
the Jacobian of the transformation is 1. Using properties of the Jacobian, we have
(Q, P )
J =
(q, p)
(Q, P ) (q, p)
=
(q, P ) (q, P )
Q p
= .
q P
and thus
Q p
= ,
q P
so we have
J = 1,
as desired.
5 Generating Functions
6
how can we define Pi so that the resulting transformation is canonical? If we define
F
Pi =
Qi
then F will be a canonical transformation. This can be shown using Theorem 3.1.
F is known as a generating function of the first kind. There are four other kinds of gener-
ating functions that generate canonical transformations, with other coordinates defined
as follows [5, 4]:
F2 F2
F = F2 (q, P ) QP , p= ,P =
q P
F3 F3
F = F3 (p, Q) + qp, q= ,P =
p Q
F4 F4
F = F4 (p, P ) + qp QP , q= ,Q = .
p P
I pledge my honor that this represents my own work in accordance with University pol-
icy.
7
References