A simple global representation for second-order

normal forms of Hamiltonian systems relative to

periodic ows
M Avenda no-Camacho, J A Vallejo and Yu Vorobjev
Departamento de Matem aticas, Universidad de Sonora, Blvd L. Encinas y Rosales
s/n Col. Centro, Ed. 3K-1 CP 83000 Hermosillo (Son) Mexico.
E-mail:
misaelave@mat.uson.mx,jvallejo@fc.uaslp.mx,yurimv@guaymas.uson.mx
Abstract. We study the determination of the second-order normal form for
perturbed Hamiltonians H

= H
0
+ H
1
+

2
2
H
2
, relative to the periodic ow of
the unperturbed Hamiltonian H
0
. The formalism presented here is global, and can
be easily implemented in any CAS. We illustrate it by means of two examples: the
Henon-Heiles and the elastic pendulum Hamiltonians.
PACS numbers: 02.40.Yy,45.10.Hj,45.10.Na
1. Introduction
In this paper we discuss some computational aspects of the normal form theory for
Hamiltonian systems on general phase spaces, that is, Poisson manifolds. According to
Deprit [9], a perturbed vector eld
A = A
0
+A
1
+

2
2
A
2
+ +

k
k!
A
k
+O(
k+1
)
on a manifold M, is said to be in normal form of order k relative to A
0
if [A
0
, A
i
] = 0
for i 1, . . . , k. In the context of perturbation theory, the normalization problem
is formulated as follows: to nd a (formal or smooth) transformation which brings a
perturbed dynamical system to a normal form up to a given order. The construction
of a normalization transformation, in the framework of the Lie transform method
[8, 12, 14, 16], is related to the solvability of a set of linear non homogeneous equations,
called the homological equations. If the homological equations admit global solutions,
dened on the whole M, we speak of a global normalization, which essentially depends
on the properties of the unperturbed dynamics.
Permanent address: Facultad de Ciencias, Universidad Autonoma de San Luis Potos, Lat. Av. S.
Nava s/n Col. Lomas, CP 78290 San Luis Potos (SLP) Mexico.
Simple representation of global second-order normal forms 2
Here we are interested in the global normalization of a perturbed Hamiltonian
dynamics relative to periodic Hamiltonian ows. In this case, a result due to Cushman
[6], states that if A is Hamiltonian, and the ow of the unperturbed vector eld A
0
is periodic, then the true dynamics admits a global Deprit normalization to arbitrary
order. The corresponding normal forms can be determined by a recursive procedure
(the so-called Deprit diagram) involving the resolution of the homological equations at
each step.
In this paper, we extend Cushmans result to the Poisson case and derive an
alternative coordinate-free representation for the second-order normal form, involving
only three intrinsic operations: two averaging operators associated to the S
1
action,
and the Poisson bracket. We give a direct derivation of this representation based on
a period-energy argument [11] for Hamiltonian systems, and some properties of the
periodic averaging on manifolds [3, 6, 19]. This formalism allows us to get an ecient
symbolic implementation for some models related to polynomial perturbations of the
harmonic oscillator with 1 : 1 resonance. In particular, we compute the second-order
normal form of the Henon-Heiles [6], and the elastic pendulum [4, 5, 10] Hamiltonians,
expressed in terms of the Hopf variables.
Let us remark that the second-order normal form plays a very important r ole in
the approximation of a perturbed dynamics by solutions of the averaged system when
a long-time scale is used [2, 20]. Our desire to study this kind of dynamics led to the
present work.
Sections 2 and 3 contain some basic properties of the action induced by the ow
of a periodic vector eld and their associated averaging operators. In Section 4 we
particularize to the case of Hamiltonian vector elds, using an energy-period relation,
and the main result is proved in Section 5. The nal section is devoted to the examples.
2. Vector elds with periodic ow
Throughout the paper, we set S
1
= R/2Z. We collect here some results regarding
the ow Fl
t
X
of a vector eld X, on an arbitrary manifold M, in the case when Fl
t
X
is
periodic. Although these results are general, later they will be applied to the case of a
Hamiltonian vector eld on a Poisson manifold (M, P).
Let X A(M) be a complete vector eld whose ow is periodic with period
function T (

(M), T > 0, that is: for any p M,

Fl
t+T
X
(p) = Fl
t
X
(p). (1)
Then, X determines an S
1
action S
1
M M given by (t, p) Fl
t/(p)
X
(p), where
:= 2/T > 0 is the frequency function, and t S
1
. Thus, the S
1
action is periodic,
with constant period 2.
The generator of this S
1
action can be readily computed:
(p) =
d
dt

t=0
Fl
t/(p)
X
(p) =
1
(p)
d
ds

s=0
Fl
s
X
(p) =
1
(p)
X(p),
Simple representation of global second-order normal forms 3
so =
1

X. Notice, from (1), that T(p) > 0 is the period of the integral curve of
X passing through p M at t = 0, c
p
: R M (which is such that c(0) = p and
c
p
(0) = X(p)). In other words, c
p
(0) = p = c
p
(T(p)). Also, each point on the image of
the integral curve c
p
, gives the same value for the period: T(p) = T(c
p
(t)), for all t R.
In terms of the ow of X, that means
((Fl
t
X
)

T)(p) = T(Fl
t
X
(p)) = T(p), for all p M.
As T is constant along the orbits of X, its Lie derivative with respect to X vanishes:
/
X
T =
d
dt

t=0
(Fl
t
X
)

T = 0.
Now, from T = 2, we get
0 = /
X
(T) = (/
X
T) +T/
X
= T/
X
.
But T > 0, so this implies that is a rst integral (or invariant) of X,
/
X
= 0. (2)
Denition 2.1. A smooth function f (

(M) is said to be an S
1
invariant if it is
invariant under the ow of the generator =
1

X, that is,
/

f = 0.
Clearly, this is equivalent to the condition (Fl
t

f = f, for all t [0, 2]. Notice that,

by (2), the frequency function is also an invariant of the S
1
action, /

=
1

/
X
= 0.
3. Averaging operators
Given a vector eld X A(M) with periodic ow, the associated S
1
action can be used
to dene two averaging operators, which we will denote by and o. In this section,
M will be an arbitrary manifold.
For any tensor eld R T
s
r
(M) (rcovariant, scontravariant), the average of R
with respect to the S
1
action on M induced by X, is the tensor eld (of the same type
as R) dened by
R :=
1
2

2
0
(Fl
t

Rdt.
The properties of the ow [1] guarantee that R is well-dened as a dierentiable tensor
eld. Also, note that if R T
s
r
(M), and X
1
, . . . , X
r
A(M),
1
, . . . ,
s

1
(M)
are arbitrary, then, for every p M, t (Fl
t

R(X
1
, . . . , X
r
,
1
, . . . ,
s
)(p) is a real
dierentiable funcion on the compact [0, 2], hence integrable. We will use this denition
mainly applied to the case of functions f (

(M) ((0, 0)tensors) and vector elds

Y A(M) ((0, 1)tensors).
Simple representation of global second-order normal forms 4
The other averaging operator that will be important in what follows, is the o operator,
o : T
s
r
(M) T
s
r
(M). It is given by
o(R) :=
1
2

2
0
(t )(Fl
t

Rdt.
Note that both, and o, are Rlinear operators. Other properties are listed below.
Lemma 3.1. For any complete vector eld Y A(M) (whose ow is not necessarily
periodic) and smooth tensor eld R T
s
r
(M), we have:
d
ds

s=0
(Fl
s
Y
)

R =
1
2

(Fl
2
Y
)

R R

,
where the averaging is taken with respect to the ow of Y , that is, R is given by
R :=
1
2

2
0
(Fl
t
Y
)

Rdt.
Proof. Start from the identities (which follow directly from the denitions of ow and
Lie derivative):
(Fl
t
Y
)

(/
Y
R) =
d
dt
(Fl
t
Y
)

R =
d
ds

s=0
(Fl
s+t
Y
)

R =
d
ds

s=0
(Fl
s
Y
)

(Fl
t
Y
)

R.
Taking the integral with respect to t between 0 and 2 on both sides, we get, on the
one hand:
1
2

2
0
(Fl
t
Y
)

(/
Y
R) dt =
d
ds

s=0
(Fl
s
Y
)

1
2

2
0
(Fl
t
Y
)

Rdt

=
d
ds

s=0
(Fl
s
Y
)

R,
and, on the other:
1
2

2
0
(Fl
t
Y
)

(/
Y
R) dt =
1
2

2
0
d
dt
(Fl
t
Y
)

Rdt =
1
2

(Fl
2
Y
)

R R

.
Proposition 1. For every R T
s
r
(M), the following properties hold:
(a) R is invariant under the ow of (that is, S
1
invariant) if and only if R = R.
(b) /

R = 0.
(c) If g (

(M) is S
1
invariant, then gR = gR.
(d) The averaging operator commutes with tensor contractions whenever one of the
tensors is S
1
invariant, that is, if S T
b
a
(M) is S
1
invariant and C
l
k
is any
contraction, then C
l
k
(R S) = C
l
k
(R S).
Proof.
Simple representation of global second-order normal forms 5
(a) If R is invariant under the ow of , then (Fl
t

R = R, for all t [0, 2], and

from this it is immediate that R = R. Reciprocally, if R = R we may apply the
preceding lemma to obtain:
d
ds

s=0
(Fl
s

R =
1
2

(Fl
2

R R

,
and from the fact that the ow of is 2periodic,
/

R =
d
dt

t=0
(Fl
t

R = 0.
(b) From the properties of the Lie derivative and the denition of R:
(Fl
t
Y
)

(/
Y
R) =
d
dt
(Fl
t

R =
d
dt
1
2

2
0
(Fl
s+t

Rds =
d
dt
1
2

t+2
t
(Fl
u

Rdu.
Now, because Fl
u

is 2periodic:
(Fl
t
Y
)

(/
Y
R) =
d
dt
1
2

2
0
(Fl
u

Rdu = 0,
so, as Fl
t

is a dieomorphism, /

R = 0.
(c) It is a straightforward computation.
(d) It is just a consequence of the commutativity between the pull-back and the tensor
contractions, and the functorial property (Fl
t

(R S) = (Fl
t

R (Fl
t

S.
Remark 1. In particular, from (d) we get that if Y A(M) and (M) is
S
1
invariant, then i
Y
= i
Y
.
Proposition 2. For any R T
s
r
(M) and g (

(M) S
1
invariant, the following
hold:
(a) o(gR) = go(R).
(b) (/

o)(R) = R R.
Proof.
(a) A straightforward computation.
(b) With an obvious change of variable, we have:
(Fl
s

o(R) =
1
2

2
0
(t )(Fl
s+t

Rdt =
1
2

s+2
s
(u s )(Fl
u

Rdu.
Dierentiating both sides of this identity with respect to the parameter s, and taking
into account the 2periodicity of the ow Fl
s

, it results:
d
ds
(Fl
s

o(R) = (Fl
s

(R R).
Simple representation of global second-order normal forms 6
The statement follows by recalling that Fl
s

is a dieomorphism, and the identity

(see [1]):
d
ds
(Fl
s

o(R) = (Fl
s

(/

o(R)).
Finally, let us give some useful properties involving the averaging operators.
Proposition 3. For all R T
s
r
(M), the operators /

, , and o, satisfy the relations:

(a) /

R = /

R = 0.
(b) o(R) = o(R) = 0.
(c) d = d, for all (M).
Proof. Straightforward computations, making use of Proposition 2 and the fact that d
commutes with pull-backs.
4. The Hamiltonian case
Let (M, P) be an mdimensional Poisson manifold, where P
2
TM is a Poisson
bivector determining a bracket f, g = P(df, dg), for all f, g (

(M). For every f, its

Hamiltonian vector eld X
f
A(M) is given by X
f
(g) := f, g, for any g (

(M),
equivalently,
X
f
= i
df
P. (3)
At any point the distribution spanned by the Hamiltonian vector elds is involutive, as a
consequence of Jacobis identity for the Poisson bracket , . Thus, these Hamiltonian
vector elds give rise to a foliation whose leaves turn out to be symplectic manifolds
(see [21]). On each leaf S, the restriction P[
S
is a non-degenerate Poisson bivector eld
which determines a symplectic structure
S
through:

S
(X
f
, X
g
) := f, g.
Indeed, by the splitting theorem due to Weinstein ([21]), the local structure of (M, P)
can be described as follows: for any p M there exists a chart (U, ) of M around p such
that, if q
1
, ..., q
k
, p
1
, ..., p
k
, y
1
, ..., y
l
are the coordinates of : U R
m
(2k + l = m),
then
P[
U
=
k

i=1

q
i

p
i
+
1
2
l

i,j=1

ij
(y
1
, ..., y
l
)

y
i

y
j
, (4)
where :
l
(U) R
l
R is smooth and
ij
(p) = 0 (
l
: R
m
= R
2k
R
l
R
l
is
the canonical projection). The non-negative integer k is called the rank of the Poisson
structure P at p M. When k = m, P induces a symplectic structure on M. Then,
the symplectic leaf S through p M, is given by the equations (y
1
, ..., y
l
) = (0, ..., 0).
When moving along the ow of a Hamiltonian vector eld, which is tangent to some
integral submanifold S, it is clear that we stay on the same symplectic leaf S. Next,
Simple representation of global second-order normal forms 7
we study what happens on these leaves when the Hamiltonian vector eld has periodic
ow.
We will need rst an auxiliary result, interesting in its own, known as the period-
energy relation (see [11]).
Proposition 4. Let X be a vector eld on the symplectic manifold (S, ) whose ow
is periodic with period function T (

(M), T > 0 (and frequency = 2/T). If X

is the Hamiltonian vector eld of a certain function f (

(M) (that is, i

X
= df),
then:
d df = 0 = dT df. (5)
Proof. By hypothesis, we have,
/
X
= i
X
d + di
X
= d
2
f = 0.
On the other hand, using the generator = X/ of the S
1
action induced by X:
/
X
= /

d df.
Recalling that , f are rst integrals of X, and hence S
1
invariants, applying the
averaging operator to the last identity, taking into account that /

= 0
(Proposition 3 (a)), and the commutativity between d and (Proposition 3 (c)), we
get:
0 = /

d df = /

d df =
1

d df.
Remark 2. Notice that, in terms of Hamiltonian vector elds, we can write the energy-
period relation (5) as follows,
X

X
f
= 0. (6)
Also, in the course of the proof we have seen that, if =
1

X is the generator of
the S
1
action induced by X:
0 = /
X
= /

d df,
so from (5) we get the following consequence.
Corollary 1. The symplectic form is S
1
invariant, /

= 0. In particular, = .
Notice that, under the hypothesis of Proposition 4, if g (

(S) is S
1
invariant,
then its Hamiltonian vector eld X
g
A(S) is also S
1
invariant. Indeed, recalling
that d commutes with the averaging (Proposition 3 (c)), Remark 1, and the preceding
Corollary, we get:
i
X
g
= dg = dg = i
Xg
= i
Xg
.
Simple representation of global second-order normal forms 8
Hence, by the non-degeneracy of , X
g
= X
g
. Now, if g is S
1
invariant, g = g,
and so X
g
= X
g
.
As a consequence, for any S
1
invariant g (

(S), we have
/

X
g
= [X

, X
g
] = 0.
Now, suppose that we are given a function H (

(M) on the Poisson manifold

(M, P) such that its Hamiltonian vector eld X
H
A(M) has periodic ow (with
frequency function (

(M), > 0). Let =

1

X
H
be the generator of the
associated S
1
action. From the results above we know that M is foliated by symplectic
leaves S in such a way that P[
S
is equivalent to a symplectic form
S
(recall (4)), and
these are invariant under Hamiltonian ows. Thus:
0 = /
X
H
P = /

P = /

i
d
P = /

P +
1

X
H
X

,
where we have used the formula /
fX
A = f/
X
A X i
df
A (valid for any function
f (

(M), vector eld X A(M) and multivector eld A (TM), see [18], p.
358), as well as (3) and the fact that > 0. From this identity and the energy-period
relation (6), we deduce that P is S
1
invariant, /

P = 0.
Moreover, if g (

(M) is S
1
invariant, the ow of its Hamiltonian vector eld X
g
leaves the integral submanifolds S invariant and, as we have seen, on each of them it
satises /

X
g
= 0, so this is also true on M. In other words, the ows of and X
g
commute on M. The following result exploits this fact.
Proposition 5. Let (M, P) be a Poisson manifold, and H (

(M) such that its

Hamiltonian vector eld X
H
A(M) has periodic ow. If f, g (

(M) and g is
S
1
invariant, then:
(a) If is the frequency function of X
H
, then, X
H
X

= 0.
(b) H, g = 0.
(c) f, g = f, g.
Proof. Item a follows from the above considerations, while (b) is proved by a
straightforward computation. Item (c) is a direct consequence of the S
1
invariance
of g and the fact that the ows of and X
g
commute.
5. The main result
Let H

= H
0
+ H
1
+
1
2

2
H
2
+ O(
3
) an dependent Hamiltonian function which
describes a perturbed Hamiltonian system on a Poisson manifold (M, P), with associated
bracket , . We will denote by X
H
= X
H
0
+X
H
1
+
1
2

2
X
H
2
+O(
3
) the corresponding
Hamiltonian vector eld. Recall that the perturbed Hamiltonian vector eld X
H
is in
(Deprit) normal form relative to X
H
0
of order k in if
[X
H
0
, X
H
i
] = 0, for all i 1, 2, ..., k. (7)
Simple representation of global second-order normal forms 9
In terms of Hamiltonian functions, (7) is satised whenever
H
0
, H
i
= 0, for all i 1, 2, ..., k.
Usually, one can bring the Hamiltonian to a normal form by means of near-to-
identity transformations. Let us recall some denitions and basic properties.
Let M be a manifold, N M be a non-empty open domain, and > 0. A smooth
mapping : (, ) N M is said to be a near-to-identity transformation if, for
each (, ), the map

: N M given by

(x) = (, x)
is such that it is a dieomorphism onto its image and, moreover,
0
= id
M
.
These transformations have the following important property: whenever we have
a time-dependent vector eld A

on M, and a near-to-identity transformation

, the
pull-back

is again an dependent vector eld on N, and it is such that,

[
=0
= A
0
.
In other words, thinking of A

as a perturbed vector eld, near-to-identity

transformations preserve the unperturbed part.
Actually, we will construct the required transformations out from the ow of a
perturbed vector eld. The following properties say that we can do that on each open
domain with compact closure.
Proposition 6. Let F : R M M a smooth mapping, sending (, x) to F

(x) =
F(, x), such that F
0
= id
M
. Then, for any open domain with compact closure
N M, there exists a > 0 such that, for each (, ), the restriction F

[
N
is a dieomorphism onto its image.
Proof. It is an immediate consequence of the fact that the closure N can be covered
by a nite number of open neighborhoods, such that the Implicit Function Theorem
applies on them.
Proposition 7. Let A

= A
0
+R

be a smooth vector eld on a manifold M. Assume

that the unperturbed vector eld A
0
is complete on M. Then, for any open domain
N M, with compact closure, and any constant > 0, there exists another constant
L > 0 such that the ow Fl
t
A
of A

, is well-dened on N for any t [0, L/] and each

(0, ].
Proof. If X, Y are vector rlds on the manifold M, their ows are related by
Fl
t
X
Fl
t
Pt
= Fl
t
Y
, (8)
where P
t
is the time-dependent vector eld given by P
t
= X + (Fl
t
X
)

Y . Now, let
(Fl
t
A
0
)

A
0
= R
t
(),
Simple representation of global second-order normal forms 10
where R
t
() = (Fl
t
A
0
)

depends smoothly on t and , and x a > 0. By the Flow-Box

Theorem and the compactness of the closure N, there exists an L > 0 such that the
ow of R
t
() is well-dened on N for any t [0, L]. Applying (8) to X = A
0
, Y = A

,
and P
t
= R
t
(), we get,
Fl
t
A
= Fl
t
A
0
Fl
t
Rt()
,
and, since Fl
t
A
0
is well-dened for all t R, the statement follows.
Denition 5.1. We say that the system described by a vector eld of the form
A

= A
0
+ R

, where A
0
has complete ow, admits a global normalization of order
k if, for each open domain N M with compact closure, there exist a > 0 and a
near-to-identity transformation F : (, ) N M, which brings A

to a normal
form of order k.
Theorem 5.2. Suppose that the ow of X
H
0
is periodic with frequency function
(

(M), > 0. Then, the perturbed Hamiltonian system admits a global normalization
of arbitrary order k. In particular, the second order normal form can be expressed as:
H

= H
0
+H
1
+

2
2

H
2
+ S

H
1

, H
1

+O(
3
). (9)
Proof. If the Hamiltonian vector eld X
H
0
has periodic ow, the existence of the near-
to-identity canonical transformation

follows from the above Propositions (see also

[3, 6, 16, 17]). Here we give a explicit formula for it.
Let

be the ow of the perturbed vector eld Z

= Z
0
+ Z
1
where Z
0
and Z
1
are the Hamiltonian vector eld of the functions G
0
=
1

o(H
1
) and G
1
=
1

o(H
2
+
o(
1

H
1
), H
1
+ H
1
), respectively. Using the Lie transform method [6, 8, 12, 14], the
second order development of H

is given by:
H

= H
0
+ (/
Z
0
H
0
+H
1
)
+

2
2

/
2
Z
0
H
0
+ 2/
Z
0
H
1
+ /
Z
1
H
0
+H
2

+O(
3
) (10)
Now, we apply the results of the preceding sections to put this Hamiltonian in the form
(9). To this end, we compute:
/
Z
0
H
0
= /
X
H
0
o(
1

H
1
) = H
1
H
1
,
/
2
Z
0
H
0
= /
X
G
0
(H
1
H
1
) =
1

o(H
1
), H
1
H
1
,
/
Z
0
H
1
= /
X
G
0
H
1
=
1

o(H
1
), H
1
,
and, nally
/
Z
1
H
0
= /
X
H
0
o(H
2
+ o(
1

H
1
), H
1
+ H
1
)
= H
2
+ o(
1

H
1
), H
1
(H
2
+ o(
1

H
1
), H
1
+ H
1
).
Substituting these identities into (10), we obtain the normal form (9).
Simple representation of global second-order normal forms 11
6. Examples
In this section we illustrate the computation of the normal form of two particular
Hamiltonians on R
2
endowed with the canonical symplectic form, = dp
1
dq
1
+dp
2
dq
1
(and the corresponding canonical Poisson bracket). If we have a system admitting an
S
1
action, described by a perturbed Hamiltonian H = H
0
+ H
1
, and such that the
Hamiltonian vector eld of H
0
, X
H
0
, has periodic ow with frequency then, as shown
in Theorem 5.2, its second-order normal form is given by:
H
0
+H
1
+

2
2

o(
H
1

), H
1

.
Example 1 (Henon-Heiles Hamiltonian). This example is taken from [7]. The
Hamiltonian is
K = K
0
+K
1
=
1
2
(p
2
1
+p
2
2
) +
1
2
(q
2
1
+q
2
2
) +

q
3
1
3
q
1
q
2
2

(note that the perturbation term is an homogeneous polynomial of degree 3). The
frequency function for the ow of X
K
0
is readily found to be constant, = 1, and,
after some computations, the second-order normal form is found to be:
p
2
2
+p
1
2
2
+
q
2
2
+q
1
2
2


2
48

5 q
2
4
+

10 q
1
2
+ 10 p
2
2
18 p
1
2

q
2
2
+56 p
1
p
2
q
1
q
2
+ 5 q
1
4
+

10 p
1
2
18 p
2
2

q
1
2
+ 5 p
2
4
+ 10 p
1
2
p
2
2
+ 5 p
1
4

It is usual to express the normal form in terms of the Hopf variables w

1
, w
2
, w
3
, w
4
, as
a previous step to carry on the reduction of symmetry process (see [6],[7]). For the case
in which H
0
is the Hamiltonian of the 2Dharmonic oscillator, these variables form
a system of functionally independent generators of the algebra of rst integrals of H
0
,
and are dened as w
1
= 2(q
1
q
2
+ p
1
p
2
), w
2
= 2(q
1
p
2
q
2
p
1
), w
3
= q
2
1
+ p
2
1
q
2
2
p
2
2
,
w
4
= q
2
1
+q
2
2
+p
2
1
+p
2
2
. Working separately with the independent term and the coecient
of
2
in the expression above, we get:
w
4
2
,
and
w
2
2
(48 + 7)
48

w
2
4
(48 + 5)
48
+w
2
3
+w
2
1
.
In the process of expressing the q
i
, p
i
variables in terms of the w
j
, a parameter appears
as a consequence of the fact that the corresponding system of equations is indeterminate.
The formulas appearing in [7] are recovered by choosing the value 0 of the parameter:
7 w
2
2
48

5 w
2
4
48
.
Thus, the second-order normal form of the Henon-Heiles system is
H

=
w
4
2
+

2
48

7w
2
2
5w
2
4

+O(
3
).
Simple representation of global second-order normal forms 12
Example 2 (The elastic pendulum). Consider the case of the Hamiltonian of a elastic
pendulum (see [5],[4],[10]):
H(q
1
, p
1
, q
2
, p
2
) =
p
2
1
+p
2
2
2
+
q
2
1
+q
2
2
2


2
q
2
1
(1 +q
2
),
which is that of a perturbed system H
0
+H
1
, where
H
0
(q
1
, p
1
, q
2
, p
2
) =
p
2
1
+p
2
2
2
+
q
2
1
+q
2
2
2
,
and
H
1
(q
1
, p
1
, q
2
, p
2
) =
q
2
1
(1 +q
2
)
2
.
Note that the perturbation term now is not homogeneous. The computation of the normal
form in the original variables gives the result:
p
2
2
+p
1
2
2
+
q
2
2
+q
1
2
2


4

q
1
2
+p
1
2

2
192

20 q
1
2
4 p
1
2

q
2
2
+ 48 p
1
p
2
q
1
q
2
+ 5 q
1
4
+

4 p
2
2
+ 10 p
1
2
+ 12

q
1
2
+ 20 p
1
2
p
2
2
+ 5 p
1
4
+ 12 p
1
2

As before, we can express in terms of the Hopf variables the independent terms and the
coecient of , getting:
w
4
2
,
and

w
4
8

w
3
8
.
Note, however, that the coecient of
2
is not a homogeneous polynomial (of degree 4):
there are two 2degree terms: (q
2
1
+p
2
1
)/16. Luckily, these terms can be easily expressed
in terms of the variables w
1
, w
2
, w
3
, w
4
(as (q
2
1
+ p
2
1
)/16 = (w
4
+ w
3
)/32) and then we
can analyse the remainder, which is a polynomial of degree 4. Again, a parameter
appears in the process:

w
2
4
(768 + 25)
768
+
w
2
3
(256 + 5)
256
+
w
2
2
(32 + 1)
32
+w
2
1

5 w
3
w
4
384
.
Let us take the simplest solution = 0:

25 w
2
4
768

5 w
3
w
4
384
+
5 w
2
3
256
+
w
2
2
32
.
The remainder in the coecient of
2
is:
w
4
32
+
w
3
32
.
Thus, we get the second-order normal form of the elastic pendulum in the Hopf variables:
H

=
w
4
2


8
(w
4
+w
3
) +

2
32

w
4
+w
3
+w
2
2

25w
4
2
24

5w
3
w
4
12
+
5w
3
2
8

+O(
3
).
Simple representation of global second-order normal forms 13
Remark 3. One of the advantages of the representation (9) for the second-order
normal form, is that it allows an easy implementation in any Computer Algebra System
(CAS), as it does not involve the resolution of the homological equations. Indeed, the
computations above were carried out with a package written in Maxima [15], available at
the URL http: // galia. fc. uaslp. mx/
~
jvallejo/ pdynamics. zip . It contains a
detailed documentation illustrating its use with the preceding examples.
References
[1] R. Abraham, J. E. Marsden and T. Ratiu, Manifolds, tensor analysis, and applications (2nd. Ed.)
Springer, New York, 1988.
[2] V. I. Arnold, V. V. Kozlov, A. I. Neistadt, Mathematical aspects of classical and celestial mechanics
(Dynamical Systems III) Springer Verlag, Berlin, 1987.
[3] M. Avenda no-Camacho and Yu Vorobjev, Homological equations for tensor elds and periodic
averaging, Russian J. Math.Phys., 18 no. 3 (2011) 243257.
[4] E. Breitenberger and R. D. Mueller, The elastic pendulum: A nonlinear paradigm, J. Math. Phys.,
22 (1981) 11961211.
[5] R. Broucke and P. A. Baxa, Periodic solutions of a spring-pendulum system, Celest. Mech., 8 2
(1973), 261267.
[6] R. Cushman, Normal forms for Hamiltonian vector elds with periodic ow, in Geometric Methods
in Mathematical Physics (S. Sternberg, ed.) D. Reidel Publ. (1984) 125144.
[7] R. Cushman, Geometry of perturbation theory, in Deterministic Chaos in General Relativity.
D. Hobill, A. Burd, A.A. Coley (eds.) Nato Science Series B, Vol. 332, Springer Verlag (1993)
89101.
[8] A. Deprit, Canonical transformation depending on a small parameter, Celest. Mech., 1 no.1 (1969)
1330.
[9] A. Deprit, Delaunay normalisations, Celest. Mech., 26 no.1 (1982) 921.
[10] I. T. Georgiou, On the Global Geometric Structure of the Dynamics of the Elastic Pendulum,
Nonlinear Dynamics, 18 Issue 1 (1999) 5168.
[11] W. B. Gordon, On the relation between period and energy in periodic dynamical systems, J. Math.
Mech., 19 2 (1970) 111114.
[12] G. Hori, Theory of general perturbations with unspecied canonical variables, Publ. Astron. Soc.
Japan, 18 (1966) 287296.
[13] D. C. Lewis, Families of periodic solutions of systems having relatively invariant line integrals,
Proc. Amer. Math. Soc., 6 (1955) 181185.
[14] A. A. Kamel, Perturbation method in the theory of nonlinear oscillations, Celest. Mech., 3 (1970)
90106.
[15] Maxima.sourceforge.net, Maxima, a Computer Algebra System. Version 5.28.0 (2012). http:
//maxima.sourceforge.net/.
[16] K. R. Meyer, Normal forms for Hamiltonian systems, Celest. Mech., 9 (1974) 517522.
[17] K. R. Meyer and G. R. Hall, Introduction to Hamiltonian dynamical systems and the N-Body
problem Springer-Verlag, New York, 1992.
[18] P. W. Michor, Topics in Dierential Geometry American Mathematical Society, Rhode Island,
2008.
[19] J. Moser, Regularization of Keplers problem and the averaging method on a manifold, Comm. on
Pure and Appl. Math., 23, Issue 4 (1970) 609636.
[20] J. Murdock, J. A. Sanders, F. Verhulst, Averaging method in nonlinear dynamical systems (2nd.
Ed.) Springer Verlag, New York, 2007.
[21] A. Weinstein, The local structure of Poisson manifolds, J. Dierential Geom., 18 n. 3 (1983)
523557.