scieee Science in your language
[en] (orig)

Poincaré-Melnikov-Arnold method for twist maps

Abstract

The Poincar\'e--Melnikov--Arnold method is the standard tool for detecting splitting of invariant manifolds for systems of ordinary differential equations close to ``integrable'' ones with associated separatrices. This method gives rise to an integral (continuous sum) known as the Melnikov function (or Melnikov integral). If this function is not identically zero, the separatrices split. Moreover, the non-degenerate zeros of this function are associated to transversal intersections of the perturbed invariant (stable and unstable) manifolds. There exists a similar theory for planar maps, and in this case the Melnikov function is not a continuous sum anymore, but an infinite and (a priori) analytically uncomputable (discrete) sum. In a previous work, we have given a method to compute explicitly this kind of sums in terms of elliptic functions, under hypotheses of meromorphicity over the functions in the sum. This method allows us to obtain a strong non-integrability criterion and to apply it to perturbations of elliptic billiards and integrable standard-like maps like the McMillan map. Explicit estimates of the splitting angles are also given. Our aim is extend this method to the study of the splitting of doubly asymptotic manifolds (separatrices) associated to hyperbolic fixed points of twist maps in arbitrary dimensions. We work with maps generated globally by a generating function. Using the variational principle satisfied by these maps, we associate the non-degenerated critical points of a scalar function (here called Melnikov potential) to the transversal intersections of the perturbed asymptotic manifolds. We want to stress the difference of this point of view with the usual one in the literature, that is based in the study of non-degenerated zeros of a vectorial function. The symplectic structure and the variational principle play a fundamental role in our construction. As a first example where this theory can be applied, we study standard-like perturbations of a $2d$-dimensional twist map given by~R. McLachlan, for $d\ge 2$. This map is a multidimensional generalization of the McMillan map. We prove, among other results, that any entire perturbation destroys the separatrix of the McLachlan map.

Read accessible full text

Poincaré-Melnikov-Arnold method for twist maps

Author: Delshams Valdés, Amadeu,Ramírez Ros, Rafael
Year: 1997
Source: https://upcommons.upc.edu/bitstream/2117/860/1/9709delsh.pdf
POINCAR

E-MELNIKOV-ARNOLD METHOD FOR TWIST MAPS
AMADEU DELSHAMS AND RAFAEL RAMREZ-ROS
1. In o duc ion
A gene al heo y o p e u ba ions o an in eg able plana map wi h a sepa a ix o a
hyp e b olic xed p oin has been de elop ed in a p e ious lec u e 5]. The spli ing o he
p e u b ed in a ian cu es was measu ed, in  s o de wi h esp ec o he pa ame e o
p e u ba ion, by means o a p e io dic
Melniko unc ion
M
dened on he unp e u b ed
sepa a ix. In he case o plana wis maps,
M
has ze o mean and he e o e he e exis s a
p e io dic unc ion
L
(called he
Melniko po en ial
) such ha
M
=
L
0
. Consequen ly,i
L
is no iden ically cons an ( esp ec i ely, has non-degene a e c i ical p oin s), he sepa a ix
spli s ( esp ec i ely, he p e u b ed cu es c oss ans e sely).
The aim o his lec u e is o p esen a simila heo y o mo e dimensions. The na u al
ame is o conside wis maps on co angen bundles. Once he sui able deni ion o
unp e u b ed sepa a ix has b een in o duced (a non- i ial p oblem in he high-dimensional
case), a scala unc ion
L
can be dened on i , in such a way ha
L
e ies he same
p op e ies han in he plana case. The de i a ion o
L
is easily ela ed o a ia ional
p inciples, and he p op e y o b eing a scala unc ion ins ead o a ec o ial unc ion like
he classical Melniko unc ion, makes i mo e use ul o compu a ions and geome ical
unde s anding. E en mo e, i allows he applica ion o Mo se heo y o es ablish he
minimal numb e o ans e se homo clinic o bi s.
The esul s o be p esen ed in his lec u e a e alid o exac symplec ic maps on
a bi a y exac symplec ic mani olds, ha is, he wis cha ac e is no essen ial. Weha e
es ic ed ou sel es o wis maps only o simplici y. Full de ails o he ideas p esen ed
he e a e con ained in 4], whe e ano he mo e gene al si ua ion (i.e., he exac symplec ic
case) is s udied. Rela ed ideas can b e ound in 1, 15, 14, 8, 7].
2. The maps
A
wis map
F
is a map om a connec ed subse
U
o he co angen bundle o a mani old
M
(which can be non-compac ) in o
U
, which comes equipp ed wi h a wis gene a ing
unc ion
L
:
MM !
R
ha sa ises
F

(
y
d
x
)
;
y
d
x
=
Y
d
X
;
y
d
x
= d
L
(
x X
)

(
X Y
)=
F
(
x y
)

whe e (
x y
) a e any co angen co o dina es on
T

M
, ha is,
x
a e co o dina es on
M
,
ex ended o co o dina es (
x y
)in heob ious way. The symplec ic o m
!
0
on
T

M
eads
2
as
!
0
= d
x
^
d
y
in co angen co o dina es. This can also be w i en in a co o dina e ee
manne . Gi en
L
, one can e ie e he map (a leas implici ly) om
y
=
;
@
1
L
(
x X
), and
Y
=
@
2
L
(
x X
). This can b e done globally (i.e.,
U
=
T

M
) only when
M
is dieomo phic
o ab e o
T

M
, o example when
M
is he co e ing space o
T
n
o a mani old o
cons an nega i e cu a u e.
Finally, le us deno e by

:
T

M!M
he canonical p o jec ion.
3. The heo y
Assume now ha we a e gi en a smo o h wis dieomo phism
F
0
on he co angen bundle
T

M
. Le
L
0
be i s wis gene a ing unc ion. We assume ha he e exis s a
hype bolic
xedpoin
z
1
0
o
F
0
,such ha i s
n
-dimensional (uns able and s able) in a ian mani olds
W
u

s
0
a e
doubled
, ha is, hey coincide:
W
:=
W
u
0
=
W
s
0
.
In he plana case, he
sepa a ix
consis s o he in e sec ion o he in a ian cu es,
excep o he hyp e b olic xed p oin , which is he only p oin whe e suchin a ian mani olds
a e no submani olds o he co angen bundle. In he high-dimensional case, he si ua ion
is mo e complica ed. We can conside h ee op ologies on he se
W
: he one induced by
he inclusion
W 
T

M
, and he wo ones induced by he inclusions
W  W
u

s
0
. We
 s dene he
bi u ca ion se
 o his p oblem as he subse o
W
o med by he p oin s
such ha he h ee op ologies do no coincide. Then, he
sepa a ix
 is dened as i s
complemen a y in
W
, i.e.,
:=
Wn

:
Wi h his deni ion, i u ns ou ha 
is
a doubly asymp o ic exac submani old o
T

M
,in a ian by
F
0
.
Nex , conside a p e u b ed wis map
F
"
, and le
L
"
=
L
0
+
"
L
1
+
O
(
"
2
) b e he wis
gene a ing unc ion o
F
"
.Fo 0
<
j
"
j
1, he e exis s a hyp e b olic xed p oin
z
1
"
o
F
"
,
close o
z
1
0
, and i is no es ic i e o no malize he wis gene a ing unc ion by imp osing
L
"
(
x
1
"
x
1
"
)=0, whe e
x
1
"
=

(
z
1
"
). In pa icula ,
L
1
(
x
1
0
x
1
0
) = 0, whe e
x
1
0
=

(
z
1
0
).
Wenow dene he
Melniko po en ial
in he same way as o plana wis maps 5]:
L
:
;!
R
 L
(
z
)=
X
k
2
Z
L
1
(
x
k
x
k
+1
)
 x
k
=

(
z
k
)
 z
k
=
F
k
0
(
z
)
 z
2

:
(1)
The Melniko heo y is based on he ollowing p op e ies o he Melniko p o en ial 4]:
;
L
:
!
R
is well-dened, smo o h and in a ian unde he ac ion o he unp e u b ed
map:
L

F
0
=
L
. Consequen ly,
L
canbedenedon he
educedsepa a ix


=
=F
0
.
;
The die en ial o he Melniko po en ial
M
= d
L
(called he
Melniko unc ion
),
measu es, in  s o de in
"
, he dis ance b e ween he p e u b ed in a ian mani olds,
and is also dened on he educed sepa a ix 

.
;
I
L
6
cons an , hen he p e u b ed in a ian mani olds
W
u

s
"
spli o 0
<
j
"
j 
1,
i.e., hey do no coincide.
;
I
L
has a c i ical p oin a
z
=
z
0
hen, o 0
<
j
"
j
1,
W
u

s
"
in e sec ans e sally
on a homo clinic p oin nea
z
0
.
;
I he unp e u b ed in a ian mani olds a e
comple ely doubled
(i.e., =
z
1
0
g
), he
educed sepa a ix is a
compac
n
-dimensional mani old wi hou b ounda y. Ac ually,
i

deno es he sign o he p o duc o he eigen alues wi h mo dulus g ea e ha one
o
DF
0
(
z
1
0
), and
S
n
;
1
s ands o he uni sphe e o
R
n
, hen 

is homeomo phic o
3
S

S
n
;
1
o

= +, and i is somewha mo e complica ed o

=
;
( o mo e de ails,
see 4]).
In his si ua ion, Mo se heo y applied o he Melniko po en ial, hough as a unc ion
o e 

,gi es he minimal numb e o ans e se homo clinic o bi s, unde condi ions
o gene ic p osi ion.
;
The e exis s a
a ia ional p inciple
,in an analogous way o he one o he plana
case 9, 6], which es ablishes ha he homo clinic o bi s o a wis map wi h wis
gene a ing unc ion
L
a e he ex emals o he
homoclinic ac ion
W

O
]:=
X
k
2
Z
L
(
x
k
x
k
+1
)

O
=(
x
k
)
k
2
Z

and a
homoclinic a ea
can b e dened o e e y pai o homo clinic o bi s
O
=(
x
k
)
k
2
Z
,
O
0
= (
x
0
k
)
k
2
Z
,and is gi en by he die ence o homo clinic ac ions 
W

O

O
0
] =
W

O
0
]
;
W

O
]. In e ms o he Melniko p o en ial, he e is also a nice exp ession o
he homo clinic a ea:

W

O

O
0
]=
"
;
L
(
z
0
0
)
;
L
(
z
0
)

+
O
(
"
2
)
:
We nish his su ey o esul s wi h wo ema ks ab ou die en , bu ela ed, se ings.
1. Rega ding Hamil onian ows, le
H
"
:
T

M
R
!
R
b e a ime-p e io dic Hamil onian
o p e io d
T
,and
F
"
= 
T
"
unde he condi ions o his sec ion, whe e 
"
(
z
) is he
solu ion o he asso cia ed Hamil onian equa ions wi h ini ial condi ion
z
a
=0. I
H
=
H
0
+
"H
1
+
O
(
"
2
), one can see 4] ha he Melniko po en ial akes he o m
(al eady known o Poinca e)
L
(
z
)=
;
Z
R
H
1
(
0
(
z
)

)d

whe e
H
1
is de e mined by imp osing
H
1
(
0
(
z
1
0
)

)

0, o simply
H
1
(
z
1
0

)

0, i
H
0
is au onomous.
2. In ela ion wi h a non-symplec ic se ing, le us assume now ha
M
=
R
n
, ha is,
T

M
=
R
2
n
. In ha case, using a die en poin o iew 13, 2], one can assume
ha he unp e u b ed map
F
0
:
R
2
n
!
R
2
n
p ossesses
n
indep enden  s in eg als
H
1
:::H
n
on he sepa a ix  (no necessa ily in in olu ion, since his concep e-
qui es a symplec ic s uc u e), and conside a p e u ba ion
F
=
F
0
+
"F
1
+
O
(
"
2
)
(no necessa ily symplec ic). Then, he Melniko ec o ial unc ion
M
:
!
R
n
can
b e w i en as (compa e wi h he plana case in 5]):
M
=(
M
1
::: M
n
)
>
 M
j
(
z
)=
X
k
2
Z
h
H
j
(
z
k
+1
)
F
1
(
z
k
)
i
 z
k
=
F
k
0
(
z
)
 z
2

:
4. The example
Le us conside
cen al
s anda d-like maps on
R
2
n
=
T

R
n
, ha is,
F
0
(
x y
)=(
y
;
x
+
V
0
(
y
)) o
L
0
(
x X
)=
;h
x X
i
+
V
0
(
X
) (2)
whe e
V
0
(
x
)=
V
c
(
k
x
k
2
) o some unc ion
V
c
:0

1
)
!
R
. Then, he angula momen a"
A
ij
(
x y
)=
x
i
y
j
;
x
j
y
i
a e  s in eg als and he (
n
+ 1)-dimensional mani old in
R
2
n
o ze o
angula momen a is
A
n
+1
0
:=
(
x y
):
A
ij
(
x y
)=0
g
=
(
qa pa
):
a
2
S
n
;
1

(
q p
)
2
R
2
g
.
4
We nowin o duce he
educed map
in
A
n
+1
0
o
F
, as he plana s anda d-like map
:
R
2
!
R
2
dened by
(
q p
)=(
p
;
q
+2
V
0
c
(
p
2
)
p
). We no e ha
(
q p
)=(
Q P
)
()
F
(
qa pa
)=(
Qa P a
)

8
(
q p
)
2
R
2
a
2
S
n
;
1

(3)
so ha he non- i ial dynamics on he sepa a ix is induced by he educed map.
To akead an age o he esul s o plana wis maps in he lec u e 5], wein o duce
now he McLachlan map 10] as he cen al s anda d-like map wi h po en ial
V
0
(
y
) =

ln(1 +
k
y
k
2
),
>
1. I has he exp ession
F
0
(
x y
)=

y
;
x
+
2
y
1+
k
y
k
2
!
>
1
:
(4)
I s educed map is no hing else bu he McMillan map 11] whose sepa a ix ; = ;
+
has
he ollowing na u al pa ame e iza ion 3]:
;=
z
0
(
)=(
q
0
(
)
p
0
(
))
g
 q
0
(
)=
p
0
(
;
h
)
 p
0
(
) = sinh
h
sech

whe e
h>
0 is de e mined by he equa ion
cosh
h
=
:
Now, i is easy o check ha
1. The o igin is a hyp e b olic xed p oin o
F
0
, and Sp ec
DF
0
(0)] =
e

h
g
.
2. The in a ian mani olds o
F
0
a e comple ely doubled, and he sepa a ix is gi en by
=
(
qa pa
):(
q p
)
2
;
a
2
S
n
;
1
g
:
3. The unc ion
z
0
:
R

S
n
;
1
;!
gi en by
z
0
(
 a
)=(
p
0
(
;
h
)
a p
0
(
)
a
)
 p
0
(
)=sinh
h
sech

(5)
is a
na u al pa ame e iza ion
o he sepa a ix, i.e.,
z
0
is a dieomo phism ha sa ises
F
0
(
z
0
(
 a
)) =
z
0
(
+
h a
), o
2
R
and
a
2
S
n
;
1
.
As exp ec ed, wenow conside a gene al p e u ba ion o (4) ha p ese es he s anda d
cha ac e , i.e.,
F
"
(
x y
)=

y
;
x
+
2
y
1+
k
y
k
2
+
"
V
(
y
)
!
>
1
"
2
R

(6)
wi h
V
:
R
n
!
R
de e mined by imp osing
V
(0) = 0. The gene a ing unc ion o
F
"
ha
anishes a he o igin is
L
"
=
L
0
+
"
L
1
, whe e
L
0
(
x X
)=
;h
x X
i
+

ln(1 +
k
X
k
2
) and
L
1
(
x X
)=
V
(
X
). The Melniko po en ial is simply
L
:
R

S
n
;
1
!
R
 L
(
 a
)=
X
k
2
Z
V
(
p
0
(
+
hk
)
a
)
 p
0
(
)=
sinh
h
cosh
:
(7)
Since
L
is
h
-p e io dic in
( his is he in a iance o he Melniko po en ial unde he
ac ion o he unp e u b ed map), wecan conside
dened mo dulo
h
, i.e.,
L
dened o e
he educed sepa a ix
S
1

S
n
;
1
.
Rep ea ing he a gumen s o he case o he plana wis maps 5], we see ha i
V
is a
non-cons an eal en i e unc ion, hen
V
(
p
0
(
)
a
) has he same isola ed singula i ies in he
complex a iable
as
p
0
(
), and i is no di cul o check ha hey emain as singula i ies
o he Melniko p o en ial, whichmus b e non-cons an . In his waywe ha e es ablished
he ollowing esul .
5
Theo em 1
I
V
is a non-cons an eal en i e unc ion, hen he pe u bed in a ian man-
i olds o he s anda d-like map (6) spli , o
0
<
j
"
j
1
.
Re e ences
1. S.V. Bolo in. Homo clinic o bi s o in a ian o i o Hamil onian sys ems. P ep in , 1994. To app ea
in Ad . So . Ma h.
2. T.C. Boun is, A. Go iely, and M. Kollmann. A Melniko ec o o
n
-dimensional mappings.
Phys.
Le . A
, 206:38{48, 1995.
3. A. Delshams and R. Ram ez-Ros. Poinca e-Melniko -A nold me ho d o analy ic plana maps.
Nonlinea i y
, 9(1):1{26, 1996.
4. A. Delshams and R. Ram ez-Ros. Melniko po en ial o exac symplec ic maps. P ep in , May
1996. To app ea in
Comm. Ma h. Phys.
5. A. Delshams and R. Ram ez-Ros and T.M. Sea a. Spli ing o sepa a ices in Hamil onian sys ems
and symplec ic maps. In Simo 12].
6. R.W. Eas on. T anspo h ough chaos.
Nonlinea i y
, 4:583{590, 1991.
7. H.E. Lomel. Saddle connec ions and he e o clinic o bi s o s anda d maps.
Nonlinea i y
, 9:649{668,
1996.
8. H.E. Lomel. Applica ions o he Melniko me ho d o wis maps in highe dimensions using he
a ia ional app oach. P ep in , Uni . o Minneso a, 1993. To appea in E go dic Theo y Dynamical
Sys ems.
9. R.S. MacKay, J.D. Meiss, and I.C. Pe ci al. T ansp o in Hamil onian sys ems.
Phys. D
, 13:55{81,
1984.
10. R.I. McLachlan. In eg able ou {dimensional maps o s anda d yp e.
Phys. Le . A
, 177:211{214,
1994.
11. E.M. McMillan. A p oblem in he s abili y o p e iodic sys ems. In E. B i in and H. Odabasi, edi o s,
Topics in mode n physics, a ibu e o E.V. Condon
, pages 219{244. Colo ado Asso c. Uni . P ess,
Boulde , CO, 1971.
12. C. Simo, edi o .
Hamil onian Sys ems wi h Th ee o Mo eDeg ees o F eedom
.NATO Ad . Sci. Ins .
Se . C Ma h. Phys. Sci. Held in S'Aga o, Spain, 19{30 June 1995. Kluwe Acad. Publ., Do d ech ,
Holland, o app ea in 1997.
13. J.H. Sun. Melniko ec o unc ion o high-dimensional maps.
Phys. Le . A
, 216:47{52, 1996.
14. E. Tabacman. Va ia ional compu a ion o homo clinic o bi s o wis maps.
Phys. D
, 85:548{562,
1995.
15. D.V. T esche . Hyp e bolic o i and asymp o ic su aces in Hamil onian sys ems.
Russ. J. Ma h.
Phys.
, 2(1):93{110, 1994.
In e ne access:
All he au ho s' quo ed and ela ed p ep in s a e a ailable a
h p://www-ma1.upc.es
in
he
P ep in s
pages, o a
p:// p-ma1.upc.es
,in he
pub/p ep in s
di ec o y.