Conley index at infinity [Elektronische Ressource] / vorgelegt von Juliette Hell

Sujets

• Mathematics

Informations

• Publié par: freie_universitat_berlin
• Publié le: 01 janvier 2010
• Nombre de lectures: 11
• Langue: English
• Poids de l'ouvrage: 1 Mo

Extrait

Inauguraldissertation
zur Erlangung des Grades eines
Doktors der Naturwissenschaften
am Fachbereich Mathematik und Informatik
der Freien Universit¨at Berlin
Conley Index at Infinity
vorgelegt von
Juliette Hell
Berlin, 20102
Betreuer und Erstgutachter: Prof. Dr. B. Fiedler
Zweitgutachter: Prof. K. Mischaikow
Tag der Disputation: 9. November 2009i
Le silence ´eternel de ces espaces infinis m’effraye.
Blaise Pascal
Le vacarme intermittent de ces petits coins me rassure.
Paul Val´eryiiContents
Introduction iii
1 The Bendixson compactification 1
1.1 Description of the transformation . . . . . . . . . . . . . . . . . . 1
1.2 Normalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Some examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.4 Some bad news . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2 The Poincar´e compactification 13
2.1 Description of the transformation . . . . . . . . . . . . . . . . . . 13
2.2 Normalization and time rescaling . . . . . . . . . . . . . . . . . . 16
2.3 Some examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Poincar´e versus Bendixson . . . . . . . . . . . . . . . . . . . . . . 21
3 Conley index: classical and at infinity 23
3.1 Classical Conley index methods . . . . . . . . . . . . . . . . . . . 24
3.1.1 Basic definitions and properties of the Conley index . . . . 24
3.1.2 Conley Index on a Manifold with Boundary . . . . . . . . 34
3.1.3 Attractor–repeller decompositions . . . . . . . . . . . . . . 39
3.1.4 Morse decompositions and connection matrices . . . . . . . 44
3.2 Conley index and heteroclines to infinity . . . . . . . . . . . . . . 55
3.3 Some examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.4 Conley index and duality . . . . . . . . . . . . . . . . . . . . . . . 61
3.5 The Conley index at infinity . . . . . . . . . . . . . . . . . . . . . 63
3.5.1 Under Bendixson compactification . . . . . . . . . . . . . . 63
3.5.2 Under Poincar´e compactification . . . . . . . . . . . . . . 83
3.6 Limitations and properties . . . . . . . . . . . . . . . . . . . . . . 101
4 Ordinary differential equations 103
4.1 Generalities on polynomial vector fields . . . . . . . . . . . . . . . 103
4.1.1 Classification results . . . . . . . . . . . . . . . . . . . . . 103
4.1.2 Critical points at infinity . . . . . . . . . . . . . . . . . . . 104
4.2 Gradient vector fields . . . . . . . . . . . . . . . . . . . . . . . . . 115
iiiiv CONTENTS
4.3 Hamiltonian vector fields . . . . . . . . . . . . . . . . . . . . . . . 117
4.3.1 Generalities on Hamiltonian vector fields . . . . . . . . . . 117
4.3.2 Planar quadratic Hamiltonian vector fields . . . . . . . . . 118
4.4 A cube at infinity . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
4.4.1 A cubic structure in the sphere at infinity . . . . . . . . . 124
4.4.2 Finite dynamic without lower order terms . . . . . . . . . 126
4.4.3 Finite dynamic with discrete Laplacian operator . . . . . . 128
4.5 The Lorenz equations . . . . . . . . . . . . . . . . . . . . . . . . . 130
5 Partial differential equations 133
5.1 Chafee–Infante structure at infinity . . . . . . . . . . . . . . . . . 133
5.2 Case of a sublinear non–linearity . . . . . . . . . . . . . . . . . . 137
5.3 Example of a bounded non–linearity . . . . . . . . . . . . . . . . 138
5.4 Abstract polynomial PDE . . . . . . . . . . . . . . . . . . . . . . 140
5.5 Blow–up and Similarity variables . . . . . . . . . . . . . . . . . . 142
5.5.1 Philosophy of the Similarity Variables . . . . . . . . . . . . 143
5.5.2 Power Nonlinearity . . . . . . . . . . . . . . . . . . . . . . 147
5.5.3 Parabolic Scalar Curvature Equation . . . . . . . . . . . . 150
Conclusions 155
Appendix 157Introduction
In this thesis we consider grow up and blow up phenomena (in forward or in
backward time direction) and interpret them as heteroclinic connections between
finite invariant sets and infinity. Under this point of view we formulate the
following question: Which bounded invariant sets admit heteroclinic connections
to infinity? There already exists methods which where developed for the analysis
of bounded global attractors. Those arise in dissipative systems, which is in fact
the assumption that we want to get rid of.
To adapt those methods for the analysis of heteroclinics to infinity and de-
scribe a non bounded attractor, we propose to make use of so called “compactifi-
cations”. AcompactificationitheprojectionofaHilbertspaceX ontoabounded
Hilbert manifold. If the space X is eventually infinite dimensional, the resulting
Hilbert manifold is bounded but not compact because of its being infinite dimen-
sional. Thereforetheword“compactification” isnotexact inthiscontext, butwe
keep it for historical reasons. Compactifications were introduced already at the
beginning of the theory of dynamical systems on the one hand under the name
of Bendixson compactification, on the other hand by Poincar´e in [31]. Those
compactifications where first introduced to compactify dynamics on the plane,
but we show in the first two chapters that they may be formulated for arbitrary
Hilbertspaces. TheBendixson compactificationisnothing morethanaonepoint
compactification where infinity is projected on the north pole of the Bendixson
sphere by a stereographic projection. The north pole or “point at infinity” is,
in many cases, so degenerated that one has to circumvent this degeneracy. This
may be achieved throughthe Poincar´e compactification. This compactification is
based ona central projection andmaps infinity onto a whole “sphere at infinity”.
However in some cases the degeneracy at infinity resists this procedure.
Poincar´e andBendixson gavealso theirnames tothefamoustheoremdescrib-
ing the longtime dynamic of planar vector fields. A globally bounded trajectory
accumulates on its ω–limit set which is a connected compact invariant set. In
the case of planar vector fields, the Poincar´e-Bendixson theorem guaranties that
ω–limit sets are one of the three following types:
1. an equilibrium,
2. a periodic orbit,
vvi INTRODUCTION
3. or a heteroclinic cycle
When the dimension of the phase space grows bigger, it is not possible to classify
those invariant sets which are crucial for the long time dynamic. Already in di-
mension three, strangeattractorsmay arise such asin the famousLorenz system.
Anattempttoanalysesophisticatedinvariantsetsandtheirinterconnectionsmay
be done with the help of the Conley index theory.
This theory was invented by Conley in the 60’s, and further developed until
today for example by Franzosa, Mischaikow or Mrozek. . The Conley index does
notdealdirectlywiththeinvariantsetsbutwithaneighbourhoodisolatingthem.
This index somehow draws a balance between the trajectories beginning in this
neighbourhood and the trajectories leaving it. If those are in balance, i. e. ev-
erything which starts in the neighbourhood also leaves it, then the Conley index
is “trivial” in the sense that it coincides with the Conley index of the empty in-
variant set. On the contrary, a neighbourhood giving rise to a non trivial Conley
index admits a non trivial invariant set in its interior. This caricature of the
Conley index shows that this tool is able to detect significant invariant sets. Fur-
thermore the Conley index theory utilizes algebraic topology in structures called
connection matrices, which are able to detect heteroclinic connections between
isolated invariant sets.
In this thesis we will combine both compactifications of the phase space and
Conley index theory. The compactifications allow us to materialize invariant
sets at infinity, which are out of reach in an unbounded phase space. On these
invariant sets at infinity we apply the Conley index methods so that heteroclinic
connections between these sets and bounded invariant sets are put into light.
Although the global idea of this strategy seems clear, one encounters many
obstacles on the way to its completion. The main obstacle was the motivation
for the most demanding part of this thesis and concerns the degeneracy of the
dynamical behaviour at infinity. Even for planar quadratic vector fields, the
invariant sets at infinity are likely to be degenerate in the sense that they are not
isolated invariant - hence out of reach for Conley index methods.
OurcontributionisthedevelopmentofaConleyindexforaclassofdegenerate
invariant sets at infinity that we denote by “invariant sets at infinity of isolated
invariant complements”. An invariant set S at infinity belong to this class if,
roughly speaking, there exists an isolated invariant set R bounded away from S
whose isolating neighbourhood may be chosen arbitrarily close toS. The precise
definition is given in 3.5.1 and 3.5.30. An equilibrium in the compactification of
2the plane R exhibiting only elliptic sectors is an example of an invariant sets
with isolated invariant complement.
Our main result consists on showing that the algebraic machinery of the con-
nection matrices extends to invariant sets at infinity with isolated invariant com-
plements. Hence heteroclinicconnections tothistypeofdegenerateinvariant sets
at infinity can bedetected by this generalized Conley index theory. This enlargesvii
significantly thehorizonofthestudyofthebehaviouratinfinityviaConleyinde