Internship proposals in mathematical physics,
Propositions de stage en physique-mathématique
Master 1, Master 2, Thèses.

Supervision by: Frédéric Faure.
Discipline: mathematical physics
Dates and Place: Spring to July, at institut Fourier, Grenoble.
Level: Master 1 or Master 2
 
For contact send me an email: frederic.faure@univ-grenoble-alpes.fr
 
Warning: To read this document, use Firefox which gives a good rendering of the equations using MathML.

Table of Contents

Part I Dynamical systems
Part II Microlocal Analysis
Part III Music and Mathematics
 
Grant for a PHD Thesis. Grant for a PHD Thesis.

Part I Dynamical systems

1 Spectral approach to classical and quantum chaos

1.1 Description

A model of deterministic chaos is a dynamical system which has a sensitivity to initial conditions, i.e. two neighboring trajectories separate at an exponential rate either in the future or in the past. This property called Anosov or hyperbolicity implies a “chaotic” behavior of the trajectories: they seem almost all unpredictable. In order to understand these patterns and nevertheless to find predictive laws, one must adopt a probabilistic approach to dynamics: study the evolution of probability distributions.
See for example this vidéo, showing the evolution of a "chaotic" individual trajectory, and this vidéo showing the evolution of a probability distribution, converging to a "steady state".
If f:MM is the dynamical system, we have to study theopérateur de transfert L : C ( M ) C ( M ) defined by L u:=uf . Its eigenvectors u j defined by L u j = λ j u j with λ j C , | λ j | 1 , are invariant distributions The eigenvalues are called Ruelle resonances. For a class of very chaotic models ("Axiom A") we show that the spectre of L is discrete (in adapted functional spaces) and that only the eigenvalue λ 0 =1 is on the unit circle, associated to a spectral projector Π 0 =|1 v 0 | . The other eigenvalues are | λ j | <1 . This has the consequence that for long times λ j n n 0 , and thus any initial distribution converges to the steady state v 0 . The rate of convergence and the transient regime is controlled by the rest of the spectrum.

1.2 Objective

We consider simple and accurate models of chaotic dynamics. The goal is to compute the Ruelle spectrum numerically and to observe the distribution of the eigenvalues. We can conjecture that there is a spectral gap and try to demonstrate it. Eventually we can observe the "equipartition of the eigenspaces" and state a conjecture of type "quantum ergodicity". One could observe a band structure of the Ruelle spectrum and compare it to the spectrum of the quantum model.
The proposed models are
  1. An S 1 extension with roof function τ ( x ) of the cat map ( 2 1 1 1 ) .
  2. Anosov flow obtained from a suspension with variable time of the cat map model.
In each of these models, we will start by realizing a program which computes and shows the spectrum ( λ j ) j . In parallel we will compute the properties of this dynamics (partially Anosov, regularity of the foliage, mixing, ergodicity,..) and the captive set. We will show that there is a discrete spectrum. We will try to prove an asymptotic spectral gap.
The student will also be asked to contribute to wikipedia on this subject.

1.3 After

the internship can be continued in a thesis, on the theme of classical chaos and quantum chaos.
Prerequisites:
Basic knowledge of spectral theory and differential geometry. Optionally, basic notions in dynamical systems theory (hyperbolic dynamics), semi-classical analysis. Motivation for physics-mathematics, and optionally motivation for numerical experimentation (e.g. C++ languages). The purpose of numerical experimentation is to illustrate the results obtained, to test new ideas or to explore new research directions.
Documentation:

1.4 Indications to start

  1. Define
    1. an extension with roof function z= τ ( x ) of fS L d Z hyperbolic. Ex: f=( 2 1 1 1 )S L 2 Z .
    2. Similarly for a flow. See suspension à temps variable de l'application du chat d'Arnold.
  2. Find examples of matrices fS L 3 Z hyperbolic and study the regularity of stable/unstable distributions. For this, try ( 1 ± 1 0 1 0 0 1 )S L 3 Z ex: ( 1 1 0 1 )( 1 0 1 1 )=( 2 1 1 1 ) .
  3. In case of a diffeomorphism (a),
    1. Let V C ( T d × T 1 ) , «potential function». Define the transfer operator L acting on C ( T d+1 ) and the reduced operator L N acting on C ( T d ) .
    2. Compute matrix elements of L N in Fourier basis. Tronquer la matrice et diagonaliser numériquement (taille M × M ). Dessin du spectre, et vérifier qu'il est stable par rapport à M .
    3. Calculer l'application symplectique T sur T * T d associée à l'opérateur de transfert L N . Identifier l'ensemble captif K . Calculer la régularité de l'ensemble captif K . Déterminer les variétés stables/instables de tout point ρ K .
    4. Par différentes methodes, montrer que L N a du spectre discret de Ruelle intrinsèque dans des espaces de Sobolev adaptés:
      1. Par les Opérateurs pseudo-différentiels (PDO) (article [13]).
      2. Par la transformation par paquets d'ondes (ou FBI), utilisant Lecturenotes,[11].
    5. Application: decroissance des fonctions de correlation à N fixé.
    6. Formule de trace d'Atiyah Bott. Relier le spectre de Ruelle de L N aux orbites fermées de f .
    7. Suivre la même démarche que avec Tobias: forme normale globale sur l'ensemble de base et sur l'ensemble captif. Expansion asymptotique et déduire une estimation du gap spectral asymptotique (pour N1 ).
    8. Application: Comptage d'orbites avec poids: N( T ) := n1 x= f n x  t.q.  l( γ x ) T 1 avec l( γ x ) := k=1 n τ ( f k ( x ) )
    9. Généralisation autant que possible des questions précédentes: au cas f:MM difféomorphisme Anosov quelconque.
  4. Dans le cas du flot (b), identifier la variété M de dimension dim M=d+1 . Expliciter une base orthonormée (ou pas) de L 2 ( M ) . Faut il considérer un opérateur Δ associé à une métrique interpolée: g( x,y,z ) = g xx ( z ) d x 2 + g yy ( z ) d y 2 + g xy ( z ) dxdy+d z 2
    1. Calculer les éléments de matrice du générateur dans cette base.
    2. Comptage d'orbites: N( T ) :={ p.o. γ  avec repetitions s.t.  | γ | T }
  5. Relation entre les spectre du diffeomorphisme (a) et du flot (b)? (i.e. comprendre la relation générale entre le spectre d'une Section de Poincaré et le spectre du flot)
  6. Essayer de discuter la possibilité de conjuguer l'opérateur de transfert à un opérateur sur H := L 2 ( T x, 1 × T z 1 ) en moyennant sur des «fibres».

2 Study of quantum chaos. Pre-quantum approach

2.1 Description

In the 70's, Souriau, Kostant and Kirillov laid the foundations of geometric quantization, which is a geometric construction of quantum dynamics, starting from classical dynamics (a Hamiltonian flow on a phase space). In the steps of the construction, they define an intermediate dynamics between classical and quantum that they call "pre-quantum", which is the evolution of wave functions on the phase space. We consider a classical "hyperbolic" dynamics, where all trajectories are unstable. The dynamics then has "chaotic" properties. An important field of research called "quantum chaos" is the study of the corresponding quantum dynamics (dynamic or spectral properties). For this, it seems natural to study also the pre-quantum dynamics.
We consider a simple model of hyperbolic (chaotic) dynamics: a linear hyperbolic application on the T 2 torus, called "Arnold's cat map". The corresponding quantum and pre-quantum models are well established. For quantum dynamics, there is a "quantum ergodicity" theorem which states that almost all eigenmodes become "equidistributed" on the torus in the semiclassical limit (i.e. when the Chern index of the fibric tends to infinity). We also know examples of "non unique ergodicity": eigenmodes which are partially localized on unstable periodic orbits.

2.2 Objectives

Inspired by the quantum ergodicity theorem described above, the objective is to establish an analogous theorem in the pre-quantum case, for the model considered, and to discuss the non unique ergodicity.

2.3 After

the internship can be continued in a thesis, on the theme of classical chaos and quantum chaos.

2.4 Prerequisite:

Basic knowledge of spectral theory and differential geometry. Optionally, basic notions in dynamical systems theory (hyperbolic dynamics), theory of complex line fibers on a Riemann surface, with connection. Motivation for physics-mathematics, and motivation for numerical experimentation (e.g. C++ or python languages). The purpose of numerical experimentation is to illustrate the results obtained, to test new ideas or to explore new research directions.

2.5 Documentation

3 Ruelle spectrum of chaotic dynamics and emergence of quantum dynamics. Theoretical analysis and numerical experiments.

3.1 Description

The general motivation of this subject is to study the long time evolution of smooth probability distributions under a deterministic but chaotic classical dynamics (uniform hyperbolic dynamics). The probability distribution converges towards a measures called equilibrium or (SRB measure). We are interested to study the small fluctuations around this limit. It has been shown that they are governed by an “effective quantum dynamics”. This phenomenon called “dynamical emergence of quantum dynamics” manifests itself in the Ruelle spectrum of the classical dynamics: the spectrum has a band structure [6, 10, 9, 12].

3.2 Objectives

The objective of this Internship is to understand this mechanism in a simple model, the “cat map” and the “perturbed cat map” and also to establish a rigorous and precise numerical algorithm to compute the Ruelle spectrum of the first band, namely of the emerging quantum operator, and finally to compare with the spectrum of an ordinary quantum operator (from Weyl quantization).
A natural extension of this work is to consider next another specific model of chaotic dynamics that is the geodesic flow on a negatively curved Riemannian manifold. A specific example being a surface with constant curvature κ =-1 .
The student will also be asked to contribute to wikipedia on this subject.

3.3 Prerequisites:

Basic knowledge of spectral theory and differential geometry. Optionally, basic notions in dynamical systems theory (hyperbolic dynamics), semi-classical analysis. Motivation for physics-mathematics, and a good motivation for numerical experimentation (e.g. C++, python or other languages adapted to scientific programming). The purpose of numerical experimentation is to illustrate the results obtained, to test new ideas or to explore new research directions.

3.4 Documentation:

4 Fractal Weyl law for Axiom A flows

Extend to Axiom A flows the results of https://arxiv.org/abs/1706.09307 that concerns Anosov flows.

5 Ruelle spectrum of Baker map and prequantum Baker map

Extend to the Baker map the results of https://arxiv.org/abs/nlin/0606063 that concerns the cat map.

6 Ruelle spectrum of Pseudo Anosov maps

Extend to the Pseudo Anosov maps the results of https://arxiv.org/abs/0802.1780 that concerns Anosov maps.

Part II Microlocal Analysis

7 Quantum revival in the wave equation

There is some strange quantum revival in the model of quantum cat map. Study if this can append in the model of the wave equation.

8 Spectrum in SU(2)

Consider two elements g 1 , g 2 SU ( 2 ) . We denote g 1 , g 2 : C ( SU ( 2 ) ) C ( SU ( 2 ) ) the pull back operator. Study the spectrum of the operator g 1 + g 2 in particular its spectral gap.

9 Fractal Weyl law in a simple model

Consider a a Hölder continuous function f:x S 1 ξ R of exponent 0< α 1 . Study the spectrum of the operator Op ( 1 { ξ f( x ) } ) , in particular the density of eigenvalues in ( 0,1 ) in terms of the exponent α .

10 Sub-Riemaniann Laplacian and quantization of contact structures

Study the possibility of doying geometric quantization of contact structures.

Part III Music and Mathematics

11 Develop a software to perform in jazz improvisation with just intonation

See the project Music on the tonnetz.

References

1N. Anatharaman, "Le théoréme d'ergodicité quantique", (2014).
2V.I. Arnold, Geometrical methods in the theory of ordinary differential equations (Springer Verlag, 1988).
3V.I. Arnold and A. Avez, Méthodes ergodiques de la mécanique classique (Paris: Gauthier Villars, 1967).
4S. Bates and A. Weinstein, Lectures on the Geometry of Quantization vol. 8, (American Mathematical Soc., 1997).
5M. Brin and G. Stuck, Introduction to Dynamical Systems (Cambridge University Press, 2002).
6F. Faure, "Prequantum chaos: Resonances of the prequantum cat map", arXiv:nlin/0606063. Journal of Modern Dynamics 1, 2 (2007), pp. 255-285.
7F. Faure and N. Roy, "Ruelle-Pollicott resonances for real analytic hyperbolic map", Nonlinearity. https://arxiv.org/abs/nlin.CD/0601010link 19 (2006), pp. 1233-1252.
8F. Faure and J. Sjöstrand, "Upper bound on the density of Ruelle resonances for Anosov flows. A semiclassical approach", Comm. in Math. Physics, Issue 2. https://fr.arxiv.org/abs/1003.0513link 308 (2011), pp. 325-364.
9F. Faure and M. Tsujii, "Band structure of the Ruelle spectrum of contact Anosov flows", Comptes rendus - Mathématique 351 , 385-391, (2013) https://arxiv.org/abs/1301.5525link (2013).
10F. Faure and M. Tsujii, "Prequantum transfer operator for symplectic Anosov diffeomorphism", Asterisque 375 (2015), https://fr.arxiv.org/abs/1206.0282link (2015), pp. ix+222 pages.
11F. Faure and M. Tsujii, "Fractal Weyl law for the Ruelle spectrum of Anosov flows", arXiv:1706.09307 https://arxiv.org/abs/1706.09307link (2017).
12F. Faure and M. Tsujii, "Microlocal analysis and Band structure of contact Anosov flows", arxiv:2102.11196 (2021).
13F. Faure, N. Roy, and J. Sjöstrand, "A semiclassical approach for Anosov Diffeomorphisms and Ruelle resonances", Open Math. Journal. https://arxiv.org/abs/0802.1780link 1 (2008), pp. 35--81.
14A. Katok and B. Hasselblatt, Introduction to the Modern Theory of Dynamical Systems (Cambridge University Press, 1995).
15M. Zworski, Semiclassical Analysis (Amer Mathematical Society, 2012).