Mon travail de thèse consiste à comprendre une géométrie introduite par Cartan en 1933 \cite{Cartan1933}. \textit{La géométrie de Finsler} présente de nombreuses analogies avec cette théorie. Nous avons étudié les grandes lignes de cette géométrie. Le point de départ de Cartan qui est analogue à celui qui conduit à la géométrie finslerienne, est d'imaginer l'espace comme étant un lieu ''d'éléments de contact'', un élément étant la donnée d'un point $M\in\mathcal{M}^n$ et d'un hyperplan $H$ passant par ce point et orienté dans l'espace tangent $T_M\mathcal{M}^n$. Nous avons ainsi défini \textit{la géométrie de Cartan fondée sur la notion d'aire} dans un premier temps, je me suis intéressé à la notion d'orthogonalité dans cette géométrie. La méthode de Cartan pour étudier le problème d'équivalence est un outil puissant qui est implicitement décrit dans cette géométrie. Nous avons ensuite appliqué cette méthode aux équations de Monge-Ampère (cas elliptique), en s'inspirant des travaux de R. Bryant, D. Grossmann et P. Griffiths. Plusieurs faits ne sont pas encore suffisamment clairs pour disposer d'un dictionnaire évident entre ces travaux et celui donné par Cartan.

We study canonical forms for Hamiltonian and symplectic matrices or pencils under equivalence transformations which keep the class invariant. In contrast to other canonical forms our forms are as close as possible to a triangular structure in the same class. We give necessary and sufficient conditions for the existence of Hamiltonian and symplectic triangular Jordan, Kronecker and Schur forms. The presented results generalize results of Lin and Ho [17] and simplify the proofs presented there.

Το κύριο στοιχείο της διατριβής είναι οι Εξελικτικοί Αλγόριθμοι. Στο πρώτο μέρος παρουσιάζονται οι Μιμιδικοί Αλγόριθμοι. Οι Μιμιδικοί Αλγόριθμοι είναι υβριδικά σχήματα που συνδυάζουν τους Εξελιτκικούς Αλγορίθμους με μεθόδους τοπικής αναζήτησης. Οι Μιμιδικοί Αλγόριθμοι συγκρίθηκαν με τους Εξελικτικούς Αλγορίθμους σε πληθώρα προβλημάτων ολικής βελτιστοποίησης και είχαν καλύτερα αποτελέσματα. Στο δεύτερο μέρος μελετήθηκαν προβλήματα μη γραμμικής δυναμικής. Αυτά ήταν η εκτίμηση της περιοχής ευστάθειας διατηρητικών απεικονίσεων, η ανίχνευση συντονισμών και ο υπολογισμός περιοδικών τροχιών. Τα αποτελέσματα ήταν ικανοποιητικά. / The main objective of the thesis was the study of Evolutionary Algorithms. At the first part, Memetic Algorithms were introduced. Memetic Algorithms are hybrid schemes that combine Evolutionary Algorithms and local search methods. Memetic Algorithms were compared to Evolutionary Algorithms in various problems of global optimization and they had better performance. At the second part, problems from nonlinear dynamics were studied. These were the estimation of the stability region of conservative maps, the detection of resonances and the computation of periodic orbits. The results were satisfactory.

The Kalman Filter has many applications in control and signal processing but may also be used to reconstruct a higher resolution image from a sequence of lower resolution images (or frames). If the sequence of low resolution frames is recorded by a moving camera or sensor, where the motion can be accurately modeled, then the Kalman filter may be used to update pixels within a higher resolution frame to achieve a more detailed result. This thesis outlines current methods of implementing this algorithm on a scene of interest and introduces possible improvements for the speed and efficiency of this method by use of block operations on the low resolution frames. The effects of noise on camera motion and various blur models are examined using experimental data to illustrate the differences between the methods discussed.

Cette thèse traite des aspects géométriques et d'intégrabilité associés aux modèles de matrices aléatoires. Son but est de présenter diverses applications des modèles de matrices aléatoires allant de la géométrie algébrique aux équations aux dérivées partielles des systèmes intégrables. Ces différentes applications permettent en particulier de montrer en quoi les modèles de matrices possèdent une grande richesse d'un point de vue mathématique. Ainsi, cette thèse abordera d'abord l'étude de la jonction de deux intervalles du support de la densité des valeurs propres au voisinage d'un point singulier. On montrera plus précisément en quoi ce régime limite particulier aboutit aux équations universelles de la hiérarchie de Painlevé II des systèmes intégrables. Ensuite, l'approche des polynômes (bi)-orthogonaux, introduite par Mehta pour le calcul des fonctions de partition, permettra d'énoncer des problèmes de Riemann-Hilbert et d'isomonodromies associés aux modèles de matrices, faisant ainsi le lien avec la théorie de Jimbo-Miwa-Ueno. On montrera en particulier que le cas des modèles à deux matrices hermitiens se transpose à un cas dégénéré de la théorie isomonodromique de Jimbo-Miwa-Ueno qui sera alors généralisé. La méthode des équations de boucles avec ses notions centrales de courbe spectrale et de développement topologique permettra quant à elle de faire le lien avec les invariants symplectiques de géométrie algébrique introduits récemment par Eynard et Orantin. Ce dernier point fera également l'objet d'une généralisation aux modèles de matrices non-hermitien (beta quelconque) ouvrant ainsi la voie à la ``géométrie algébrique quantique'' et à la généralisation de ces invariants symplectiques pour des courbes ``quantiques''. Enfin, une dernière partie sera consacrée aux liens étroits entre les modèles de matrices et les problèmes de combinatoire. En particulier, l'accent sera mis sur les aspects géométriques de la théorie des cordes topologiques avec la construction explicite d'un modèle de matrices aléatoires donnant le dénombrement des invariants de Gromov-Witten pour les variétés de Calabi-Yau toriques de dimension complexe trois utilisées en théorie des cordes topologiques. L'étendue des domaines abordés étant très vaste, l'objectif de la thèse est de présenter de façon la plus simple possible chacun des domaines mentionnés précédemment et d'analyser en quoi les modèles de matrices peuvent apporter une aide précieuse dans leur résolution. Le fil conducteur étant les modèles matriciels, chaque partie a été conçue pour être abordable pour un spécialiste des modèles de matrices ne connaissant pas forcément tous les domaines d'application présentés ici. / This thesis deals with the geometric and integrable aspects associated with random matrix models. Its purpose is to provide various applications of random matrix theory, from algebraic geometry to partial differential equations of integrable systems. The variety of these applications shows why matrix models are important from a mathematical point of view. First, the thesis will focus on the study of the merging of two intervals of the eigenvalues density near a singular point. Specifically, we will show why this special limit gives universal equations from the Painlevé II hierarchy of integrable systems theory. Then, following the approach of (bi) orthogonal polynomials introduced by Mehta to compute partition functions, we will find Riemann-Hilbert and isomonodromic problems connected to matrix models, making the link with the theory of Jimbo, Miwa and Ueno. In particular, we will describe how the hermitian two-matrix models provide a degenerate case of Jimbo-Miwa-Ueno's theory that we will generalize in this context. Furthermore, the loop equations method, with its central notions of spectral curve and topological expansion, will lead to the symplectic invariants of algebraic geometry recently proposed by Eynard and Orantin. This last point will be generalized to the case of non-hermitian matrix models (arbitrary beta) paving the way to ``quantum algebraic geometry'' and to the generalization of symplectic invariants to ``quantum curves''. Finally, this set up will be applied to combinatorics in the context of topological string theory, with the explicit computation of an hermitian random matrix model enumerating the Gromov-Witten invariants of a toric Calabi-Yau threefold. Since the range of the applications encountered is large, we try to present every domain in a simple way and explain how random matrix models can bring new insights to those fields. The common element of the thesis being matrix models, each part has been written so that readers unfamiliar with the domains of application but familiar with matrix models should be able to understand it. / Travail réalisé en cotutelle avec l'université Paris-Diderot et le Commissariat à l'Energie Atomique sous la direction de John Harnad et Bertrand Eynard.

Dans cette thèse, nous étudions les solutions des équations Hamilton-Jacobi. Plus précisément, nous comparons la solution de viscosité, obtenue comme limite de solutions de l'équation perturbée par un petit terme de diffusion, et la solution minmax, définie géométriquement à partir d'une fonction génératrice quadratique à l'infini. Dans la littérature, il y a des cas bien connus où les deux coïncident, par exemple lorsque le hamiltonien est convexe ou concave, le minmax pouvant alors être réduit à un min ou un max. Mais les solutions minmax et de viscosité diffèrent en général. Nous construisons des "minmax itérés" en répétant pas à pas la procédure de minmax et démontrons que, quand la taille du pas tend vers zéro, les minmax itérés tendent vers la solution de viscosité. Dans une deuxième partie, nous étudions les lois de conservation en dimension un d'espace par le méthode de "front tracking". Nous montrons que dans le cas où la donnée initiale est convexe, la solution de viscosité et le minmax sont égaux. Et comme application, nous décrivons sur des exemples la manière dont sont construites les singularités de la solution de viscosité. Pour finir, nous montrons que la notion de minmax n'est pas aussi évidente qu'il y paraît.

Multisymplectic integration is a relatively new addition to the field of geometric integration, which is a modern approach to the numerical integration of systems of differential equations. Multisymplectic integration is carried out by numerical integrators known as multisymplectic integrators, which preserve a discrete analogue of a multisymplectic conservation law. In recent years, it has been shown that various discretisations of a multi-Hamiltonian PDE satisfy a discrete analogue of a multisymplectic conservation law. In particular, discretisation in time and space by the popular symplectic Runge–Kutta methods has been shown to be multisymplectic. However, a multisymplectic integrator not only needs to satisfy a discrete multisymplectic conservation law, but it must also form a well-defined numerical method. One of the main questions considered in this thesis is that of when a multi-Hamiltonian PDE discretised by Runge–Kutta or partitioned Runge–Kutta methods gives rise to a well-defined multisymplectic integrator. In particular, multisymplectic integrators that are explicit are sought, since an integrator that is explicit will, in general, be well defined. The first class of discretisation methods that I consider are the popular symplectic Runge–Kutta methods. These have previously been shown to satisfy a discrete analogue of the multisymplectic conservation law. However, these previous studies typically fail to consider whether or not the system of equations resulting from such a discretisation is well defined. By considering the semi-discretisation and the full discretisation of a multi-Hamiltonian PDE by such methods, I show the following: • For Runge–Kutta (and for partitioned Runge–Kutta methods), the active variables in the spatial discretisation are the stage variables of the method, not the node variables (as is typical in the time integration of ODEs). • The equations resulting from a semi-discretisation with periodic boundary conditions are only well defined when both the number of stages in the Runge–Kutta method and the number of cells in the spatial discretisation are odd. For other types of boundary conditions, these equations are not well defined in general. • For a full discretisation, the numerical method appears to be well defined at first, but for some boundary conditions, the numerical method fails to accurately represent the PDE, while for other boundary conditions, the numerical method is highly implicit, ill-conditioned and impractical for all but the simplest of applications. An exception to this is the Preissman box scheme, whose simplicity avoids the difficulties of higher order methods. • For a multisymplectic integrator, boundary conditions are treated differently in time and in space. This breaks the symmetry between time and space that is inherent in multisymplectic geometry. The second class of discretisation methods that I consider are partitioned Runge– Kutta methods. Discretisation of a multi-Hamiltonian PDE by such methods has lead to the following two major results: 1. There is a simple set of conditions on the coefficients of a general partitioned Runge– Kutta method (which includes Runge–Kutta methods) such that a general multi- Hamiltonian PDE, discretised (either fully or partially) by such methods, satisfies a natural discrete analogue of the multisymplectic conservation law associated with that multi-Hamiltonian PDE. 2. I have defined a class of multi-Hamiltonian PDEs that, when discretised in space by a member of the Lobatto IIIA–IIIB class of partitioned Runge–Kutta methods, give rise to a system of explicit ODEs in time by means of a construction algorithm. These ODEs are well defined (since they are explicit), local, high order, multisymplectic and handle boundary conditions in a simple manner without the need for any extra requirements. Furthermore, by analysing the dispersion relation for these explicit ODEs, it is found that such spatial discretisations are stable. From these explicit ODEs in time, well-defined multisymplectic integrators can be constructed by applying an explicit discretisation in time that satisfies a fully discrete analogue of the semi-discrete multisymplectic conservation law satisfied by the ODEs. Three examples of explicit multisymplectic integrators are given for the nonlinear Schr¨odinger equation, whereby the explicit ODEs in time are discretised by the 2-stage Lobatto IIIA– IIIB, linear–nonlinear splitting and real–imaginary–nonlinear splitting methods. These are all shown to satisfy discrete analogues of the multisymplectic conservation law, however, only the discrete multisymplectic conservation laws satisfied by the first and third multisymplectic integrators are local. Since it is the stage variables that are active in a Runge–Kutta or partitioned Runge– Kutta discretisation in space of a multi-Hamiltonian PDE, the order of such a spatial discretisation is limited by the order of the stage variables. Moreover, the spatial discretisation contains an approximation of the spatial derivatives, and thus, the order of the spatial discretisation may be further limited by the order of this approximation. For the explicit ODEs resulting from an r-stage Lobatto IIIA–IIIB discretisation in space of an appropriate multi-Hamiltonian PDE, the order of this spatial discretisation is r - 1 for r = 10; this is conjectured to hold for higher values of r. For r = 3, I show that a modification to the initial conditions improves the order of this spatial discretisation. It is expected that a similar modification to the initial conditions will improve the order of such spatial discretisations for higher values of r.

Mapas simpléticos são bem conhecidos por preservarem o volume do espaço de fase em dinâmica Hamiltoniana e são particularmente apropriados para problemas que requerem longos tempos de integração. Nesta tese nós desenvolvemos abordagens baseadas em mapas simpléticos para o acoplamento de multi sub-sistemas/domínios astrofísicos/códigos de simulação, para integração eficiente de sistemas de N-corpos auto-gravitantes com grandes variações nas escalas de tempo características. Nós estabelecemos uma família de 48 novos mapas simpléticos baseados numa separação Hamiltoniana recursiva, que permite que o acoplamento ocorra de uma maneira hierárquica, contemplando assim todas as escalas de tempo das interações envolvidas. Nossa formulação é geral o suficiente para permitir que tal método seja utilizado como receita para combinar diferentes fenômenos físicos, que podem ser modelados independentemente por códigos especializados. Nós introduzimos também uma separação Hamiltoniana baseada em Hamiltonianos de Kepler, para resolver o problema gravitacional geral de N-corpos como uma composição de N2 problemas de 2-corpos. O método resultante é exato para cada problema de 2-corpos individual e produz resultados rápidos e precisos para sistemas de N-corpos quase- Keplerianos, como sistemas planetários ou um aglomerado de estrelas que orbita um buraco-negro supermassivo. O método é também apropriado para integração de sistemas de N-corpos com hierarquias intrínsecas, como um aglomerados de estrelas com binárias compactas. Nós apresentamos a implementação dos algoritmos mencionados e descrevemos o nosso código tupan, que está publicamente disponível na seguinte url: https://github.com/ggf84/tupan. / Symplectic maps are well know for preserving the phase space volume in Hamiltonian dynamics and are particularly suited for problems that require long integration times. In this thesis we develop approaches based on symplectic maps for the coupling of multi sub-systems/astrophysics domains/simulation codes for efficient integration of self-gravitating N-body systems with large variation in characteristic time-scales. We establish a family of 48 new symplectic maps based on a recursive Hamiltonian splitting, which allow the coupling to occur in a hierarchical manner, thus contemplating all time-scales of the involved interactions. Our formulation is general enough to allow that such method be used as a recipe to combine different physical phenomena which can be modeled independently by specialized simulation codes. We also introduce a Keplerian-based Hamiltonian splitting for solving the general gravitational Nbody problem as a composition of N2 2-body problems. The resulting method is precise for each individual 2-body solution and produces quick and accurate results for near-Keplerian N-body systems, like planetary systems or a cluster of stars that orbit a supermassive black-hole. The method is also suitable for integration of N-body systems with intrinsic hierarchies, like a star cluster with compact binaries. We present the implementation of the mentioned algorithms and describe our code tupan, which is publicly available on the following url: https://github.com/ggf84/tupan.

Pour toute variété affine W munie d'une opération d'un groupe réductif G, le schéma de Hilbert invariant est un espace de modules qui classifie les sous-schémas fermés de W, stables par l'opération de G, et dont l'algèbre affine est somme directe de G-modules simples avec des multiplicités finies préalablement fixées. Dans cette thèse , on étudie d'abord le schéma de Hilbert invariant, noté H, qui paramètre les sous-schémas fermés GL(V)-stables Z de W=n1 V oplus n2 V^* tels que k[Z] est isomorphe à la représentation régulière de GL(V) comme GL(V)-module. Si dim(V)<3,on montre que H est une variété lisse, et donc que le morphisme de Hilbert-Chow gamma: H -> W//G est une résolution des singularités du quotient W//G. En revanche, si dim(V)=3, on montre que H est singulier. Lorsque dim(V)<3, on décrit H par des équations et aussi comme l'espace total d'un fibré vectoriel homogène au dessus d'un produit de deux grassmanniennes. On se place ensuite dans le cadre symplectique en prenant n1=n2 et en remplaçant W par la fibre en 0 de l'application moment mu: W -> End(V). On considère alors le schéma de Hilbert invariant H' qui paramètre les sous-schémas contenus dans mu^{-1}(0). On montre que H' est toujours réductible, mais que sa composante principale Hp' est lisse lorsque dim(V)<3. Dans ce cas, le morphisme de Hilbert-Chow est une résolution (parfois symplectique) des singularités du quotient mu^{-1}(0)//G. Lorsque dim(V)<3, on décrit Hp' comme l'espace total d'un fibré vectoriel homogène au dessus d'une variété de drapeaux. Enfin, on obtient des résultats similaires lorsque l'on remplace GL(V) par un autre groupe classique (SL(V), SO(V), O(V), Sp(V)) que l'on fait opérer d'abord dans W=nV, puis dans la fibre en 0 de l'application moment. / Let W be an affine variety equipped with an action of a reductive group G. The invariant Hilbert scheme is a moduli space which classifies the G-stable closed subschemes of W such that the affine algebra is the direct sum of simple G-modules with previously fixed finite multiplicities. In this thesis, we first study the invariant Hilbert scheme, denoted H. It parametrizes the GL(V)-stable closed subschemes Z of W=n1 V oplus n2 V^* such that k[Z] is isomorphic to the regular representation of GL(V) as GL(V)-module. If dim(V)<3, we show that H is a smooth variety, so that the Hilbert-Chow morphism gamma: H -> W//G is a resolution of singularities of the quotient W//G. However, if dim(V)=3, we show that H is singular. When dim(V)<3, we describe H by equations and also as the total space of a homogeneous vector bundle over the product of two Grassmannians. Then we consider the symplectic setting by letting n1=n2 and replacing W by the zero fiber of the moment map mu: W -> End(V). We study the invariant Hilbert scheme H' which parametrizes the subschemes included in mu^{-1}(0). We show that H' is always reducible, but that its main component Hp' is smooth if dim(V)<3. In this case, the Hilbert-Chow morphism is a resolution of singularities (sometimes a symplectic one) of the quotient mu^{-1}(0)//G. When dim(V)=3, we describe Hp' as the total space of a homogeneous vector bundle over a flag variety. Finally, we get similar results when we replace GL(V) by some other classical group (SL(V), SO(V), O(V), Sp(V)) acting first on W=nV, then on the zero fiber of the moment map.

Mapas simpléticos são bem conhecidos por preservarem o volume do espaço de fase em dinâmica Hamiltoniana e são particularmente apropriados para problemas que requerem longos tempos de integração. Nesta tese nós desenvolvemos abordagens baseadas em mapas simpléticos para o acoplamento de multi sub-sistemas/domínios astrofísicos/códigos de simulação, para integração eficiente de sistemas de N-corpos auto-gravitantes com grandes variações nas escalas de tempo características. Nós estabelecemos uma família de 48 novos mapas simpléticos baseados numa separação Hamiltoniana recursiva, que permite que o acoplamento ocorra de uma maneira hierárquica, contemplando assim todas as escalas de tempo das interações envolvidas. Nossa formulação é geral o suficiente para permitir que tal método seja utilizado como receita para combinar diferentes fenômenos físicos, que podem ser modelados independentemente por códigos especializados. Nós introduzimos também uma separação Hamiltoniana baseada em Hamiltonianos de Kepler, para resolver o problema gravitacional geral de N-corpos como uma composição de N2 problemas de 2-corpos. O método resultante é exato para cada problema de 2-corpos individual e produz resultados rápidos e precisos para sistemas de N-corpos quase- Keplerianos, como sistemas planetários ou um aglomerado de estrelas que orbita um buraco-negro supermassivo. O método é também apropriado para integração de sistemas de N-corpos com hierarquias intrínsecas, como um aglomerados de estrelas com binárias compactas. Nós apresentamos a implementação dos algoritmos mencionados e descrevemos o nosso código tupan, que está publicamente disponível na seguinte url: https://github.com/ggf84/tupan. / Symplectic maps are well know for preserving the phase space volume in Hamiltonian dynamics and are particularly suited for problems that require long integration times. In this thesis we develop approaches based on symplectic maps for the coupling of multi sub-systems/astrophysics domains/simulation codes for efficient integration of self-gravitating N-body systems with large variation in characteristic time-scales. We establish a family of 48 new symplectic maps based on a recursive Hamiltonian splitting, which allow the coupling to occur in a hierarchical manner, thus contemplating all time-scales of the involved interactions. Our formulation is general enough to allow that such method be used as a recipe to combine different physical phenomena which can be modeled independently by specialized simulation codes. We also introduce a Keplerian-based Hamiltonian splitting for solving the general gravitational Nbody problem as a composition of N2 2-body problems. The resulting method is precise for each individual 2-body solution and produces quick and accurate results for near-Keplerian N-body systems, like planetary systems or a cluster of stars that orbit a supermassive black-hole. The method is also suitable for integration of N-body systems with intrinsic hierarchies, like a star cluster with compact binaries. We present the implementation of the mentioned algorithms and describe our code tupan, which is publicly available on the following url: https://github.com/ggf84/tupan.