Institut de recherche mathématique avancée

Résumé : Dans cet exposé, je vous propose de partir à la découverte du jeu de Hex, un jeu de plateau à cases hexagonales pour 2 joueurs, dont la richesse mathématique n’a d’égale que sa simplicité. La première apparition du jeu de Hex revient à Piet Hein en 1942, puis il sera redécouvert de manière indépendante en 1948 par John Nash, qui en étudiera la mathématique. Après avoir présenté les règles, nous répondrons à plusieurs questions : Existe-t-il une stratégie gagnante ? Si oui, pour quel joueur ? Peut-il y avoir égalité ? Existe-t-il une configuration où les deux joueurs sont gagnants en même temps ? Toutes ces réponses nous permettront de nous servir du jeu pour démontrer un célèbre théorème de topologie : le théorème de Brouwer.

Résumé : Consider the eigenvalue problem for the discrete Laplacian on finite regular graphs. One can relate three types of phase space distributions associated to the resulting eigenfunctions, namely Patterson–Sullivan, invariant Ruelle, and Wigner distributions. It has been shown that, on hyperbolic surfaces, these three phase-space distributions are asymptotically equivalent, naturally raising the question of whether a similar relationship holds for finite graphs. In this talk, I will show how discrete analogues of these distributions can be defined and how they relate to each other.

Résumé : Let A be an abelian variety defined over a number field k. The connected monodromy field k(eA) is the minimal extension of k over which every \ell-adic Galois representation attached to A has connected image, or equivalently, the minimal extension over which all Tate classes on all self-products A^r are defined. When k(eA)/k(End A) has positive degree, "exotic" Tate classes arise on certain powers A^r — classes not explained by the endomorphism algebra alone. I will explain how to compute k(eA) when A is the Jacobian of a curve with complex multiplication. We will exploit CM theory to describe the algebra of Tate classes and make the Galois action on this algebra explicit in terms of periods — suitable integrals of algebraic differential forms. Though periods are generally transcendental, those attached to Tate classes are algebraic, so computing k(eA) reduces to identifying these periods as exact algebraic numbers.

Résumé : Persistent homology is a field that has emerged in data analysis over the past two decades. Its primary goal is to recover the topology of a point cloud. In this talk, we explore the applications of persistent homology to another type of data that is ubiquitous in neuroscience : spike trains. A spike train is a collection of binary time series corresponding to neuronal activity. We define the persistent homology of a spike train via a filtration based on the frequencies of cofiring neurons, and we explore the performance of this method with both simulated and real data.

Résumé : TBA