Séminaire de topologie, géométrie et algèbre (archives)

The prototypical question in metric systolic geometry is to bound the length of a shortest closed geodesic on a closed Riemannian manifold by the volume of the manifold. This question has been extensively studied for non simply connected manifolds, but in the recent years there has been some progress also for simply connected manifolds, on which closed geodesics cannot be found simply by minimizing the length. This progress involves extending systolic questions to Reeb flows, a class of dynamical systems generalising geodesic flows. On the one hand, this extension and the use of symplectic techniques provide some answers to classical questions within metric systolic geometry.

In a 4-manifold, the Whitney trick seeks to remove a pair of oppositely signed intersection points between immersed surfaces. I will discuss recent joint work with Chris Davis and JungHwan Park where we describe a relative Whitney trick. The relative Whitney trick seeks to remove a single double point in a properly immersed surface, at the expense of changing the boundary of the surface by a homotopy. Our main application is to prove that any link in a homology 3-sphere is homotopic to a link that bounds a collection of locally flatly embedded discs in a contractible topological 4-manifold. In other words, every link in a homology sphere is homotopic to a topologically slice link.

To describe the behavior of the iterates of a holomorphic germ f on a complex surface X fixing a (possibly singular) point x0, we are led to study the lifts fπ to birational models Xπ over (X,x0). In general fπ has indeterminacy points: when the fπ-orbits eventually avoid these indeterminacy points, we say that X_π is algebraically stable. In a joint work with William Gignac, we show the existence of algebraically stable models in this setting (but for one class of exceptions, where no such models exist). The proof relies on fixed point theorems for the dynamics induced on suitable valuation spaces, following Favre and Jonsson.

Les singularités en épissure sont la classe la plus vaste que l'on connaisse de singularités intersections complètes de surfaces complexes dont le bord est une sphère d'homologie entière. Neumann et Wahl conjecturèrent que leurs fibres de Milnor peuvent se construire à partir de fibres de Milnor de singularités plus simples du même type. Je présenterai le contexte dans lequel ils formulèrent cette conjecture et j'expliquerai les étapes de la preuve obtenue en collaboration avec Maria Angelica Cueto et Dmitry Stepanov.

Les nombres de Markov sont des entiers positifs qui apparaissent dans les solutions d'une équation diophantienne, la cubique de Markov x2 + y2 + z2−3xyz = 0. Sujet classique de la théorie des nombres, ces nombres sont liés à de nombreux domaines des mathématiques tels que la combinatoire, la géométrie hyperbolique, la théorie de l'approximation et les algèbres amassées.

Dans les années 50, H. Cohn a découvert une relation entre les nombres de Markov et les longueurs de géodésiques simples fermées sur le tore percé. Dans les années 90, avec Igor Rivin, nous avons introduit une méthode qui permet d'estimer le nombre de nombres de Markov inférieurs à $ L> 0 $. L'ingrédient clé en était l'utilisation d'une norme sur la première homologie du tore.

All oriented links $L$ have special diagrams. Based on such a diagram we construct a sutured handlebody $M$ which embeds in the branched double cover of $L$. From the sutured Floer homology of $M$ we recover the Alexander polynomial $\Delta$ of $L$ via a simple forgetful map. More surprisingly, in cases when the diagram is also positive (so that $L$ is a special alternating link), $\mathrm{SFH}(M)$ can be used to compute those coefficients of the HOMFLY polynomial of $L$ whose sum is the leading coefficient of $\Delta$. To extract this information algebraically, we develop the notion of the interior polynomial of a bipartite graph. The talk involves joint results with A. Juh\'asz, H. Murakami, A. Postnikov, J. Rasmussen, and D. Thurston.

I will explain a geometric construction of an Enriques surface fibration over P^1 of even index. This answers a question of Colliot-Thelene and Voisin, and provides new counterexamples to the Integral Hodge conjecture. This is joint work with Fumiaki Suzuki.

Given a partial differential equation (PDE), its solutions can be difficult, if not impossible, to compute. The purpose of the Fundamental theorem of differential tropical (partial) algebraic geometry is to extract from the equations certain properties of the solutions. More precisely, this theorem proves that the support of the solutions (in $k[[t1, \cdots, tm]]$ with $k$ a field of characteristic zero) of a system of algebraic PDE can be obtained by solving a so-called tropicalized differential system.