Soit $X$ le produit direct de deux grassmanniennes des sous-espaces de dimensions $k$, $l$ d'un espace vectoriel $V$. Nous étudions les orbites d'un sous-groupe de Borel $B$ de GL($V$) opérant diagonalement dans $X$, et les adhérences de Zariski de ces orbites, en analogie avec les cellules et les variétés de Schubert dans les grassmanniennes. On vérifie sans pein que ces orbites sont en nombre fini. Elles ont été décrites de façon combinatoire par P. Magyar, J. Weyman et A. Zelevinsky. Nous obtenons un critère pour l'inclusion d'une orbite dans l'adhérence d'une autre orbite, et nous construisons une résolution de ces adhérences d'orbites, analogue aux désingularisations de Bott-Samelson des variétés de Schubert.

The obstruction to the existence of Fredholm problems for elliptic differentail operators on manifolds with edges is a topological invariant of the operator. We give an explicit general formula for this invariant. As an application we compute this obstruction for geometric operators.

No description available.

We find formulas for the Segre-MacPherson classes of Schubert cells in T-equivariant cohomology and the motivic Segre classes of Schubert cells in T-equivariant K-theory. In doing so we look at the pushforward of the projection map from the Bott-Samelson (Kempf-Laksov) desingularization to the Grassmannian. We find that the Segre-MacPherson classes are stable under pullbacks of maps embedding a Grassmannian into a bigger Grassmannian. We also express these formulas using certain Demazure-Lusztig operators that have previously been used to study these classes. / Doctor of Philosophy / Schubert calculus was first introduced in the nineteenth century as a way to answer certain questions in enumerative geometry. These computations relied on the multiplication of Schubert classes in the cohomology ring of Grassmannians, which parameterize k-dimensional linear subspaces of a vector space. More recently Schubert calculus has been broadened to refer to computations in generalized cohomology theories, such as (equivariant) K-theory. In this dissertation, we study Segre-MacPherson classes and motivic Segre classes of Schubert cells in Grassmannians. Segre-MacPherson classes are related to Chern-Schwartz-MacPherson classes, which are a generalization to singular spaces of the total Chern class of the tangent bundle. Motivic Segre classes are similarly related to motivic Chern classes, which are a K-theory analogue of Chern-Schwartz-MacPherson classes. This dissertation also studies the relationship between Schubert varieties and their Bott-Samelson desingularizations, specifically their (T-equivariant) cohomology and K-theory rings. Since equivariant cohomology (or K-theory) classes can be represented by polynomials, we can represent the Segre-MacPherson (or motivic Segre) classes as rational functions. Furthermore, we use certain operators that act on such polynomials (or rational functions) to find formulas for the rational function representatives of the aforementioned classes.

Soit M une variété lisse fermée et considérons sont fibré cotangent T*M muni de la structure symplectique usuelle induite par la forme de Liouville. Une hypersurface S de T*M$ est dite étoilée fibre par fibre si pour tout point q de M, l'intersection Sq de S avec la fibre au dessus de q est le bord d'un domaine étoilé par rapport à l'origine 0q de la fibre T*qM. Un flot est naturellement associé à S, il s'agit de l'unique flot généré par le champ de Reeb le long de S, le flot de Reeb. <p><p>L'existence d'une orbite orbite fermée du flot de Reeb sur S fut annoncée par Weinstein dans sa conjecture en 1978. Indépendamment, Weinstein et Rabinowitz ont montré l'existence d'une orbite fermée sur les hypersurfaces de type étoilées dans l'espace réel de dimension 2n. Sous les hypothèses précédentes, l'existence d'une orbite fermée fut démontrée par Hofer et Viterbo. Dans le cas particulier du flot géodésique, l'existence de plusieurs orbites fermées fut notamment étudiée par Gromov, Paternain et Paternain-Petean. Dans cette thèse, ces résultats sont généralisés. <p><p>Les résultats principaux de cette thèse montrent que la structure topologique de la variété M implique, pour toute hypersurface étoilée fibre par fibre, l'existence de beaucoup d'orbites fermées du flot de Reeb. Plus précisément, une borne inférieure de la croissance du nombre d'orbites fermées du flot de Reeb en fonction de leur période est mise en évidence. /<p><p>Let M be a smooth closed manifold and denote by T*M the cotangent bundle over M endowed with its usual symplectic structure induced by the Liouville form. A hypersurface S of T*M is said to be fiberwise starshaped if for each point q in M the intersection Sq of S with the fiber at q bounds a domain starshaped with respect to the origin 0q in T*qM. There is a flow naturally associated to S, generated by the unique Reeb vector field R along S ,the Reeb flow. <p><p>The existence of one closed orbit was conjectured by Weinstein in 1978 in a more general setting. Independently, Weinstein and Rabinowitz established the existence of a closed orbit on star-like hypersurfaces in the 2n-dimensional real space. In our setting the Weinstein conjecture without the assumption was proved in 1988 by Hofer and Viterbo. The existence of many closed orbits has already been well studied in the special case of the geodesic flow, for example by Gromov, Paternain and Paternain-Petean. In this thesis we will generalize their results.<p><p>The main result of this thesis is to prove that the topological structure of $M$ forces, for all fiberwise starshaped hypersurfaces S, the existence of many closed orbits of the Reeb flow on S. More precisely, we shall give a lower bound of the growth rate of the number of closed Reeb-orbits in terms of their periods. / Doctorat en Sciences / info:eu-repo/semantics/nonPublished

We generalize Forman’s discrete Morse theory, on one end by developing a discrete analogue of Morse-Bott theory for CW complexes, motivated by Morse-Bott theory in the smooth setting. On the other, motivated by J-N. Corvellec’s Morse theory for continuous functionals, we generalize Forman’s discrete Morse-floer theory by considering a vector field more general than the one extracted from a discrete Morse function, and defining a boundary operator from which the Betti numbers of the CW complex are obtained. We also do some Conley theory analysis.