Séminaires

Nous développons une nouvelle technique pour minorer des sommes d’exponentielle de la forme $\sum f(n) e^{2i\pi\alpha n}$ pour tout $\alpha.$
Nous montrerons en particulier que la somme $\sum_{p\le x} e^{2i\pi\alpha p}$ est non bornée pour tout $\alpha,$ et plus précisément diverge au moins comme $x^{1/6-\varepsilon}$ pour une suite de $x$ tendant vers l’infini, uniformément en $\alpha.$

The geometry of the flag manifold G/P associated to a parabolic subgroup P of a semisimple Lie group G gives rise to a G-equivariant complex of elliptic operators (the BGG complex) which satisfy the property of maximal hypoellipticity, as shown by the work of Dave and Haller. In the case of the group SU(n,1), the BGG complex is the Rumin complex associated to the contact structure on the sphere S^{2n-1}. We shall describe the quaternionic analogue (i.e. for G=Sp(n,1)), and compute the bundles and operators involved in the BGG complex thanks to the Kostant theorem of 1961 (generalized by Cap and Slovak) on the cohomology of the maximal nilpotent subalgebra of a parabolic subalgebra of a semisimple Lie algebra.
We shall also explain that, in the context of Non Commutative Geometry, the BGG complexes are a crucial ingredient for the construction of the so called Kasparov gamma-element, which is the obstruction to the subjectivity of the Baum-Connes map.

The phenomenon known as ‘the Mackey analogy’ is a correspondance between the tempered dual of a Lie group $G$ and the unitary dual of its associated motion group $G_0$.  The precise statement, formulated and proven by Higson in the case of complex groups, and by Afgoustidis in the general case of real groups, has long been known to be intimately connected to the Connes-Kasparov isomorphism relating the K-theory of the reduced C$^\ast$-algebras of $G$ and $G_0$ via a deformation argument.
In this talk, I will report on joint work with Higson and Román, aimed at understanding the Mackey analogy directly at the level of the group C$^\ast$-algebras.  The main result is an embedding of the C$^\ast$-algebra of $G_0$ into the reduced C$^\ast$-algebra of $G$, which is proven to characterize the bijection of Afgoustidis and to induce the Connes-Kasparov isomorphism.

Anton Kapustin, Nikita Sopenko and others have studied the group of “locally generated automorphisms” of a UHF C*-algebra whose matrix algebra factors are labelled by the points of a coarse space.  This talk will be about a related Lie algebra of “locally generated derivations.”  Specifically it will be about the problem of computing the homology of this Lie algebra.  I shall describe a set of techniques that may be employed to solve the problem, at least for lattices in Euclidean space and similar spaces. Some are from the coarse Baum-Connes conjecture, and some, from homological algebra, are much older. This is joint work with Tsuyoshi Kato.

We construct a Poisson algebra bundle whose distributional sections are suitable to represent multilocal observables in classical field theory. To do this, we work with vector bundles over the unordered configuration space of a manifold M and consider the structure of a 2-monoidal category given by the usual (Hadamard) tensor product of bundles and a new (Cauchy) tensor product which provides a symmetrized version of the usual external tensor product of vector bundles on M.

In « $L_\infty$-algebras and higher analogues of Dirac structures and Courant algebroids » (arXiv:1003.1004), Marco Zambon constructed an explicit $L_\infty$-morphism between the Courant $r=1$-Lie algebra of a smooth manifold $M$, twisted by a closed 2-form $\sigma$, and the untwisted Courant $r=2$-Lie algebra of $M$. He left open the question of whether similar canonical $L_\infty$-morphisms exist in higher degrees — that is, between the Courant $r$-Lie algebra twisted by a closed $(r+1)$-form $\sigma$ and the untwisted Courant $(r+1)$-Lie algebra. In this talk, we present a general framework that naturally accounts for the existence of such morphisms for arbitrary $r$. This is joint work with Domenico Fiorenza.

In this talk I will describe a way to implement the resonance method in problems of analytic number theory which are not necessarily multiplicative in nature.
This extension of the method not only produces improved extreme results wherever Dirichelt’s approximation theorem has been usually employed but it also highlights its connection to Bohr’s and Jessen’s proof of Kronecker’s approximation theorem.

En se laissant guider par l’exemple des ensembles de Sidon (ensembles de nombres dont les sommes de deux éléments sont uniques, très étudiés en combinatoire additive), je présenterai des résultats récents, en collaboration avec R. Riblet, où des techniques de théorie des ensembles permettent de construire des ensembles « grands » en certains sens (cardinalité, mesure ou dimension) tout en étant « épars » car évitant des configurations prescrites (pas de relation linéaire, ou ne contenant pas de parallélogramme, etc.). Des questions subtiles en lien avec l’axiome du choix seront évoquées.

Motivé par ses travaux et ceux de Behrend dans les années 30 concernant les ensembles primitifs d’entiers, Erdős conjectura en 1948 que si $\mathcal{A}$ est un ensemble dénombrable de réels $>1$, tel que $\limsup_{x\to +\infty} \frac{1}{\log x}\sum_{\alpha\leq x, \alpha\in \mathcal{A}}\frac{1}{\alpha} >0$, alors pour tout $\varepsilon>0$, il existe une infinité de triplets $(\alpha, \beta, n)\in \mathcal{A}^2\times \mathbb{N}$ tels que $\alpha\neq \beta$ et $|n\alpha-\beta|<\varepsilon.$ Très peu de temps avant sa mort en 1996, il avait offert 500$ pour la résolution de ce problème de nature diophantienne.

Dans cet exposé, je présenterai un travail récent, en collaboration avec Dimitris Koukoulopoulos et Jared Lichtman, où l’on démontre cette conjecture.