Date/heure
14 avril 2022
14:15 - 15:15
Oratrice ou orateur
Peter Hochs (Nijmegen)
Catégorie d'évènement Séminaire Théorie de Lie, Géométrie et Analyse
Résumé
If we have a flat vector bundle E over a compact manifold M, then we can form the de Rham cohomology H(M; E) of the M twisted by E. Given a Riemannian metric on M, this cohomology can be realised as the kernel of a Laplacian, via the Hodge theorem. Using the same Laplacian, In 1971, Ray and Singer defined the analytic torsion of M (now also assumed oriented), twisted by E. If H(M; E) vanishes, then analytic torsion is independent of the Riemannian metric used to define the Laplacian, and therefore a smooth invariant. Ray and Singer’s motivation for this definition was to give an analytic way to realise Reidemeister-Franz torsion, a combinatorial invariant of finite cell complexes. In 1978, Cheeger and Müller proved independently that these two notions of torsion are indeed equal. Fried’s conjecture from 1986 is a relation between analytic torsion and the Ruelle dynamical zeta function of a flow on M with suitable properties. That is a quantity involving the lengths of closed flow curves. So analytic torsion gives us a link between analysis, topology and dynamical systems. Most of this talk is a survey of analytic torsion. I will also mention some recent work with Hemanth Saratchandran on an equivariant version of analytic torsion for proper group actions.