Home

Je suis membre de l'équipe Topologie et Géométrie Algébriques, à l'Institut de Mathématiques de Jussieu (UMR 7586).

En collaboration avec Charles De Clercq, on développe une axiomatisation de la théorie de Kummer: c'est notre project sur les groupes profinis lisses.
La motivation de ce travail est de fournir une preuve compréhensible du théorème de Rost-Voevodsky, grâce à des résultats de relèvement de la cohomologie des groupes profinis lisses.

The new corrected texts of the project "Smooth profinite groups" are now available. Notably, the second paper has been deeply modified. For further information, you can read the introduction of "Smooth profinite groups, I". See section "Travaux" of this webpage.

Update (May 23): I have spotted a problem in Definition 12.10 of the second paper. There, the line bundles det(V)^p and det(V^(1)) are tacitly identified, but they are in general non-isomorphic. [Of course, they are canonically isomorphic modulo p].
This can be corrected easily by adding suitable (Witt) line bundles twists, but they cause trouble in some computations in paper III.
I have a precise idea how to fix this, by refining the "uplifting pattern" (with the help of complete flags).
Modifications will be implemented in June, that should hopefully lead to simpler computations in paper III.

Update (August 5): Modifications are going well, but taking longer that expected. I'll do my best to finish this summer:)

Update (September 30): My best was not enough... the work is still is progress;) Nice new ideas should simplify the presentation, while (hopefully) leading to a stonger result in the third paper.
Namely, for a (1,1)-smooth profinite group G, any cohomology class in H^n(G,F_p) should be a sum of quasi-symbols. (For details, see property 2B of Remark 5.10 of the current version of `smooth profinite groups', III).
This should be proved by a quite explicit recursive process, along a completely filtered n-extension representing the cohomology class.
Still, a key feature of the strategy is to build an "integral model" of Kummer theory, using Hochschild cohomology of linear algebraic groups.