Artin a résolu le 17ème problème de Hilbert en démontrant qu'un polynôme positif en n variables à coefficients réels est une somme de carrés de fractions rationnelles, et Pfister a montré que 2^n carrés suffisent. Dans cet exposé, on étudiera quand le théorème de Pfister peut être amélioré. On montrera qu'un polynôme réel positif de degré d en n variables est une somme de (2^n)-1 carrés si d<2n, et dans certains cas si d=2n.

I will explain some geometric ideas (mostly due to de Jong-Starr) one can use to study the Hasse principle for varieties defined over funvtion fields. I will illustrate these methods by giving a new proof of the classical result of Hasse -Minkowsky on quadrics.

It is well-known that weak approximation is birational invariant between smooth varieties by the implicit function theorem. For strong approximation, such property is no longer true. However one can expect that strong approximation is invariant between smooth varieties up to a closed sub-variety of codimension at least 2. Indeed, this result is proved for affine spaces in a joint work with Yang Cao which is applied to show strong approximation for toric varieties. Such result is also proved by Dasheng Wei by using a different method. In this talk, I'll […]