Sur l’inégalité de Brunn-Minkowski pour les mesures log-concaves
Arnaud Marsiglietti
Caltech, USA
https://people.clas.ufl.edu/amarsiglietti/
Date(s) : 13/02/2017 iCal
10h00 - 11h00
On the Brunn-Minkowski inequality for log-concave measures
The Brunn-Minkowski inequality states that for all compact subsets A,B of the Euclidean space n, and any scalar λ
[0,1], V ol((1 – λ)A + λB)1∕n ≥ (1 – λ)V ol(A)1∕n + λV ol(B)1∕n, where V ol denotes the Lebesgue measure. This inequality has important consequences in geometry since it implies isoperimetric inequality, but also in analysis and probabilities, especially in measurement concentration, since it implies Gross log-Sobolev inequality. Moreover, the class of log-concave measures is rich and contains for example the Gaussian, exponential, and uniform measures on a convex set. In this talk, we show that the Brunn-Minkowski inequality extends to log-concave measures. We thus deduce new isoperimetric inequalities.
https://hal.archives-ouvertes.fr/hal-01081764
Catégories