On weighted sampling without replacement
schedule le lundi 20 janvier 2020 de 17h00 à 18h00
Organisé par : F. Bechtold, W. Da Silva , A. Fermanian, S. Has, Y. Yu
Intervenant : Othmane Safsafi (LPSM)
Lieu : Jussieu, Salle Paul Lévy (16-26-209)
Sujet : On weighted sampling without replacement
Consider n real numbers that we call weights. Each weight $w_i$ has a probability $p_i$ of being sampled. There are two different ways of sampling such weights. Either we take them with replacement, meaning a same weight can be taken multiple times, or without replacement. When taken with replacement the sequence of such random weights is independant and indentically distributed (iid). We have then a lot of informations about such a sequence. However when taken without replacement, the weights are not iid . In this presentation we investigate whether some properties of the sampling with replacement can still hold for the sampling without replacement. This question can arise in multiple settings, and in particular random graphs.