Balázs Szegedy: On the graph limit approach to random regular graphs

Speaker: Balázs Szegedy

Title: On the graph limit approach to random regular graphs

Abstract: We use graph limit techniques combined with information theory to study the structure of random regular graphs. Motivated by recent results in random matrix theory we show that the entries of delocalized eigenvectors of random regular graphs have a Gaussian distribution.  Our theorems work in the more general framework of almost-eigenvectors. Using this we show the Gaussianity of factor of i.i.d eigenvector processes on the d-regular tree if d>2. Joint work with A. Backhausz.