📖
Grandes valeurs propres de graphes aléatoires dilués
Simon Coste

Grandes valeurs propres de graphes aléatoires dilués

Author / Organization:
Science & Mathematics Published 2019 Verification Pending

Overview & Synopsis

A random n \times n matrix is \emph{diluted} when the number of non-zero entries is of order n; adjacency matrices of d-regular graphs or adjacency matrices of Erdös-Rényi graphs with fixed average degree d are diluted. This dissertation is about the spectrum of diluted random matrices. In the first chapter I show an upper bound on the second eigenvalue of the transition matrix on a diluted directed graph model, the \textit{directed configuration model}, in which the degree (in and out) of each vertex is specified. We also get an important generalization of Friedman's theorem: the second eigenvalue of the adjacency matrix of a directed d-regular graph is less than \sqrt{d}+o(1). A second short chapter, from a collaboration with Charles Bordenave, gives a generalization of the Erdös-Gallai theorem. The third chapter, a collaboration with Justin Salez, solves a problem raised in 2004 by Bauer and Golinelli: the existence (or not) of extended states in the limiting spectrum of Erdös-Rényi graphs with parameter d/n. We show the absence of extended states at zero when de. Our results extend to the spectra of unimodular Galton-Watson tree. I also prove the absence of extended states at zero in the spectrum of the skeleton tree. The last chapter is a collaboration with Charles Bordenave and Raj Rao Nadakuditi. We study the eigenvalues of the adjacency matrix A of a directed Erdös-Rényi graph with parameter d/n, in which the edges are weighted by the entries of a symmetric matrix P. We show a spectacular phase transition: there is a threshold T depending on P and d such that the largest eigenvalues of (n/d)A converge to the eigenvalues of P which are greater than T. The associated eigenvectors of A are aligned with those of P.

Indexed from repository: HAL Open Science (hal.science)
Loading document from verified source...
Keyboard Shortcuts: ← / → for pages • + / - for zoom • 0 to fit width Powered by Mozilla PDF.js v3.11

Document Information

Document Title Grandes valeurs propres de graphes aléatoires dilués
Author / Organization Simon Coste
Subject & Discipline Multidisciplinary Academic Open Research
Categories Science & Mathematics
Publication Year 2019
License / Rights Open Access / Creative Commons
Resource Verification Listed PDF (Verification Pending)
Original Host Website HAL Open Science (hal.science)
File Format application/pdf (Portable Document Format)
Language EN
Catalog ID DOC-132270
Original Source Link Visit Source Page on hal.science »