Localization and delocalization of eigenvectors for heavy-tailed random matrices

成果类型:
Article
署名作者:
Bordenave, Charles; Guionnet, Alice
署名单位:
Centre National de la Recherche Scientifique (CNRS); Universite de Toulouse; Universite Toulouse III - Paul Sabatier; Centre National de la Recherche Scientifique (CNRS); Ecole Normale Superieure de Lyon (ENS de LYON)
刊物名称:
PROBABILITY THEORY AND RELATED FIELDS
ISSN/ISSBN:
0178-8051
DOI:
10.1007/s00440-012-0473-9
发表日期:
2013
页码:
885-953
关键词:
semicircle law UNIVERSALITY spectrum statistics
摘要:
Consider an Hermitian random matrix with, above the diagonal, independent entries with -stable symmetric distribution and . We establish new bounds on the rate of convergence of the empirical spectral distribution of this random matrix as goes to infinity. When and , we give vanishing bounds on the -norm of the eigenvectors normalized to have unit -norm. On the contrary, when , we prove that these eigenvectors are localized.
来源URL: