Skip to Main content Skip to Navigation
Theses

Grandes déviations et convergence du spectre de matrices aléatoires

Abstract : One of the main objects of random matrix theory is the spectrum of matrices of large dimension and whose entries are random variables. In this thesis, we concern ourselves with two questions : the large deviations of the largest eigenvalue and the convergence of the empirical measure. Apart from the case of coefficients with Gaussian or heavy-tailed distributions., few large deviation principles are proved in random matrix theory, for the empirical measure or the largest eigenvalue. In the first part of this thesis, we prove a series of large deviation principles for the largest eigenvalue of Wigner matrices, Wishart matrices and matrices with variance profiles with coefficients whose distributions satisfy a sub-Gaussian bound.In the second part of this thesis, we examine the question of the convergence of the empirical measure of polynomials of random matrices. In the case of self-adjoint polynomials in independent Wigner matrices, thanks to the results of Voiculescu, we can see the limit measure as the spectral measure of a polynomial in independently free semi-circular elements. But in the general case, it is necessary to control the smallest singular values in order to conclude. We present such a control for polynomials of degree 2 in Ginibre matrices and we prove the convergence of the empirical measure.
Document type :
Theses
Complete list of metadatas

Cited literature [10 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02497186
Contributor : Abes Star :  Contact
Submitted on : Tuesday, March 3, 2020 - 2:59:08 PM
Last modification on : Wednesday, March 4, 2020 - 1:34:47 AM
Long-term archiving on: : Thursday, June 4, 2020 - 3:39:18 PM

File

HUSSON_Jonathan_2019LYSEN067_T...
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02497186, version 1

Collections

Citation

Jonathan Husson. Grandes déviations et convergence du spectre de matrices aléatoires. Probabilités [math.PR]. Université de Lyon, 2019. Français. ⟨NNT : 2019LYSEN067⟩. ⟨tel-02497186⟩

Share

Metrics

Record views

144

Files downloads

68