Grant: Discovery Grants Program - Individual
Université de Montréal
$27,000.00
Eigenvectors of Random Matrices
FY 2024-25
Summary
Random matrix theory has its roots both in the world of statistics from the work of Wishart and in mathematical physics with the seminal works of Wigner. The main idea from the latter was to model the Hamiltonian of highly correlated systems of heavy nuclei by high-dimensional self-adjoint matrices whose entries are random. The statistics of eigenvalues of such models were a surprisingly good match for this physical system and are now mysteriously ubiquitous both in theoretical mathematics as well as real world systems.
The study of eigenvectors were not part of Wigner's original study but was first considered by Porter and Thomas to study nuclear reaction widths. In the initial analysis of disordered quantum systems, two main behaviours have been conjectured to encompass most non trivial systems: an insulating phase corresponding to localized eigenvectors with Poisson local statistics of eigenvalues, and a conducting phase corresponding to delocalized eigenvectors with strongly dependent eigenvalues.
This proposal seeks to refine the knowledge of the conducting phase for general mean-field models of random matrices as well as studying a new phase of eigenvectors, called the "bad metal" phase or non-ergodic delocalized phase, by considering new models of random matrices.
The analysis of the delocalized phase for general mean-field model of random matrices has seen an extensive progress over the past decade with the seminal works of Erdoes and Yau, and Tao and Vu proving the Wigner-Dyson-Mehta universality conjecture. While this conjecture has first been stated for eigenvalues, a similar one can be stated for eigenvectors: How close to a Haar-distributed matrix on the orthogonal group is the eigenbasis of an i.i.d symmetric random matrix? Statistics such as entry size entry distribution, thermalization (or QUE) are all evidence that eigenvectors behave as if there were close to uniformly distributed on the sphere. This proposal aims to study extreme statistics of eigenvectors such as the largest entry. The only universal property we have so far is a sharp (for the Gaussian case) upper bound with high probability. But questions remain: can we have the corresponding lower bound? Is the whole distribution universal?
There has recently been evidence that the dichotomy of localized and delocalized phase is too coarse to capture the complexity of many body disordered quantum systems. A new phase, called non-ergodic delocalized, where quantum states have unbounded support (delocalized) while their amplitudes are not uniformly distributed (non ergodic) has emerged. The first mathematical model where this phase emerged was the Rosenszweig-Porter model which is now well understood. Physicists are now building more complex (and realistic) models of random matrices where this phase exists in order to investigate intricate operators such as the Anderson model on random regular graphs where the existence of the phase is debated.
Program
Discovery Grants Program - Individual
Recipient
Location
Quebec, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
RGPIN-2023-03882
Name nom
Benigni, Lucas
Department département
Département de mathématiques et de statistique
Organization id
63
Institution établissement
Université de Montréal
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2023
Award amount
27000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1508
Committee name en
Mathematics and Statistics
Committee name fr
Mathématiques et statistique
Area of application code
1205
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Mathematical sciences
Area of application fr
Sciences mathématiques
Research subject code
2900
Research subject group en
Pure mathematics
Research subject group fr
Mathématiques pures
Research subject en
Pure mathematics
Research subject fr
Mathématiques pures
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Eigenvectors of Random Matrices
Keywords
probability; random matrix theory; universality; delocalization; extreme statistics; non-ergodic delocalization; rosenzweig-porter model; levy matrices; beta-ensembles
Application summary
Random matrix theory has its roots both in the world of statistics from the work of Wishart and in mathematical physics with the seminal works of Wigner. The main idea from the latter was to model the Hamiltonian of highly correlated systems of heavy nuclei by high-dimensional self-adjoint matrices whose entries are random. The statistics of eigenvalues of such models were a surprisingly good match for this physical system and are now mysteriously ubiquitous both in theoretical mathematics as well as real world systems.
The study of eigenvectors were not part of Wigner's original study but was first considered by Porter and Thomas to study nuclear reaction widths. In the initial analysis of disordered quantum systems, two main behaviours have been conjectured to encompass most non trivial systems: an insulating phase corresponding to localized eigenvectors with Poisson local statistics of eigenvalues, and a conducting phase corresponding to delocalized eigenvectors with strongly dependent eigenvalues.
This proposal seeks to refine the knowledge of the conducting phase for general mean-field models of random matrices as well as studying a new phase of eigenvectors, called the "bad metal" phase or non-ergodic delocalized phase, by considering new models of random matrices.
The analysis of the delocalized phase for general mean-field model of random matrices has seen an extensive progress over the past decade with the seminal works of Erdoes and Yau, and Tao and Vu proving the Wigner-Dyson-Mehta universality conjecture. While this conjecture has first been stated for eigenvalues, a similar one can be stated for eigenvectors: How close to a Haar-distributed matrix on the orthogonal group is the eigenbasis of an i.i.d symmetric random matrix? Statistics such as entry size entry distribution, thermalization (or QUE) are all evidence that eigenvectors behave as if there were close to uniformly distributed on the sphere. This proposal aims to study extreme statistics of eigenvectors such as the largest entry. The only universal property we have so far is a sharp (for the Gaussian case) upper bound with high probability. But questions remain: can we have the corresponding lower bound? Is the whole distribution universal?
There has recently been evidence that the dichotomy of localized and delocalized phase is too coarse to capture the complexity of many body disordered quantum systems. A new phase, called non-ergodic delocalized, where quantum states have unbounded support (delocalized) while their amplitudes are not uniformly distributed (non ergodic) has emerged. The first mathematical model where this phase emerged was the Rosenszweig-Porter model which is now well understood. Physicists are now building more complex (and realistic) models of random matrices where this phase exists in order to investigate intricate operators such as the Anderson model on random regular graphs where the existence of the phase is debated.