External Lecturer:
Thomas Leblé
Course Type:
PhD Course
Academic Year:
2026-2027
Period:
16-27 November
Duration:
12 h
Description:
Lecturer's webpage:
Description:
The Sine-beta point process describes eigenvalues of large random Hermitian matrices belonging to the universality class of Gaussian beta-ensembles, at microscopic scale, in the bulk of the spectrum. Its mere existence was only proven in the late 2000s, thanks to independent breakthroughs by Valkó–Virág and Killip–Stoiciu.
It is known since the works of Dyson in the 1960s that the eigenvalues of Gaussian ensembles can be interpreted as particles of a certain one-dimensional toy model of statistical physics, the 1d Log-gas. This opens the door to a "statistical physics" perspective on their behavior, which has turned out to be useful. However, this perspective has so far taken the existence of the limiting process for granted.
The goal of these lectures is to sketch how a fully "physical" proof of the existence of Sine-beta would go. This should take us in several directions:
1. Obtaining local laws for the log-gas, down to the microscopic scale, by a careful bootstrap.
2. Deriving the Dobrushin–Lanford–Ruelle (DLR) equations for every limit point of the microscopic process.
3. Proving the "DLR variational principle" and connecting DLR equations to minimizers of a certain "free energy functional".
4. Showing that free energy minimizers are unique, by a convexity argument involving optimal transportation of measures in infinite dimension.
This course is part of the Probability Trimester at SISSA: https://sites.google.com/view/probsissa
Note: the first week is in room 134, the second is in room 136
Research Group:
Location:
A-136
