
September 14
Oleg Zaboronski (Warwick)
What is the probability that a large random matrix has no real eigenvalues?
Abstract. In this talk I will discuss the calculation of the probability p(n) that
a large real matrix with independent normal entries has no real eigenvalues.
I will show how to guess the answer using the link between the law of real
eigenvalues and a certain interacting particle system. I will then outline the steps
of the proof of the result based on the determinantal expression for p(n).

September 21
Roland Bauerschmidt (Harvard)
Local eigenvalue statistics for random regular graphs
Abstract. The universality conjecture for random matrices states that the local
spectral statistics are universal for broad classes of random
matrices. It has been proved for random matrix ensembles that are
invariant under large continuous symmetry groups (invariant ensembles)
and for random matrices with i.i.d. entries (Wigner matrices). The
adjacency matrix of a random regular graph is a random matrix with
hard constraints and in neither of these classes. Nonetheless it has
been conjectured and numerically confirmed that the eigenvalues of
random regular graphs obey local random matrix statistics. We prove
that the local semicircle law holds at the optimal scale and moreover
that the local eigenvalue distribution in the bulk of the spectrum is
indeed given by that of the GOE, both under suitable assumptions on
the degrees of the graphs. This is joint work with J. Huang, A.
Knowles, and H.T. Yau.

September 28
Alexey Bufetov (MIT)
Asymptotics of particle systems governed by Schur functions
Abstract. We study the global limit behavior of probabilistic particle systems
whose combinatorics is related to Schur functions. Our main examples are the
decompositions of the tensor products of irreducible representations of classical Lie
groups, the random lozenge tilings, the random domino tilings, and the decompositions
of extreme characters of the infinitedimensional unitary group. We prove the Law of
Large Numbers and the Central Limit Theorem for all these systems.
The talk is based on a joint work with V. Gorin.

Friday October 2
Charles River Lectures.
See link
for information and free registration.

October 5
Mariya Shcherbina (Kharkov)
Fluctuations of eigenvalues of random matrices with independent or
weakly dependent entries
Abstract. We will discuss the method, which allows to prove automatically
CLT for linear eigenvalue statistics (LES) of smooth test functions for almost all
classical models of random matrix theory: deformed Wigner and sample covariance
matrices, sparse matrices, diluted random matrices, matrices with heavy tails,
etc. In particular, the application of the method allows us to prove
CLT for LES of random band ndimensional matrices, whose bandwidth b is assumed to
grow with n in such a way that b/n \to 0, without any additional restrictions. Thus we
remove the main technical restriction n >> b >> n^{1/2} of all the papers, where
LES of band matrices was studied before.

Tuesday (due to Columbus Day) October 13
E17136 (not usual room)
Philippe Rigollet (MIT)
Aggregation in highdimensional statistics
Abstract. Originally introduced as a tool for adaptation to smoothness in
nonparametric statistics, aggregation has become a powerful tool in diverse areas of
statistics and machine learning, primarily in highdimensions, where it can be used
to adapt to the underlying sparsity pattern of the problem. In this talk, I will give
an overview of recent advances in the context of sparse linear regression where
aggregation has been used successfully to derived the sharpest available oracle
inequalities.

October 19
Nicholas Cook (UCLA)
Random regular digraphs: singularity and spectrum
Abstract. We consider two random matrix ensembles associated to large random regular digraphs:
(1) the 0/1 adjacency matrix, and (2) the adjacency matrix with iid bounded edge
weights. Motivated by universality conjectures, we show that the spectral distribution
for the latter ensemble is asymptotically described by the circular law, assuming the
graph has degree linear in the number of vertices. Towards establishing the same
result for the unweighted adjacency matrix, we prove that it is invertible with high
probability, even for sparse digraphs with degree growing only polylogarithmically.

October 26
Peter Winkler (Dartmouth)
Permutons
Abstract. Permutons are doubly stochastic measures, i.e., probability measures
on the unit square with uniform marginals. They function as limit structures
for finite permutations, much as graphons do for graphs. As in the graph
case, a variational principle holds, making it possible to count and describe
certain classes of permutations. Unlike the graph case, most classdescribing
permutons are smooth and thus amenable to standard analytic techniques.
I'll show how both rigorous and empirical methods can help understand
permutations with given pattern densitiesan objective whose surface has
so far only been scratched.
Joint work with Rick Kenyon (Brown), Dan Kral (Warwick) and Charles Radin (Texas).

November 2
Xin Sun (MIT)
BiLaplacian Gaussian field and uniform spanning forests
Abstract. In this talk, I will first review Gaussian free field in $\R^d$ and its
generalization called fractional Gaussian field, which includes log correlated field
and biLaplacian Gaussian field as examples. Fractional Gaussian field arises
naturally as scaling limits of spin models, i.e. Ising model and phi^4 model,
at or above their critical dimension for the mean field behavior. We describe a
simple spin model from uniform spanning forests in $\Z^d$ whose critical dimension is
4 and prove that the scaling limit is the biLaplacian Gaussain field for $d\ge 4$.
At dimension 4, there is a $log n$ correction for the spinspin correlation and the
biLaplacian Gaussian field is a log correlated field. Based on a joint work with
Greg Lawler and Wei Wu and a survey with Asad Lodhia, Sam Watson and Scott Sheffield.

November 9
Shirshendu Ganguly (UW)
Competitive erosion is conformally invariant
Abstract. We study a graphtheoretic model of interface dynamics called
competitive erosion. Each vertex of the graph is occupied by a particle, which
can be either red or blue. New red and blue particles are emitted alternately from
their respective sources and perform random walk. On encountering a particle of the
opposite color they remove it and occupy its position. This is a finite, competitive
version of the celebrated Internal DLA growth model first analyzed by
Lawler, Bramson and Griffeath in 1992.
We establish conformal invariance of competitive erosion on discretizations of
smooth, simply connected planar domains. This is done by showing that at stationarity,
with high probability the red and the blue regions are separated by an orthogonal
circular arc on the disc and more generally by a hyperbolic geodesic. The proof
relies on convergence of solutions of the discrete Poisson problem with Neumann
boundary conditions to their continuous counterparts and robust IDLA estimates.
(Joint work with Yuval Peres, available at
arXiv:1503.06989).

November 16
Hoi Nguyen (Ohio)
Anti concentration of random walks and eigenvalue repulsion of random matrices
Abstract. I will address recent characterization results on random walks
(in both abelian and nonabelian groups) which stick to a small region unusually long.
As an application, we show a Wegnertype estimate for the number of eigenvalues
inside an extremely small interval for Wigner matrices of discrete type.

November 23
Guillaume Barraquand (Columbia)
Random walks in Beta random environment
Abstract. We consider a model of random walks in spacetime random environment,
with Betadistributed transition probabilities. This model is exactly
solvable, in the sense that the law of the (finite time) position of the
walker can be completely characterized by Fredholm determinantal
formulas. This enables to prove a limit theorem towards the TracyWidom
distribution for the second order corrections to the large deviation
principle satisfied by the walker, thus extending the scope of KPZ
universality to RWRE. We will also discuss a few similar results about
degenerations of the model: a first passage percolation model which is
the "zerotemperature" limit, and a certain diffusive limit which leads
to wellstudied stochastic flows. (Work in collaboration with Ivan Corwin).

November 30
Damien Gayet (Institut Fourier)
Universal components of random nodal hypersurfaces
Abstract. Let S be a compact hypersurface of R^n, and M be a compact
Riemannian manifold of dimension n. I will explain that if we take at
random a linear combination of eigenfunctions of the Laplacian on M with
eigenvalues less than L, then, in average, at least c_S Vol(M) L^{n/2}
diffeomorphic copies of S appear in the vanishing locus of this sum, as
L grows to infinity, where c_S is a positive constant depending only on
S. This is a joint work with JeanYves Welschinger.

December 7
Eyal Lubetzky (Courant)
Effect of initial conditions on mixing for spin systems
Abstract. Recently, the "information percolation" framework was introduced as
a way to obtain sharp estimates on mixing for spin systems at high temperatures, and
in particular, to establish cutoff for the Ising model in three dimensions up to
criticality from a worst starting state. I will describe how this method can be used
to understand the effect of different initial states on the mixing time, both
random (''warm start'') and deterministic.
Joint work with Allan Sly.