A key question in population biology is understanding the conditions under which the species of an ecosystem persist or go extinct. Theoretical and empirical studies have shown that persistence can be facilitated or negated by both biotic interactions and environmental fluctuations. We study the dynamics of n interacting species that live in a stochastic environment. Our models are described by n dimensional piecewise deterministic Markov processes. These are processes (X(t), r(t)) where the vector X denotes the density of the n species and r(t) is a finite state space process which keeps track of the environment. In any fixed environment the process follows the flow given by a system of ordinary differential equations. The randomness comes from the changes or switches in the environment, which happen at random times. We give sharp conditions under which the populations persist as well as conditions under which some populations go extinct exponentially fast. As an example we look at the competitive exclusion principle from ecology, which says in its simplest form that two species competing for one resource cannot coexist, and show how the random switching can facilitate coexistence.

Description of some work with Elizabeth Meckes at SEPC 2021

An emerging way to protect privacy is to replace true data by synthetic data. Medical records of artificial patients, for example, could retain meaningful statistical information while preserving privacy of the true patients. But what is synthetic data, and what is privacy? How do we define these concepts mathematically? Is it possible to make synthetic data that is both useful and private? I will tie these questions to a simple-looking problem in probability theory: how much information about a random vector X is lost when we take conditional expectation of X with respect to some sigma-algebra? This talk is based on a series of papers with March Boedihardjo and Thomas Strohmer.

SEPC 2021 in honor of Elizabeth Meckes. Slides from the talks and more information are available <a href="https://services.math.duke.edu/~rtd/SEPC2021/SEPC2021.html">here</a>

Branching Brownian motion is a classical process in probability theory belonging to the class of “Log-correlated random fields”. We study the structure of extreme level sets of this process, namely the sets of particles whose height is within a fixed distance from the order of the global maximum. It is well known that such particles congregate at large times in clusters of order-one genealogical diameter around local maxima which form a Cox process in the limit. We add to these results by finding the asymptotic size of extreme level sets and the typical height and shape of those clusters which carry such level sets. We also find the right tail decay of the distribution of the distance between the two highest particles. These results confirm two conjectures of Brunet and Derrida.(joint work with A. Cortines, O Louidor)

Data lying in a high dimensional ambient space are commonly thought to have a much lower intrinsic dimension. In particular, the data may be concentrated near a lower-dimensional subspace or manifold. There is an immense literature focused on approximating the unknown subspace and the unknown density, and exploiting such approximations in clustering, data compression, and building of predictive models. Most of the literature relies on approximating subspaces and densities using a locally linear, and potentially multiscale, dictionary with Gaussian kernels. In this talk, we propose a simple and general alternative, which instead uses pieces of spheres, or spherelets, to locally approximate the unknown subspace. I will also introduce a curved kernel called the Fisher–Gaussian (FG) kernel which outperforms multivariate Gaussians in many cases. Theory is developed showing that spherelets can produce lower covering numbers and mean square errors for many manifolds, as well as the posterior consistency of the Dirichlet process mixture of the FG kernels. Time permitting, I will also talk about an ongoing project about stochastic differential geometry.

The large deviations analysis of solutions to stochastic differential equations and related processes is often based on approximation. The construction and justification of the approximations can be onerous, especially in the case where the process state is infinite dimensional. In this work we show how such approximations can be avoided for a variety of infinite dimensional models driven by some form of Brownian noise. The approach is based on a variational representation for functionals of Brownian motion. Proofs of large deviations properties are reduced to demonstrating basic qualitative properties (existence, uniqueness, and tightness) of certain perturbations of the original process. This is a joint work with P.Dupuis and V.Maroulas.

We introduce some recent advances in the study of nonlinear stochastic heat equations, and related stochastic PDEs. Special attention will be paid to the local structure of the solution. In particular, we show that, frequently, the solution exhibits a form of intermittency. Time permitting, we discuss related connections to classical potential theory and mathematical physics as well.

Points fall into the unit interval according to a certain rule, splitting it up into fragments. An example rule is the following: at each step, two points are randomly drawn from the unit interval and the one that falls into the smaller (or larger) interval is discarded, while the other one is kept. This process is inspired by the so-called "power of choice" paradigm originating in the computer science literature on balanced load allocation models. The question of interest is how much the rule affects the geometry of the point cloud. With Elliot Paquette [1] we introduced a general version of this interval fragmentation model and showed that the empirical distribution of rescaled interval lengths converges almost surely to a deterministic probability measure. I will report on this work as well as on work in progress [2] where we show that the empirical measure of the points converges almost surely to the uniform distribution. The proofs involve techniques from stochastic approximation, non-linear integro-differential equations, ergodic theory for Markov processes and perturbations of semigroups on L^p spaces, amongst other things. [1] Maillard, P., & Paquette, E. (2016). Choices and intervals. Israel Journal of Mathematics, 212(1), 337–384. [2] Maillard, P., & Paquette, E. (in preparation). Interval fragmentations with choice: equidistribution and the evolution of tagged fragments

Consider factor (graphical) models on sparse graph sequences that converge locally to a random tree T. Using a novel interpolation scheme we prove existence of limiting free energy density under uniqueness of relevant Gibbs measures for the factor model on T. We demonstrate this for Potts and independent sets models and further characterize this limit via large-deviations type minimization problem and provide an explicit formula for its solution, as the Bethe free energy for a suitable fixed point of the belief propagation recursions on T (thereby rigorously generalize heuristic calculations by statistical physicists using ``replica'' or ``cavity'' methods). This talk is based on a joint work with Andrea Montanari and Nike Sun.

For several classical matrix models the joint density of the eigenvalues can be written as an expression involving a Vandermonde determinant raised to the power of 1, 2 or 4. Most of these examples have beta-generalizations where this exponent is replaced by a parameter beta>0. In recent years the point process limits of various beta ensembles have been derived. The limiting processes are usually described as the spectrum of certain stochastic operators or with the help of a coupled system of SDEs. In the bulk beta Hermite case (which is the generalization of GUE) there is a nice geometric construction of the point process involving a Brownian motion in the hyperbolic plane, this is the Brownian carousel. Surprisingly, there are a number of other limit processes that have carousel like representation. We will discuss a couple of examples and some applications of these new representations. Joint with Balint Virag.

First passage percolation is the study of a random metric space generated by replacing each edge in a graph by an edge of a random length. The distance between two vertices u and v is the length of the shortest path connecting u and v. An infinite path P is a geodesic if for any two vertices u and v on P the shortest path between them in the random graph is along P. It is easy to show that in the nearest neighbor graph with vertices Z^2 that there exists at least one (one sided) infinite geodesic starting at any given vertex. It is widely expected that there are infinitely many such one sided infinite geodesics that begin at the origin, with (at least) one in every direction. But it turns out to be very difficult to prove that there are even two with positive probability. We will discuss some recent results which get closer to proving this widely held belief.