The goal of this talk is to price American-type financial contracts in the presence of Knightian uncertainty. In particular, instead of choosing a particular probabilistic model to represent the price process of some underlying asset (on which an American option is written), we first restrict our attention to the whole class of models that are…
Dense linear systems in large-scale kernel approximation in machine learning, discretization of boundary integral equations in mathematical physics, and low-rank approximation of Schur complements in large sparse matrix factorization often employ a multilevel structure of low-rank off-diagonal blocks. To solve such systems efficiently, we present a GPU-based parallel recursive skeletonization solver that utilizes batched dense…
As with weather, there is a limit to how far into the future we can accurately forecast the future course of epidemics. Yet, we often need to make plans further into the future than we can reasonably forecast, and in many cases our actions themselves will change the course of an epidemic. One solution to this conundrum is the construction of "planning scenarios" where we project the likely course of an epidemic well beyond the…
In this talk, I will introduce a novel Riemannian geometric model for wave propagation in gas giant planets. Terrestrial planets and gas giants have one key difference: in gas, density goes to zero at the surface, and seismic waves come to a full stop. We model the sound speed in a planet by a Riemannian…
In Stanley’s seminal work “Cohen-Macaulay Complexes”, Stanley conjectured that all h vectors of matroid complexes are pure O-sequences. We constructed coparking functions on matroids with extra restrictions and showed that the degree sequences of coparking functions are the same as h vectors of matroid complexes. By this construction, we proved that Stanley’s conjecture is true…
In this talk, I shall present a relaxation formula and duality theory for the multi-marginal Coulomb cost that appears in optimal transport problems arising in Density Functional Theory. The related optimization problems involve probabilities on the entire space and, as minimizing sequences may lose mass at infinity, it is natural to expect relaxed solutions which…
We consider weakly compressible flows coupled with a cloud system that models the dynamics of warm clouds. Our goal is to explicitly describe the evolution of uncertainties that arise due to unknown input data, such as model parameters and initial or boundary conditions. The developed stochastic Galerkin method combines the space-time approximation obtained by a…
We start by giving a short introduction to geometric inverse problems. Then we present our recent results regarding inverse problems for the minimal surface equation. Minimal surfaces are solutions to a quasilinear elliptic equation. We determine a minimal surface up to an isometry from the corresponding Dirichlet-to-Neumann map in dimension 2. Applications of the results…
Quantum hardware has advanced to the point where it is now possible to perform simulations of small physical systems. Although the current capabilities are limited, given the rapid advancement it is an opportune time to develop novel algorithms for the simulation of quantum matter, and to develop those that make it possible to make connections…
Feasibility models are a powerful approach to many real-world problems where simply finding a point that comes close enough to meeting many, sometimes contradictory demands is enough. In this talk I will outline the theoretical foundations for the convergence analysis of fixed point iterations of expansive mappings, and show how this specializes to fundamental algorithms…
Affect (e.g., beliefs, attitudes, emotions) plays a crucial role in mathematics learning, but reliance on verbal and written responses (from surveys, interviews, etc.) can limit students’ expression of their affective states. As a complement to existing methods, we explore how asking students to graph their affect can be used to study their mathematical experiences. I…
In this talk we present both numerical analysis and experiments to study a few basic computational issues in practice: (1) the numerical error one can achieve given a finite machine precision, (2) the learning dynamics and computation cost to achieve a given accuracy, and (3) stability with respect to perturbations. These issues are addressed for…
Consider a dense Erdös-Rényi random graph with parameters n and p, with p fixed in (0,1). Let H_n(p) be the hitting time between two distinct vertices: run simple random walk from a vertex w, wait until it hits another vertex v, and average over the random walk. What does the distribution of H_n(p) look like? Heuristic…
PyOED is a highly extensible scientific package that enables developing and testing model-constrained optimal experimental design (OED) for inverse problems. Specifically, PyOED aims to be a comprehensive Python toolkit for model-constrained OED. The package targets scientists and researchers interested in understanding the details of OED formulations and approaches. It is also meant to enable researchers…
This talk will focus on point-cloud data, their analysis and biological applications in which they naturally arise. In particular, I will introduce "optimal transport", which has evolved as one of the major frameworks to meaningfully compare point-cloud data and explain how it can be incorporated into classical machine learning algorithms for further downstream analysis. This…
Consider the affine Lie algebra $\mathfrak{g}$ associated with the simple Lie algebra $sl(n)$ consisting of $n\times n$ trace zero matrices over the field of complex numbers. For every dominant integral weight $\lambda$ there is a unique (upto isomorphism) irreducible highest weight $\mathfrak{g}$ module $V(\lambda)$. Although there are infinitely many weights of this module, certain important…
A hollow vortex is a region of constant pressure bounded by a vortex sheet and suspended inside a perfect fluid; we can think of it as a spinning bubble of air in water. In this talk, we present a general method for desingularizing non-degenerate steady point vortex configurations into collections of steady hollow vortices. The…
Physics-Informed Neural Networks (PINNs) have received much attention recently due to their potential for high-performance computations for complex physical systems, including data-based computing, systems with unknown parameters, and others. The idea of PINNs is to approximate the equations and boundary and initial conditions through a loss function for a neural network. PINNs combine the efficiency…