Applied Math Seminar

Title: Mathematical modeling of polymerization processes in physiology Abstract: Polymerization, or aggregation, is essential for many physiological systems. For example, the emergence of a fibrin polymer mesh during the formation of a blood clot is required for a stable clot and long-term, sustained intracellular transport in neurons rely on persistent yet dynamic polymers that comprise the […]

Title: Randomized Sums of Graph Spectra Abstract: The adjacency matrix of an Erdős-Rényi-Gilbert graph is a random symmetric matrix whose entries are Bernoulli random variables. These entries, modulo the constraints imposed by symmetry, are independent. We aim to understand the asymptotic behavior of randomized sums of the spectra and singular spectra of these matrices. In […]

Title: Data-driven large eddy simulation of atmospheric turbulence Abstract: Over the last few years, machine learning has been critical in science and engineering and emerged as a data-driven turbulence model. However, machine learning depends on training data from previous experiments on turbulent flows. Typically, training data capture only a fraction of the active scales of […]

Title: Scattering Resonances Through Subwavelength Holes and Their Applications in Imaging and Sensing Abstract: The so-called extraordinary optical transmission (EOT) through metallic nanoholes has triggered extensive research in modern plasmonics and their applications in bio-sensing, imaging, etc. This talk aims to provide quantitative mathematical theories to understand a variety of resonances that induce the EOT […]

Title: Using Mutual Information of Hypergraph Compressions for Clustering Abstract: Hypergraphs are often used to represent higher order observed relationships between subjects of study. In particular, the vertices of a hypergraph could represent the basic elements of study, and edges represent observed relationships between the vertices. Implicitly, the assumption is that observed edges are more […]

Title:Geometric Scattering on Measure Spaces Abstract: Geometric Deep Learning is an emerging field of research that aims to extend the success of convolutional neural networks (CNNs) to data with non-Euclidean geometric structure. Despitebeing in its relative infancy, this field has already found great success in many applications such as recommender systems, computer graphics, and traffic […]

Title: Identification of the order of semilinear subdiffusion with memory Abstract: See attached abstract

Title:Inferring birth and death rates from population size time series data Abstract: Models of population dynamics are usually formulated and analyzed with net growth rates. However, separately identifying birth and death rates is significant in various biological applications such as disambiguating (1) exploitation vs. interference competition in ecology, (2) bacteriostatic vs. bactericidal antibiotics in clinical […]

Title: A common pathway to cancer: oncogenic mutations abolish p53 oscillations. Abstract: The tumor suppressor p53 oscillates in response to DNA double-strand breaks, a behavior that has been suggested to be essential to its anti-cancer function. Nearly all human cancers have genetic alterations in the p53 pathway; a number of these alterations have been shown to […]

Title: PLSS: A Projected Linear Systems Solver (joint work with Michael Saunders) Abstract: Iteratively solving linear systems has proven to be useful for many large applications. Projection methods use sketching matrices (possibly randomized) to generate a sequence of small projected subproblems, but even the smaller systems can be costly. We develop a method in which […]

Title: Modeling size distributions and collisions in cloud microphysics Abstract: Feedbacks between a warming atmosphere, emission of aerosols, and clouds and precipitation are one of the most difficult aspects for climate models to accurately capture. While these models operate at resolutions of tens or hundreds of kilometers, many of the physics that determine how and […]

Title: Towards Understanding the Success of First Order Methods in Training Mildly Overparameterized Networks Abstract: For most problems of interest the loss landscape of a neural network is non-convex and contains a plethora of spurious critical points. Despite this first order methods such as SGD and Adam are in practice remarkably successful at finding optimal, […]