Past Seminars

Abstract:

Optimal transport is a powerful tool for measuring the distances between
signals. However, the most common choice is to use the Wasserstein
distance where one is required to treat the signal as a probability
measure. This places restrictive conditions on the signals and although
ad-hoc renormalisation can be applied to sets of unnormalised measures
this can often dampen features of the signal. The second disadvantage is
that despite recent advances, computing optimal transport distances for
large sets is still difficult. In this talk I will focus on the
Hellinger--Kantorovich distance, which can be applied between any pair
of non-negative measures. I will describe how the distance can be
linearised and embedded into a Euclidean space. The Euclidean distance
in the embedded space is approximately the Wasserstein distance in the
original space. This method, in particular, allows for the application
of off-the-shelf data analysis tools such as principal component
analysis.

This is joint work with Bernhard Schmitzer (TU Munich).

Matthew is a research fellow in the Cantab Capital Institute for the
Mathematics of Information at the University of Cambridge. Prior to that
he was a postdoctoral associate at Carnegie Mellon University and a PhD
student at the University of Warwick. From this coming March he will be
a lecturer (US equivalent Assistant Professor) in Applied Mathematics at
the University of Manchester.

Abstract:

Abstract: This talk is about stochastic homogenization of linear elliptic equations in divergence form. Let $a(x)=h(G(x))$ be a diffusion coefficient field, where $h$ is a Lipschitz function and $G$ is a Gaussian field (with possibly thick tail). Solutions $u_\varepsilon$ of elliptic equations $-\nabla \cdot a(\cdot/\varepsilon) \nabla u_\varepsilon = \nabla \cdot f$ in $\mathbb R^d$ with such random heterogeneous coefficients $a$ both oscillate spatially and fluctuate randomly at scale $\varepsilon >0$. I will show how suitable quantitative two-scale expansions allow one to reduce the analysis of oscillations and fluctuations of solutions to bounds on the corrector and fluctuations of the homogenization commutator, respectively. The main probabilistic ingredient is Malliavin calculus, and the main analytical ingredient is large-scale elliptic regularity. This is based on joint works with Mitia Duerinckx, Julian Fischer, Stefan Neukamm, and Felix Otto.

Abstract:

Firearm deaths and injuries are a significant problem in the United States. Indeed, the American Medical Association recently called firearm violence “a public health crisis” and called for a comprehensive public health response and solution. Gun violence in America exacts a significant toll on our society in both human and economic terms. Some argue that Americans have a moral obligation to address the issue of gun violence. But even from a more concrete perspective, the economic cost of firearms directly impacts the financial outcomes of insurers and taxpayers. There is a clear need for unbiased and objective research on the societal and economic impact of firearms. Actuaries are well positioned to study the mortality and morbidity related to firearms, both to quantify the risk and to inform governmental and public health interventions to mitigate the risk associated with firearms. Yet there is little on the topic in the actuarial and insurance literature. In this talk, I will provide a brief overview on the scope of firearm deaths and injuries and examine the extent to which actuaries and insurance professionals have studied or addressed the issue. I will compare firearm risk to risks that are considered in the underwriting process for life and homeowners insurance. I will describe some existing insurance products related to firearm risk as well as proposed legislation regarding gun liability insurance. In a different vein, if time permits, I will discuss preliminary work on a dynamical systems model of gun violence within a population. We are studying how, in an idealized model, changes to various policy parameters affect the long-term behavior of a system. Finally, I will describe some of the many open questions related to gun violence that are amenable to study by actuaries and mathematicians. Bio: https://sites.lsa.umich.edu/ksmoore/

Abstract:

Young diagrams are fundamental combinatorial objects in representation theory and algebraic geometry. Many constructions that rely on these objects depend on variations of a straightening process that expresses a filling of a Young diagram as a sum of semistandard tableaux subject to certain relations. It has been a long-standing open problem to give a non-iterative formula for this straightening process. In this talk I will give such a formula. I will then use this non-iterative formula give a proof that the coefficient of the leading term in the straightening is either 1 or -1, generalizing a theorem of Gonciulea and Lakshmibai.

Abstract:

Mirror symmetry is a phenomenon which relates the symplectic geometry of one space X to the algebraic geometry of another space Y. One consequence is that a canonical basis of regular functions on Y can be defined in terms of certain counts of holomorphic curves in X. I'll discuss the application of this to (quantum) cluster algebras --- certain combinatorially defined algebras whose definition was motivated by the appearance of canonical bases in representation theory and Teichmüller theory.

Abstract:

In this talk, I will introduce a new unsupervised learning framework for data points that lie on or close to a smooth manifold naturally equipped with a group action. In many applications, such as cryo-electron microscopy image analysis and shape analysis, the dataset of interest consists of images or shapes of potentially high spatial resolution, and admits a natural group action that plays the role of a nuisance or latent variable that needs to be quotient out before useful information is revealed. We define the pairwise group-invariant distance and the corresponding optimal alignment. We construct a graph from the dataset, where each vertex represents a data point and the edges connect points with small group-invariant distance. In addition, each edge is associated with the estimated optimal alignment group. Inspired by the vector diffusion maps proposed by Singer and Wu, we explore the cycle consistency of the group transformations under multiple irreducible representations to define new similarity measures for the data. Utilizing the representation theoretic mechanism, multiple associated vector bundles can be constructed over the orbit space, providing multiple views for learning the geometry of the underlying base manifold from noisy observations. I will introduce three approaches to systematically combine the information from different representations, and show that by exploring the redundancy created artificially across irreducible representations of the transformation group, we can get drastically improved nearest neighbor identification, when a large portion of the true edges are corrupted. I will also show the application in cryo-electron microscopy image analysis.

Abstract:

In the applied sciences one is often confronted with free boundaries, which arise when the solution to a problem consists of a pair: a function u (often satisfying a partial differential equation (PDE)), and a set where this function has a specific behavior. Two central issues in the study of free boundary problems and related problems in calculus of variations and geometric measure theory are:(1) What is the optimal regularity of the solution u?
(2) How smooth is the free boundary (or how smooth is a certain set related to u)?

In this talk, I will overview recent developments in obstacle type problems and almost minimizers of Bernoulli-type functionals, illustrating techniques that can be used to tackle questions (1) and (2) in various settings.

The study of the classical obstacle problem - one of the most renowned free boundary problems - began in the ’60s with the pioneering works of G. Stampacchia, H. Lewy and J. L. Lions. During the past five decades, it has led to beautiful and deep developments in the calculus of variations and geometric partial differential equations. Nowadays obstacle type problems continue to offer many challenges and their study is as active as ever.
While the classical obstacle problem arises from a minimization problem (as many other PDEs do), minimizing problems with noise lead to the notion of almost minimizers. Interestingly, though deeply connected to "standard" free boundary problems, almost minimizers do not satisfy a PDE as minimizers do, requiring additional tools from geometric measure theory to address (1) and (2).

Abstract:

Topology studies the shape of spaces. Arithmetic statistics studies the behavior of random algebraic objects such as integers and polynomials. I will talk about a circle of ideas connecting these two seemingly unrelated areas. To illuminate the connection, I will focus on three concrete examples: (1) the Burau representation of the braid groups, (2) analytic number theory for effective 0-cycles on a variety, and (3) cohomology of the space of multivariate irreducible polynomials. These projects are parts of a broader research program, with numerous contributions by topologists, algebraic geometers, and number theorists in the past decade, and lead to many future directions yet to be explored.
(PS. This will be a rehearsal of a job talk accessible to the general audience. Any comments or suggestions are appreciated.)

Abstract:

If we want the solution to the Schrodinger equation to converge to its initial data pointwise, what's the minimal regularity condition for the initial data should be? I will present recent progress on this classic question of Carleson. This pointwise convergence problem is closely related to other problems in PDE and geometric measure theory, including spherical average Fourier decay rates of fractal measures, Falconer's distance set conjecture, etc. All these problems essentially ask how to control Schrodinger solutions on sparse and spread-out sets, which can be partially answered by several recent results derived from induction on scales and Bourgain-Demeter's decoupling theorem.

Abstract:

In this talk, we will discuss 1-2-1 networks, which offer a simple yet informative model for mmWave networks. In such networks, it is assumed that two nodes can communicate only if they point beams at each other, otherwise the signal is received well below the thermal noise floor. The focus of the talk will be on single unicast 1-2-1 networks, where the communication from a source to a destination is assisted by a number of relays. In the first part of the presentation, we will characterize the maximum flow of information over such 1-2-1 networks. In particular, we will show that the Shannon capacity can be approximated by routing information along a polynomial (in the network size) number of paths between the source and the destination, and that the scheduling of the node beam orientations can be efficiently performed. In the second part of the presentation, we will analyze the security aspect of such 1-2-1 networks in the presence of an external eavesdropper who wiretaps a set of edges of her choice. In particular, we will derive secure capacity results, which highlight fundamental differences between the traditional secure network coding and security over 1-2-1 networks.

Martina Cardone is currently a tenure-track Assistant Professor within the Electrical and Computer Engineering department at the University of Minnesota. She received her B.Sc. and her M.Sc. in Telecommunications Engineering from Politecnico di Torino, Italy in 2009 and 2011, respectively. As part of a double degree program, she also received a M.Sc. degrees in Telecommunications Engineering from Télécom ParisTech, France in 2011. In 2015, she received her Ph.D. in Electronics and Communications from Télécom ParisTech (with work done at Eurecom in Sophia Antipolis, France), where she worked with Professor Raymond Knopp and Professor Daniela Tuninetti. From July 2015 to August 2017 she was a post-doctoral research fellow in the Electrical and Computer Engineering department at the University of California, Los Angeles, where she worked with Professor Christina Fragouli. From November 2017 to January 2018, she was a post-doctoral associate in the Electrical and Computer Engineering department at the University of Minnesota. She regularly serves on the Technical Program Committee of IEEE workshops and conferences. Her main research interests are in network information theory, wireless communications, network privacy and secrecy, network coding and distributed computing. She was the reci

Abstract:

For the action of a group on the plane, the group equivalence problem for curves can be stated as: given two curves, decide if they are related by an element of the group. The signature method, using differential invariants, to answer the local group equivalence problem for smooth curves and its application to image science has been extensively studied. For planar algebraic curves under subgroups of the general linear group, we show that this provides a method to associate a unique algebraic curve to each equivalence class, the algebraic curve's signature curve. However, computing the implicit equation of the signature curve is a challenging problem. In this talk we consider signatures of algebraic curves, show how to compute the degree without computing its defining polynomial explicitly, and present some results on the structure of signature curves for generic algebraic curves of fixed degree. Additionally we show that this leads to a method to solve the group equivalence problem for algebraic curves using numerical algebraic geometry.

Abstract:

The Pieri rule for the product of a Schur function and a single row Schur function is notable for having an elegant bijective proof that can be intuited by the rule’s concise diagrammatic interpretation, to wit, by appending cells to a Young diagram. Now, key polynomials generalize Schur polynomials to a basis of the full polynomial ring, in which they also refine the Schubert basis via a nice formula. In this talk, I will describe a Pieri rule for the product of a key polynomial and a single row key polynomial that can be analogously interpreted as appending cells to a key diagram, albeit potentially dropping some cells in between each cell addition. I will also outline the main points of the rule’s bijective proof, and in the process hopefully illustrate the utility of understanding the rule from a diagrammatic perspective. Joint work with Sami Assaf.

Abstract:

The capacity to quantify crop architecture and morphology is foundational to the development of higher yielding cultivars via hybridization and genetic engineering. However, at the mass scale required by the science, manual plant phenotyping with physical instruments is arduous, time consuming, subjective, and a leading cause of undergraduate burnout in the UMN plant genetics department. While the process has been slightly improved with manual image analysis, such is almost as time consuming and remains subject to human error. Thereby, in efforts to further expedite phenotyping processes, circumvent human error, and provide more detailed analyses, we aim to completely automate plant phenotyping processes for the UMN plant genetics department and beyond. Working from over 15,000 soybean plants, we’ve advanced robust image processing platforms for measuring petiole and stem length, leaf area, leaf shape via signature curves, and branch angles via energy minimization in 2D, and begun preliminary work at 3D reconstructions from 2D data for 3D branch angles and further analyses. After data extraction and verification, we plan to implement clustering algorithms and machine learning to automatically group plant phenotypes as well as conduct principal component analysis to assemble an allometry space and identify the primary genes of influence.

Abstract:

How many critical points does a smooth random function on a high-dimensional space typically have at a given height? how are their distances distributed? what is the volume or geometry of the level sets? can we design efficient optimization algorithms for the random function? For the spherical spin glass models, those questions are closely related to the structure of the Gibbs measures, which have been extensively studied in physics since the 70s.

I will start with an overview of the celebrated Parisi formula and ultrametricity property. I will then describe an alternative method to analyze the Gibbs measure using critical points, in the setting of the pure spherical models. Finally, I will explain how the latter can be extended to all spherical models, using another (soft) geometric approach, while at the same time making rigorous and generalizing the famous Thouless-Anderson-Palmer approach from physics.

Abstract:

https://www.fitmetrix.io/webportal/schedulemobile/f9719b20-4914-e911-a97...(Legs%20%2B%20Butt)&dateRangeFrom=2019-11-21T07%3A00%3A00&dateRangeTo=2019-11-21T07%3A55%3A00&locationid=5854&classID=31918145

Abstract:

Density functional theory (DFT) has been a very successful effective theory of many-body quantum mechanics. In particular, the Kohn-Sham (KS) equations of DFT serve as an accurate model for the electron densities. The KS equations are a case of the Schrodinger-Poisson equations whose electron-electron effective interaction potential only depends on the density of electrons. When the number of electrons are limited, the KS equation can be solved quickly by numerical method at temperature T = 0. Since physically interesting settings are at T > 0, we study the KS equations at positive temperature and give an iterative scheme to construct solutions.

One important class of electronic structures described by the KS equations is a crystalline lattice. At positive temperature, we show that a local perturbation to a crystalline structure induces an electric field governed by the Poisson equation. The latter equation emerges as an effective equation of the KS equations. This is a joint work with Israel M. Sigal.

Abstract:

Modern data are usually high-dimensional with noise and corruption. A useful representation of data has to be robust and address the data structure. In this talk, I will first present a class of robust models called the scattering transform that can be used to generated features from graph data. In graph scattering transforms, the representation is generated in an unsupervised manner based on graph wavelets. It is approximately invariant to permutations and stable to signal or graph manipulations. Numerical results show that it works effectively for classification and community detection problems. Next, I will address how the structure of data can be found using autoencoders. Indeed, in the framework of autoencoders, graph scattering transform can be applied to the important task of graph generation. It shows state-of-the-art performance in link prediction and can be used to generate molecular samples.

Abstract:

This talk will focus on the Casselman-Shalika formula for GL_2 over a non-Archimedean local field, which is an explicit formula for the values of the spherical Whittaker function. A good amount of time will be dedicated to explaining the necessary background including Whittaker models and spherical vectors, which come together to form the spherical Whittaker function. We will then be ready to discuss the formula, Casselman's method of calculating it, and its significance.

Abstract:

Data Scientist has emerged as one of the hottest and most talked about jobs in the world today.  In my talk, I will provide an overview of how data science has emerged in the insurance industry. I will give some examples of how data science is being applied in life insurance and describe how the Actuarial profession is adapting. Bio: https://mcfam.dl.umn.edu/people/gary-hatfield

Abstract:

Mathematical models can support biomedical research through identification of key mechanisms, validation of experiments, and simulation of new therapeutic approaches.

We investigate the evolution of melanomas under adoptive cell transfer therapy with cytotoxic T-cells. It was shown in experiments that phenotypic plasticity, more precisely an inflammation-induced, reversible dedifferentiation, is an important escape mechanism for the tumor. Recently, the effects of possible mutation to a permanently resistant genotype were studied by introducing knockout melanoma cells into the wildtype tumor.

We use a stochastic individual-based Markov process to describe the evolution of the tumor under various therapeutic approaches. It is an extension of the model introduced in the paper of Baar et al in 2016 and further includes the effects of T-cell exhaustion and some limited spatial component which results in additional non-linearities. The model is implemented as a hybrid algorithm that combines Gillespie-type stochastic calculations and a deterministic approximation to speed up simulations while keeping the effects of random events.

Numerical simulations confirm the resistance to therapy via phenotypic switching as well as genotypic mutation. T-cell exhaustion is identified as an important mechanism that is crucial in fitting the model to the experimental data. We gain further insights into how originally unfit knockout cells can accumulate under therapy, shield the wild type cells from the T-cells, and thus cause an earlier relapse. Going beyond the experiment, the possibility of naturally occurring rare mutations, in contrast to artificially introduced knockout cells, is explored in simulations and produces the same effects. Thus, the clinical relevance of the experimental findings can be confirmed.

Abstract:

Mathematical models can support biomedical research through identification of key mechanisms, validation of experiments, and simulation of new therapeutic approaches.

We investigate the evolution of melanomas under adoptive cell transfer therapy with cytotoxic T-cells. It was shown in experiments that phenotypic plasticity, more precisely an inflammation-induced, reversible dedifferentiation, is an important escape mechanism for the tumor. Recently, the effects of possible mutation to a permanently resistant genotype were studied by introducing knockout melanoma cells into the wildtype tumor.

We use a stochastic individual-based Markov process to describe the evolution of the tumor under various therapeutic approaches. It is an extension of the model introduced in the paper of Baar et al in 2016 and further includes the effects of T-cell exhaustion and some limited spatial component which results in additional non-linearities. The model is implemented as a hybrid algorithm that combines Gillespie-type stochastic calculations and a deterministic approximation to speed up simulations while keeping the effects of random events.

Numerical simulations confirm the resistance to therapy via phenotypic switching as well as genotypic mutation. T-cell exhaustion is identified as an important mechanism that is crucial in fitting the model to the experimental data. We gain further insights into how originally unfit knockout cells can accumulate under therapy, shield the wild type cells from the T-cells, and thus cause an earlier relapse. Going beyond the experiment, the possibility of naturally occurring rare mutations, in contrast to artificially introduced knockout cells, is explored in simulations and produces the same effects. Thus, the clinical relevance of the experimental findings can be confirmed.

Abstract:

Latent factor models are widely used to measure unobserved latent traits in social and behavioral sciences, including psychology, education, and marketing. Motivated by the applications of latent factor models to large-scale measurements which consist of many manifest variables (e.g. test items) and a large sample size, we study the properties of latent factor models under an asymptotic setting where both the number of manifest variables and the sample size grows to infinity. In this talk, I will introduce generalized latent factor models under exploratory and confirmatory settings. For the exploratory setting, we propose a constrained joint maximum likelihood approach for model estimation and investigate its theoretical properties. For the confirmatory setting, we study how the design information affects the identifiability and estimability of the model, and propose a rate-optimal estimator when the model is identifiable. The estimators can be efficiently computed through parallel computing. Our results provide insights on the design of large-scale measurement and have important implications on measurement validity.

Abstract:

Rankin-Selberg method has been one of the most powerful techniques for studying the Langlands program. In this talk, we will start with the original simplest example of the Rankin-Selberg method, and then come to a more general case of the Rankin-Selberg method on GL_m*GL_n where we can reduce the global integral to the more accessible lovely local integrals so that we can establish some of the important analytic properties of the L-functions.

Abstract:

The Deligne category Rep(S_t) can be thought of as "interpolating" the representation categories of symmetric groups. After describing this category, I will explain how a calculation in the Deligne category can be used to prove stability properties of permutation patterns within conjugacy classes (joint with Christian Gaetz).

Abstract:

We propose a strategy for improving the existing methods for solving synchronization problems that arise from various computer vision tasks. Specifically, our strategy identifies severely corrupted relative measurements based on cycle consistency information. To the best of our knowledge, this paper provides the first exact recovery guarantees using cycle consistency information. This result holds for a noiseless but corrupted setting as long as the ratio of corrupted cycles per edge is sufficiently small. It further guarantees linear convergence to the desired solution. We also establish stability of the proposed algorithm to sub-Gaussian noise.

Abstract:

Both the Langlands-Shahidi method of studying automorphic L-functions and approach via the theory of Weyl group multiple Dirichlet series to studying moments of L-functions now require new classes of groups with which to work. In this talk, I will explain our progress on extending these techniques to certain infinite-dimensional Kac-Moody groups, namely loop groups (and their metaplectic covers). Of note in our work is the presence of two quite different types of Eisenstein series that exist on the same group and which need to be considered in conjunction with one other. This is a report on joint work in progress with H. Garland, S.D. Miller, and A. Puskas.

Abstract:

In this talk, several topics from High-dimensional probability shall be discussed. This fascinating area is rich in beautiful problems, and several easy-to-state questions will be outlined. Further, some connections between them will be explained throughout the talk.

I shall discuss several directions of my research. One direction is invertibility properties of inhomogeneous random matrices: I will present sharp estimates on the small ball behavior of the smallest singular value of a very general ensemble of random matrices, and will briefly explain the new tools I developed in order to obtain these estimates.

Another direction is isoperimetric-type inequalities in high-dimensional probability. Such inequalities are intimately tied with concentration properties of probability measures. Among other results, I will present a refinement of the concavity properties of the standard gaussian measure in an n-dimensional euclidean space, under certain structural assumptions, such as symmetry. This result constitutes the best known to date estimate in the direction of the conjecture of Gardner and Zvavitch from 2007.

The above topics will occupy most of the time of the presentation. In addition, I shall briefly mention other directions of my research, including noise-sensitivity estimates for convex sets, or, in other words, upper bounds on perimeters of convex sets with respect to various classes of probability distributions. If time permits, I will discuss my other results, such as small ball estimates for random vectors with independent coordinates, and partial progress towards Levi-Hadwiger illumination conjecture for convex sets in high dimensions.

Abstract:

Sparse recovery finds numerous applications in different areas, for example, engineering, computer science, business, applied mathematics, and statistics. Sparse recovery is often formulated as relatively large-scale and challenging constrained (convex or nonconvex) optimization problems. Constraints are ubiquitous and important in many applications of sparse recovery, but they make analysis and computation nontrivial and require novel techniques to handle them. The goal of this talk is to present numerical and analytical techniques for constrained sparse recovery using convex analysis and optimization tools. Three topics are investigated in the realm of constrained sparse recovery.

First, we analyze quantitative adverse properties of different $p$-norm-based optimization problems with $p>1$, such as generalized basis pursuit, basis pursuit denoising, ridge regression, and elastic net. We show that their optimal solutions are least sparse for almost all measurement matrices and measurement vectors. Second, we study the solution uniqueness of an individual feasible vector of a class of convex optimization problems involving convex piecewise affine functions and subject to general polyhedral constraints. We apply these solution uniqueness results to a broad class of $\ell_1$-minimization problems in constrained sparse optimization, such as basis pursuit, LASSO, and polyhedral gauge recovery. Third, we propose a constrained matching pursuit algorithm for constrained sparse recovery and develop uniform conditions for exact support and vector recovery on constraint sets. The exact recovery via this algorithm not only depends on a measurement matrix but also critically relies on a constraint set. Hence, we identify an important class of constraint sets, called coordinate projection admissible. We then use the conic hull structure of these sets together with constrained optimization techniques to establish sufficient conditions for uniform exact recovery via this algorithm on coordinate projection admissible sets. These conditions are expressed in terms of the restricted isometry-like and the restricted orthogonality-like constants

Abstract:

For a Coxeter group W, a W-graph is a graph which produces a nice basis of the corresponding representation of W and also describes the action of W on the basis elements. Even when W is finite and its irreducible characters are known, W-graphs are still useful for understanding representations of W. In this talk, I will talk about W-graphs when W is an (extended) affine symmetric group, especially when these graphs are associated with “two-row partitions”. Also I will discuss the connection between them and Lusztig’s periodic W-graph. This work is joint with Pavlo Pylyavskyy.

Abstract:

This is the first of a series of three talks on the spectral sequence of a filtered complex.
This material is by now classical and is an important part of homological algebra.
The main difficulty in dealing with spectral sequences is that there are a lot of indexes involved and this is a considerable obstacle to understanding what is going on. The goal of these talks is to present this material, including most proofs, in an accessible manner.

Abstract:

Data generated as part of a real-life experiment is often quite organized. So much so that, in many cases, projecting the data onto a random line has a high probability of uncovering a clear division of the data into two well-separated groups. In other words, the data can be clustered with a high probability of success using a hyperplane whose normal vector direction is picked at random. We call such data ``highly likely clusterable.” The clusters obtained in this fashion often do not seem compatible with a cluster structure in the original space. In fact, the data in the original space may not contain any cluster at all. This talk is about this surprising phenomenon. We will discuss empirical ways to detect it as well as how to exploit it to cluster datasets, especially datasets consisting of a small number of points in a high-dimensional space. We will also present a possible mathematical model that would explain this observed phenomenon. This is joint work with Alden Bradford (Purdue Math), Sangchun Han (Purdue ECE, now at Google) and Tarun Yellamraju (Purdue ECE, now at Qualcomm).

Mireille (Mimi) Boutin graduated with a bachelor’s degree in Physics-Mathematics from the University of Montreal. She received the Ph.D. degree in Mathematics from the University of Minnesota under the direction of Peter J. Olver. She joined Purdue University after a post-doctorate with David Mumford, David Cooper, and Ben Kimia at Brown University, Rhode Island, followed by a post-doctorate with Stefan Muller at the Max Plank Institute for Mathematics in the Sciences in Leipzig, Germany. She is currently an Associate Professor in the School of Electrical and Computer Engineering, with a courtesy appointment in the Department of Mathematics. Her research is in the area of signal processing, machine learning, and applied mathematics. She is a three-time recipient of Purdue’s Seed for Success Award. She is also a recipient of the Eta Kappa Nu Outstanding Faculty Award, the Eta Kappa Nu Outstanding Teaching Award and the Wilfred “Duke” Hesselberth Award for Teaching Excellence.

Abstract:

After introducing the theory of FI-modules in 2012, the collaborative unit consisting of Thomas Church, Jordan Ellenberg and Benson Farb applied their framework to asymptotically stable counting problems in a certain classes of FI-varieties over finite fields in their 2013 paper Representation stability in cohomology and asymptotics for families of varieties over finite fields. The paper serves as a proof-of-concept, unifying a number of previously-known combinatorial results. The key to their method is the Grothendieck-Lefschetz fixed-point theorem with twisted statistics, which relates the rational cohomology of an algebraic variety over the complex numbers with the trace of the Frobenius map applied to the étale cohomology with coefficients in an $\ell$-adic sheaf of that variety over a finite field. In this talk, we shall introduce the Grothendieck-Lefschetz formula and its associated machinery as well as FI-modules and representation stability, then use these ideas to give an exposition of some results of Church, Ellenberg and Farb as they relate to configuration spaces and the braid group.

Abstract:

An integrable Hamiltonian system is a dynamical system with "enough conserved quantities" to guarantee that it can, in principle, be solved, or "integrated". I will give some basic definitions of Poisson algebras and what it means to be integrable in this context. I will then show, by way of an example (namely the "pentagram map"), how some combinatorial techniques using weighted directed graphs can be used to model the system and demonstrate its integrability. This method also hints at connections with cluster algebras and Postnikov's constructions related to stratifications of the positive Grassmannian. Time permitting, I will also discuss recent work generalizing this example.

Abstract:

The Boeing Inertial Upper Stage (IUS) launch vehicle was used to launch spacecraft from 1982 until 2004. The Gamma Guidance algorithm was used on board the IUS to select the ignition times, durations, and directions of engine firings. I will discuss the mathematics employed in Gamma Guidance as well as collateral on-board ;processes in order to illuminate how mathematics is applied in the launch setting.

Dr. Gary Green holds mathematics degrees from the Universities of Idaho, Michigan State, and Pennsylvania State. He taught mathematics at California State College Stanislaus before joining the Aerospace Corporation (www.aerospace.org), where he served as an applied mathematician and systems engineer in several capacities: employing numerical analysis to model launch vehicles, overseeing algorithm and software development for space systems, evaluating space system performance, and analyzing threats against space systems.

Abstract:

Algebraic K-theory is an extremely rich but notoriously difficult invariant to compute. In order to make calculations tractible, topological Hochschild homology and topological cyclic homology were introduced, along with the Dennis and cyclotomic trace maps. A natural question to consider is whether or not these invariants are sheaves for various topologies arising in algebraic geometry. In fact, it turns out that topological Hochschild homology is a sheaf for the fpqc topology on connective commutative ring spectra. In this talk, I plan to introduce the language necessary and sketch the argument of this result.

Abstract:

In this talk I will consider extreme behavior of the extremal eigenvalues of white Wishart matrices, which play an important role in multivariate analysis. I will focus on the case when the dimension of the feature p is much larger than or comparable to the number of observations n, a common situation in modern data analysis. I will discuss asymptotic approximations for the tail probabilities of the extremal eigenvalues. In addition, I will discuss the construction of an efficient Monte Carlo importance sampling algorithm to estimate the tail probabilities. Simulation results show that our method has the best performance among known approximation approaches, and furthermore provides an efficient and accurate way for evaluating the tail probabilities in practice. Based on joint work with Tiefieng Jiang and Gongjun Xu.

Abstract:

Reflection groups are an object in classical geometry with deep connections to Lie theory, representation theory, algebraic geometry, invariant theory, and combinatorics. The first half of the talk will give a quick introduction to various flavors of reflection groups. In the second half of the talk I will state a lemma about a restricted class of reflection groups, and discuss an idea for how incorporating cyclotomic fields may allow us to remove that restriction.

Abstract:

Immanants are matrix functionals that generalize the determinant. One notable family of immanants are the Kazhdan-Lusztig immanants. These immanants are indexed by permutations and are defined as sums involving Kazhdan-Lusztig polynomials. One notable property of Kazhdan-Lustzig immanants is that they are nonnegative on totally positive matrices. We give a condition on permutations that allows us to extend this theorem to the setting of k-positive matrices. We also conjecture a larger class of permutations for which our theorem holds true.

Abstract:

UnitedHealth Group Research and Development (UHG R&D) is working on creating, validating, testing, and refining innovations in the healthcare industry. Our data science team provides technical solutions by applying various statistical and machine learning methods to a vast amount of data in order to answer research questions. In this talk, I will give a brief introduction to my day-to-day experience as a data scientist. I will also discuss the methodology involved in several analyses of treatment outcomes for prostate cancer.

Abstract:

We consider the solution to a tight-binding, periodic Schrödinger equation with a random potential evolving stochastically in time. If the potential evolves according to a stationary Markov process we obtain a positive, finite diffusion constant for the evolution of the solution. More generally, we show that the square amplitude of the wave packet, after diffusive rescaling, converges to a solution of the heat equation. (Joint work with J. Schenker and Z. Tilocco at the Michigan State University).

Abstract:

Discontinuous vector fields arise naturally in some applications. In this presentation, a simple classical model of ocean circulation is introduced as an example of how discontinuities give rise to nonunique solutions. Standard bifurcation techniques often fail when the vector field is not smooth, and certainly fail when the vector field is discontinuous. However, some topological techniques seem to carry over, and a crude birfurcation theory can be extended to a large class of discontinuous systems.

Abstract:

Modern approaches to simulation, involving Monte Carlo methods and randomized procedures of decision-making, are usually dated back to the mid-20th century and the arrival of the computer era. Deeper history goes back to the 19th and even 18th centuries and involves such devices as Galton’s board and Buffon’s needle. However, one can argue that long before the invention of computers, older devices such as dice and their predecessors have been effectively used for games and divination. The idea of this paper is to review the use of ancient randomizing devices to trace the history of simulation and random rules of decision-making. Special attention will be paid to some contemporary cultures, which have preserved some unique elements of their ancient history: native cultures of the Americas, the Celtic civilizations of Ireland and Scotland, and the indigenous peoples of Northern and Central Asia (Altai and Siberia).Bio: https://www.stthomas.edu/mathematics/faculty/arkady-shemyakin.html

Abstract:

We will show the sharp estimate on the behavior of the smallest singular value of random matrices under very general assumptions. One of the key steps in the proof is a result about the efficient discretization of the unit sphere in an n-dimensional euclidean space. This result allows us to work with very general random matrices. The proof of the result will be outlined. Partially based on the joint work with Tikhomirov and Vershynin.

Abstract:

Let Q be a regular local ring. An ideal I in Q is said to be licci if it is in the linkage class of a complete intersection. Christensen, Veliche and Weyman conjectured that a perfect ideal of grade 3 in Q is licci if and only if its free resolution corresponds to Dynkin diagrams.

After introducing the theory of linkage, I will talk about the conjecture, how Dynkin diagrams are involved, and their approach in showing one direction of the conjecture in arXiv:1712.04016.

Abstract:

Consider a collection of n rigid, massive bodies interacting according to their mutual gravitational attraction. A relative equilibrium motion is one where the entire configuration rotates rigidly and uniformly about a fixed axis — all of the bodies are phase locked. Such a motion is possible only for special positions and orientations of the bodies. A minimal energy motion is one which has the minimum possible energy in its fixed angular momentum level. While every minimal energy motion is a relative equilibrium motion, the main result here is that a relative equilibrium motion of n >= 3 disjoint rigid bodies is never an energy minimizer. Since energy minimizers are the expected final states produced by tidal interactions, phase locking of 3 or more bodies will not occur.

Abstract:

“Mysterious Duality” was discovered by Iqbal, Neitzke, and Vafa in 2001. They noticed that toroidal compactifications of M-theory lead to the same series of combinatorial objects as del Pezzo surfaces do, along with numerous mysterious coincidences: both toroidal compactifications and del Pezzo surfaces give rise to the exceptional series E_k; the U-duality group corresponds to the Weyl group W(E_k), arising also as a group of automorphisms of the del Pezzo surface; a collection of various M- and D-branes corresponds to a set of divisors; the brane tension is related to the “area” of the corresponding divisor, etc. The mystery is that it is not at all clear where this duality comes from. In the talk, I will present another series of mathematical objects: certain versions of multiple loop spaces of the sphere S^4, which are, on the one hand, directly connected to M-theory and its combinatorics, and, on the hand, possess the same combinatorics as the del Pezzo surfaces. This is a report on an ongoing work with Hisham Sati.

Abstract:

Principal components analysis has become widely used in a variety of fields. In finance and, more specifically, in the theory of interest rate derivative modeling, its use has been pioneered by R. Litterman and J. Scheinkman. Their key finding was that a few components explain most of the variance of treasury zero-coupon rates and that the first three eigenvectors represent level, slope and curvature changes on the curve. This result has been, since then, observed in various markets.Over the years, there have been several attempts at modeling correlation matrices displaying the observed effects as well as trying to understand what properties of the those matrices are responsible for the effect. Using recent results of the theory of positive matrices we characterize these matrices and, as an application, we shed light on some of the critiques to this methodology.Bio: http://mcfam.math.umn.edu/people/carlos-tolmasky

Abstract:

The finite circular beta-ensembles and their point process scaling limit can be represented as the spectra of certain random differential operators. These operators can be realized on a single probability space so that the point process scaling limit is a consequence of an operator level limit. The construction allows the derivation of the scaling limit of the normalized characteristic polynomials of the finite models to a random analytic function. I will review these representations and constructions, and present a couple of applications.

Joint with B. Virág (Toronto).

Abstract:

This talk will continue to follow the "Parliament of Polytopes" paper by Di Rocco, Jabbusch, and Smith. I'll discuss the result that the T-equivariant generators for the global sections of a toric vector bundle E correspond to the lattice points in the parliament of polytopes for E and look to understand this result for some examples, including direct sums of line bundles.

Abstract:

We propose a new dynamic programming principle related to the Dirichlet problem for the homogeneous p-Laplace equation in connection with the Tug of War games with noise. We also discuss similar approximations in presence of the Robin boundary conditions. For the proofs, we use martingale techniques involving various couplings of random walks and yielding estimates on the involved probabilistic representations.

Abstract:

The phase retrieval problem is the problem of reconstructing an unknown signal from its Fourier intensity function. This problem has a long history in physics and engineering and occurs in contexts such as X-ray crystallography, speech processing and computational biology. As stated, the phase retrieval problem is ill-posed as there may be up to $2^N$ non-equivalent signals (called ambiguities) with the same Fourier intensity function. To enforce uniqueness additional constraints must be imposed.

In this talk we discuss the geometry of the space of ambiguities obtained by varying the signal. By understanding this geometry we prove a general result characterizing constraints that enforce a unique solution to the phase retrieval problem. This result was applied in work with Tamir Bendory and Yonina Eldar on blind phaseless short-time Fourier transform recovery.

Abstract:

The cyclic sieving phenomenon of Reiner, Stanton, and White says that we can often count fixed points for a cyclic group acting on a combinatorial set by plugging roots of unity into a polynomial related to this set. One of the most impressive instances of the cyclic sieving phenomenon is a theorem of Rhoades asserting that the set of plane partitions in a rectangular box under the action of promotion exhibits cyclic sieving. In Rhoades's result the sieving polynomial is the size generating function for these plane partitions, which has a well-known product formula due to MacMahon. We extend Rhoades's result by studying the interaction of promotion with symmetries of plane partitions. We obtain cyclic sieving-like formulas in this context where the relevant polynomial is the size generating function for symmetric plane partitions, whose product formula was conjectured by MacMahon and proved by Andrews. We then go on to consider the way the symmetries interact with rowmotion, another operator acting on plane partitions which is closely related to promotion. We end by explaining the connection of our work to some earlier conjectures we made concerning rowmotion acting on the P-partitions of various “triangular” posets P.

Abstract:

The Langlands Program, connecting algebra, analysis and geometry in diverse ways, is foundational to modern number theory. I will introduce this program and indicate some recent progress. As we shall see, a great deal still remains to be done.

Abstract:

For functions acquired through point evaluations, is there an optimal way to estimate a quantity of interest or even to approximate the functions in full? We give an affirmative answer to this question under the novel assumption that the functions belong to a model set defined by approximation capabilities. In fact, we produce implementable linear algorithms that are optimal in the worst-case setting. We present applications of the abstract theory in atmospheric science and in system identification.

Dr. Simon Foucart earned a Masters of Engineering from the Ecole Centrale Paris and a Masters of Mathematics from the University of Cambridge in 2001. In 2006, he received his Ph.D. in Mathematics at the University of Cambridge, specializing in Approximation Theory. After two postdoctoral positions at Vanderbilt University and Université Paris 6, he joined Drexel University in 2010 before moving to the University of Georgia in 2013. Since 2015, he has been with Texas A&M University, where is now professor. His recent work focuses on the field of Compressive Sensing, whose theory is exposed in the book ‘A Mathematical Introduction to Compressive Sensing’ he coauthored with Holger Rauhut. Dr. Foucart’s research was recognized by the Journal of Complexity, from which he received the 2010 Best Paper Award. Dr. Foucart’s current interests also include the mathematical aspects of Metagenomics, Optimization, Deep Learning and Data Science at Large

Abstract:

Like Andy's talk last week, this week will be another introductory talk; the topic is L-functions. We will start with a discussion of Dirichlet's use of L-series to show the infinitude of primes in arithmetic progressions, then proceed to how L-functions have become a major area of investigation. We will discuss only a sampling of topics, with the goal of motivating interest and setting the stage for future talks.

Abstract:

The cohomology of a category has many properties that extend those that are familiar when the category is a group. Second cohomology classifies equivalence classes of category extensions, first cohomology parametrizes conjugacy classes of splittings, first homology is the abelianization of the fundamental group, and second homology has a theory that extends that of the Schur multiplier. Defining restriction and corestriction maps on the homology of categories is problematic: most attempts to do this require induction and restriction functors to be adjoint on both sides, and this typically does not happen with categories. We describe an approach to defining these maps that includes all the situations where they can be defined in group cohomology, at least when the coefficient ring is a field. The approach uses bisets for categories, the construction by Bouc and Keller of a map on Hochschild homology associated to a bimodule, and the realization by Xu of category cohomology as a summand of Hochschild cohomology.

Abstract:

Learn from current students and MFM alumni in industry about the benefits of this professional master program. Panelists will talk about what attracted them to the program, how to make the most of your time in the MFM and why the practical, real-world learning helps you land jobs in quantitative risk analysis, hedging, trading, portfolio management, fintech, data analytics and other jobs that are hot in the field right now.

Abstract:

We discuss critical behavior of percolation on finite random networks. In a seminal paper, Aldous (1997) identified the scaling limit for the component sizes in the critical window of phase transition for the Erdos-Renyi random graph (ERRG). Subsequently, there has been a surge in the literature, revealing several interesting scaling limits of these critical components, namely, the component size, diameter, or the component itself when viewed as a metric space. Fascinatingly, when the third moment of the asymptotic degree distribution is finite, many random graph models have been shown to exhibit a universality phenomenon in the sense that their scaling exponents and limit laws are the same as the ERRG. In contrast, when the asymptotic degree distribution is heavy-tailed (having an infinite third moment), the limit law turns out to be fundamentally different from the ERRG case and in particular, becomes sensitive to the precise asymptotics of the highest degree vertices.

In this talk, we will focus on random graphs with a prescribed degree sequence. We start by discussing recent scaling limit results, and explore the universality classes that arise from heavy-tailed networks. Of particular interest is a new universality class that arises when the asymptotic degree distribution has an infinite second moment. Not only it gives rise to a completely new universality class, it also exhibits several surprising features that have never been observed in any other universality class so far.

This is based on joint works with Shankar Bhamidi, Remco van der Hofstad, Johan van Leeuwaarden and Sanchayan Sen.

Abstract:

Most applied problems we encounter can be naturally written as nonconvex optimization, for which
obtaining even a local minimizer is computationally hard in theory, never mind the global minimizer. In practice, however, simple numerical methods often work surprisingly well in finding high-quality solutions, e.g., training deep neural networks.

In this talk, I will describe our recent effort in bridging the mysterious theory-practice gap for nonconvex optimization, in the context of solving practical problems in signal processing, machine learning, and scientific imaging. 1) I will highlight a family of smooth nonconvex problems that can be solved to global optimality using simple numerical methods, independent of initialization. 2) The discovery, however, does not cover nonsmooth functions, which are frequently used to encode structural objects (e.g., sparsity) or achieve robustness. I will introduce tools from nonsmooth analysis, and demonstrate how nonsmooth, nonconvex problems can also be analyzed and solved in a provable manner. 3) Toward the end, I will provide examples to show how innovative problem formulation and physical design can help to tame nonconvexity.

Ju Sun is an assistant professor at the computer science & engineering department, University of Minnesota at Twin Cities. Prior to this, he was a postdoctoral scholar at Stanford University,
working with Professor Emmanuel Cand?s. He received his Ph.D. degree from Electrical Engineering of Columbia University in 2016 (2011--2016) and B.Eng. degree in Computer Engineering (with a minor in Mathematics) from the National University of Singapore in 2008 (2004--2008). His research interests span computer vision, machine learning, numerical optimization, signal/image processing, and high-dimensional data analysis. Recently, he is particularly fascinated by why simple numerical methods often solve nonconvex problems surprisingly well (on which he maintains a bibliographic webpage: http://sunju.org/research/nonconvex/ ) and the implication on representation learning. He won the best student paper award from SPARS'15 and honorable mention of doctoral thesis for the New World Mathematics Awards (NWMA) 2017.

Abstract:

We will show that the space of complex irreducible polynomials of degree d in n variables satisfies two forms of homological stability: first, its cohomology stabilizes as d increases, and second, its compactly supported cohomology stabilizes as n increases. Our topological results are inspired by counting results over finite fields due to Carlitz and Hyde.

Abstract:

In the talk we will investigate a natural braid group action on factorizations of elements in reflection groups. In particular, we will consider reflection factorizations of Coxeter elements in Coxeter groups of finite rank and state conditions whether two factorizations lie in the same orbit under the natural action. The presented result extends the investigation of Lewis - Reiner (arXiv:1603.05969) to arbitrary Coxeter groups of finite rank. (j.w. Patrick Wegener)

Abstract:

Anders Björner and Joseph Bernstein raised the question of finding regular CW complexes naturally arising from representation theory having the intervals of Bruhat order as their posets of closure relations. Sergey Fomin and Michael Shapiro conjectured a solution, namely the link of the identity in the totally nonnegative real part of the unipotent radical of a Borel in a semisimple, simply connected algebraic group defined and split over the reals together with a family of related spaces indexed by the different Coxeter group elements. The Fomin-Shapiro conjecture indeed proved to be true, with the proof utilizing an interpretation of these stratified spaces as images of an intriguing family of maps - maps also arising in work of Lusztig related to canonical bases. I will start by reviewing some highlights of this story, then turn to recent joint work with Jim Davis and Ezra Miller regarding the structure of the fibers of these same maps.

Abstract:

We continue our crash course in finite group representation theory by looking at some important operations on representations.

We start by defining the group algebra of a finite group; group representations naturally biject with modules over the group algebra.

After that, we'll talk through a variety of ways to construct new representations from old, such as restriction, induction, inflation, tensor product, and we may even squeeze in symmetric and exterior powers.

Abstract:

A partition is an aa-core partition if none of its hook lengths are divisible by aa. It is well known that the number of aa-core partitions is infinite and the number of simultaneous (a,b)(a,b)-core partitions is a generalized Catalan number if aa and bb are relatively prime. In the first half of the talk, we give an expression for the number of simultaneous (a1,a2,…,ak)(a1,a2,…,ak)-core partitions that is equal to the number of integer points in a polytope. In the second half, we discuss objects closely related to core partitions, called numerical semigroups, which are additive monoids that have finite complements in the set of non-negative integers. For a numerical semigroup SS, the genus of SS is the number of elements in ??SN?S and the multiplicity is the smallest nonzero element in SS. In 2008, Bras-Amorós conjectured that the number of numerical semigroups with genus gg is increasing as gg increases. Later, Kaplan posed a conjecture that implies Bras-Amorós conjecture. In this talk, we prove Kaplan's conjecture when the multiplicity is 4 or 6 by counting the number of integer points in a polytope. Moreover, we find a formula for the number of numerical semigroups with multiplicity 4 and genus gg.

Abstract:

We describe a recent construction of self-similar blow-up solutions of the incompressible Euler equation. A consequence of the construction is that there exist finite-energy $C^{1,a}$ solutions to the Euler equation which develop a singularity in finite time for some range of $a>0$. The approach we follow is to isolate a simple non-linear equation which encodes the leading order dynamics of solutions to the Euler equation in some regimes and then prove that the simple equation has stable self-similar blow-up solutions.