Colloquia 2012-2013
Mathematics Colloquium
All colloquia are on Fridays at 4:00 pm in Van Vleck B239, unless otherwise indicated.
Fall 2012
Spring 2013
| date | speaker | title | host(s) |
|---|---|---|---|
| Jan 25 | Alexander Fish (Sydney) | TBA | Gurevich |
| Feb 1 | Anne Thomas (Sydney) | TBA | Dymarz |
| Feb 15 | Eric Lauga (UCSD) | TBA | Spagnolie |
| Feb 22 | Tentatively Scheduled | Stovall | |
| Mar 1 | Kirsten Wickelgren (Harvard) | TBA | Street |
| March 8 | Svetlana Jitomirskaya (UV Irvine) | TBA | Kiselev |
| March 15 | Kurusch Ebrahimi-Fard (Madrid) | TBA | Gurvich |
| March 22 | Neil O'Connell (Warwick) | TBA | Timo Seppalainen |
| March 29 | Spring Break | No Colloquium | |
| April 5 | John Jones (ASU) | TBA | Boston |
| April 19 | Andrew Snowden (MIT) | TBA | Street |
| May 3 | Davesh Maulik (Columbia) | TBA | Street |
| May 10 | Steve Gelbart (Weizmann Institute) | TBA | Gurevich |
Abstracts
Sept 14: Jordan Ellenberg (UW-Madison)
FI-modules: an introduction (joint work with T Church, B Farb, R Nagpal)
In topology and algebraic geometry one often encounters phenomena of _stability_. A famous example is the cohomology of the moduli space of curves M_g; Harer proved in the 1980s that the sequence of vector spaces H_i(M_g,Q), with g growing and i fixed, has dimension which is eventually constant as g grows with i fixed.
In many similar situations one is presented with a sequence {V_n}, where the V_n are not merely vector spaces, but come with an action of S_n. In many such situations the dimension of V_n does not become constant as n grows -- but there is still a sense in which it is eventually "always the same representation of S_n" as n grows. The preprint
http://arxiv.org/abs/1204.4533
shows how to interpret this kind of "representation stability" as a statement of finite generation in an appropriate category; we'll discuss this set-up and some applications to the topology of configuration spaces, the representation theory of the symmetric group, and diagonal coinvariant algebras. Finally, we'll discuss recent developments in the theory of FI-modules over general rings, which is joint work with (UW grad student) Rohit Nagpal.
Sept 21: Joyce R. McLaughlin (Rensselaer Polytechnic Institute)
Mathematics for Imaging Biomechanical Parameters in Dynamic Elastography
Elastography, the imaging of biomechanical parameters in tissue, is motivated by the doctor’s palpation exam where the doctor presses against the skin to detect stiff and abnormal tissue changes. In dynamic elastography experiments, the tissue is in motion with displacement amplitudes on the order of tens of microns. The displacement(s) are determined with sequences of MR data sets or sequences of ultrasound RF/IQ data sets within the tissue; and the data shows a dispersive effect indicating that tissue is viscoelastic. A choice of viscoelastic model must be made. For each model the biomechanical parameters satisfy a first order, linear or nonlinear, partial differential equation (system) with real or complex coefficients. We discuss the mathematical properties of these equations and how those properties lead to successful interpretation of the data, and to successful algorithms and images. We show biomechanical images of breast cancer and prostate cancer and compare those images to ultrasound images and histology slides with marked cancerous inclusions.
Sept 28: Eric Marberg (MIT)
Supercharacters for algebra groups: applications and extensions
The group U_n(F_q) of unipotent upper triangular matrices over a finite field belongs to the same list of fundamental examples as the symmetric or general linear groups. It comes as some surprise, therefore, that the group's irreducible characters are unknown, and considered in some sense unknowable. In order to tackle problems normally requiring knowledge of a group's irreducible characters, Diaconis and Isaacs developed the notion of the supercharacters of an algebra group, generalizing work of Andre and Yan. Algebra groups form a well-behaved class of p-groups including U_n(F_q) as a prototypical example, and supercharacters are certain reducible characters which form a useful approximation to the set of irreducible characters. In this talk I will survey several equivalent definitions of the supercharacters of an algebra group, and discuss some applications and extensions of these approaches. On one end of things, I intend briefly to introduce the recent discovery of how certain representation theoretic operations on the supercharacters of U_n(F_q) naturally define a Hopf algebra structure, which has been studied under a different name by combinatorialists. In another direction, I will explain how one can view the supercharacters of an algebra group as the first step in a more general reduction process, which can be used to shed light on some mysterious properties of U_n(F_q).
Oct 04 Howard Masur (Chicago)
Winning games for badly approximable real numbers and billiards in polygons
Wolfgang Schmidt invented the notion of a winning subset of Euclidean space in a game between two players. Winning sets have nice properties such as full Hausdorff dimension. The basic example of a winning set considered by Schmidt are those reals badly approximated by rationals. An equivalent formulation by Artin is that badly approximable reals correspond to geodesics that stay in a bounded set in the modular curve. There is also an equivalent formulation in terms of sets of directions for the linear flow on a square torus or equivalently, billiard trajectories in a square. In joint work with Yitwah Cheung and Jon Chaika we extend this notion of winning to flows on flat surfaces of higher genus with applications to billiards in rational angled polygons. My intention in this talk is to give the background on the Schmidt game, describe the classical results before introducing the more recent work.
Wed, Oct 10: Bas Lemmens (University of Kent)
From hyperbolic geometry to nonlinear Perron-Frobenius theory
In a letter to Klein Hilbert remarked that the logarithm of the cross-ratio is a metric on any open, bounded, convex set in Euclidean space. These metric spaces are nowadays called Hilbert geometries. They are a natural non-Riemannian generalization of Klein's model of the hyperbolic plane, and play a role in the solution of Hilbert's fourth problem.
In the nineteen fifties Garrett Birkhoff and Hans Samelson independently discovered that one can use Hilbert's metric and the contraction mapping principle to prove the existence and uniqueness of a positive eigenvector for a variety of linear operators that leave a closed cone in a Banach space invariant. Their results are a direct extension of the classical Perron-Frobenius theorem concerning the eigenvectors and eigenvalues of nonnegative matrices. In the past decades this idea has been further developed and resulted in strikingly detailed nonlinear extensions of the Perron-Frobenius theorem. In this talk I will discuss the synergy between metric geometry and (nonlinear) operator theory and some of the recent results and open problems in this area.
Thur, Oct 11: Irene Gamba (University of Texas-Austin)
Analytical and numerical issues associated with the dynamics of the non-linear Boltzmann equation
The non-linear Boltzmann equation models the evolution of a statistical flow associated with particle systems in a rarefied or mesoscopic regimes. Its analytical local as well as long time behavior depends strongly on the growth conditions of the collisional kernels, as functions of the intermolecular potentials and scattering mechanisms. We will present recent analytical results, such as classical convolutional inequalities and sharp moments estimates, which imply propagation of $W^{k,p}$ norms and exponential decay of high energy tails, and their consequences on the existence, regularity, and stability of solutions for initial value problems as well as control of decay rates to equilibrium. In particular we present a numerical approximation to the non-linear Boltzmann problem by a conservative spectral scheme, and show spectral accuracy as well as error estimates.
Oct 12: Joachim Rosenthal (Univ of Zurich)
Linear Random Network Codes, a Grassmannian Approach
Elastography, the imaging of biomechanical parameters in A novel framework for random network coding has been introduced by Koetter and Kschischang. In this framework information is encoded in subspaces of a given ambient space over a finite field. A natural metric is introduced where two subspaces are `close to each other' as soon as their dimension of intersection is large. This framework poses the challenge to come up with new codes with optimal or near optimal distance and to develop efficient decoding algorithms.
In a first part of the talk we will provide a survey. In a second part of the talk we report on progress constructing spread codes and orbit codes. The decoding problem of orbit codes can be interpreted as a problem in Schubert calculus over a finite field.
Fri, Oct 19: Saverio E. Spagnolie (Madison)
Elastic slender bodies in fluids and slender bodies in elastic fluids
Abstract:
The scientific study of elastic materials dates back to Galileo, and fluid mechanics to Archimedes, but the interaction of elastic bodies and viscous fluids remains a topic at the frontier of modern research. We will discuss two problems on this topic of recent interest. First, when a flexible filament is confined to a fluid interface, the balance between capillary attraction, bending resistance, and tension from an external source can lead to a self-buckling instability. We will walk through an analysis of this elastocapillary instability, and analytical formulae will be shown that compare favorably with the results of detailed numerical computations. Second, we will discuss the motility of a swimming helical body in a viscoelastic fluid, wherein the fluid itself exhibits an elastic response to deformation. The helical geometry is exploited to generate a highly accurate numerical method, and we will show that the introduction of viscoelasticity can either enhance or retard the swimming speed depending on the body geometry and the properties of the fluid (through a dimensionless Deborah number). Our findings bridge the gap between studies showing situationally dependent enhancement or retardation of swimming speed, and may help to clarify phenomena observed in systems from spermatozoan swimming to mechanical drilling.
Fri, Oct 26: Luke Oeding (UC Berkeley)
The Trifocal Variety
Abstract: In Computer Vision one considers many cameras looking at the same scene. From this setup many interesting geometric and algebraic questions arise. In this talk we will focus on the case of 3 cameras and study the so called trifocal tensors. Trifocal tensors are constructed from a bilinear map defined using the trifocal setup. A natural question is, given a particular tensor, how can one determine if it is a trifocal tensor? This question can be answered by finding implicit defining equations for the trifocal variety. From an algebraic standpoint, it is also interesting to know the minimal generators of the defining ideal of the trifocal variety.
In this talk I will explain our use of symbolic and numerical computations aided by Representation Theory and Numerical Algebraic Geometry to find the minimal generators of the ideal of the trifocal variety. This is joint work with Chris Aholt (Washington). Our work builds on the work of others (such as Hartley-Zisserman, Alzati-Tortora and Papadopoulo-Faugeras) who have already considered this problem set-theoretically.
My goal is to make most of the talk accessible to anyone with a modest background in Linear Algebra.
Tues, Oct 30: Andrew Majda (Courant)
Data Driven Methods for Complex Turbulent Systems
An important contemporary research topic is the development of physics constrained data driven methods for complex, large-dimensional turbulent systems such as the equations for climate change science. Three new approaches to various aspects of this topic are emphasized here. 1) The systematic development of physics constrained quadratic regression models with memory for low-frequency components of complex systems; 2) Novel dynamic stochastic superresolution algorithms for real time filtering of turbulent systems; 3) New nonlinear Laplacian spectral analysis (NLSA) for large-dimensional time series which capture both intermittency and low-frequency variability unlike conventional EOF or principal component analysis. This is joint work with John Harlim (1,2), Michal Branicki (2), and Dimitri Giannakis (3).
Thurs, Nov 1: Lenya Ryzhik (Chicago)
The role of a drift in elliptic and parabolic equations
The first order partial differential equations are closely connected to the underlying characteristic ODEs. The second order elliptic and parabolic equations are as closely connected to the Brownian motion and more general diffusions with a drift. From a variety of points of view, the drift does not really matter - all diffusions look more or less similar. As a caveat to this reasonable line of thought, I will describe a menagerie of problems: linear and nonlinear, steady and time-dependent, compressible and incompressible, where the drift and diffusion confirm the Mayakovsky thesis "Woe to one alone!" Together, they lead to enhanced mixing and improved regularity that are impossible for each one of them to attain.
Nov 2: Vladimir Sverak (Minnesota)
On scale-invariant solutions of the Navier-Stokes equations
The solutions of the Navier-Stokes equations which are invariant under the scaling symmetry of the equations provide an interesting window into non-linear regimes which are not accessible by perturbation theory. They appear to give valuable hints concerning the old question about the uniqueness of weak solutions. In the lecture we outline a recent proof of the result that for every scale-invariant initial data there is a global scale-invariant solution (smooth for positive times), and we explain connections to the uniqueness problem. This is joint work with Hao Jia.
Nov 9: Uri Andrews (Madison)
Computable Stability Theory
Stability theory attempts to classify the underlying structure of mathematical objects. The goal of computable mathematics is to understand when mathematical objects or constructions can be demonstrated computably. I'll talk about the relationship between underlying structure and computation of mathematical objects.
Monday, Nov 12: Charles Smart (MIT)
Regularity and stochastic homogenization of fully nonlinear equations without uniform ellipticity
I will discuss regularity of fully nonlinear elliptic equations when the usual uniform upper bound on the ellipticity is replaced by bound on its $L^d$ norm, where $d$ is the dimension of the ambient space. Our estimates refine the classical theory and require several new ideas that we believe are of independent interest. As an application, we prove homogenization for a class of stationary ergodic strictly elliptic equations.