Seminars and Colloquia by Series

Coupling at infinity

Series
Stochastics Seminar
Time
Thursday, March 10, 2011 - 15:05 for 1 hour (actually 50 minutes)
Location
Skiles 005
Speaker
Jonathan MattinglyDuke University, Mathematics Department
I will discuss how the idea of coupling at time infinity is equivalent to unique ergodicity of a markov process. In general, the coupling will be a kind of "asymptotic Wasserstein" coupling. I will draw examples from SDEs with memory and SPDEs. The fact that both are infinite dimensional markov processes is no coincidence.

Cantor Boundary Behavior of Analytic Functions

Series
Analysis Seminar
Time
Thursday, March 10, 2011 - 15:00 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Ka-Sing LauHong Kong Chinese University
There is a large literature to study the behavior of the image curves f(\partial {\mathbb D}) of analytic functions f on the unit disc {\mathbb D}. Our interest is on the class of analytic functions f for which the image curves f(\partial {\mathbb D}) form infinitely many (fractal) loops. We formulated this as the Cantor boundary behavior (CBB). We develop a general theory of this property in connection with the analytic topology, the distribution of the zeros of f'(z) and the mean growth rate of f'(z) near the boundary. Among the many examples, we showed that the lacunary series such as the complex Weierstrass functions have the CBB, also the Cauchy transform F(z) of the canonical Hausdorff measure on the Sierspinski gasket, which is the original motivation of this investigation raised by Strichartz.

First-Fit is Linear on (r+s)-free Posets

Series
Graph Theory Seminar
Time
Thursday, March 10, 2011 - 12:05 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Kevin MilansUniversity of South Carolina
First-Fit is an online algorithm that partitions the elements of a poset into chains. When presented with a new element x, First-Fit adds x to the first chain whose elements are all comparable to x. In 2004, Pemmaraju, Raman, and Varadarajan introduced the Column Construction Method to prove that when P is an interval order of width w, First-Fit partitions P into at most 10w chains. This bound was subsequently improved to 8w by Brightwell, Kierstead, and Trotter, and independently by Narayanaswamy and Babu. The poset r+s is the disjoint union of a chain of size r and a chain of size s. A poset is an interval order if and only if it does not contain 2+2 as an induced subposet. Bosek, Krawczyk, and Szczypka proved that if P is an (r+r)-free poset of width w, then First-Fit partitions P into at most 3rw^2 chains and asked whether the bound can be improved from O(w^2) to O(w). We answer this question in the affirmative. By generalizing the Column Construction Method, we show that if P is an (r+s)-free poset of width w, then First-Fit partitions P into at most 8(r-1)(s-1)w chains. This is joint work with Gwena\"el Joret.

Exact asymptotic behavior of the Pekar-Thomasevich functional

Series
Math Physics Seminar
Time
Wednesday, March 9, 2011 - 16:30 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Rafael D. BenguriaPhysics Department, Catholic University of Chile
An explicit asymptotic expression for the ground-state energy of the Pekar-Tomasevich functional for the N-polaron is found, when the positive repulsion parameter U of the electrons is less than twice the coupling constant of the polaron. This is joint workwith Gonzalo Bley.

Energy estimates for the random displacement model

Series
Analysis Seminar
Time
Wednesday, March 9, 2011 - 14:00 for 1 hour (actually 50 minutes)
Location
Skiles 005
Speaker
Michael LossSchool of Mathematics, Georgia Tech
This talk is about a random Schroedinger operator describing the dynamics of an electron in a randomly deformed lattice. The periodic displacement configurations which minimize the bottom of the spectrum are characterized. This leads to an amusing problem about minimizing eigenvalues of a Neumann Schroedinger operator with respect to the position of the potential. While this configuration is essentially unique for dimension greater than one, there are infinitely many different minimizing configurations in the one-dimensional case. This is joint work with Jeff Baker, Frederic Klopp, Shu Nakamura and Guenter Stolz.

The mathematics of service processes

Series
Research Horizons Seminar
Time
Wednesday, March 9, 2011 - 12:00 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Ton DiekerISYE - Georgia Institute of Technology

Please Note: Hosts: Amey Kaloti and Ricardo Restrepo

This talk gives an overview of the mathematics of service processes, with a focus on several problems I have been involved in. In many service environments, resources are shared and delays arise as a result; examples include bank tellers, data centers, hospitals, the visa/mortgage application process.I will discuss some frequently employed mathematical tools in this area. Since randomness is inherent to many service environments, I will focus on stochastic processes and stochastic networks.

2-dimensional TQFTs and Frobenius Algebras

Series
Geometry Topology Student Seminar
Time
Wednesday, March 9, 2011 - 11:00 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Alan DiazGeorgia Tech
An n-dimensional topological quantum field theory is a functor from the category of closed, oriented (n-1)-manifolds and n-dimensional cobordisms to the category of vector spaces and linear maps. Three and four dimensional TQFTs can be difficult to describe, but provide interesting invariants of n-manifolds and are the subjects of ongoing research. This talk focuses on the simpler case n=2, where TQFTs turn out to be equivalent, as categories, to Frobenius algebras. I'll introduce the two structures -- one topological, one algebraic -- explicitly describe the correspondence, and give some examples.

Efficiently Learning Gaussian Mixtures

Series
ACO Seminar
Time
Tuesday, March 8, 2011 - 16:30 for 1 hour (actually 50 minutes)
Location
KACB 1116
Speaker
Greg ValiantUniversity of California, Berkeley
Given data drawn from a mixture of multivariate Gaussians, a basic problem is to accurately estimate the mixture parameters. This problem has a rich history of study in both statistics and, more recently, in CS Theory and Machine Learning. We present a polynomial time algorithm for this problem (running time, and data requirement polynomial in the dimension and the inverse of the desired accuracy), with provably minimal assumptions on the Gaussians. Prior to this work, it was unresolved whether such an algorithm was even information theoretically possible (ie, whether a polynomial amount of data, and unbounded computational power sufficed). One component of the proof is showing that noisy estimates of the low-order moments of a 1-dimensional mixture suffice to recover accurate estimates of the mixture parameters, as conjectured by Pearson (1894), and in fact these estimates converge at an inverse polynomial rate. The second component of the proof is a dimension-reduction argument for how one can piece together information from different 1-dimensional projections to yield accurate parameters.

Math Modeling of Biological Memory

Series
Mathematical Biology Seminar
Time
Tuesday, March 8, 2011 - 11:00 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Vadim L. StefanukRussian Academy of Sciences
Some properties of biological memory are briefly described. The examples of short term memory and extra long term memory are drawn from psychological literature and from the personal experience. The short term memory is modeled here with the two types of mathematical models, both models being special cases of the Locally Organized Systems (LOS). The first model belongs to Prof. Mikhail Tsetlin of Moscow State University. His original ?pile of books? model was independently rediscovered a new by a number of scientists throughout the World. Tsetlin?s model demonstrates some very important properties of a natural memory organization. However mathematical study of his model turned out to be rather complicated. The second model belongs to the present author and has somewhat similar properties. However, it is organized in a completely different manner. In particular it contains some parameters, which makes the model rather interesting mathematically and pragmatically. The Stefanuk?s model has many interpretations and will be illustrated here with some biologically inspired examples. Both models founded a number of practical applications. These models demonstrate that the short term memory, which is heavily used by humans and by many biological subsystems is arranged reasonably. For humans it helps to keep the knowledge in the way facilitating its fast extraction. For biological systems the models explain the arrangement of storage of various micro organisms in a cell in an optimal manner to provide for the living.

Isospectral Graph Reductions, Estimates of Matrices' Spectra, and Eventually Negative Schwarzian Systems

Series
Dissertation Defense
Time
Tuesday, March 8, 2011 - 09:00 for 1 hour (actually 50 minutes)
Location
Skiles 006
Speaker
Benjamin WebbSchool of Mathematics, Georgia Tech
Real world networks typically consist of a large number of dynamical units with a complicated structure of interactions. Until recently such networks were most often studied independently as either graphs or as coupled dynamical systems. To integrate these two approaches we introduce the concept of an isospectral graph transformation which allows one to modify the network at the level of a graph while maintaining the eigenvalues of its adjacency matrix. This theory can then be used to rewire dynamical networks, considered as dynamical systems, in order to gain improved estimates for whether the network has a unique global attractor. Moreover, this theory leads to improved eigenvalue estimates of Gershgorin-type. Lastly, we will discuss the use of Schwarzian derivatives in the theory of 1-d dynamical systems.

Pages