University of Georgia
Mathematics Department
Applied Mathematics Seminar
|
All talks are given in Room 326, Boyd Graduate Studies Building
on Fridays, 12:20pm--1:10pm,
unless otherwise noted. There will be some pizzas and drinks
available before the seminar.
Talk Date: 20070202
Multivariate splines for numerical solution of PDE
Ming-Jun Lai
Abstract
Multivariate splines are smooth piecewise polynomial functions defined over tria
ngulated domains. I will^M
explain how to use them for numerical solution of some^M
linear and nonlinear PDE without constructing finite element method or locally s
upported basis functions. ^M
A special iteration method for the associted linear systems will be introduced t
o show that the method converges. Some numerical examples in MATLAB for Poisson^
M
More about speaker, check here.
Talk Date: 20070209
Mathematical Biology and Ecological Systems
Caner Kazanci
Abstract
In this talk, I present some of the mathematical problems arising in biological sciences in general. I will then focus on ecological applications, and introduce Ecological Network Analysis, a graph theoretical approach to study energy or material flow in ecosystems. I will discuss a new stochastic method that we developed, called Particle Tracking Algorithm. Similar to a microscope, this method enables us to observe each and individual energy (or mass) packets flow in the network; and we can simultaneously label and track all of their movements. Unlike agent or individual based models (ABM, IBM) where particles move according to a pre-defined algorithm, particle flow occurs ‘naturally’ (compatible with master equation) without human intervention. This brings the possibility of investigating all predefined ecological network properties, such as cycling index, residence time, dominance of indirect effects, and evaluate how well these algebraic definitions reflect their meaning.
More about speaker, check here.
Talk Date: 20070216
The Replicator Equations from Evolutionary Game Theory
Malcolm Adams
Abstract
The replicator equations are a system of ordinary differential equations that describe the evolution of n competing populations in terms of the relative fitness of the various components of the population. I will give a brief derivation of these equations and discuss some properties and classical examples. If time permits, I will also describe recent results of myself and Sornborger on the long term behavior of solutions for a certain class of these equations.
More about speaker, check here.
Talk Date: February 23
New challenges of Information Theory.
Alexander Petukhov
Abstract
Mathematical aspects of fundamental problems of Information Theory will be discussed, which became hot topics
during last two years. Many leading mathematicians, statisticians, and computer scientists switched to these
problems. A very incomplete list of them includes D. Donoho, T. Tao, R. Coifman, I. Daubechies, V. Tarokh, E. Candes.
The topics include Compression (and Compessed Sensing as a special case), Transmission in Noisy Channels,
and Cryptography for real-valued (or, more generally,
metric space) data. Instant applications of these problems are digital broadcasting, mobile telephony, the storage
of multimedia data.
Those problemes for discrete (finite group) data where solved in the 90's.
We show that all mentioned problems in real-valued data can be reduced
to finding sparse solutions of systems of linear equations with rectangular matrices.
It is known that in a general setting, the problem has non-polynomial computational complexity.
The goal of this talk is to discuss existing sub-optimal methods and possible approaches in finding optimal methods.
More about speaker, check here.
Talk Date: March 2, 2007
Variation Models and PDE Techniques in Wavelet Inpainting
Haomin Zhou
Abstract:
In this talk, I will present our work (collaborated with Tony Chan (UCLA)
and Jackie Shen (Minnesota)) on image inpainting in wavelet domain.
The problem is closely related to the classical image inpainting, with the
difference being that the inpainting regions are in the wavelet domain, that
brings new challenges to the reconstructions, as there is no geometrically
well defined inpainting region in the pixel domain, and the damage is
inhomogeneous. We propose new variational models, especially total variation
minimization in conjunction with wavelets for the wavelet inpainting
problems. The models lead to PDE's, which are Euler-Lagrange equations
of the variational formulations, in the wavelet domain and can be solved
numerically. The proposed models can have effective and automatic control
over geometric features of the inpainted images including sharp edges, even
in the presence of substantial loss of wavelet coefficients, including in
the low frequencies.
More about speakeri, check
here.
Talk Date: 20070323
Trading a mean reverting asset: buy low and sell high
Qing Zhang
Abstract
This paper is concerned with an optimal trading (buy and sell)
rule. The underlying asset price is governed by a mean-reverting
model. The objective is to buy and sell the asset so as to
maximize the overall return. Slippage cost is imposed on each
transaction. The associated HJB equations (variational
inequalities) are used to characterize the value functions. It is
shown that the solution to the original optimal stopping problem
can be obtained by solving two algebraic equations which are much
simpler to solve. Sufficient conditions are given in the
form of a verification theorem. A numerical example is reported to
demonstrate the results.
More about speaker, check here.
Talk Date: March 30, 2007
High-order operator splitting methods for unitary and parabolic evolution
Andrew Sornborger
Abstract
In this talk, I will discuss operator splitting methods and present methods that we developed for unitary quantum simulations on quantum computers. I will also discuss their application to parabolic differential equations and show that a conjecture that these methods are unstable in the parabolic case is incorrect.
More about speaker, check here.
Talk Date: 20070406
Darcy's Law and porous medium flow
Ronghua Pan
Abstract
The motion of isentropic flow through porous media can be
decribed by compressible Euler equtions with frictional damping.
Time asymptotically, the density is conjectured to satisfy the porous
media equation and the momentum obeys the classical Darcy's law.
Previous results are valid for small smooth flow away from vacuum.
In this talk, we report a proof for all physical flows.
More about speaker, check here.
Talk Date: 20070420
Approximation of functional regression models with bivariate splines
Serge Guillas
Abstract
We consider the functional linear regression model where the explanatory variable is a random surface and the response is a real random variable, with no noise. Bivariate splines over triangulations represent the random surfaces. We use this representation to construct least squares estimators of the regression function with or without a penalization term.
Under the assumptions that the regressors in the sample are bounded and span a large enough space of functions, bivariate splines approximation properties yield the consistency of the estimators. Simulations illustrate the quality of the asymptotic properties in various situations.
More about speaker, check here.
Talk Date: 20070427
SOME ASYMPTOTICS OF SELF-INTERACTING STOCHASTIC CHAINS
Dan Kannan
Abstract
The protein folding, dark matter, chemotaxis, and
polymers in solution are a few of the examples of
self-interacting systems. Probabilists modeled,
independently, some nearest neighbor interactions
on lattices using random walks in random environmens,
(recently applying such random walks to some of the
above systems). We present, in this lecture, a more
general foundation of self-interacting stochastic
chains (SISC) and consider one or more of the following symptotic properties of the SISC: 1. Stability,
2. Functional limit theorems, and 3. Large deviation
principle.
More about speaker, check here.