Yermal Sujeet Bhat (University of Minnesota) 
Homogenization of a nonlinear elliptic boundary value problem related to corrosion modeling 
Abstract: We study a nonlinear elliptic boundary value problem motivated by a corrosion model used in the electrochemistry community in the study of heterogeneous electrode surfaces. The boundary condition is of an exponential type with periodically varying parameters. We treat the problem from the point of view of homogenization theory by constructing a firstorder asymptotic approximation. We establish convergence estimates for both the two and threedimensional case and provide twodimensional numerical experiments. 
Ludovica Cecilia CottaRamusino (University of Minnesota) 
Looping probability densities of elastic rods 
Abstract: We exploit elastic rod models to evaluate DNA looping probabilities, adopting a path integral formalism to compute approximations to the probability density function for the location and orientation of one end of a continuum elastic rod at thermodynamic equilibrium with a heat bath. 
Glenn Fung (Siemens Medical Solution) 
Learning classifiers for computer aided diagnosis using
local correlations 
Abstract: computer aided
diagnosis (CAD) applications the goal is to detect structures
of interest to physicians in medical images: e.g. to identify
potentially malignant lesions in an image (mammography, lung
CT, Colon CT, heart ultrasound, etc.). In an almost universal
paradigm, this problem is addressed by a 5 stage system:
1. Segmentation to identify/extract the general area of
interest; 2. Candidate generation which identifies suspicious
unhealthy candidate regions of interest (ROI) from a medical
image; 3. feature extraction that computes descriptive features
for each candidate; 4. classification that differentiates
candidates based on candidate feature vectors; 5. visual
presentation of CAD findings to the radiologist in order for
him to accept or reject the CAD findings.
For the fourth stage, many standard algorithms (such as
support vector machines (SVM), backpropagation neural nets,
kernel Fisher discriminants) have been used to learn
classifiers for detecting malignant structures. However, these
generalpurpose learning methods either make implicit
assumptions that are commonly violated in CAD applications, or
cannot effectively address the difficulties arisen when
learning a CAD system.
NonIID Data Traditional learning methods almost universally
assume that the training samples are independently drawn from
an identical albeit unobservable underlying distribution (the
IID assumption), which is often not the case in CAD systems.
Due to spatial adjacency of the regions identified by a
candidate generator, both the features and the class labels of
several adjacent candidates are highly correlated.
In this talk we present two recent proposed machine learning
algorithms that successfully takes into account the correlation
among candidates to significantly improve classification
performance. 
Peter R. Kramer (Rensselaer Polytechnic Institute) 
Stochastic mathematical and computational models in microbiology 
Abstract: I shall discuss three areas of current research involving the use of
stochastic methods for the physical modeling for microscopic processes
in physiology. First, I exhibit a variation of the immersed boundary
method designed, in joint work with Paul Atzberger (UCSB) and Charles
Peskin (NYU) for simulating microbiological systems where thermal
effects play a significant role, such as molecular motors, DNA and
other polymer dynamics, and gel swelling. Statistical mechanical
principles indicate that the thermal fluctuations should manifest
themselves through a random force density in the fluid component of
the immersed boundary equations. Secondly, I briefly review the
mathematical procedure, currently being developed with Juan Latorre
and Grigorios Pavliotis (Imperial), for coarsegraining stochastic
molecular motor models. Finally, I shall discuss recent explorations
with Adnan Khan (Lahore) and Shekhar Garde (Rensselaer, Biochemical
Engineering) concerning the parameterization of a simple stochastic
model for the behavior of water molecules near a solute surface which
has the potential for improving substantially upon Brownian dynamics
models more conventionally used in engineering applications. We use
exactly solvable mathematical models as a testbed for addressing some
basic datadriven parameterization issues. 
Christopher J. Lee (University of California) 
Working Seminar: Probabilistic methods in bioinformatics 
Abstract: Purpose: to discuss challenges arising from the analysis of massive
datasets such as highthroughput genomics or proteomics data, and
probabilistic methods for analyzing them.
Chris will provide some useful introduction to various topics in each
session, but leave the time open for informal discussion.
Initial discussion: Introduction to the
challenges of highthroughput data analysis in "postgenomic"
biology, and methods of statistical inference used to solve these
problems.
lab webpage: http://www.bioinformatics.ucla.edu/leelab
blog: http://thinking.bioinformatics.ucla.edu 
Christopher J. Lee (University of California) 
Working seminar: Probabilistic methods in bioinformatics 
Abstract: Purpose: to discuss challenges arising from the analysis of massive datasets such as highthroughput genomics or proteomics data, and probabilistic methods for analyzing them.
Chris will provide some useful introduction to various topics in each session, but leave the time open for informal discussion.
Initial discussion: Introduction to the challenges of highthroughput data analysis in "postgenomic" biology, and methods of statistical inference used to solve these problems.
lab webpage: http://www.bioinformatics.ucla.edu/leelab
blog: http://thinking.bioinformatics.ucla.edu 
Christopher J. Lee (University of California) 
Working seminar: Probabilistic methods in bioinformatics 
Abstract: Purpose: to discuss challenges arising from the analysis of massive datasets such as highthroughput genomics or proteomics data, and probabilistic methods for analyzing them.
Chris will provide some useful introduction to various topics in each session, but leave the time open for informal discussion.
Initial discussion: Introduction to the challenges of highthroughput data analysis in "postgenomic" biology, and methods of statistical inference used to solve these problems.
lab webpage: http://www.bioinformatics.ucla.edu/leelab
blog: http://thinking.bioinformatics.ucla.edu 
Christopher J. Lee (University of California) 
Working seminar: Probabilistic methods in bioinformatics 
Abstract: Purpose: to discuss challenges arising from the analysis of massive datasets such as highthroughput genomics or proteomics data, and probabilistic methods for analyzing them.
Chris will provide some useful introduction to various topics in each session, but leave the time open for informal discussion.
Initial discussion: Introduction to the challenges of highthroughput data analysis in "postgenomic" biology, and methods of statistical inference used to solve these problems.
lab webpage: http://www.bioinformatics.ucla.edu/leelab
blog: http://thinking.bioinformatics.ucla.edu 
Christopher J. Lee (University of California) 
Mapping evolutionary pathways of HIV1 drug resistance using
conditional selection pressure 
Abstract: Can genomics provide a new level of strategic intelligence about rapidly evolving pathogens? We have developed a new approach to measure the rates of all possible evolutionary pathways in a genome, using conditional Ka/Ks to estimate their “evolutionary velocity”, and have applied this to several datasets, including clinical sequencing of 50,000 HIV1 samples. Conditional Ka/Ks predicts the preferred order and relative rates of competing evolutionary pathways. We recently tested this approach using independent data generously provided by Shafer and coworkers (Stanford HIV Database), in which multiple samples collected at different times from each patient make it possible to track which mutations occurred first during this timecourse. Out of 35 such mutation pairs in protease and RT, conditional Ka/Ks correctly predicted the experimentally observed order in 28 cases (p=0.00025). Conditional Ka/Ks data reveal specific accessory mutations that greatly accelerate the evolution of multidrug resistance. Our analysis was highly reproducible in four independent datasets, and can decipher a pathogen’s evolutionary pathways to multidrug resistance even while such mutants are still rare. Analysis of samples from untreated patients shows that these rapid evolutionary pathways are specifically associated with drug treatment, and vanish in its absence. 
Anton Leykin (University of Minnesota) 
Applications of numerical algebraic geometry 
Abstract: Numerical Algebraic Geometry provides a collection of new methods to
treat the solutions of systems of polynomial equations. The numerical
homotopy continuation technique forms a base for higher level algorithms
in the area.
This talk exposes three topics. First is a recent application of
homotopy continuation to a problem in enumerative algebraic geometry:
computation of Galois groups of Schubert problems. Second is a deflation
method that restores the convergence of the Newton's method at a
singular isolated solution of a polynomial system. Third is a new
approach to detecting embedded components of an underlying complex
variety dubbed numerical primary decomposition.

Roger Y. Lui (Worcester Polytechnic Institute) 
Three topics in the mathematics of molecular and cellular
biology 
Abstract: In this talk, I will discuss three topics in the mathematics of molecular
and cellular biology. They are Protein Folding, Biochemical Network, and
Cell Motility. I am an analyst by training so you are going to see a lot of
equations in my talk. But I will try to make things interesting and easy
to understand. 
Laura Lurati (University of Minnesota) 
Design under uncertainty using stochastic collocation 
Abstract: Optimization methods generally treat objectives, constraints and parameters as deterministic "perfectly known" values. However, this is often not the case for real problems. Uncertainty may enter the design process as early as the conceptual design phase, through manufacturing, as well as in the use/operation of the final product. Design optimization under uncertainty seeks to minimize the impact of random parameters on the design. Stochastic collocation methods are proposed as the underlying statistical method for robust/reliability design optimization using direct search methods. Examples demonstrate the ease of use of the method as well as its flexibility. Test problems include the robust design of an airfoil over a range of Mach numbers and robust/reliability design of a cantilever beam under manufacturing uncertainty. Possible modifications to the method for efficient representation of multiple random variables are discussed. 
Alfio Quarteroni (Politecnico di Milano) 
Mathematical modeling in medicine, sports, and the environment 
Abstract: Mathematical models are enabling advances in increasingly complex areas of engineering and technology. Recent developments in multiscale geometrical modeling have opened the way to progress in modeling such complex systems as the human circulatory system and the climate system. Professor Quarteroni leads a team which has harnessed mathematical modeling to design improved cardiac surgical interventions and to optimize the design of the twice winning America's cup yacht Alinghi. He will talk about this work, and their efforts to confront some of the great environmental challenges that face us.

David Umulis (University of Minnesota) 
Computational analysis of BMPmediated embryonic patterning in Drosophila melanogaster 
Abstract: The principal aim of developmental biology is to delineate how genes are turned on and off at the correct point in time and space to produce the multitude of specialized cell types present in the mature organism. The complexity of many developmental processes precludes an intuitive understanding of regulation at the systems level, making it difficult to construct new hypotheses and design experiments to reveal the molecular function of novel regulators. To address these challenges, we developed a unified approach that couples experimental methods such as fluorescent in situ hybridization (FISH), immunostaining, and in vitro kinetics with sophisticated 3D computational models to analyze early developmental processes in Drosophila melanogaster. In addition to elucidating molecular function, we used these mechanistic models to study the following questions, such as: How robust are developmental systems to perturbations in the underlying network structure and the quantities of the molecular regulators? How do different organisms within a species preserve proportion even though they vary substantially in body size? And, finally, how do cells respond to dynamic and noisy signals during development? 