Date 
Speaker(s) 
Title(s) 
Abstract 
Sept. 16 
Department faculty 
Introductions 
Come meet department faculty, newcomers and veterans, and learn about their interests in and out of the classroom. Find out about department activities for the year. Pizza and beverages served at noon in 416 Clapp. Presentations
12:201:00 pm.

Sept. 23 
MHC Students from 20102012 
Summer Opportunities: What? Where? How? 
Students will briefly describe their summer experiences (including REUs), how they found out about them, and what math or stat preparation they required. Pizza and beverages served at noon in 416 Clapp. Panel 12:20 
1:00 p.m.
in 402 Clapp 
Sept 26, 9:30 am  5pm 
Mia Minnes, MIT Anna Lysyanskaya, Brown 
Women in Mathematics In New England conference 
Plenary talks, short talks by students, lunch, panel on being a graduate student. See http://maven.smith.edu/~jhenle/wimin09 
Sept. 30 
Giuliana Davidoff and department faculty plus Fatema Burhani '10

So, you want to go to grad school?

Math and stat majors have many options for graduate
study: mathematics (pure and applied), statistics, allied fields that use
mathematics and statistics, professional master’s degrees of various
kinds. How should you prepare yourself with coursework? With
outside the classroom experiences? (Fatema Burhani '10 will speak about the Budapest Semesters in Mathematics program.) What is required for an
application? Department faculty, led by Giuliana Davidoff, will discuss the answers to these questions 12:201:00 p.m. in 416 Clapp (or in 402 if we need more room). Pizza
and beverages will be served at noon in 416 Clapp.

Oct 6
12:201:00 pm

Thomas Kerler, OSU 
IInformation on graduate study at OSU in 402 Clapp 
Thomas Kerler is ViceChair for Graduate Studies in Mathematics at Ohio State University.
What are good reasons to pursue a graduate degree in mathematics? How can you prepare early during your college sophomore or junior years? What are strategies to get informed about and to choose graduate programs to apply to? What does an effective application look like? What is the relevance of recommendation and intent letters, grades, GRE scores, research experience, etc.? How do you negotiate admissions and declinations and make your final decision? This presentation will shed light on these and more questions from the point of view of a graduate recruiter from a large rsearch university. We will introduce, in particular, the graduate program in mathematics at The Ohio State Uiversity. Pizza will be provided at noon in 416, courtesy of OSU.

Oct. 7 
Tom Moore, Grinnell College

Baboon "mothering": Using permutation tests to uncover patterns in how female baboons handle the infants of other females

Female baboons, some with infants, were observed and counts made of interactions in which females handled the infants of other females (socalled infanthandling). Independent of these observations, each baboon was assigned a dominance rank of "low," "medium," or "high." Researchers hypothesized that females tend to handle infants of females ranked below them. The data set that emerged from this observational study contained challenges to common methods of significance testing. Permutation tests were developed in the 1930s, ironically before computers were available to make them a practical tool. Here we show how to use such tests in the analysis of the infanthandling data. This is the inaugural lecture in an annual series on applied statistics, in honor of George Cobb. Talk 12:20  1:00 p.m. in 407 Clapp. Pizza
and beverages will be served at noon in 416 Clapp.

Oct 7
58 pm 
Actuary Career Fair at UMass 
Insurance companies from New England 
More information on the Math/Stat Club bulletin board and online at
http://www.math.umass.edu/Program/actuarial_science.html 
Oct. 14 
Margaret Robinson, MHC

Generating functions and counting polynomial roots upon division by p^e 
According to Herbert Wilf, a generating function is a clothesline on which we hang up a sequence of numbers for display. Given a sequence of numbers a_0, a_1, a_2, ..., a_n, ..., we can form its generating function as f(t) = sum_{n=0} ^{\infty} a_n t^n. This talk will introduce some famous problems about generating functions and then talk about the generating function that counts roots of polynomials modulo powers of a prime p. Pizza and beverages served at noon in
416 Clapp. Talk 12:201:00 pm.

Oct. 21 
Michelle Lastrina '06, Iowa State University 
Can you paint yourself into a corner? 
Let G=(V,E) be a planar graph and let P be a subset of V. If the vertices of P are precolored and all other vertices are assigned lists of size 5, under what conditions on P is there a proper coloring of G from the lists? We will look at some conditions that allow such a precoloring to extend, as well as some that do not. Before doing this, we will go through the background information needed, including some basics of graph theory, graph coloring and list coloring. Along the way we will present some wellknown results and look at many examples. Pizza and beverages served at noon in
416 Clapp. Talk 12:201:00 pm

Oct. 28 
Geometry Center at Univ Minnesota

Not Knot, a short film 
A guided tour into computer animated hyperbolic space, it proceeds from the world of knots to the complementary space, what's not a knot. Profound theorems of recent mathematics show that most knot complements carry the structure of hyperbolic geometry, a geometry in which the sum of three angles of a triangle is always less than 180 degrees and in which there is so much room that, with unit of length one meter, a hyperbolic hemispherical swimming pool 25 meters in diameter contains 23 times the (ordinary) volume of the earth. The video shows the gometry of the knot complement, the space around the knot, changing into hyperbolic space, and then you see what it is like to "fly through" the hyperbolic space.
Pizza and beverages served at noon in 416 Clapp. Talk
12:201:00 pm

Oct 29 12:20 in 203 Kendade 
Dylan Shepardson, MHC 
Inverting fundamental equations from neuroscience 
Since the 1950s, when AL Hodgkin and AF Huxley developed a system of equations describing the electrical activity of the squid giant axon, neuroscientists have been equipped with a powerful theoretical frlamework for understanding neuron function. These equations take as input data representing the internal state of the neuron and simulate the behavior of the neuron. Since the behavior of the neuron is relatively easy to observe, the the internal details of the neuron are hard to discover experimentally, it would be useful to be able to run the HodgkinHuxley equations in reverse  to take some output behavior and determine the details of the neuron that produced it. Dr. Shepardson will present some recent work on solving this inverse problem. Sigma Xi Science Cafe; bring your lunch. Dessert provided. 
Nov. 4 
Department faculty 
Information session on 300level
courses spring 2010

Come learn about
300level courses in mathematics and statistics for spring 2010 in 402 Clapp, including
both MHC courses and offerings in the Valley. Pizza
and beverages served at noon in 416 Clapp.

Nov .11 
Muluwork Geremew '03, Corning Inc.

From MHC to a Career at a Research and Development Corporation 
Muluwork was honored as Most Promising Engineer at the 2009 Black Engineer of the Year Awards Conference. In this talk, she will be sharing her career and transiting experience from school to a R&D corporation. A brief introduction to Corning Incorporated and the Modeling and Simulation group within the corporation will be provided. She will also give a high level descript
ion of some of her projects along with the mathematical methods and tools utilized.
Pizza and beverages served at noon in 416 Clapp. Talk
12:201:00 pm

Nov. 18 
Milka Doktorova '10, Faye Stevens '11

"Formation and dynamics of objective structures of the nanotube group" and "Can you hear the shape of a torus?" 
Millka’s abstract: This summer I participated in an REU program at the Institute for Mathematics and Its Applications at the University of Minnesota. My project was to study the formation and dynamics of socalled “objective structures,” which are mathematical structures that model the tail sheaths of viruses and other physical systems demonstrating a particular type of symmetry in both their structure and force interactions. Together with two other students I developed MATLAB programs to simulate the movement of the structures by applying molecular dynamics techniques. In my talk I will explain how to use familiar math concepts from linear algebra to generate one particular type of objective structures. I will also demonstrate some of the MATLAB “movies” illustrating their motion.
Faye's abstract: In 1966, Mark Kac published a mathematical article asking if it was possible to "hear" the shape of a drum. In other words, if one knows the frequencies (eigenvalues) of the sound a drum makes, can one determine the shape of the drumhead? In 1992, Gordon, Webb, and Wolpert answered the question in the negative by constructing two differently shaped regions in the plane with identical eigenvalues. This summer, I explored the same question in three dimensions
.Pizza and beverages served at noon in 416
Clapp. Talk 12:201:00 pm

Dec 1 
Scott Schwartz, Duke Univ. 
Bayesian analysis and computation for finite mixture models, censored/missing data, and intermediate variables, with connections to nonparametric Dirichlet processes 
In this talk, I will describe a data modeling problem involving birthweight and gestational age. I will use this example to introduce Bayesian posterior analysis and finite mixture models. Special attention will be paid to the Bayesian treatment of missing and censored data. In addition, the birthweight and gestational age example provides an opportunity to consider the implications of intermediate variables in conditional analyses, e.g., birthweight conditional on gestational age analyses. As I will show, conditioning on intermediate variables produces posttreatment selection bias, thus invalidating causal estimands. This realization will bring us to principal stratification, which may be used to adjust for intermediate variables without introducing bias. I will conclude by describing my current nonparametrics work  very closely related to finite mixture models  in the principal stratification setting using a cholesterol reduction study with partial compliance. Pizza and beverages served at noon in 416 Clapp. Talk 12:201:05 in 402 Clapp. There is also a tea 4:00  5:00 p.m. in 416 Clapp. 
Dec. 2 cancelled

Viveka Borum, Columbia Univ Teacher's College

Black women in mathematics: Taking a closer look at diversity in the field 
Talk cancelled because of illness. To be rescheduled in the spring.

Dec. 9 
Jing (Maria) Zhang, Statistics, Harvard Univ.

Bayesian Inference of Interaction in Biological Problems

Recent development of biotechnologies such as microarrays and highthroughput sequencing has greatly accelerated the pace of genetics experimentation and discoveries. As a result, large amounts of highdimensional genomic data are available in population genetics and medical genetics. With millions of biomarkers, it is a very challenging problem to search for the diseaseassociated or treatmentassociated markers, and infer the complicated interaction (correlation) patterns among these markers. In this talk, we present Bayesian inference of interactions in two biological research areas: wholegenome association studies of common diseases, and HIV drug resistance studies. We have developed a Bayesian model for simultaneously inferring haplotypeblocks and selecting SNPs within blocks that are associated with the disease, either individually, or through epistatic interactions with others. Simulation results show that this approach is uniformly more powerful than other epistasis mapping methods. Applying this method to Type 1 Diabetes data, we landscape the interaction patterns in MHC region for this disease. We have investigated the HIV drug resistance problem from a new perspective. By probabilistically modeling mutations in the HIV1 proteases isolated from drugtreated patients, we have derived a statistical procedure that first detects potentially complicated mutation combinations and then infers detailed interacting structures of these mutations. Pizza and beverages served at noon in 416 Clapp. Talk 12:20  1:05 in 402 Clapp. There is also a tea on TUES DEC 8, 4:00  5:00 p.m. in 416 Clapp.

Dec 10
THURS 
Robin Young, Biostatistics, Boston Univ 
Generalized Additive Models and Power of Smoother Hypothesis Tests 
Generalized additive models (GAMs) have distinct advantages over generalized linear models as they allow investigators to make inferences about associations between outcomes and predictors without placing parametric restrictions on associations. The variable of interest is often smoothed using a locally weighted regression (LOESS) and the optimal span (degree of smoothing) can be determined by minimizing the Akaike Information Criterion. Such models can be applied in spatial statistics using a bivariate smoother to determine whether there is an association between geographic location and disease status. A natural hypothesis when using GAMs is to test whether the smoothing term is necessary or if a simpler model would suffice. An approximate chisquare test is available but known to be biased. Permutation tests are a reasonable alternative. This research uses simulated data generated under simple null and alternative hypotheses to evaluate the properties of the approximate chisquare and three permutation testing methods. The methods are compared in their type I error rates, powers, theoretical appropriateness, and computational efficiency.Pizza and beverages served at noon in 416 Clapp. Talk 12:20  1:05 in 402 Clapp. There is also a tea 4:00  5:00 p.m. in 416 Clapp.

Dec 16 
Ji Young Kim, Univ of Illinois 
Robust EM Clustering through a Mixture in Multivariate Regression 
Clustering is one of the fundamental data mining techniques to assign data points to subgroups. Among a wide variety of clustering techniques, mixture models are commonly used because of their usefulness to model heterogeneous data. The attention has first focused on the use of normal distributions because of their computational convenience. However, for many applied problems, normal distributions are often not appropriate to describe the clusters with heavier tails due to the existence of outliers. We propose a robust clustering method based on the mixtures in linear regression, designed to reduce the effects of the outliers on the clustering analysis. The model is fitted via the EM (expectation and maximization) algorithm with a new iterative algorithm to solve the optimization problem. In simulation studies, we find that the proposed method often works better than existing clustering methods. The superior performance of the proposed robust clustering method can be attributed to the robustness and the use of covariates. Along with other existing methods, the proposed method provides an elaborate approach to clustering heterogeneous data.Pizza and beverages served at noon in 416 Clapp. Talk 12:20  1:05 in 402 Clapp. There is also a tea on TUES DEC 15, 4:00  5:00 p.m. in 416 Clapp.

(Dec 18 FRI) CANCELLED 
Stacey Hancock, Reed College 
Uncovering Changes in Time Series Data 
Many time series data sets exhibit structural breaks in a variety of ways, the most obvious being a mean level shift. In this case, the mean level of the process is constant over periods of time, jumping to different levels at times called "changepoints". These jumps may be due to outside influences such as changes in government policy or manufacturing regulations. Structural breaks may also be a result of changes in variability or changes in the spectrum of the process. The goal is to estimate where these structural breaks occur and to provide a model for the data within each stationary segment. The program AutoPARM (Automatic Piecewise AutoRegressive Modeling procedure), developed by Davis, Lee, and RodriguezYam (2006), uses the minimum description length principle to estimate the number and locations of changepoints in a time series by fitting autoregressive models to each segment. In this talk, we will start with a review of time series data structures, then consider the changepoint problem in the context of AutoPARM. Pizza and beverages served at noon in 416 Clapp. Talk 12:20  1:05 in 402 Clapp. There is also a tea on THURS DEC 17, 4:00  5:00 p.m. in 416 Clapp.
