Events for Spring 2007:

Date	Speaker(s)	Title(s)	Abstract
Jan.31	Dept. faculty and students	Interested in summer research? Graduate school?	Department faculty and student veterans will offer advice on  applying for undergraduate research programs for summer 2006 and on getting ready to apply to graduate programs for 2008-2009.  Pizza and beverages  at 12:15 in 416 Clapp.
Feb. 7	Rebecca Horowitz, NC State	Multiple Regression and Diabetes	Once data has been collected and basic descriptive statistics computed, how do we create a model that can be used to make "good" predictions?  In this talk I will describe one such method, multiple regression, and some variable selection methods that are commonly used to create such models.  As a motivating example these methods were applied to a study on diabetes patients to determine which factors affect the progression of the disease.  Once a model has been formed, we can use it to make predictions for future patients. Pizza and beverages at 12:15 in 416 Clapp.
Feb 13 Amherst College	Ben Brubaker, MIT	On Beyond Zeta: More infinite series in number theory Undergrad. Conn. Valley Colloqium pizza 6pm, Seeley Mudd 208 talk 7pm, Seeley Mudd 206 RSVP for pizza: awtorrey@amherst.edu	The Riemann Hypothesis (RH) is probably the most famous unsolved problemin mathematics. According to most pop-culture references, anyone who attemptsto solve it is doomed to insanity (or at least, to cutting out lots of newspaper clippings). The RH concerns a certain infinite series known as the (Riemann) zeta function. We'll talk about the zeta function, why it is of interest to number theorists, some of the reasons why the Riemann Hypothesis is so hard, and what other functions are lurking out there with similar properties to the zeta function. Our hunt for these functions will lead us in some surprising directions, including such mathematically fashionable objects as "modular forms" and "elliptic curves." No prior knowledge of these topics will be assumed in the talk.
Feb 15 Willits 3-5pm (liv. rm) and 6:30-8:30 (Morrison room)	UMass, Harvard, Yale reps	Public Health Career Fair PLEASE REGISTER by Wednesday, February 14, at: www.alumnae.mtholyoke.edu/go/publichealth	Mount Holyoke invites all five-college area students interested in learning about careers in public health to a joint career fair and panel discussion next Thursday, February 15. Career Fair: Meet with representatives from UMass, Harvard, and Yale Schools of Public Health to learn about their graduate programs. From 6:30 to 8:30 p.m., join us for a panel discussion where schoolrepresentatives and alumnae panelists will share their advice with studentsconsidering post-graduate education and careers in public health.Morrison Room, Willits-Hallowell Center. (Refreshments will be served.) Featured alumnae panelists: Julie M. Flahive '97, MPH student at UMASS and lecturer in statistics at MHC Margaret L. Harvey '84; Epidemiologist for Connecticut Dept. of Public Health Brittany L. McCrary '04, Health Counselor at Tapestry Health, Amherst, MA .Representatives from UMass, Harvard, and Yale Schools of Public Health
Feb. 14	Janet Rosenbaum, Harvard U.	Do teenagers always tell the truth? Estimating the prevalence of adolescent risk behaviors from inconsistent self-report	Adolescents do not always answer survey questions truthfully.  Researchers cannot detect lies directly, but they can detect logical contradiction, such as when an adolescent gives different answers to the same question in a short time frame.  This study identifies question properties associated with higher inconsistency and estimates the increased error due to inconsistency.  The data are from a reliability study of the Youth Risk Behavior Survey (YRBS), in which a diverse convenience sample of 4628 high school students answered the same72 questions at an interval of two weeks.  In the two week interval, the prevalence of 41 of the 72 risk behaviors changed significantly.  Linear regression showed that questions about substance use are more consistent than questions about other topics, and questions about lifetime risk behaviors are more consistent than questions about behaviors in other time frames.  A Bayesian method estimated that the standard error of the prevalence of 72 risk behaviors is at median 3.3 times larger than conventional estimates of standard error  (95% CI (1.9, 5.0)).  The majority of questions on the YRBS do not yield reliable estimates of risk behavior prevalence, but questions which cannot be interpreted in multiple ways may be more reliable than ambiguous questions. Conventional measures of uncertainty underestimate the error on the prevalence of adolescent risk behaviors. Pizza and beverages at 12:15 in 416 Clapp.
Feb. 21	Chris Mun,  U.Wisconsin	A non-deletion approach to detect discordant subjects in repeated measurements.	<>As will be explained, functional data can be seen in many fields such as medical research, engineering research, linguistic research, etc, but they have not been studied as much as other types of data, and the diagnostics for functional data is still in its infancy.   Some authors have attempted to adjust the typical diagnostic tools in regression to functional data but those modified statistics are mainly based on the change of coefficients.  And other diagnostics previously proposed for growth curve data do not generalize well to random effects and nonlinear models. Functional data are different from the data in regression in the sense that they have curves as their components.  So a potential discordant subject(s) should be dealt with in a different manner.  Two types of deviation are taken into consideration and a technique is presented to simultaneously diagnose subjects in more than one sense.  The new method re-expresses residual sum of squares at each level and then “studentize” them. As secondary applications, the proposed method can be facilitated to examine the influence of each point in a given curve, and can play a graphical measure of goodness-of-fit.  This method will be the first to look at two dimensions of the outlier problem at once and promises to reveal more information about the shape and location of discordant subjects. The performance of the proposed method can be seen in the simulation example and in real data from biomedical engineering and orthodontic growth data. Pizza and beverages  at 12:15 in 416 Clapp.
Feb.28	Margaret Robinson, MHC	An introduction to the p-adic numbers	"16 is a REALLY even number," said six year old Phoebe.  What's it called when there are a lot of threes in a number?" Answer:  "The number is 3-adically very small." Come learn about the 2-adics, the 3-adics, the 5-adics and, you got it, the p-adics. Pizza and beverages at 12:15 in 416 Clapp.
Mar. 7	Don O'Shea, MHC	The Poincare Conjecture: in search of the shape of the universe	Pizza and beverages at 12:15 in 416 Clapp.
Mar 7 4:00 p.m.	Kobi Abayomi, Columbia University & Haverford College	TBA	Talk at 4:00 p.m. in 402 Clapp.  Cookies and cider available at 3:45 in 416 Clapp.
Mar. 14	Harriet Pollatsek, MHC	Continuous Symmetry:  Calculus meets abstract algebra	This talk will give a glimpse of the rich theory named for the nineteenth century Norwegian mathematician Sophus Lie.  Lie studied continuous groups of symmetries, as a language to describe geometry and as geometries in their own right.  We’ll look at a few examples illustrating this “double” geometry and showing how calculus gets into the act.  Regard this as shameless advertising for Math 319 (Topics in algebra: Lie groups) next fall -- prerequisite linear algebra. Pizza and beverages at 12:15 in 416 Clapp.
Mar. 28	Dept. faculty	Information session on 300-level courses, fall 2007	Come learn about 300-level courses in mathematics and statistics for fall 2007, including both MHC courses and offerings in the Valley.  Pizza and beverages at 12:15 in 416 Clapp.
Apr. 3 Smith College 4:30 pm McConnell404	Peter Shor, MIT	Quantum Computing	Quantum computers are hypothetical devices which use the principles of quantum mechanics to perform computations. For some difficult computational problems, including the cryptographically important problems of prime factorization and finding discrete logarithms, the best algorithms known for digital computers are exponentially slower than the algorithms known for quantum computers. Although they ave not yet been built, quantum computers do not appear to violate any fundamental properties of physics. I give a mathematical model of quantum computation, explain how quantum mechanics provides this extra computational power, and briefly describe the algorithm for efficient prime factorization. Tea at 4pm on 4th floor of  Burton.
Apr. 4	Marjorie Senechal, Smith	3-dimensional Penrose Tilings -- an unsolved problem	Pizza and beverages at 12:15 in 416 Clapp..
Apr. 11	Allan Rossman, Cal .Polytech. State Univ. at San Luis Obispo	Choosing the Best: Decision Making under Uncertainty	A classic probability problem known as the "Secretary Problem" illustrates the utility of a clever strategy for making decisions under uncertainty.  The problem is easy to state: Your task is to hire a new employee for your company.  You know how many candidates have applied, you interview them one-at-a-time in random order, and you can rank the candidates after you have interviewed them.  But the rules are that you have to decide immediately whether or not to hire a candidate (i.e., you can't invite one back later), and you must hire THE best candidate (i.e., if you hire the second best one, you have failed).  We will devise an optimal strategy for choosing the best candidate and explore how well it performs with both small and large numbers of candidates. Pizza and beverages at 12:15 in 416 Clapp.
Apr. 18	Stephanie Jones, Harvard Med School	Mathematics in Biomedical Imaging: Using computational neural models to understand human brain activity.	In this talk I will give a brief introduction to the biomedical imaging technologies used for clinical and basic science research at MGH, and how mathematics is essential in the development and use of these technologies. The main focus of the talk will be on MEG/EEG (magnetoencepholography/electroencepholography) brain imaging and how computer simulations of neural network models can be used to understand the neural dynamics creating the measured brain signals. The development of the neural models is based on the mathematics of dynamical systems theory. Pizza and beverages at 12:15 in 416 Clapp.
Apr. 21 Sat. HRUMC Siena C.	Georgia Benkart, U Wisconsin	Ladies of the Rings Plenary Lecture	Benkart's talk is "a tale of science non-fiction about two women and their magical mathematics."  Student speakers need an MHC faculty sponsor; abstracts are due by February 25.  See http://www.skidmore.edu/academics/mcs/pages/hrumc14in.htm
Apr. 25	Richard Jordan, MHC	How Differential Equations Broke My Heart: Mathematical Models of (Dysfunctional) Relationships	Pizza and beverages at noon in 416 Clapp.
May 2	Math/Stat students	Math/Stat "Talk Back"	This is an opportunity for you to give the department advice about opportunities and events outside the classroom.  (Seniors will get a detailed questionnaire --- watch for it!)  What are your reactions to opportunities and events in 2006-2007?  What advice/suggestions do you have for 2007-2008? Come share your ideas and join the conversation. .  Pizza and beverages at 12:15 in 416 Clapp.