BOWLING GREEN STATE UNIVERSITY DEPARTMENT OF MATHEMATICS AND STATISTICS CALENDAR Week of November 4-8 Monday, November 4 3:30 PM Coffee 3:45 PM COLLOQUIUM - Room 459 MSC Valen E. Johnson, Institute of Statistics and Decision Sciences, Duke University "An Alternative to GPA-based Evaluation of Student Performance" Abstract: In the fall of 1995, the Academic Affairs Committee of the Arts and Sciences Council of Duke University was charged with studying the problem of grade inflation. Based on a preliminary analysis of several years of undergraduate grade data, the Committee on Grades concluded that anxiety over grade inflation stemmed as much from disparities in the grading policies of different departments and instructors as it did from the increasing trend in average grade. It was clear that these disparities caused a systematic bias against many of the University's students. In this talk, I describe an alternative measure of student performance designed to overcome the inherent problems associated with raw grades and derived grade point average (GPA). This alternative measure has been endorsed by the Academic Affairs Committee as a replacement for GPA-based class rank, and will be reviewed for university-wide adoption by the Duke University Arts and Sciences Council in the forthcoming academic year. Tuesday, November 5 2:30 PM MAPLE WORKSHOP - SciCompLab on the classroom (south) side John Gresser, Dept. of Mathematics and Statistics, BGSU. Topic: Starting today we will focus our attention on how this technology can be used in the class room. We can always discuss new commands, and strategies as they come up in problems. I will go over some initial problems that students coming out of Math 131 can look at during the first few weeks of a Maple program. Polynomials are not always so easy to graph. Maple's integration command can be suppressed so that we can manipulate integrals just like we do with pencil-paper techniques. Maple can be used in a really powerful way to enhance the ideas we present to students about setting up definite integrals for area, volume, force on a dam, etc. Wednesday, November 6 2:30 PM ANALYSIS SEMINAR - Room 459 MSC Craig Zirbel, Dept. of Mathematics and Statistics, BGSU. "Statistical Mechanics for Magnetohydrodynamics, Part III" Abstract: We will discuss the classical ideal gas as an example of the microcanonical and canonical ensembles of statistical mechanics. Then we will return to MHD, considering a model due to Fyfe and Montgomery (1976), and the inadequacies that have become apparent with this model. 3:30 PM STATISTICS SEMINAR - VAX lab in the basement of Hayes Hall. Jim Albert, Dept. of Mathematics and Statistics, BGSU. "Modeling using S-Plus" Second of two talks. 7:00 Pm CAREER NIGHT James Gabel, Nationwide Insurance Co. Actuarial science Thursday, November 7 1:30 PM SCIENTIFIC COMPUTATION SEMINAR - Room 459 MSC Paul Endres, Department of Chemistry, BGSU. "Genetic algorithms in chemical physics" Abstract: Continuing the previous discussions on the use of Genetic Algorithms, Dr. Paul Endres will discuss two recent papers in the area of Chemical Physics. This will be a very informal presentation and there will certainly be time for discussing other applications... The first application is finding the optimal set of parameters in an approximation scheme for solving molecular problems. (This is basically the Semiempirical Molecular Orbital scheme, discussed earlier this semester, applied in an area where the traditional parameters aren't appropriate.) The second paper uses a modified GA to find the lowest energy geometry of a cluster of N-atoms (n=4-30.) This is actually a hybrid method that uses a GA to find a new generation, but then uses a conjugate gradient method to minimize the energy (coarsely) of each individual. (This is no longer true Darwinian evolution, but we must not take our metaphors too seriously.) 3:30 PM ALGEBRA SEMINAR - Room 459 MSC Debra Waugh, University of Michigan "Quotients of Coxeter groups under the weak order" Abstract: We determine which quotients of affine Coxeter groups are lattices under the weak order. By a result of Bjorner, we know that quotients of arbitrary Coxeter groups are meet semi-lattices under the weak order, so we need only consider the question of existence of upper bounds within the quotients. Let W_a be an affine Coxeter group with corresponding finite Weyl group W. We will use certain regions of the Shi arrangement of hyperplanes (the regions we will use are translations of the Weyl chambers of W) both to establish the existence of the upper bound in the quotients which are lattices and to demonstrate that there are pairs of elements with no upper bound in the quotients which are not lattices. Friday, November 8 3:30 PM Coffee 3:45 PM COLLOQUIUM - Room 459 MSC Mark Becker, Dept. of Biostatistics, University of Michigan "EM Algorithms for Nonmissing Data" Abstract: In this talk we review the EM algorithm and the general principle of optimization transfer; i.e., optimizing a surrogate function rather than the original function of interest. Very many problems in computational statistics involve the optimization of an objective function, with notable examples being the maximization of log--likelihood functions and posterior distributions. Important features of the EM algorithm that derive from the optimization transfer principle are numerical stability, the substitution of simple optimization problems for difficult ones, and a well understood theory of local and global convergence properties. Several examples of optimization transfer algorithms useful in computational statistics are used to illustrate the broad generality of these features of the EM algorithm. This announcement and a schedule of future colloquia are available on the Worldwide Web; see http://www.bgsu.edu/departments/math/. If you would like to place a link to this calendar on your page, use html code Department of Mathematics and Statistics calendar If you wish to be placed on the e-mail distribution list, or have comments or material for the calendar, send email to