AN16-LS16 Abstracts
Total Page:16
File Type:pdf, Size:1020Kb
AN16-LS16 Abstracts Abstracts are printed as submitted by the authors. Society for Industrial and Applied Mathematics 3600 Market Street, 6th Floor Philadelphia, PA 19104-2688 USA Telephone: +1-215-382-9800 Fax: +1-215-386-7999 Conference E-mail: [email protected] Conference web: www.siam.org/meetings/ Membership and Customer Service: 800-447-7426 (US & Canada) or +1-215-382-9800 (worldwide) 2 2016 SIAM Annual Meeting Table of Contents AN16 Annual Meeting Abstracts ................................................4 LS16 Life Sciences Abstracts ...................................................157 SIAM Presents Since 2008, SIAM has recorded many Invited Lectures, Prize Lectures, and selected Minisymposia from various conferences. These are available by visiting SIAM Presents (http://www.siam.org/meetings/presents.php). 2016 SIAM Annual Meeting 3 AN16 Abstracts 4 AN16 Abstracts SP1 University of Cambridge AWM-SIAM Sonia Kovalevsky Lecture - Bioflu- [email protected] ids of Reproduction: Oscillators, Viscoelastic Net- works and Sticky Situations JP1 From fertilization to birth, successful mammalian repro- Spatio-temporal Dynamics of Childhood Infectious duction relies on interactions of elastic structures with a Disease: Predictability and the Impact of Vaccina- fluid environment. Sperm flagella must move through cer- tion vical mucus to the uterus and into the oviduct, where fertil- ization occurs. In fact, some sperm may adhere to oviduc- Violent epidemics of childhood infections such as measles tal epithelia, and must change their pattern of oscillation provide a particularly clear illustration of oscillatory to escape. In addition, coordinated beating of oviductal ’predator-prey’ dynamics. We discuss limits on the pre- cilia also drives the flow. Sperm-egg penetration, trans- dictability of these systems, both in the era before vacci- port of the fertilized ovum from the oviduct to its implan- nation and at present, where vaccine hesitancy limits the tation in the uterus and, indeed, birth itself are rich ex- effectiveness of vaccination programs in many countries. amples of elasto-hydrodynamic coupling. We will discuss We also discuss the impact of viral evolution on predictabil- successes and challenges in the mathematical and compu- ity and the design of vaccination programs, with particular tational modeling of the biofluids of reproduction. In addi- reference to influenza and rotavirus. tion, we will present reduced models that evoke intriguing questions in fundamental fluid dynamics. Bryan Grenfell Princeton University Lisa J. Fauci [email protected] Tulane University Department of Mathematics [email protected] CP1 Analysis of Block Methods for Toeplitz Matrices SP2 Toeplitz matrices and matrices from related matrix alge- The John von Neumann Lecture: Satisfiability and bras arise in a variety of applications including signal pro- Combinatorics cessing and partial differential equations. The analysis of iterative solvers for linear systems is usually based on The Satisfiability Problem, which asks whether or not a studying the generating symbol of the iteration matrix. given Boolean formula can be satisfied for some values of Block methods can be studied using proper decomposi- its variables, has long been thought to be computationally tions of the matrices, resulting in analyses that fit in the hopeless. Indeed, SAT is the well-known ”Poster Child” established framework used to analyze non-blocked meth- for NP-complete problems. But algorithmic breakthroughs ods. These techniques will be presented in this talk. have made it possible for many important special cases of the problem to be solved efficiently. Industrial-strength Matthias Bolten ”SAT solvers” have become a billion-dollar industry, and University of Wuppertal they play a vital part in the design of contemporary com- Department of Mathematics puters. The speaker will explain how the new SAT tech- [email protected] nology also helps us to solve a wide variety of problems that belong to combinatorial mathematics. CP1 Donald E. Knuth A Feast Eigenvalue Algorithm Without Solving Stanford University Linear Systems mam(at)theory.stanford.edu The FEAST Eigenvalue algorithm uses a shift-invert strat- egy that requires solving multiple linear systems in the SP3 complex plane. Similarly to other shift-invert algorithms, W.T. and Idalia Reid Prize in Mathematics solving the linear systems represents the most challenging part of the computation. We present an innovative mod- Not available at time of publication. ification of FEAST that finds interior eigenpairs without solving linear systems, instead using only a few iterations Ioannis Kevrekidis of iterative refinement per subspace update, thus removing Dept. of Chemical Engineering the need for using matrix factorizations or preconditioner Princeton University matrices. [email protected] Brendan Gavin ECE Department SP4 University of Massachusetts, Amherst I. E. Block Community Lecture: Toy Models [email protected] Would you like to come see some toys? ’Toys’ here have a special sense: objects of daily life which you can find or Eric Polizzi make in minutes, yet which, if played with imaginatively, University of Massachusetts, Amherst, USA reveal surprises that keep scientists puzzling for a while. [email protected] We will see table-top demos of many such toys and visit some of the science that they open up. The common theme CP1 is singularity. Combining Krylov Subspace Recycling with Recy- Tadashi Tokieda cling Preconditioners for Sequences of Linear Sys- AN16 Abstracts 5 tems the utility of the Lavalette distribution in several dataset and discuss estimation and goodness of fit methods. Preconditioners are generally essential for fast convergence of iterative solvers. For sequences of systems, it may be Oscar Fontanelli, Pedro Miramontes advantageous to recycle (update and reuse) precondition- Facultad de Ciencias, Universidad Nacional Aut´onoma de ers. Recycling Krylov subspaces from previous systems M´exic is a complementary method for reducing computational [email protected], [email protected] cost. We examine the combination of recycling subspaces with recycling preconditioners, providing theoretical and Germinal Cocho numerical analysis. Applications include model reduction, Instituto de Fsica, Universidad Nacional Aut´onoma de tomography, and others. M´exico cocho@fisica.unam.mx Arielle K. Grim-McNally, Eric De Sturler Virginia Tech Wentian Li [email protected], [email protected] The Feinstein Institute for Medical Research [email protected] CP1 Neighbor Discovery for Algebraic Multigrid and CP2 Matrix Migration Optimizing Radioactive Contamination Monitor- ing Using Utility Functions in a Bayesian Network The idea of a neighbor — processors to whom point-to- point messages get sent/received — is core to MPI-based After an incident involving radionuclides release, the level parallelism. Finding those communication patterns re- of radioactive contamination of individuals is assessed from quires neighbor discovery, which can be quie expensive. results of bioassay monitoring. Such retrospective assess- We will focus on two computational kernels involved in al- ment is prone to uncertainty which Bayesian methods gebraic multigrid — migration of sparse matrices between are well-adapted to evaluate. An algorithm based on a processors and sparse matrix-matrix multiplication. We Bayesian network (BN) providing an initial quantification will show that for these kernels, passing additional infor- of the contamination and its associated uncertainty was de- mation during point-to-point communication can allow one veloped to optimize bioassay monitoring. The best follow- to perform neighbor discovery more efficiently. up program is then determined through a prospective util- ity function integrated in the BN. Christopher Siefert Sandia National Laboratories Leila Gharsalli [email protected] IRSN, France [email protected] CP1 Mathilde Vincent Heavy Ball Minimal Residual Method for Least- Ancient IRSN Squares Problem [email protected] The heavy ball minimal residual (HBMR) method is pre- sented for solving overdetermined least-squares problem Estelle Davesne, Eric Chojnacki, Eric Blanchardon Ax − b ,whereA is a sparse matrix. HBMR method IRSN 2 [email protected], [email protected], seeks optimal approximate solutions of the least- squares T [email protected] problem by minimizing the residual norm A r 2 over both the Krylov subspace obtained by the restarted Golub- Kahan bidiagonalization process and the information of the CP2 Krylov subspaces in the previous cycles. Numerical exper- iments are reported to show the advantages of the HBMR Modeling Extreme Weather Events method. The 2016 Blizzard brought disastrous outcomes to the Mei Yang Northeastern U.S. Using extreme value analysis to ana- University of Texas at Arlington lyze NYCs historical daily extreme snowfall data up to [email protected] the 2006-2007 winter, we will determine if predicting the Blizzard was reasonable. Initial analysis shows that time- independent extreme value models did not predict the Bliz- Ren-Cang Li zard, suggesting models need to take climate changes into University of Texas - Arlington account. We also will use data up to January 2016 to assess [email protected] return levels for future extreme events. Mintaek Lee CP2 Boise State University New Insights into Scaling Laws: The Lavalette Dis- [email protected] tribution and Its Properties We introduce a novel family of probability