[BANANA] Stanford SCREAM this Saturday

[David Gleich]

As a courtesy to Stanford's SIAM Student Chapter, if you plan to
attend the symposium, could you please send email to

sssc-scream-2007 at lists.stanford.edu

with your name and affiliation?

========================================================================

Please join the Stanford SIAM Student Chapter for a Symposium on
Current Research in Engineering and Applied Mathematics (SCREAM 2007).
The goal of the symposium is to promote student presentations on
current research at Stanford University.

The symposium will be held

* Saturday, May 5th
* from 10:45am until 4:20pm in the
* Gates Building, Rm. 104.

It consists of

* a guest lecture,
* lunch, two breaks, and
* eight student talks.

10:45 AM : Introductions by Prof. Margot Gerritssen

11:00 AM : Guest Lecture by Prof. James Demmel of UC Berkeley on
            "Recent Progress in Fast and Accurate Linear Algebra"
            (see abstract below)

12:00 PM : Lunch provided by the SIAM Student Chapter

01:00 PM : John Carlsson - Finding Equitable Convex Partitions and
            Resource Allocation Applications
01:20 PM : Esteban Arcaute - On Threshold Behavior in Query
            Incentive Networks
01:40 PM : Jeremy Kozdon - Simulating Gas Injection into a Porous Medium

02:00 PM : Break 1

02:20 PM : Benjamin Armbruster - Contact Tracing to Control
            Infectious Disease
02:40 PM : Paul Constantine - PageRank and Polynomial Chaos

03:00 PM : Break 2

03:20 PM : Anwei Chai - Deconvolution: A Wavelet Frame Approach
03:40 PM : Will Fong - Stability of Asynchronous Variational Integrators
04:00 PM : Thomas Callaghan - IMRT Inverse Planning with Voxel-Based
            Penalty Scheme

We hope to see you at what promises to be an exciting symposium!

========================================================================

Abstract for Professor James Demmel's guest lecture

We describe recent progress in designing fast and accurate algorithms
and software for linear algebra operations. First we review recent
work in designing fast matrix multiply routines, and show how they can
be used to solve all standard dense linear algebra problems (linear
systems, least squares, eigenvalue problems) with the same asymptotic
complexity as well as stably. Second, we discuss the automatic
generation of fast software for sparse linear algebra.  The challenge
is that the fastest way to implement basic operations like
sparse-matrix vector-multiplication (SpMV) can depend both on the
computer and the matrix sparsity pattern, which may not be known until
run-time. We show how to combine off-line benchmarking and fast
run-time statistical sampling of a sparse matrix to automatically
construct the fastest possible SpMV routine.

========================================================================

Thanks to SIAM and the GSC for providing funding for this event.