title:
=====

"From Combinatorics to Scientific Computing"

abstract:
========

Scientific computing is usually perceived as a field that focuses on numerical
computations and, indeed, it was born because scientists and engineers needed
numerical answers to mathematical problems. Nevertheless, this field is not
necessarily concerned just with computations of numerical nature, a large
number of computations being actually discrete. This happens either because the
original problem is continuous and must be discretized in order to be solved on
a computing platform or because the original problem itself is discrete.
Regardless of the reason, a significant number of discrete computations shapes
the subfield of combinatorial scientific computing, which is concerned with the
design, analysis and implementation of algorithms for combinatorial (discrete)
  problems that arise in scientific computing.

Because of my background in combinatorial algorithms I approach scientific
computing from a combinatorial perspective and I consider myself as part of the
combinatorial scientific computing subfield, although I have also worked on
problems of purely numerical nature.  In this talk I discuss two different but
related topics in order to illustrate one of the various connections between
combinatorics and scientific computing. In the first part of the talk I focus
on the purely combinatorial problem of graph matching which has applications in
sparse matrix computations. Such computations are heavily used in scientific
computing problems, therefore the connection between the two computational
fields. In order to fully motivate this discussion I switch to a particular
scientific computing problem in the second part of the talk. I conclude with a
list of research topics that are currently of interest in combinatorial
scientific computing.

biography:
=========

Florin Dobrian did his undergraduate studies in Romania, at the Polytechnic
Institute of Bucharest, in the Department of Computer Engineering. After
graduation he worked for several years in the software industry, before joining
a doctoral program in the Department of Computer Science at Old Dominion
University in Norfolk, Virginia. Florin obtained his Ph.D. degree in 2001, his
doctoral dissertation focusing on algorithms for sparse matrix computations.
This work also led to the implementation of a new and robust sparse linear
solver package that is based on modern software techniques. From 2001 to 2006
Florin was employed by DOE's "Scientific Discovery through Advanced Computing"
(SciDAC) program, first as a postdoctoral fellow at Old Dominion University,
then as an associate research scientist at Columbia University in New York,
his projects focusing both on software infrastructure (sparse linear solvers)
and on scientific computing applications. At present he continues his work
for DOE in SciDAC 2, a five year continuation of the original SciDAC. Florin
belongs to SciDAC's "Combinatorial Scientific Computing and Petascale
Simulations" institute, a collaboration between Old Dominion University,
Sandia National Laboratories, Argonne National Laboratory and Ohio State
University, his attributions including the design, analysis and
implementation of efficient parallel matching algorithms for petascale
computers.