CSE RESEARCH SYMPOSIUM
KEYNOTE ADDRESS
SPEAKER: DR. IAN FOSTER
Mathematics & Computer Science Division
Argonne National Laboratory, Argonne, IL
DATE: Friday, April 18, 1997
TIME: 10:00 AM
PLACE: 335 Grainger Engineering Library
TITLE: Designing and Building Parallel Climate Models
ABSTRACT
The impossibility of meaningful experiments makes numerical modeling a
particularly important tool in studies of the earth's climate system.
Driven by the need for greater physical resolution, physical realism,
and timescales, climate models have historically been among the first
applications to exploit new computer architectures. In recent years,
a number of groups have addressed the problem of transitioning climate
models from vector multiprocessors to scalable parallel computers. In
this talk, I describe one such effort, a joint project of Argonne and
other laboratories and universities. This project has produced new numerical
methods and parallel algorithms suitable for use on parallel computers,
and incorporated these new techniques in models capable of executing with
high efficiency on multiple parallel platforms. My goal is to describe
not only the numerical and algorithmic techniques required to build
parallel climate models, but also the software engineering issues that
arise when dealing with systems of this complexity. In many respects,
the latter problems are harder than the former.
BIOGRAPHY
Ian Foster received his BSc (Hons I) in Computer Science from the
University of Canterbury, New Zealand, and his PhD in Computer Science
from Imperial College, England. He holds a joint appointment as a
Scientist at Argonne National Laboratory and Associate Professor in the
Department of Computer Science at the University of Chicago. He is the
author of three books and over 100 articles and reports on various topics
relating to parallel and distributed computing. Dr. Foster
is a member of the Technical Steering Committee for the NSF Center for
Research on Parallel Computation and of the Science Team of the DOE's
CHAMMP climate modeling program. He also coleads the Globus project,
which is developing software infrastructure for high-performance
distributed computing.