SIAM Workshop on Object Oriented Methods for Inter-operable Scientific
and Engineering Computing
October 21-23, 1998
IBM Research
Yorktown Heights, New York
Background:
There is a growing awareness in the universities, and industrial
and governmental labs, that object oriented methods have the potential
for greatly improving the usefulness of computers in science and
engineering. There are already many efforts underway to redesign and
reimplement large codes that were written in the 70's and 80's to take
advantage of the improvement in maintainability and flexibility that
OO designs offer.
There still remains a large opportunity to improve the amount of reuse
within the community. Repositories such as Netlib and indices like
GAMS have improved our ability to share code, but making the shared
code useful requires widespread agreement about how the code is
structured and how scientific and engineering codes should
interoperate.
The world is changing. It used to be that large companies had their
own internal development teams, which implemented techniques from the
open literature or developed their own proprietary methods.
University researchers looked for methods for new classes of problems,
and pushed the limits of problem size. Software companies provided
application packages aimed at solving common problems, bundled
with everything from front-ends, to mesh generation and back-ends.
Now, as many large companies are cutting back on in-house software
development, universities and government labs seem to be doing increasingly
more software development and software vendors are springing up to provide
special purpose codes.
What are the new roles for academics, software companies, and industry?
Can a university department support a software product? How does a
company test and fire-harden research code? How can software
companies quickly incorporate new methods into a professional quality
product? Who owns what, and who supports what?
This interdisciplinary workshop will bring together representatives from
academia, software vendors, industry, and government labs, to identify
current and future challenges to implementing and using mathematical
algorithms in scientific and engineering computing.<p>
Site:
The Thomas J. Watson Research Center, established in 1961, serves as the
worldwide headquarters for IBM Research. With facilities in Yorktown
Heights and Hawthorne in New York's Westchester County, the lab houses more
than 1,200 scientists focusing research in the areas of semiconductors,
physical and computer sciences, and mathematics.
Yorktown Heights is located 50 miles north of New York City.
Organizing Committee:
Mike Henderson, IBM Research
Chris Anderson, UCLA
Steve Lyons, Mobil Technology Co.
Presentations:
The workshop will consist of 3 days of contributed and invited
presentations and discussions including, not not limited to, the
broad topics of:
The current state of the art.
- Tools that work together well (or poorly).
- The environments in which codes are used.
- How new techniques make their way to industrial use.
Developing interoperable scientific codes.
- Packaging codes.
- Balancing interoperability with performance.
- Maintenance and ownership of codes.
Future directions.
- Language of the future.
- Design techniques.
- Standards for the interoperation of mathematical software?
Case studies, position papers proposing topics for discussion, and
technical talks describing particular approaches are solicited.
The organizers are looking for a broad spectrum of participants,
including those involved in the development of industrial codes,
and industrial end users.
How to Contribute:
A report will be issued covering the workshop. Participants will be
required to submit a document for inclusion in this report.
Those wishing to participate should submit an abstract of a
case study, position paper or technical talk by July 1, 1998.
Abstracts should be sent to mhender@watson.ibm.com.
Notifications of acceptance will be sent by July 30, 1998.
Important Dates to Remember:
July 1, 1998, deadline for submission of extended abstract.
July 30, 1998, notification of acceptance.
August 31 1998, deadline for submission of final paper.