NA Digest Friday, February 27, 1987 Volume 87 : Issue 5

This weeks Editor: Gene Golub

Today's Topics:


Date: Thu, 26 Feb 87 10:39:15 cst
From: dongarra@anl-mcs.ARPA (Jack Dongarra)
Subject: Argonne's ACRF Class

Argonne National Laboratory has set up an Advanced Computing
Research Facility (ACRF) for the study of parallel computing. It
currently features an 8-processor Alliant FX/8, a 20-processor
Encore Multimax, a 24-processor Sequent Balance 21000, a
32-processor Intel iPSC hypercube machine, and a 16-processor Intel
iPSC hypercube with vector processors. Local projects utilizing
the ACRF include investigations in parallel logic programming and
parallel linear algebra and the development of portable parallel
programming methodologies.

To encourage use of the ACRF as a national user facility,
Argonne is sponsoring various classes to familiarize potential
users with the ACRF multiprocessors and with parallel programming
in general. The next class will be held on March 11 - 13 and another
April 29 - May 1. The first two days will cover parallel programming
on the Alliant, Encore, Sequent, and Intel computers; the third day
will be devoted to consideration of each attendee's particular project.
Fortran will be emphasized as the primary programming language, although
C will be discussed. This will be a hands-on class; at its completion
participants will have written and run a number of programs on each
machine, and should be familiar with the ACRF environment.

Those interested in the March or April class should contact

Teri Huml
Mathematics and Computer Science Division
Argonne National Laboratory
Argonne, IL 60439-4844

(312) 972-7163

There will be no charge for this class, nor is any financial support
for attendees available.


Date: Fri, 27 Feb 87 14:28:30 snt
From: Sven Holmquist <enea!volvo!sven@seismo.CSS.GOV>
Subject: invitation

Att: Gene Golub

By suggestion from Axel Ruhe at Chalmers in Goteborg I'm sending
you a copy of the invitation sent to usnet for further distribution
as you find suitible

Best regards

Tony Elmroth
Volvo Data

/Thru Sven Holmquist:
Sven Holmquist, Volvo Data AB, DVF, S-405 08 GOTEBORG
Tel: +46 31 669176

UUCP: sven@volvo.UUCP or

ARPA: enea!volvo!


Keywords: CAD, approximation, shape control

25-26 Mars 1987



Volvo, Torslanda, Goteborg

The Department of Mathematics at Chalmers University of
Technology, Volvo Data AB and the Swedish Institute of Applied
Mathematics have the pleasure of inviting you to a symposium on
held at Volvo, Torslanda on March 25 and 26, 1987.

A reception will be arranged on the evening of March 24. The
Chairman and Chief Executive Officer of AB Volvo, Med Dr h c
Pehr G. Gyllenhammar will take part in the initial deliberations
on March 25.

The purpose of the symposium is to bring together users of
computer aided design with mathematicians working in the fields of
optimization or approximation theory. The further advance of
CAD-technology requires the development of automatic procedures
for the generation of curves and surfaces. This leads to difficult
mathematical and computational problems, especially when one wants
the object designed to have a given quality.

This symposium will provide a status report on this exciting
interface between CAD and mathematics.

Guest speakers include:

Prof. Xavier Benveniste
Ecole Polytechnique and Dassault Systemes

Prof. Dr. Wolfgang Dahmen
Universitat Bielefeld

Prof. John A. Gregory
Brunel University

Prof. Michael A. Lachance
University of Michigan

Prof. Tom Lyche
University of Oslo

Prof. Charles A. Micchelli
IBM Thomas J. Watson Research Center

Lennart Johansson
Product Design Department
Volvo Car Corporation

Prof. Bjorn E. J. Dahlberg
University of Goteborg and Chalmers University of Technology

Dr Roger Andersson
Volvo Data AB


RECEPTION : March 24, 7.00 - 9.00 PM, Park Avenue Hotel

SYMPOSIUM : March 25, 8.15 AM, buses leave Park Avenue Hotel.
Proceedings at Exhibition Hall, Volvo, Torslanda
from 9.00 AM to 6.30 PM, when buses leave for
Park Avenue Hotel.

March 26, 8.15 AM, buses leave Park Avenue Hotel.
Proceedings at Exhibition Hall from 9.00 AM to 4.30 PM
when buses leave for Park Avenue Hotel and Landvetter

BANQUET : March 25, 8.00 PM, Park Avenue Hotel.

General information:

COST : Symposium, including lunch March 25-26, SEK 250.
Banquet, March 25, SEK 250.

HOTEL : Park Avenue Hotel is recommended. Single room per night
SEK 630 incl. Breakfast.

Registration and further information: call +46-31-66 17 07
or write to:

Central Services
Dept 38210, HA BV S
AB Volvo
S 405 08 Goteborg

Information on the scientific programme at the symposium can be
obtained from:

Dr. Roger Andersson
Volvo Data AB
Tel: +46-31-66 91 40. MEMO-id: VD.2930RA

Prof. Bjorn E. J. Dahlberg
Deparment of Mathematics
University of Goteborg and
Chalmers University of Technology
Tel. +46-31-16 79 77. MEMO-id: VD.CTHBD


From: "Dr. Mel Ciment" <>
Subject: Dear colleague letter
Date: Fri, 27 Feb 87 08:56:52 -0500

Attached is a copy of a "Dear Colleague Letter" which is
self-explanatory. In essence it is announcing NSF's initiative in
Computational Science and Engineering and CISE's involvement in that
initiative. Please bring this information to the attention of all
interested parties.

Questions should be referred to:

Dr. Mel Ciment or Dr. Al Harvey
(202) 357-9776 (202) 357-7727

Computer and Information Science and Engineering
1800 G Street, NW
Washington, DC 20550

January 30, 1987
Dear Colleague:

This letter is being sent to you in order to inform the scientific
community of important activities presently taking place at the National
Science Foundation. The fiscal 1987 Budget for the National Science
Foundation includes funds of several million dollars in support of an
initiative known as: "Computational Science and Engineering" (CSE). These
funds are being distributed among the various disciplines: Biological,
Behavioral and Social Sciences; Computer and Information Science and
Engineering; Mathematical and Physical Sciences; Science and Engineering
Education; Engineering; and Geosciences. It is anticipated that this new
program will stimulate activity at the interface between the sciences and
advanced computer technology. The NSF strongly urges investigators to
inquire further about the details of the initiative with the various
program directors at the Foundation. Enclosed with this letter is a
program announcement (NSF 86-91) that describes the goals of the overall
NSF/CSE programs.

Many of you may know that there have been a number of organizational
changes at NSF. One is the creation of a new Directorate for Computer and
Information Science & Engineering (CISE), which combines several
preexisting computer activities from other directorates, the Division of
Computer Research, the Division of Information Science and Technology, and
programs in Computer Engineering, Communications and Signal Processing, and
the Office of Advanced Scientific Computing.

CISE supports research in computer science, information systems and
processing, robotics, networking and communications, microelectronics,
advanced scientific computing and intelligent systems. The overall goal of
the effort is to improve the knowledge base, research infrastructure and
professional labor force needed to understand and improve the nature,
synthesis and use of computing and information processing devices and
systems. The current structure of CISE includes 5 divisions:

* Computer & Computation Research
* Advanced Scientific Computing
* Information, Robotics & Intelligent Systems
* Networking & Communications Research & Infrastructure
* Microelectronic Information Processing Systems


Although many of the efforts described below can be performed by single
investigators, and will be, to some extent, supported in that form, this
new initiative will emphasize strong inter-disciplinary approaches to the
enhanced computing capability and environment of the scientist and
engineer. Proposals involving computer scientists, mathematicians,
scientists and engineers, and specialists in such areas as computer
graphics, might be integrated in such a way as to form an
inter-disciplinary group or team, addressing specific problems of
importance to one or more scientific or engineering disciplines. For
example, such proposals might be strongly coupled with the efforts of
innovators of state-of-the-art algorithms and software for application on
machines with highly parallel architecture. Such approaches could develop
new paths for entire disciplines to follow. They will be coordinated
among CISE programs and the NSF scientific and engineering disciplines.


Proposals with a strong interdisciplinary approach are being encouraged in
the following computational areas, although this list is not intended to
be complete:

* Software and Algorithm Development
* Application of Advanced Technologies to problem solving
* Visualization, Graphics and Image Processing
* Formation of Novel Computational Strategies
* Network and Communication Systems
* Performance Evaluation of Computer Systems and Software
* Distributed and Parallel Processing and Vectorization

Software and Algorithm Development: It is widely accepted that software
lags hardware development. This fact is especially true for supercomputers
and other machines of advanced architectures. Within CISE, the
Computational Science and Engineering Initiative will focus on research
that addresses the development of novel algorithms and their implementation
into useful software packages, the creation of friendly working
environments, and the automatic production of fast, efficient code for
scientists and engineers working on advanced computers. Innovations in
languages, user-friendly interfaces, software tools, etc. might address
issues related to the speed up of code development and therefore the
productivity of researchers. Methods which assist in the portability of
code across a variety of advanced machines will also be considered.

Visualization, Graphics and Image Processing: More powerful visualization
capability is being demanded to take advantage of the most powerful
machines. Substantial insights are already being gained from graphics,
which is the only way to understand many scientific phenomena. Among the
many research topics in graphics and image processing are: extemporaneous,
interactive steering of numerically intensive calculations; dynamic
visualization of fields in higher dimensions; high bandwidth graphics,
networks and protocols; massive data set handling and standards;
vectorization and parallelized algorithms for visualization;
workstation-driven remote use of supercomputers; standard graphics-oriented
scientific programming environments.

Performance Evaluation: A recent NAS/NRC report on "An Agenda for Improved
Evaluation of Supercomputer Performance" remarks on the severe lack of
scientific foundation, regarding our ability to evaluate the performance of
advanced computers. Investigations into the definition and techniques for
performance evaluation of parallel or other computer systems are encouraged
either as the principal subjects of proposals, or as components of other
research projects in this initiative.

Distributed and Parallel Processing and Vectorization: The direction of
advanced scientific computing is clearly headed toward parallelism to
achieve increased capacity. Since the complexities of programming in
parallel environments with optimally vectorized code place even more
challenging demands on software and algorithm development, the
Computational Science and Engineering Initiative will emphasize means to
provide effective scientific computing in vector and highly parallel
environments. For example, the initiative will consider methods for
automatically parallelizing existing scientific codes or rewriting them for
efficient use on machines of advanced architectures. Also, software tools
for increasing productivity of the programming environment on parallel and
distributed architectures will be encouraged especially, for vector and
multiprocessor computers.

Advanced Technologies: The Science and Engineering Initiative welcomes
proposals concerned with areas of technology that have a strong impact on
the conduct of future computing. Examples include high capacity and/or
high performance mass storage coupled with appropriate file and data base
management systems, optical computing, neural networks, non-binary
computing, or any such ideas that could influence the nature of advanced
scientific computing. The CSE Initiative will cooperate with other
programs on the potential application of advanced computing technologies
and systems to scientific and engineering problems. Proposals of this type
will be coordinated as appropriate both within and outside the Foundation.

Formulation of Novel Computational Strategies: New computer architectures,
communications technologies, languages, and other software or hardware
advances becoming available offer promise of greatly enhanced speed,
flexibility, or cost-effectiveness in performing scientific and
engineering research. However, the hope for significant increases in
insight to discipline specific problems may demand a fundamental revision
in the strategic approach taken toward solving problems to make effective
use of these options. Investigations into alternate ways of formulating
and computing important scientific and engineering problems are encouraged.

Network and Communication Systems: Recently increased accessibility of
advanced computing resources opens possibilities for new,
computationally-based, advances in the understanding - i.e., analysis and
especially design/synthesis - of computer networks and communication
systems generally. This Initiative will entertain proposals for
computational research in such problem areas as: event-based, Monte
Carlo, or other simulation methodology applied to very large scale computer
networks with attention to realistic detail; protocol design based on
computational studies of state-machine models of networks with state spaces
so large as to render such studies hitherto impracticable; specialized,
interdisciplinary studies of Presentation- and Application-layer protocols;
knowledge-based or other expert aids for intelligent dynamic network
management; and research using symbolic computation in studies of algebraic
coding theory. Proposals in these and other appropriate topical areas will
emphasize the innovative computational nature of the proposed
investigations, and may include the use of advanced (e.g., highly parallel)
architectures in the research.


C. Gordon Bell, Assistant Director
Computer and Information Science and Engineering

For further information write or call the program director of the program
most related to your area of interest or Dr. Mel Ciment, Division of
Advanced Scientific Computing (202-357-9776).


End of NA Digest