NA Digest Thursday, December 29, 1988 Volume 88 : Issue 51

Today's Editor: Cleve Moler

Today's Topics:

-------------------------------------------------------


From: Gaston H Gonnet <attcan!utgpu!watmath!watdragon!ghgonnet@uunet.uu.net>
Date: 7 Dec 88 17:07:19 GMT
Subject: ISSAC-89 Conference Announcement

ACM - SIGSAM

1989 International Symposium on Symbolic and
Algebraic Computation

CALL FOR PAPERS

ISSAC-89

July 17-19, 1989
Portland, Oregon

The 1989 International Symposium on Symbolic and Algebraic
Computation will be held in Portland, Oregon on July 17-19,
1989.
The 1989 conference will feature original research contribu-
tions, demonstrations of hardware and software and invited
lectures. Typical, but not exclusive, topics of interest
include:

- symbolic algebraic and analytical algorithms,
- automatic theorem proving and programming,
- computational analytical geometry,
- problem representation,
- languages and systems for symbolic and algebraic
computation
- applications to science, engineering and education.

Papers presenting original research on theoretical and applied
aspects of this theme are being sought. Submissions will be
refereed by a program committee. Authors should send nine
copies of a detailed abstract (5 to 10 double-spaced pages) by
February 17, 1989 to the Programme Committee Chairman:

Dr. Gaston Gonnet
Department of Computer Science
University of Waterloo
Waterloo, Ontario
Canada, N2L 3G1

Late submissions risk rejection without consideration.
Authors will be notified of acceptance or rejection by March
24, 1989. A working draft of the paper, not exceeding 15
pages, will be due by April 24, 1989 for inclusion in proceed-
ings which will be made available at the conference.

PROGRAM COMMITTEE (Preliminary)

Manuel Bronstein Bob Caviness
Tetsuo Ida Anthony Hearn
Daniel Lazard Teo Mora
Franz Winkler Gaston Gonnet

General Conference Chairman: Dr. Kamal Abdali
Local Arrangements Chairman: Dr. Bruce Jensen
Exhibition Chairman: Dr. Dennis Arnon
Publicity Chairmen: Benton Leong


------------------------------

From: Michael Todd <miketodd@gvax.cs.cornell.edu>
Date: Tue, 20 Dec 88 15:30:09 EST
Subject: Complexity of Approximating Eigenvalues

Does anyone know of references to precise bounds on the bit (or arithmetic)
complexity of approximating the eigenvalues of a real symmetric matrix?
I would appreciate any pointers.

-- Mike Todd (miketodd@gvax.cs.cornell.edu)


------------------------------

From: Erkki Aalto <mcvax!enea!kth!draken!tut!hydra!hylka!aalto@uunet.uu.net>
Date: 19 Dec 88 08:53:31 GMT
Subject: Exponential and Logarithm of Power Series

My research involves solving a large system of nonlinear equations, and
I am having trouble with an instability that does not seem to originate
from the original equations. The trouble seems to be in two functions F
and S, where F = exp(S). Solution of a related problem that involved
calculating the power series of S from that of F showed that S seems to
be remarkably stable against variations in F caused by taking more and more
terms into expansion of F. Unfortuntely this gives a reason to expect that
the inverse calculation is higly unstable and that seems to be the source
of my problems. Does anyone know a stable algorithm for computing the power
series of a logarithm of a function?

Erkki Aalto
University of Helsinki


------------------------------

From: George Cybenko <gc@uicsrd.csrd.uiuc.edu>
Date: Wed, 21 Dec 88 13:10:29 CST
Subject: New Address for George Cybenko

I have recently changed jobs. My new address is

George Cybenko
Center for Supercomputing Research and Development
University of Illinois,
Urbana, IL 61801

My phone number is (217) 244-4145. My academic appointment is in
Electrical and Computer Engineering.

George


------------------------------

From: Sanzheng Qiao <qiao@amvax.TN.CORNELL.EDU>
Date: Mon, 26 Dec 88 15:18:52 EST
Subject: Address Change for Sanzheng Qiao

As of Jan. 1, 1989, my address will be:
Dept. Computer Science and Systems
McMaster University
Hamilton, Ontario CANADA
L8S 4K1
e-mail address:
utai!gpu.utcs.toronto.edu!maccs!qiao@uunet.UU.NET

Sanzheng Qiao

------------------------------

From: Apostolos Gerasoulis <gerasoul@aramis.rutgers.edu>
Date: Wed, 21 Dec 88 16:35:37 EST
Subject: Fourth Parallel Circus

The fourth "Parallel Circus" was hosted by Rutgers University and
co-chaired by Gene Golub (Dept. of Computer Science, Stanford
University) and Apostolos Gerasoulis (Dept. of Computer Science,
Rutgers University) on December 2/3 at the Hyatt Regency in New
Brunswick. This meeting provides a forum for researchers interested
in parallel computing to exchange ideas in an informal fashion. The
first Parallel Circus had been hosted by Yale and chaired by Martin
Schultz in May of 1987 and was followed by meetings at Cornell
University and IBM Kingston (organized by Charles Van Loan and Vijay
Sonnad, respectively).

The rising interest in parallel computing was documented by the about
80 attendees that made this meeting the largest and most diversified
ever. The twenty nine talks that were presented touched virtually
every aspect of scientific computing. Some distinct advantages are
the exposure to as yet unpublished work or work in progress and the
informal atmosphere fostering exchange of ideas.

Joe Flaherty (flaherje@turing.cs.rpi.edu) of the Computer Science
Department of Rensellaer Polytechnic Institute was volunteered to host
the next meeting which is tentatively planned for around April of
1989.

If you need additional info of the latest circus contact
gerasoulis@aramis.rutgers.edu.

The talks at the Fourth Parallel Circus were:

<Egbert Ammicht, AT&T, whuts!eea@att.att.com>:
Wavefront Beamforming on the AT&T Systolic Processor

<Christian Bischof, ANL, bischof@mcs.anl.gov>:
Incremental Condition Estimation and Applications

<Rob Bjornson, Yale, bjornson@cs.yale.edu>:
The Linda project:Parallelism to the People

<Anne Elster, Cornell, elster@svax.cs.cornell.edu>:
Some Basic Parallel Numerical Algorithms for the Polymorphic
Torus

<Anne Greenbaum, NYU, greenbau@cmcl2.nyu.edu>:
LAPACK: A Linear Algebra Library for High-Performance
Computers. (joint work w/ Jim Demmel and ANL)

<Ching-Tien Ho, Yale, ho@cs.yale.edu >:
Optimal Communication Primitives on Boolean Cubes

<Robert Kelly, Grumman>:
Techniques to relax synchronization in Parallel algorithms.

<Linda Kaufman, Bell Lab.>:
Cholesky factor updating techniques for rank 2 matrix
modifications

<David Keyes, Yale, keyes-david@yale.arpa>:
Domain Decomposed Solution ofReacting Flows

<Doyle Knight, JvNC, knight@jvncc.csc.org>:
John von Neumann National Supercomputer Center(JvNC): Status,
Plans and Research.

<Anita Mayo, IBM Watson, amayo@ibm.com>:
A new class of particle mesh methods

<Greg McRae,CMU, "mcrae%cpwpsca.bitnet"@forsythe.stanford.edu >
Solving Very Large Systems of Differential Equations.

<Greg McRae,CMU, "mcrae%cpwpsca.bitnet"@forsythe.stanford.edu >:
Parallel Algorithms for Combinatorial Optimization-Applications
to scheduling

<Bill Mitchell, General Electric, wmitchell%atl.decnet@ge-crd.arpa >:
Distributed and Centralized Parallel Algorithms for Selective
Scheduling Problems

<Ramesh Natarajan, IBM Watson, ramesh@ibm.com>:
Parallel Eigenvalue Computations using a Fetch-and-Add
Synchronization

<Izzy Nelken, israel@aramis.rutgers.edu>:
Scheduling for Message Passing Architectures with Applications
to linear Algebra (joint work w/ Apostolos Gerasoulis)

<Andrew Odlyzko, Bell Labs, amo@research.att.com>:
Rapid computation of zeros of the Riemann zeta function

<R. Panda, IBM Kingston>:
Parallel Pseudo-spectral Methods

<Rodgers, Penn State, pardalos@shire.cs.psu.edu>:
Parallel Branch and Bound Algorithms for Quadratic Integer
Programming(joint work w/ Panos Pardalos)

<Richard Pelz, Rutgers, pelz@jove.rutgers.edu>:
Spectral Methods on Hypercubes

<Joseph Peters, Rutgers, peters@occlusal.rutgers.edu>:
A Reconfigurable Compiler for Parallel and Pipelined Computers
(joint work w/ Stanley Dunn)

<P. Sadayappan, Ohio State, saday@tut.cis.ohio-state.edu>:
A circuit simulation on vector supercomputers

<Vivek Sarkar, IBM T. J. Watson, vivek@ibm.com>:
The programming problem for a general purpose multiprocessor

<Stig Skelboe, U. of Copenhagen, stig@diku.dk>:
A multilevel parallel solver for banded linear systems

<Binay Sugla, AT&T, sugla@vax135.uucp >:
An Introduction to the CAPER Concurrent Application
Programming System

<Anthony Terrano,Rutgers, terrano@caip.rutgers.edu >:
1. The Coherent Parallel Computer Project 2. On the derivation
of optimal problem partitioning
for Multiprocessor computers

<Sesh Venugopal, Rutgers>:
PYRROS: A Parallel scheduling system for linear algebra
problems. (joint work w/ Apostolos Gerasoulis)

<Zhiwei Xu,Rutgers, zxu@caip.rutgers.edu >:
A language Construct for Parallel Programming

<Feng Zhao, MIT, fz@ai.ai.mit.edu>:
The Parallel Multipole Method on the Connection Machine


------------------------------

From: Mike Grigoriadis <grigoria@zeno.rutgers.edu>
Date: Fri, 23 Dec 1988 2:06:40 EST
Subject: DIMACS and Special Year
Message-Id: <CMM.0.87.598864000.grigoria@zeno.rutgers.edu>
Status: RO

The following is a TEX file containing an announcement of the
newly-formed Discrete Mathematics and Theoretical Computer Science
Center and of its first ``special year'' on Discrete and Computational
Geometry.

Please be kind enough to post it and pass it on.

Thank you.

\nopagenumbers
%\hsize3.5in\hfuzz2pt
\magnification=\magstep1
\tolerance=1600
\centerline{\bf CENTER FOR DISCRETE MATHEMATICS}
\centerline{\bf AND THEORETICAL COMPUTER SCIENCE}
\bigskip
\centerline{\bf Special Year 1989-90}
\centerline{\bf Discrete and Computational Geometry}
\bigskip

Applications are invited for visiting and post-doctoral positions in
the Center for Discrete Mathematics and Theoretical Computer Science
(DIMACS). This center is supported through the NSF Science and
Technology Centers Research Program. The participating institutions
are Rutgers University, Princeton University, AT\&T Bell Laboratories
and Bell Communications Research. Research facilities are located at
the Rutgers and Princeton campuses.

The purpose of the center is to address a generally recognized need to
understand fundamental mathematical issues of computation. Applicants
are sought in all areas of discrete mathematics and theoretical
computer science, including (but not limited to) analysis of
algorithms, combinatorics, complexity, computational algebra, discrete
and computational geometry, discrete optimization and graph theory.
The Center will be able to offer long- and short-term visiting
positions. In addition, some regular positions at the participating
institutions may also be available.

A primary activity of the Center is to sponsor year-long research
programs on specific topics of current interest. The topic for
1989-90 is {\bf Discrete and Computational Geometry}. People with
expertise in this area are particularly encouraged to apply. During
this year the Center will sponsor a number of long-term visitors with
research interests in discrete and computational geometry, as well as
short-term research workshops in these areas to which a larger number
of participants will be invited. In addition, the Center is planning
a number of other research and educational activities, which will be
announced at a later time.

Postdoctoral and junior applicants must demonstrate superior research
and scholarship potential. Senior applicants must have an exceptional
record of research achievement. Successful candidates will pursue an
active research program and participate in the activities of the
Center.

Direct all inquiries to:
\medskip
\centerline{\vbox{\halign{&#\hfil\cr
Professor Daniel Gorenstein, Director\cr
DIMACS\cr
Hill Center for the Mathematical Sciences\cr
Rutgers University\cr
New Brunswick, NJ 08903\cr
Arpanet:~dimacs@aramis.rutgers.edu\cr}}}\medskip

All participating institutions are equal opportunity\slash affirmative
action employers.

\end


------------------------------

From: Michael Overton <overton@csd38.nyu.edu>
Date: Fri, 23 Dec 88 17:21:34 EST
Subject: Symmetric Band Generalized Eigenvalue Problems

There is an algorithm by Martin/Peters/Wilkinson for solving band symmetric
generalized eigenvalue problems using bisection and exploiting band structure.
This algorithm is the one recommended in the book by Strang and Fix for
solving finite element eigenvalue problems; in addition to the two papers
by Peters and Wilkinson referenced in Strang and Fix there is a paper by
Martin and Wilkison on the same subject which appeared in Numer. Math.
labeled as a "handbook contribution". However, this algorithm did not
appear in the final version of the Wilkinson and Reinsch "Handbook", and it
did not appear in EISPACK. Does anyone know why? More importantly, is
there a general purpose reliable implementation which is generally available?
Thank you for any information you may have.

Michael Overton.


------------------------------

From: Dennis Gannon <gannon@iuvax.cs.indiana.edu>
Date: Mon, 26 Dec 88 16:06:03 EST
Subject: Intnl. Conf. on Supercomputing 89

Last Call for Papers
ACM SIGARCH and Computer Technology Institute, Greece
1989 International Conference on Supercomputing
In cooperation with SIAM, AICA, BCS, INRIA, IPSJ and GI
June 5-9, Crete, Greece
Capsis Beach Hotel, Heraklion
Conference Co-Chairmen
George Paul, IBM USA and T. Papatheodorou, CTI Greece
Program Committee Directors
D. Gannon, Indiana and E. N. Houstis, Purdue
F. Hossfeld, KFA Chairman Europe and Africa,
Y. Muraoka, Waseda Chairman Japan and Far East,
J. Sopka, DEC Chairman North and South America
Program Committee
Japan and Far East Europe and Africa North and South America
K. Asai, Tokyo J.-P- Verjus, Grenoble D. Degroot, TI
M. Amamiya, Kyushu J. Lenfant, Rennes J. Dongarra, Argonne
Y. Kanada, Tokyo P. Lallemand, Paris J. Fox, Cal Tec
S. Suzuki, Fujitsu E. Gelenbe, Paris, S. Gallopoulos, CSRD
N. Suzuki, Japan IBM A. Lichnewski, INRIA Y. Patt, U. C. Berkeley
H. Tanaka, Tokyo J. Gurd, Manchester M. Simmons, LANL
H. Terada, Osaka P. Sguazzero, Rome C. Polychronopoulos, CSRD
H. Horikoshi, HItachi R. Perrot, Belfast J. Panetta, Brazil
M. Yamamoto, NEC I. Duff, Harwell F. Darema, IBM
T. Yuba, ETL W. Giloi, Bonn F. Baskett, Stanford
U. Trottenberg, Suprenum
W. Jalby, Rennes/INRIA

The proceedings will be published by ACM and papers are solicited
in the following areas.

Applications of Supercomputing including
studies in the physical science, artificial intelligence,
symbolic computation, visualization, mathematical software,
and numerical algorithms.
Software Systems including:
operating systems, parallel languages, compilers,
performance evaluation, programming environments
and high level problem solving systems
Architecture:
MIMD, SIMD and Data Flow systems,
memory system design (distributed, shared or multilevel),
interconnection networks and instruction architecture (RISC, CISC, etc.)

Authors should send five (5) copies of a full paper or an extended
abstract (8 to 15 pages) to the program chairman of their region.
The deadline for submissions is February 1, 1989.
The addresses for submissions are:

Europe and Africa:
Dr. Friedel Hossfeld
KFA Julich
ZAM Postfach 1913
D-5170 Julich
Fed. Rep. of Germany

North and South America:
John R. Sopka
Digital Equip. Corp.
BXB 1-1/F11
85 Swanson Rd.
Boxboro, Mass.
01719

Japan and Far East
Yoichi Muraoka
Dept. of Electrical Engineering
Waseda University
3-4-1 Okubo, Shinjuku-ku
Tokyo, Japan

Further details including a list of invited speakers and registration
information will follow. Stay tuned to this network.


------------------------------

From: D. G. Wilson <wilson@msr.EPM.ORNL.GOV>
Date: Wed, 28 Dec 88 13:41:25 EST
Subject: Simulating Vehicular Motion

I have received the following query from a friend in Florence. Can
anyone supply information? George Wilson

Have you any information on computer packages for simulating
the motion of a vehicle (in particular car, bus, train)?
What I am interested in is the simulation of the motion of a
car body when the wheels are running on an uneven surface.
The case of a train is also interesting. A more accurate
approximation could include the motion of passenger seats to
evaluate passenger comfort in the vehicle. Are you aware of
any comparative evaluations of what exists on the market, or
at least descriptions that are not mere advertising?

Mario Primicerio bitnet address: fismat@ifiidg


------------------------------

From: Richard Sincovec <sincovec@riacs.edu>
Date: Wed, 28 Dec 88 16:05:12 PST
Subject: RIACS/NASA Postdoc Fellowship in Parallel Processing

1989 RIACS/NASA POSTDOCTORAL FELLOWSHIP IN
PARALLEL PROCESSING

The Research Institute for Advanced Computer Science
(RIACS) at the NASA Ames Research Center, Moffett Field,
California, offers a one-year in-residence fellowship in
massively parallel algorithms and architectures. The
stipend is $50,000.

The winning fellow will perform research in the parallel
systems division of RIACS and the Numerical Aerodynamic
Simulation (NAS) systems division of NASA. The parallel
systems division works in partnership with NASA to explore
algorithms that will exploit computing architectures with
thousands of processors that can produce breakthroughs in
the NASA objective of achieving a thousand-fold increase in
computational capability by 1996.

RIACS and NASA have a large number of staff working in
this area who will be collaborators with the winning Fellow
including David Bailey (NASA), Paul Frederickson (RIACS),
Creon Levit (NASA), Youcef Saad (RIACS), Rob Schreiber
(RIACS), Horst Simon (Boeing), and Rich Sincovec (RIACS).
In addition, RIACS generally supports a number of visitors
in this area. Recent visitors were George Adams (Purdue),
Mikhail Atallah (Purdue), Tony Chan (UCLA), Jack Dennis
(MIT), and Walter Tichy (Karlsruhe).

The computing environment in which this research is
conducted includes a 32,768 processor Connection Machine
(CM-2), a Cray Y-MP with 8 processors, a Cray-2 with 4
processors, an Alliant FX-8 with 8 processors, a Sequent
Balance with 16 processors, and an Encore Multimax with 4
processors. RIACS also has access via network to a
Multiflow 7-200. The computing environment also includes
advanced graphics workstations by Ardent and Stellar, and
the NAS environment includes workstations by Silicon
Graphics.

To apply for the fellowship, you must have a PhD in
computer science or closely related field awarded or
expected between January 1987 and June 1989. Submit these
items by January 30, 1989:

1. Resume.

2. Copies of one or more papers published (or
accepted for publication).

3. Copy of your thesis abstract, introduction, and
conclusion.

4. Names of at least three well-known researchers
whom you have requested to provide letters of
reference in support of your application (see
below).

5. Description of your vision of what you can
accomplish at RIACS during your stay, and a
research plan for your proposed accomplishment.

As part of the application process, you must arrange for
at least three well-known researchers to send us letters of
reference. It is your responsibility to ensure that the
letters reach us by January 30, 1989.

Submit to: Dr. Richard Sincovec, Assistant Director for
Parallel Systems, RIACS, NASA Ames Research Center, 230-5,
Moffett Field, CA 94035.

All candidates will be notified by February 15, 1989, and
three finalists will be invited to Ames during March 1989
for discussions and a seminar on their research. The award
will be made in April and the winner will be expected to
begin research in residence by September 1989.

____________________________________________________________

RIACS is chartered to provide preeiminent leadership in
basic and applied computer science research as partners with
NASA in support of NASA's goals and missions. RIACS is a
not-for-profit institute operated by the Universities Space
Research Association, a consortium of 64 universities with
graduate programs in aerospace sciences. RIACS receives
funds through cooperative agreements with NASA; the parallel
systems division's funds come from the NAS systems division
and from DARPA. All research by RIACS is unclassified and
in the public domain. RIACS is an equal opportunity
employer. Inquiries about RIACS should be directed to Dr.
Peter J. Denning, Director.


------------------------------

From: SIAM <SIAM@wharton.upenn.edu>
Date: Thu, 22 Dec 88 11:35 EST
Subject: Season's Greetings from SIAM


A L L T H E B E S T


T O A L L O F Y O U


F O R T H E H O L I D A Y S E A S O N


A N D


T H E N E W Y E A R!



x
x x
x x x
x x x x
x x x x x
x x x x x x x
x x x x x x x x
x x x x x x x x x x
x x x x x x x x x x x
x x x x x x x x x x x x x
x x x
x x x
x x x



SOCIETY for INDUSTRIAL and APPLIED MATHEMATICS (SIAM)

------------------------------

End of NA Digest

**************************
-------