NA Digest Monday, August 10, 1998 Volume 98 : Issue 29

Today's Editor:
Cleve Moler
The MathWorks, Inc.

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to

URL for the World Wide Web:

From: Ron Boisvert <>
Date: Sat, 8 Aug 1998 02:19:38 -0400
Subject: Jama, A Java Matrix Package

JAMA is an elementary linear algebra package for Java. It provides
basic user-level classes for defining and computing with real dense
matrices. Among its capabilities are the following:

Object manipulation: constructors, get, set, copy, clone Elementary
operations: addition, multiplication, norm, transpose, etc.
Decompositions: Cholesky, LU, QR, SVD, eigenvalue Linear equation
solution: regular and least squares Derived quantities: condition,
determinant, rank, inverse

JAMA is intended to serve as _the_ standard matrix class for Java,
and will be proposed as such to Sun via the Java Grande Forum. A
straightforward public-domain reference implementation has been
developed as a strawman proposal for such a class. We are releasing
this version in order to obtain public comment. There is no guarantee
that future versions of JAMA will be compatible with this one.

The initial version of JAMA was developed by Joe Hicklin, Cleve Moler
and Peter Webb of the MathWorks and Ron Boisvert, Bruce Miller,
Roldan Pozo and Karin Remington of NIST.

Further information on JAMA, including complete documentation and
source code, can be obtained at


From: Christoph Helmberg <>
Date: Mon, 10 Aug 1998 15:35:25 +0200
Subject: ICM'98 Opening Ceremony Live in the Internet

ICM'98 Opening Ceremony Live in the Internet

Dear Colleague:

On Tuesday, August 18, the Opening Ceremony of the International
Congress of Mathematicians (ICM'98) in Berlin will be broadcast
live on the internet.

A highlight of the Opening Ceremony is the awarding of the
Fields Medals and the Nevanlinna Prize, sometimes called
the "Nobel" prizes of Mathematics.

For the first time in ICM history mathematicians throughout
the world will have a chance to participate in one of the most
important events in mathematics. Please inform your friends and
colleagues who do not have the possiblity to experience the
Opening Ceremony in person about this service.

The transmission will be done via MBone under the title
"ICM'98 Opening Ceremony" during the following
periods of time

Tuesday, August 18 10:00 - 12:30 Central European Summertime (CEST)
14:30 - 16:00 Central European Summertime (CEST)

This covers the whole Opening Ceremony and the speeches on the
achievements of the winners of the Fields Medal and the Nevanlinna Prize.

To give you an orientation,
Berlin 10:00 (this is CEST) corresponds to

Los Angeles 1:00 am
New York 4:00 am
Johannesbourg 10:00 am

Moscow 12:00 pm
Bombay 1:30 pm
Bangkok 3:00 pm
Beijing 4:00 pm
Tokyo 5:00 pm
Sydney 6:00 pm

To give mathematicians in America a chance to view the ICM opening
during daytime, we will REPEAT THE TRANSMISSION from 18:00 to 22:00
Central European Summertime (CEST) (leaving out the break).

For more information on this Mbone transmission, on the program
of the Opening Ceremony and on ICM'98 in general see the ICM'98 server,

We hope that you will join us in Berlin!


Martin Groetschel
President of the ICM'98 Organizing Committee


From: Arun Verma <verma@CS.Cornell.EDU>
Date: Mon, 10 Aug 1998 17:02:39 -0400 (EDT)
Subject: Automatic Differentiation for MATLAB

Dear Colleagues,

ADMAT -- an Automatic Differentiation toolbox for MATLAB
is now available. ADMAT can compute derivatives of programs written
in MATLAB (i.e., M-files), e.g. the gradient, the Jacobian or the Hessian
matrix. ADMAT is very easy to use, it can compute derivatives of
most M-files and very easy to plug-into other MATLAB toolboxes.

Our research web page
has more information about our Automatic differentiation project
and efficient computation of sparse and structured derivative
matrices using AD technology.

To download get the following tarred and compressed file:

ADMAT runs under MATLAB version 5.0 and above. Read the file ADMAT/README
for information on documentation and user guide and some test applications.



From: Kenneth Parker <>
Date: Thu, 6 Aug 1998 10:34:06 -0400
Subject: Undigestify for Microsoft Outlook

If you use emacs to read NA Net you can "Undigestify" the mail. That is
there exists a command for breaking the single incoming message into
several smaller messages, one per contributor. This is useful if you
want to save one or two of messages, perhaps so that you can reply to

Does anyone know of the existence of similar functionality for Microsoft
Outlook ... Perhaps some sort of plug in ...



From: Ke Chen <>
Date: Wed, 5 Aug 1998 21:44:50 +0100 (BST)
Subject: Help on Gibbs-Poole-Stockmeyer_King Algorithms

Dear Colleagues,

May I ask if anyone has used TOMS Algorithm 582 which implements
Gibbs-Poole-Stockmeyer_King Algorithms? Basically the copy I
downloaded from netlib does not compile on Unix. The error message
says "Subscript number 1 has value 0 in array LVLLST". I located
the problem in LEND=... when LEVEL=DEPTH i.e. PLSTRT-1=0 as you see
... ...
... ...
But what is the fix? Any help/suggestions on the solution or
alternative public domain programs will be appreciated. Thanks.

Ke Chen


From: John Crow <>
Date: Thu, 06 Aug 1998 16:01:17 -0500
Subject: New Contact Info for John A. Crow

Howdy -

This is to let you know that as of Aug. 10 I will no longer
be with SGI/Cray. New contact info is below. For the time being
I will concentrate on building up a custom technical software
business with some colleagues.

- John

John A. Crow
Minnesota Research Associates
P. O. Box 21504 (651) 454-7313
Eagan, MN 55121


From: Erricos Kontoghiorghes <>
Date: Thu, 06 Aug 1998 10:35:58 +0200
Subject: Special Issue on Parallel Processing and Statistics




Guest Editor: Erricos J. Kontoghiorghes


Submissions are invited for a special issue of Computational
Statistics & Data Analysis on Parallel Processing and Statistics. The
aim of the special issue is to make statisticians aware of recent
developments in parallel processing that can be used efficiently to
tackle problems that previously have been considered unsolvable. At
the same time the special issue will bring to the attention of the
parallel processing community the various computer intensive and
large-scale statistical problems with the ultimate aim of promoting
research in parallelism and statistics.

The special issue will consider papers addressing the design, analysis
and implementation of parallel algorithms for solving statistical
problems. Papers dealing with theoretical and practical issues
associated with parallel statistical algorithms, the impact of
parallelism on statistics and specific applications involving
parallelism and statistics will be considered for the peer review
sections of the journal. Software tutorial and reviews, products from
software organizations and supercomputer vendors, book reviews and
other relevant information on parallelism and statistics will be
considered for publication in the the fourth section of the journal
SSN (Statistical Software Newsletter).

Important Dates

* Submission Deadline: September 1, 1998.
* Acceptance Decision: February 1, 1999.
* Final Version Due: March 1, 1999.
* Publication Date: July, 1999 (tentative)


The focus of the papers submitted to the special issue must contain a
parallel computing and statistical component. Authors who are
uncertain as to the suitability of their papers for the special issue
should contact the special issue editor. All papers submitted must
contain original unpublished work that is not being submitted for
publication elsewhere. Instructions to authors and general
information about the Computational Statistics & Data Analysis, the
official journal of the International Association of Statistical
Computing (IASC), which is published by Elsevier North-Holland can be
found in

Submission electronically is encouraged. Please e-mail a postscript
file of your manuscript together with a plain text cover letter to Authors unable to submit
electronically may send four copies of their manuscript to:

Erricos John Kontoghiorghes (CSDA)
Institut d'informatique
Universite de Neuchatel
Emile-Argand 11
CH-2007 Neuchatel



From: Rolf Jeltsch <>
Date: Fri, 7 Aug 1998 10:05:22 +0200 (MET DST)
Subject: 50th Anniversary of the Seminar for Applied Math ETH Zuerich

50 Years of the Seminar for Applied Mathematics

50 years ago professor E. Stiefel founded the Seminar for Applied
Mathematics,SAM, (it was called Institute f\"ur Angewandte Mathematik
from 1948 to 1969). We would like to celebrate this occasion together with
you by looking a little bit at the history of SAM and with a symposium on
numerics in partial differential equations, the field most researches of
SAM are working today.

The event starts on Wednesday Nov. 18 around 3.30 pm. The lectures end
on Friday Nov. 20 at 5.15 pm. However there will be a dinner in honour of
the recipient of the honory degree of ETH Friday evening

In addition, ETH will give an honorary degree to an applied mathematician
on Saturday, November 21.

There is no registration fee for attending. If you wish to attend
the reception on Wednesday (which is free) please register in order
that we can plan the event. You can obtain the registration form from the
web page:
or by requesting it by e-mail from

Marcela Kraemer

The titel of the talks and the progragm can be found on the same

Speakers will be

A. Speiser, The early years of the institute: Aquisition
and operation of the Z4, planing of the ERMETH
M. Gutknecht, CSCS, ETH Z\"urich:
Ph. G. Ciarlet, Universite Pierre et Marie Curie,Paris
F. Brezzi, Universit\`a di Pavia, Pavia
A. Quarteroni, EPF Lausanne
M. F. Wheeler, TICAM, University of Texas, Austin
D. Braess: Ruhr-University, Bochum
B. Engquist, UCLA, Los Angeles and Royal Institute of
Technology, Stockholm
H. O. Kreiss, UCLA, Los Angeles
G. L. Browning, OAR MASC, Boulder
B. Gustafsson, University of Uppsala
J. Moser, FIM, ETH Z\"urich

We at SAM hope that many people will celebrate with us!

Rolf Jletsch


From: Trini Flores <>
Date: Mon, 10 Aug 98 13:53:45 -0500
Subject: SIAM Conference on Parallel Processing

Ninth SIAM Conference on Parallel Processing for Scientific Computing
March 22-24, 1999
Adam's Mark San Antonio-Riverwalk Hotel
San Antonio, Texas

SEPTEMBER 1, 1998 -- Deadline for submission of 2-page extended
abstract and 75-word abstract for presentation in lecture or poster format.

Submissions should be sent electronically to by
using the LaTeX macros for both extended and short abstracts available

To learn more about the conference, visit


From: Rick Caron <>
Date: Thu, 6 Aug 1998 14:48:32 EDT
Subject: Canadian Operational Research Society

The 41st National Conference of the Canadian Operational
Research Society (CORS) will be held June 7-9, 1999 in
Windsor, Ontario. The theme of the conference is
"Operational Research In Motion" and the confirmed plenary
speakers are Harvey Greenberg, Carl Harris, and Gilbert Laporte.
The deadline for abstract submission is March 1, '99.
Early registrations fees are applicable until May 1, 99.
For more information visit <>
or email Richard Caron <>.


From: Tianruo Yang <>
Date: Mon, 10 Aug 1998 11:33:52 +0200
Subject: Worshop on Scientific and Engineering Computing in India

Special Worshop on Scientific and Engineering Computing with Applications


The above special workshop is organized inside the 6th International
Conference on Advanced Computing, Pune, December 14-16, India.

Scientific and engineering computing has become a key
technology which will play an important part in determining, or at
least shaping, future research and development activities in many
academic and industrial branches.
This workshop is to bring together computer scientists, applied
mathematicians and researchers to present, discuss and exchange
idea, results, work in progress and experience of research in the
area of parallel and distributed computing for numerical and
optimization problems in science and engineering applications.

Among the main topics (are not limited) are:
1. development of advanced parallel and distributed numerical
and optimization methods,
2. parallel and distributed computing techniques for above methods,
3. applications to numerical fluid mechanics, material sciences,
applications to signal and image processing, dynamic systems,
semiconductor technology, and electronic circuits and systems
design etc.

Authors should send three copies of papers to the workshop organizers
via electronic mail or airmail. Contributions will be reviewed for
relevance and technical contents on basis of papers. Selection of
papers will be reviewed for possible publication in the Journal of
Parallel and Distributed Computing Practice.

Further information about the conference proceedings and
registration fee can be found by responding to the e-mail address:, or or check either of the WWW-page:

Deadline Schedule:
Paper submission Due: 10 of September, 1998
Notification of Acceptance: 20 of September, 1998
Final camera-ready paper: 05 of October, 1998

Workshop organizers:
Tianruo Yang
Department of Computer and Information Science
Linkoping University, 581 83 Linkoping, Sweden
Email: Tel: 0046-13-281763 Fax: 0046-13-282666=20


From: David Bailey <dhb@Nersc.GOV>
Date: Thu, 6 Aug 1998 15:53:50 -0700 (PDT)
Subject: Postdoctoral Position at LBL/NERSC

The National Energy Research Scientific Computing Center (NERSC),
which is a division of the Lawrence Berkeley National Laboratory, is
seeking postdoctoral researchers in the area of performance scaling
analysis. This Postdoctoral Fellow position is one-year appointment
with the possibility of renewal and/or advancement. The position is
with the NERSC Scientific Computing Group, which does research in
numerical algorithms and applications in service to the NERSC user
community. Any interested person should contact:

David H. Bailey
NERSC, Lawrence Berkeley National Laboratory
Tel: 1-510-495-2773


Essential --

The duties for this particular position will include studying the
performance scaling behavior of some of the key algorithms and
applications in use by the NERSC research community. The intent here
is to better understand the performance achieved by these applications
on current systems, and to accurately project the performance that
they will achieve on future systems, in particular systems with
thousands of processors. This research will require analyzing source
code, understanding memory hierarchy behavior, measuring communication
performance, developing performance models, and testing performance
models on a variety of available parallel platforms. This position
will also require some travel and presentions at professional


Essential --

1. A recent graduate possessing a Ph.D. degree or equivalent with
strong emphasis on computing or computational science.

2. Demonstrated ability to handle large-scale scientific application
codes. Familiar with numerical methods in solving partial
differential equations.

3. Significant experiences in programming Fortran.

4. Good knowledge about computer architectures and computations in

5. Effective communication and personal interaction skills -- must be
able to collaborate effectively with other scientists.

Marginal --

1. Working knowledge of C and C++, UNIX, scientific libraries and
tools such as Matlab, graphics packages, and related software.

2. Prior experience with highly parallel computing systems is


From: Georgios Makrakis <>
Date: Fri, 07 Aug 1998 11:11:23 -0700
Subject: Postdoctoral Positions at FORTH, Crete

Institute of Applied and Computational Mathematics
Foundation for Research and Technology - Hellas
Herakleion, Crete


The Institute of Applied and Computational Mathematics (IACM) expects to
offer 1-2 post-doctoral and short term visiting positions (two-three months)
in the field of Viscosity Solutions and related topics for the academic year
1998-1999. These positions are available for one year or a shorter period up
to a maximum of two years and they are financed by the TMR Research Network
on "Viscosity Solutions and their Applications"
( Additional funds
exist to finance short term visits of the post-docs to other universities
participating in the TMR.

The Institute of Applied and Computational Mathematics which is one of the
Institutes of Foundation for Research and Technology - Hellas has strong links
with the University of Crete, and it is oriented towards applied as well
as theoretical research.

Young scientists with interests in the area of nonlinear partial differential
equations, numerical analysis and/or scientific computation are encouraged to
apply. Priority will be given to those with research interests in:

1) Theoretical and numerical aspects of the theory of viscosity solutions
2) Theoretical and numerical aspects of hyperbolic equations
3) Phase transitions and front propagation
4) Multiphase geometrical optics and high frequency asymptotics

Candidates must be nationals of a European Union Member State other than
Greece or a State associated with the TMR Programme (Iceland, Liechtenstein,
Norway, Israel) and must not be over 35 years of age (an allowance to this age
limit may be made for military service and child-care).
People who are interested should submit their applications
(Curriculum Vitae, a list of published works, research interests and
two letters of recommendation). It is expected to make decisions by
September 10, 1998. Late applications may be considered.

Correspondence should be sent to:

TMR Postdoc Programme - Viscosity Solutions
Attn: George Kossioris
Foundation for Research and Technology - Hellas
Institute of Applied and Computational Mathematics
P.O.Box 1527, 71110 Heraklion, Crete, GREECE
fax: +30(81) 391801


From: Jun Zhang <>
Date: Fri, 7 Aug 1998 10:36:48 -0400 (EDT)
Subject: Postdoctoral Position at the University of Kentucky

Postdoctoral Research Associate Position

A postdoctoral research associate position is available in the
Department of Computer Science of the University of Kentucky to
work with Professor Jun Zhang on applied scientific and parallel
computing. This position is funded by the Center for Computational
Sciences and the Department of Computer Science of the University
of Kentucky. The postdoctoral research associate may start sometime
after August 16, 1998.

Research topics will generally be in applied iterative methods,
those include but are not limited to multi-level or multigrid
methods, robust preconditioning techniques, numerical solution of
partial differential equations (e.g., convection-diffusion equation),
computational fluid dynamics. (Check the URL address at Programming skill (in Fortran or C)
is essential and experience with parallel computers is desirable.

Interested person should e-mail a curriculum vitae (with full
publication list and e-mail addresses of three referees) in
postscript or ASCII to Jun Zhang at or fax it
to (606)323-1971. The last means is to send a postal mail to:

Professor Jun Zhang
Department of Computer Science
University of Kentucky
773 Anderson Hall
Lexington, KY 40506--0046

Please do NOT have the reference letters sent, they will be requested
if needed.


From: Luc Giraud <>
Date: Thu, 6 Aug 1998 14:40:49 +0200 (MET DST)
Subject: Research Position at CERFACS, France

A research position on domain decomposition techniques is open
in the Parallel Algorithms Project at CERFACS. The main objective is
to develop preconditioning techniques for the solution of 2D and 3D
transport-diffusion equations and study their implementation on parallel
distributed plateforms. Those techniques will be applied in particular
in the framework of device modelling simulations.
In this context we are looking for a PhD student with motivation,
interest, and background in parallel scientific computing.
The position starts in September 1998 and has to be filled
as soon as possible.

The position is in the Parallel Algorithms Project led by I.S. Duff,
where the main research activities are :
- Large sparse matrix calculations:
- direct linear solvers
- iterative linear solvers, preconditioning
- eigensolvers
- Domain decomposition
- Large scale nonlinear systems and optimization
- Reliability of numerical software

For detailed information about CERFACS
URL address :
For more information about Toulouse, the city where CERFACS is located
URL address :

Technical questions should be addressed to Luc Giraud
( Applications should be sent to Mrs Campassens
according to the CERFACS recruitment procedure described at the
URL address :


From: Henry Gabb <>
Date: Thu, 06 Aug 1998 11:09:58 -0500
Subject: Position at Nichols Research Corp., Vicksburg

Science/engineering programming position in high performance computing

Nichols Research Corporation has positions in the area of high performance
computing (HPC) within the Computational Migration Group (CMG). Members of
the CMG collaborate with U.S. Department of Defense researchers to migrate
computational projects from Cray Research vector platforms to modern
scalable HPC systems.

This work will be conducted at the U.S. Army Corps of Engineers Waterways
Experiment Station, High Performance Computing, Major Shared Resource
Center in Vicksburg, Mississippi. This site is managed by Nichols Research
Corporation for the U.S. Department of Defense. The CEWES MSRC ranks among
the top ten most powerful high performance computing centers in the world.
We currently operate a CRAY C90 (16 CPUs), a CRAY T3E (336 CPUs), a SGI
Origin 2000 (128 CPUs), and two IBM SP supercomputers (256 and 126 CPUs).

Minimum requirements:

Fortran programming experience in a UNIX environment
B.S. in science, engineering, or related discipline
Exposure to parallel computation
U.S. citizenship

Desired experience:

C or C++ programming
Parallel programming (i.e., MPI, PVM, HPF, OpenMP, Pthreads)
Vector programming on Cray Research platforms

Interested parties may contact:

Henry A. Gabb
Computational Migration Group
1165 Porters Chapel Road
Vicksburg, MS 39180
601-634-2507 tel
601-634-3808 fax


From: Vladik Kreinovich <>
Date: Wed, 5 Aug 98 18:14:46 MDT
Subject: Contents, Reliable Computing

Reliable Computing, 1998, Vol. 4, No. 4, table of contents

Mathematical Research

Wrapping Effect and Wrapping Function
Roumen Anguelov, Svetoslav Markov

Reliable Collision Detection for Time-Dependent Parametric Surfaces
Anthony Leclerc, Jeff Ely

An NP-Hardness Result for Nonlinear Systems
Christian Jansson

Optimal Enclosure of Quadratic Interval Functions
Misha Koshelev, Luc Longpr\'e, Patrick Taillibert

Verified Integration of ODEs and Flows Using Differential Algebraic
Methods on High-Order Taylor Models
Martin Berz Kyoko Makino


A Note on Epsilon-Inflation
Siegfried M. Rump

Short Communications

Intersection Subdistributivity and Interval Hull Superdistributivity
with Respect to the Interval Maps
Gregory G. Menshikov

Solving Linear Interval Systems Is NP-Hard Even If We Exclude Overflow
and Underflow
Gerhard Heindl, Vladik Kreinovich, Anatoly V. Lakeyev

Why Intervals? Because If We Allow Other Sets, Tractable Problems Become
Monica Nogueira, Amarendra Nandigam

Letters to the editor

Hyper-Spectral Satellite Images: Interval Methods May Be Helpful
Erzs\'ebet Mer\'enyi, Scott Starks, Karen Villaverde

Hilbert Problems (Almost) 100 Years Later (From the Viewpoint of
Interval Computations)
Olga M. Kosheleva


Review of the book V.Kreinovich, A.Lakeyev, J.Rohn, P.Kahl.
Computational Complexity and Feasibility of Data Procesing
and Interval Computations. Kluwer Academic Publishers,
Dordrecht etc., 1998


1998 SIAM Annual Meeting: Interval Highlights

ICRA99 International Conference on Rational Approximation

The contents of all the issues is placed on the Journal part of the
interval computations website


End of NA Digest