NA Digest Sunday, May 30, 1999 Volume 99 : Issue 22

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: Rajat Todi <>
Date: Mon, 24 May 1999 12:38:06 -0500 (CDT)
Subject: Request for Parallel Codes


I am a doctorate candidate at Iowa State University. I am
looking to evaluate the difference between performance of
shared and distributed memory parallel computers. I am in
need of some benchmark programs or scientific code written
in both MPI/PVM as well as in Pthread. Any help in this
regard would be appreciated.


Rajat Todi
Scalable Computing Laboratory,
Ames Laboratory,
237 Wilhelm, ISU, Ames, IA 50011-3020
Ph: (515)-294-1918
Fax: (515)-294-4491


From: Marek Anthony Wlasak <>
Date: Tue, 25 May 1999 13:40:00 +0100 (BST)
Subject: Direct Solver for a Biharmonic Equation on a Sphere

I am looking for a direct solver which can solve a generalised biharmonic
equation on a spherical shell of the form
__4 __2 __

a,b,c are scalar functions of latitude and longitude,
d is a vector function dependent on longitude and latitude and
psi is the variable to be found.

Boundary conditions are considered to be periodic both in psi and in its
own first derivative.

Any help would be much appreciated, including information as to the
uniqueness/existence of solutions to this problem.

Marek Wlasak


From: Nemanja Stanarevic <>
Date: Sat, 29 May 1999 20:10:04 -0400 (EDT)
Subject: Seeking Programs in C for ODEs

Dear Colleagues,
I am working on the research at Physics Department of the Kean University.
My task is to translate existing fortran 77 calculation program into C.
The problem I have is obtaining a good library of C functions for solving
systems of ordinary differential equations. If you have any information
regarding my problem, please contact me at


Nemanja Stanarevic


From: Mike Epton 425-865-3512 <>
Date: Fri, 28 May 1999 13:15:40 -0700 (PDT)
Subject: Question About a Condition Number Bound

In an effort to understand the roots of ill-conditioning in discretized
integral equations, I have developed a bound on the condition number
of a matrix built up by adjoining two smaller matrices. Specifically
let the matrix C be defined by C = [A|B], let largest and smallest
singular values of A and B be denoted s(A,max), s(A,min), s(B,max),
s(B,min) and let 'th' denote the angle between the subspaces spanned
by A and B:

th = angle[ S_A, S_B ]; S_A = range(A), S_B = range(B)

where the angle defined by the relation:

angle[ S_A, S_B ] = min min angle(a,b)
a in S_A b in S_B

can be computed from: arccos( s_max[ P' Q ] ) -- where P and Q form
orthogonal bases for S_A and S_B respectively. The bound I've obtained
takes the form:
max{ s(A,max), s(B,max) }
cond[C] <= -------------------------------------
min{ s(A,min), s(B,min) } * sin(th/2)

My question is this: Is this bound "well" known, and could someone
provide a reference for where it appears?

Michael A. Epton
Computational Physics Group voice: 425-865-3512 / fax: 425-865-2966

U.S. Mail: Overnight Service:
The Boeing Company Mathematics and Computing Technologies
P.O. Box 3707, M/S 7L-21 33-07 Bldg, 2760 160th S.E.
Seattle WA 98124-2207 Bellevue WA 98008


From: Ramayya Krishnan <>
Date: Sun, 30 May 1999 14:57:14 -0400
Subject: INFORMS Computing Society Prize

We are pleased to announce the winners of the 1999 INFORMS Computing
Society Prize. Yair Censor and Stavros A. Zenios were awarded "The 1999
INFORMS Computing Society Prize for Research Excellence In The Interface
Between Operations Research And Computer Science" for their book
"Parallel Optimization: Theory, Algorithms, and Applications", Oxford
University Press, New York, 1997. The prize was awarded at the National
Spring Meeting of INFORMS (Institute for Operations Research and the
Management Sciences), held May 2-5, 1999, in Cincinnati, Ohio, USA.

The book is a volume in the series: "Numerical Mathematics and
Scientific Computation", Hardcover, 576 pages, ISBN 0-19-510062-X. For
information on the book please visit the publisher on the internet at or contact the authors.

For further information about the INFORMS, the INFORMS Computing Society
(ICS), their missions and activities as well as details of how to join
the ICS and more information about the prize please visit the INFORMS
site on the internet at

Ramayya Krishnan
Professor of Management Science and Information Systems
The Heinz School
Carnegie Mellon University
Pittsburgh, PA 15213
fax: 412-268-7036
phone: 412-268-2174


From: SIAM <>
Date: Thu, 27 May 99 14:40:14 -0500
Subject: Electronic Submissions to SIAM Journal of Optimization

SIAM Journal on Optimization Encourages You to Submit Electronically

The Editors of SIAM Journal on Optimization (SIOPT) remind you that
the journal accepts submissions in PostScript form.

Electronic submission is the preferred method of submission for SIOPT.
Authors immediately benefit from submitting papers electronically --
electronic submissions are processed more quickly and efficiently than
paper submissions.

Complete information about and instructions for electronic submissions
to SIOPT can be found at or by request

Traditional paper submissions will continue to be accepted, but are
intended primarily for those authors whose work is not conveniently
available in standard PostScript format.

SIOPT contains research articles on the theory and practice of
optimization. The areas addressed include linear and quadratic
programming, convex programming, nonlinear programming,
complementarity problems, stochastic optimization, combinatorial
optimization, integer programming, and convex, nonsmooth and
variational analysis. Contributions may emphasize optimization theory,
algorithms, software, computational practice, applications, or the
links between these subjects.

Michael L. Overton, Courant Institute of Mathematical Sciences,

Major Mathematical Abstracting and Indexing Sources: ACM Computing
Reviews, Current Contents, INSPEC, Math Reviews/Math SciNet, Science
Citation Index, Zentralblatt fur Mathematik.

Impact Factor: 1.202

SIOPT is published electronically well in advance of the print
edition; see


From: Pat Worley <worley@haven.EPM.ORNL.GOV>
Date: Fri, 28 May 1999 11:57:45 -0400 (EDT)
Subject: MPICL MPI Instrumentation Library Now Available

MPICL MPI Instrumentation Library Now Available

MPICL is a subroutine library for collecting information on communication and
user-defined events in message-passing parallel programs written in C or
FORTRAN. For MPI programs it uses the MPI profiling interface to
automatically intercept calls to MPI communication routines, eliminating the
need to add more than a few statements to the source code in order to collect
the information.

MPICL is typically used in one of two ways. It can be used to collect profile
data, summarizing the number of occurrences, the associated volume
statistics, and the time spent in communication and user-defined events for
each processor. It can also be used to collect detailed traces of each
event, which can then be viewed using the ParaGraph visualization tool,
developed by Michael Heath and Jennifer Finger and available from the website .

MPICL also represents an update to the Portable Instrumented Communication
Library (PICL), a software package that provided a portable message-passing
interface in the days before the MPI standard. While an MPI user need know
nothing about PICL message-passing, this does mean that MPICL can be used to
collect performance data for non-MPI programs.

MPICL source code and documentation are available from the website

Note that MPICL is research software, with no guarantees, that is being made
available at this time to coordinate with the release of the new MPI-aware
version of ParaGraph. However, the author has been using MPICL and PICL
in a variety of performance evaluation studies since its initial
incarnation in 1989 and expects to continue developing and maintaining the


Contact Pat Worley at or (423) 574-3128.


From: Michael Heath <>
Date: Fri, 28 May 1999 11:09:25 -0500 (CDT)
Subject: New MPI Version of ParaGraph

New MPI Version of ParaGraph Parallel Performance Visualization Tool

ParaGraph is a graphical display tool for visualizing the behavior and
performance of parallel programs that use message passing. The visual
animation of the parallel program is based on execution trace data
gathered during an actual run of the program on a message-passing
parallel computer system. The resulting trace data are replayed
pictorially to provide a dynamic depiction of the detailed behavior of
the parallel program, as well as graphical summaries of its overall
performance. The same performance data can be viewed from many
different visual perspectives to gain insights that might be missed by
any single view.

A new release of ParaGraph is now available that incorporates support
for the full functionality of MPI (Message-Passing Interface)
prescribed in the MPI Standard, including process groups and collective
communication among them; new features added in MPI-2 are not yet
supported, however. A number of other general improvements and new
displays are also included in this new version of ParaGraph, which
remains backward compatible with the older PICL message-passing model
and the tracefiles it generates.

The necessary execution trace data for the new version of ParaGraph are
produced by the MPICL tracing package, developed by Pat Worley of Oak
Ridge National Laboratory, which uses the profiling interface of MPI to
provide timestamped records of MPI events. MPICL is available from the

Source code, documentation, and sample tracefiles for the new version
of ParaGraph are available from the following website:

For more information, contact Michael Heath,


From: Marcin Paprzycki <>
Date: Fri, 28 May 1999 00:07:33 -0500 (CDT)
Subject: Special Issue on Industrial Applications of Parallel Computing


Special Issue of the "International Journal of Computer Research"
( on


Parallel scientific and engineering computing is becoming of paramount
importance in several industrial applications, especially when the
solution of large and complex problems must cope with harder and harder
time scheduling.

In this special issue, we would like to report on relevant research
representing the state-of-the-art in the following areas of parallel

1. parallel and distributed combinatorial and global optimization methods,
as applied to industrial/practical problems

2. parallel and distributed computing techniques and software systems, as
applied to industrial/practical problems

The application areas are to be understood very broadly and include, but
are not limited to: computational fluid mechanics, structural engineering,
computational chemistry, electronic and electromagnetic circuits, signal
and image processing, etc.

Electronic submission to one of the editors of the special issue (please
contact the editor to discuss the technical details of submission) is
preferred. In case of hard-copy submission, please contact an editor for
more details.

We also would like to indicate that in case when an extremely large number
of high quality submissions will be obtained, a publication of a volume in
the "Advances in Computation. Theory and Practice" book-series will be

Important deadlines:
Paper submission September 15th, 1999
Final papers due December 31st, 1999

Special Issue editors:

Marcin Paprzycki
University of Southern Mississippi

Luciano Tarricone
University of Perrugia

Tianruo Yang
Delft University of Technology


From: Mike Powell <>
Date: Wed, 26 May 99 17:16:54 BST
Subject: Conference on System Modelling and Optimization

19th IFIP TC7 Conference on
System Modelling and Optimization
Cambridge, England
July 12-16, 1999

Readers of NANET will be very welcome at the TC7 Conference in Cambridge,
England in July. You can obtain a registration form and further information
from the world wide web ( or by sending
an e-mail to These biennial meetings cover several
fields. Special attention will be given to optimization algorithms this
time, by providing papers on this subject throughout the parallel sessions
for submitted talks, and by including the following plenary lectures:

R.E. BIXBY MIP: Closing the Gap between Theory and Practice,
A.L. DONTCHEV Lipschitzian Stability of Newton's Method,
N.I.M. GOULD SQP Methods for Large-Scale Nonlinear Programming,
H.H. WOLKOWICZ Semidefinite Relaxations for Hard Combinatorial Problems,
S.J. WRIGHT Recent Developments in Interior-Point Methods.

The titles of most of the submitted contributions are on the web.

Mike Powell.


From: Zohar Yosibash <>
Date: Thu, 27 May 1999 11:26:18 +0000
Subject: Conference on Finite Element Methods

A conference in honor of the 65th birthday of PROF. BARNA SZABO
May 31 - June 2, 2000
Washington University, St. Louis, USA

The international conference, ``p and hp Finite Element Methods:
Mathematics and Engineering Practice,'' is dedicated to Professor Barna A.
Szabo on the occasion of his sixty-fifth birthday.
Being one of the founders of the p-version of the FEM, Professor Szabo
has been working on and advocating these methods for the past 30 years.
He has played a key role in emphasizing the need for estimating the
reliability of computed data, and in the effective
implementation of p and hp FEM in industrial computer codes
used in daily engineering practice.
The main objective of the conference is to bring together researchers
with interests in the mathematical, engineering practice and computational
aspects of the p and hp (and high-order/spectral) methods for the
solution of partial differential equations (mainly associated with
solid and fluid mechanics).
It is anticipated that the conference will provide a forum for
interaction between a broad group of
scientists and engineers with diverse interests in the mathematical and
engineering application of p and hp methods.

A-priori and a-posteriori error estimation.
Efficient solution algorithms for system of equations
generated by p and hp methods.
Application of methods to nonlinear problems.
Practical application of high-order methods in engineering practice.
Connection of high-order methods to dimension reduction and
plate/shell models.
High-order methods and their coupling with CAD systems.
High-order time-space methods.
Treatment of singularities by high-order methods.
High-order methods applied to bio-mechanical systems.
p and hp-BEM.

M. Suri, Univ. of Maryland - Baltimore County, USA
Z. Yosibash, Ben-Gurion Univ. of the Negev, Israel

Borje Andersson, Aeronautical Research Institute of Sweden, Sweden
Ivo Babuska, Univ. of Texas at Austin, USA
Ted Belytschko, Northwestern Univ., USA
George Karniadakis, Brown Univ., USA
Saeed Paydarfar, Boeing Defense and Space Group, USA
Christoph Schwab, ETH Zurich, Switzerland
J.Tinsley Oden, Univ. of Texas at Austin, USA
Juhani Pitkaranta, Helsinki Univ. of Technology, Finland
Michael Wheeler, Parametric Technology Corp., USA

The web site of the conference contains up-to-date information including
electronic forms for submission of mini-symposium proposals and individual

Zohar Yosibash
Mechanical Engineering Dept., Ben-Gurion University of the Negev,
P O Box 653, Beer-Sheva 84105, ISRAEL
Tel: + 972 - 7 - 6477103


From: Mike Pettipher <>
Date: Fri, 28 May 1999 13:54:41 +0100
Subject: Summer School in Distributed Memory Programming and Visualization

Summer School for HPC in Distributed Memory Programming and Scientific
6th - 17th September 1999, Manchester, UK

This is the first annual two-week residential event being organised by the
Manchester Research Centre for Computational Science. It will be a unique
opportunity for you to develop skills so that you can make effective use of
distributed programming environments, as typified by the Cray T3E, and to
learn how to visualise the results of applications run on such supercomputers.
You will get to use the most powerful computer available to academia in the
world, meet the international experts who will teach you and develop contacts
with other participants.

The list of topics and speakers from Europe and the USA is:

Professor Jack Dongarra, University of Tennessee -
Systems & Architectures for Distributed Memory Programming

Professor Michael Resch, RUS Stuttgart -
Distributed Memory Programming and Tools for Profiling & Debugging

Dr Lawrence Mulholland, Nag, UK -
Parallel Numerical Libraries for Distributed Memory Systems

Dr Len Freeman, CNC, University of Manchester -
Parallel Algorithms

Professor Frits Post, TU Delft -
Visualization for Multidimensional & Scientific Data Visualization of
CFD data (an extended case study)

Mr Terry Hewitt, Manchester Visualization Centre -
Systems & architectures for Visualization Computational Steering

Further Information and Application Form

For more general information on MRCCS and CSAR: or

If you have any problems finding information or require further
assistance, please contact the CSAR helpdesk:

email or telephone 0161 275 5997/6824.


From: Richard Allen <>
Date: Fri, 28 May 1999 09:54:51 -0600
Subject: Conference on Graduate Education in the Computational Sciences

The Conference on Graduate Education in the Computational Sciences (CGECS)
will be held in Albuquerque, NM, August 5-6, 1999. The event is sponsored by
the NSF National Computational Science Alliance, The Partnership for Advanced
Computing Infrastructure Education Outreach and Training (PACI-EOT), and The
Albuquerque High Performance Computing Center at the University of New Mexico.

The Conference will provide a forum for faculty, students and representatives
from industry to meet, discuss and recommend graduate education programs in the
computational sciences. A comprehensive report on existing computational
science programs at US colleges and universities will be presented. Papers are
solicited on new and innovative ideas in graduate computational science
education and training and in the broader area of Informatics.

Please check the following web site for more information about the Conference,
registration details, abstracts, and papers:

Questions to be addressed in the conference include:

Are we addressing the right issues in our various graduate programs? Are we
meeting our customer needs? Do our students have the expertise that they need
to enter the work force or academia?

What is the effectiveness of the various approaches to computational science
education: separate degree programs, certificate programs, integrated single
discipline programs, etc.? Are practicums useful adjuncts to academic programs?

Should academic course work be more integrated, not compartmentalized?
Should our Programs emphasize "teaming" to solve problems?

How should we balance breadth vs depth in our academic training in the
computational sciences?

Dick Allen


From: Kelly Black <>
Date: Mon, 24 May 1999 07:54:26 -0400 (EDT)
Subject: Faculty Position at University of New Hampshire

University of New Hampshire
Department of Mathematics
Durham, NH 03824
Non Tenure-Track Faculty Position

Position Description:

Faculty In Residence
The Department of Mathematics of the University of New Hampshire
invites applications for a temporary position (1-3 years).
Preference will be given to candidates whose research areas include
control theory, dynamical systems, or numerical analysis.

Candidates are expected to have a Ph.D. in Mathematics by August
1999 and demonstrated research ability. Strong commitment to
teaching is expected. The teaching load is expected to be two large
lecture courses per semester. The position will begin in the fall
of semester 1999. Review of applications will begin on June 20, 1999,
but applications will be considered until the position is filled.

Please submit resume, e-mail address if possible, and three letters of
recommendation to:
Kenneth Appel
Chair, Department of Mathematics
University of New Hampshire
Durham, NH 03824.

UNH is committed to excellence through diversity of its faculty
and strongly encourages women and minorities to apply.


From: Penny Davies <>
Date: 26 May 1999 15:46:09 +0100
Subject: Research Position at University of Strathclyde

Department of Mathematics
University of Strathclyde
Glasgow, Socotland

Applications are invited for a one-year EPSRC funded post to work on
the project entitled "Computing the current induced on a thin wire by a
transient electromagnetic field" (ref: R26/99). The aim of the project
is to develop and analyse efficient and accurate numerical methods for
computing the current induced on a thin wire by an electromagnetic
pulse by solving systems of integral and differential equations.

Applicants should have (or be close to completing) a PhD in numerical
analysis of differential or integral equations, or have other relevant
experience. They should ideally be competent programmers in Matlab
and/or Fortran.

Further details may be obtained from
and informal enquiries can be made to Dr Penny Davies

Salary in range 15,735 - 20,107 pounds (Grade 1A scale), depending on
qualifications, age and experience.

Starting date: 1 October 1999 or by agreement.

For application form (available on request in alternative format for
applicants with a disability) contact the Personnel Office, University
of Strathclyde, Glasgow, UK, G1 1QX. Tel +44 (0)141 553 4133 (24 hour
voicemail service). Overseas applicants are encouraged to contact the
Investigator directly by email.

Closing date for applications: 15 June 1999.


From: Ekkehard Sachs <>
Date: Thu, 27 May 1999 13:43:45 +0200
Subject: Graduate and Postdoctoral Positions at University of Trier

Graduate and Postdoctoral positions
in the
Graduate Program
Mathematical Optimization
University of Trier

Scholarships and postdoc positions in
Mathematical Optimization will become available on

September 15, 1999 or earlier

The main research areas in the graduate program are

- Global Optimization, Multiple Objective Optimization
(Prof. Dr. Reiner Horst)

- Computational Complexity

- VLSI-Design and Formal Verification
(both Prof. Dr. Christoph Meinel)

- Numerical Methods in Optimization and Optimal Control
(Prof. Dr. Ekkehard Sachs)

- Ill-posed Variational Problems
(Variational Inequalities, Semi-infinite Optimization)
(Prof. Dr. Rainer Tichatschke).

The scholarships are DM 1.800 per month (or higher for
qualified applicants) according to the
regulations of the German Research Foundation (DFG).

The salary for postdocs is DM 2.790 per month.

The letter of application should include

- curriculum vitae, copy of degree certificates,
- short description of prospective research program,
- copies of scientific papers (thesis, etc.),
- letter of recommendation.

The applications should be sent until July 31, 1999 to

Prof. Dr. E. Sachs
Sprecher des Graduiertenkollegs
Mathematische Optimierung
Universitaet Trier
D-54286 Trier

For further information see

or contact the office:

Telephone +49 651 201-3477
Telefax +49 651 201-3973


From: Phyllis Stokes <>
Date: Fri, 28 May 1999 08:58:04 -0700
Subject: Position at Silvaco International

Silvaco International is a market leader in TCAD/ECAD software
for the microelectronics industry. Rapid expansion has created
an exciting opportunity with the High Performance Computing Division:


location: Santa Clara, California.


Design and implement numerical approximation methods for 3D PDEs.
Use finite elements method to solve elliptic and parabolic PDEs.
Particular interest is on Laplace, Poisson and continuity equations.

* Ph.D in mathematics or physics + 3 years experience.
MSc Maths/Physics + 7 years experience considered.

* Experience with:

- 3D finite elements method
- non linear solvers
- Domain decomposition methods
- Multilevel solution methods
- Iterative linear solvers

* Strong C++ programming background on Unix platforms.

To apply for this position, please submit your resume by mail, fax,
or email (PostScript or plain text) to:

SILVACO International
Attn. HPC Manager
4701 Patrick Henry Drive bldg#2
Santa Clara, CA 95054
Fax: (408) 567-1009


From: Barb Taub <>
Date: Fri, 28 May 1999 20:01:29 -0500
Subject: Position at Wolfram Research

Numerical Optimization Developer (Wolfram Research).

Wolfram Research, makers of Mathematica, is seeking an outstanding
candidate to add to its research and development team in the area of
numerical optimization. A successful applicant should have a PhD in
numerical analysis, operations research or other area closely related to
optimization. In addition practical implementation experience in C/C++ and
well rounded computational skills are needed. Mathematica skills are a
strong plus.

You can contact us:
* by filling out a web employment application at
* by sending email to
* by sending a fax to 217-398-0747
* by sending regular mail to Human Resources Department, Wolfram Research,
100 Trade Center Drive, Champaign, IL 61820-7237, USA

Barb Taub
Director of Human Resources
Wolfram Research


From: Vladik <>
Date: Fri, 28 May 1999 09:14:38 -0600 (MDT)
Subject: Contents, Reliable Computing

Reliable Computing
volume 5, issue 3
Special issue, Proceedings of the SCAN-98
Guest editor: Tibor Csendes


Comparing Partial Consistencies
Helene Collavizza, Francois Delobel, Michel Rueher

Verified Computation of Fast Decreasing Polynomials
Neli S. Dimitrova, Svetoslav M. Markov

An Accurate Distance-Calculation Algorithm for Convex Polyhedra
Eva Dyllong, Wolfram Luther, Werner Otten

Verified Error Bounds for Linear Systems Through the Lanczos Process
Andreas Frommer, Andre Weinberg

A Representation of the Interval Hull of a Tolerance Polyhedron
Describing Inclusions of Function Values and Slopes
Gerhard Heindl

A Few Results on Table-Based Methods
Jean-Michel Muller

An Interval Hermite-Obreschkoff Method for Computing Rigorous Bounds on
the Solution of an Initial Value Problem for an Ordinary Differential
Nedialko S. Nedialkov
Kenneth R. Jackson

The Interval-Enhanced GNU Fortran Compiler
Michael J. Schulte, Vitaly Zelov, Ahmet Akkas, James Craig Burley

Outer Estimation of Generalized Solution Sets to Interval Linear Systems
Sergey P. Shary

A Real Polynomial Decision Algorithm Using
Arbitrary-Precision Floating Point Arithmetic
Adam Strzebonski

A Numerical Verification Method of Solutions
for the Navier-Stokes Equations
Yoshitaka Watanabe, Nobito Yamamoto
Mitsuhiro T. Nakao


End of NA Digest