NA Digest Sunday, October 6, 1996 Volume 96 : Issue 36

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: NA Digest <>
Date: Sun Oct 6 13:10:13 EDT 1996
Subject: NA Digest Calendar

The Netlib Conferences Database is on the Web at:

NA Digest Calendar
Date Topic Place NA Digest #

Oct. 1- 4 European Multigrid Conference Stuttgart, Germany 20
Oct. 3- 4 Parallel Computing Minneapolis, MN 29
Oct. 7- 9 ICASE/LaRC Industry Roundtable Williamsburg, VA 27
Oct. 9-11 Workshop on Scientific Computing Braunschweig, Germany 24
Oct. 9-11 SIAM Conference on Sparse Matrices Coeur d'Alene, ID 16
Oct. 10-11 International Meshing Roundtable Pittsburgh, PA 32
Oct. 11-12 Modeling in Biochemical Engineering Minneapolis, MN 32
Oct. 11-12 Department Chairs Colloquium Washington, DC 32
Oct. 20-23 High Performance Computing Tempe, AZ 01
Oct. 21-22 Computational Science and Engineering Purdue, IN 24
Oct. 31 ParkBench Knoxville, TN 38
Oct. 21-25 Evolutionary Algorithms Minneapolis, MN 29
Oct. 24-26 Materials Studies Workshop University Park, PA 27
Oct. 31... Innovative Time Integrators Amsterdam, Netherlands 40

Nov. 7- 8 BLAS Technical Forum Eagan, MN 35
Nov. 11-12 Computational Mechanics Codes London, England 35
Nov. 18-21 Overset Grids Symposium Los Alamos, NM 27

Dec. 11-13 Carleman Estimate and Inverse Problems Kyoto, Japan 30
Dec. 17-19 Mathematics in Signal Processing Warwick, England 48


Jan. 5- 7 Discrete Algorithms New Orleans, LA 15
Jan. 5-12 Computational Mathematics Rio de Janeiro, Brazil 08
Jan. 5-12 Numerical Linear Algebra Rio de Janeiro, Brazil 13
Jan. 8-10 Numerical and Mathematical Elasticity Kyoto, Japan 30
Jan. 15-18 Numerical Linear Algebra, Optimization Parana, Brazil 34
Jan. 24-26 Multi-Scale Problems Kiel, Germany 38
Jan. 27-31 Maths-in-Industy Study Group Melbourne, Australia 11

Feb. 24-28 Optimization and Optimal Control Lambrecht, Germany 37

Mar. 10-12 Scientific Computing Hong Kong 25
Mar. 12-14 Algorithms and Complexity Rome, Italy 24
Mar. 14-17 SIAM Parallel Processing Minneapolis, MN 32
Mar. 16-21 Approximation and Optimization Caracas, Venezuela 21
Mar. 20-22 Multiwavelets Huntsville, TX 37
Mar. 21-22 AMS Session on Approximation Theory Memphis, TN 11

Apr. 1- 3 Monte Carlo Methods Brussels, Belgium 16
Apr. 9-13 Copper Mt. Multigrid Copper Mountain, CO 33
Apr. 14-18 Computational Issues in Drug Design Minneapolis, MN 32
Apr. 17-18 Meeting Honoring Bill Morton Oxford, England 26

May 12-14 Materials Science Philadelphia, PA 32
May 19-21 Applications of Dynamical Systems Snowbird, UT 27
May 21-24 Macromolecular Modelling Berlin, Germany 31
May 26-30 Computational Heat Transfer Cesme, Turkey 05
May 27-28 Computational Science and Engineering Hefei, China 38

June 1- 5 Computer Science Education Uppsala, Sweden 38
June 16-18 Computer Methods in Water Resources Byblos, Lebanon 35
June 16-18 Mathematical Issues in Geosciences Albuquerque, NM 18
June 16-21 Iterative Methods Milovy, Czech Rep. 37
June 18-21 Principles + Practice of Parallel Prog. Las Vegas, NV 27
June 24-27 Dundee NA Conference Dundee, Scotland 13

July 3- 4 CFD in Minerals, Metal & Power Melbourne, Australia 33
July 4- 5 Honor Lothar Collatz Hamburg, Germany 32
July 9-11 Computational Fluid Dynamics Twente, Netherlands 38
July 9-12 Iterative Methods Laramie, WY 36
July 13-18 SIAM Annual Meeting Stanford, CA 36
July 14-18 Theoretical and Computational Acoustics New York, NY 14

Aug. 10-14 Domain Decomposition Boulder, CO 35
Aug. 18... Radial Basis Functions Asilomar, CA 32
Aug. 24-29 IMACS World Congress Berlin, Germany 07
Aug. 24-29 Fast Algorithms Berlin, Germany 37

Sep. 1- 5 Numerical Solution of ODEs Halle, Germany 13
Sep. 10-12 Computer Arithmetic Lyon, France 37
Sep. 15-18 Boundary Integral Methods Manchester, England 27
Sep. 24-26 Dutch Numerical Mathematicians Zeist, Netherlands 38
Sep. 15-19 Scientific Computing & Diff. Eqns. Grado, Italy 26
Sep. 29-.. ENUMATH-97 Heidelberg, Germany 50

Oct. 13-16 Computational Methods, Function Theory Nicosia, Cyprus 34


From: Gene Golub <golub@sccm.Stanford.EDU>
Date: Sat, 5 Oct 96 17:03:58 -0700
Subject: Seymour Cray

The following article was posted Saturday on the HPCwire, an electronic
news service specializing in high performance computing.

-- Gene


Colorado Springs, Colo. -- Seymour Cray, 70, died Saturday at 2:53 AM in
Penrose Hospital from injuries he sustained Sept. 23 as a result of a
multiple-car accident. A hospital spokesperson listed complications from
massive head injuries as the cause of death. Cray founded supercomputer-maker
Cray Research Inc in 1972.

A native of Chippewa Falls, Wis., Cray earned his bachelor's degree in
electrical engineering from the University of Minnesota in 1950. His
professional career began with UNIVAC I. In 1957, he was one of the founders
of Control Data Corp.

by Norris Parker Smith, editor at large

October 5, 1996 -- Seymour Cray is supercomputing.

The present tense is appropriate, even though Seymour Cray died today. His
identification with the concept and industry that he shaped will endure.

His was a life of wonderful simplicity. Seymour Cray was dedicated to
clearly-formed, strongly-held opinions about the best way to make computers
that were far more powerful than any that had existed before.

The world is full of technologists with very good ideas. A much smaller
number have outstanding ideas -- and Seymour Cray certainly belongs in this

Seymour Cray was exceptional, however, because he devoted the same
exceptional creative intelligence to the practical tasks of turning his ideas
into machines that worked.

Even more remarkable, he founded and guided a company that built and sold
those machines, transforming supercomputing from an esoteric specialty into
a significant sector within computing.

Inventors and entrepreneurs of this versatility -- and impact on the world
-- are rare. One recalls Edison, Alexander Graham Bell, Marconi, but then one
has to grope for additional names.

Seymour Cray belongs in the same class. Supercomputing may have a less
obvious impact on ordinary life than the mass-market industries created by
those other household names.

Yet, through its indispensability for many of the most advanced spheres of
science and engineering, the consequences of supercomputing as Seymour Cray
conceived it may be as widespread.


Many single-minded persons with extraordinary creativity and
entrepreneurial determination are difficult and sometimes not very likable

Seymour Cray was different.

If his life had taken a different turn, if he had spent a conventional
career as a thoroughly competent engineer in some large enterprise, he would
still have been respected, admired, and trusted by hundreds, perhaps
thousands of people.

Because Seymour Cray was an exceptional man -- in dimensions that reach
far beyond technological skills or the ability to start and sustain a
successful company.

Graceful, in movement, thought, and in personal relations, Seymour Cray
was dignified, confident, even authoritative, but gentle and patient. He
spoke forcefully and persuasively, especially in face-to-face conversations,
but he also listened carefully.

These qualities explain the exceptional strength and persistence of the
widespread loyalty and respect of people who worked for decades with Seymour
Cray. He also had an unusual impact on those, like this reporter, who met
him only once, for a brief interview with other journalists.


Seymour Cray became a famous man, but he did not respond to fame like most
other celebrities. He refused to be celebrated.

Many of the obituaries that will appear during the next few days will use
the word "reclusive." The writers may attach an implication that he was
eccentric, perhaps a bit weird.

That would be misleading. To be sure, he took little interest in marketing,
beyond the circle of personal friends whom he regarded as collaborators
rather than customers. He had the utmost faith in his products, and he was
confident that their excellence was sufficient eloquence.

He valued his privacy and had little taste for the penalty of fame in this
century -- which is to be forced to expose one's personal life and nature to
the world, through the often-distorting lenses and preconceptions of the news

For those who knew Seymour Cray, he was a very normal man -- except for
his remarkable personal qualities as well as his originality.


Time passed. In supercomputing, other ways to achieve exceptionally high
performance began to equal, and then to surpass, the architectural
philosophies that Seymour Cray believed in.

During the last few years, he endured repeated disappointments -- although,
characteristically, he launched a new venture only a few months before his
tragic and untimely death.

Nevertheless, his computational watchwords of simplicity, balance, and
efficient memory management remain as sound today as at the beginning of
Seymour Cray's lustrous career.

Many creative minds are contributing to the contemporary development of
supercomputing in a number of directions -- many of them distant from the
solutions that Seymour Cray originated.

Systems with performance in the hundreds of gigaflops are becoming
routine. Teraflop performance is now a realistic and imminent goal. Equally
important, products capable of several gigaflops are becoming available at
prices less than the pricetag on a well-equipped family sedan.

Yet all of those responsible for these modern marvels must acknowledge
that they are building upon the rich testament of an extraordinary man.

In death as in life, Seymour Cray is supercomputing.


From: Per Grove Thomsen <>
Date: Mon, 30 Sep 1996 14:11:35 +0100 (MET)
Subject: Thoger Busk

I regret to inform you that Professor Thoger Busk of the Dept. for
Numerical Analysis, DTU, Lyngby, Denmark, passed away on Wednesday,
September 25, after a few months illness.
Thoger retired eight years ago from the chair of Numerical Analysis at
DTU and in later years spent time on his passion of collecting ancient
coins and medals of which he was compiling a register for the Danish
National Museum.
Thoger is survived by his wife, Mary, and their son John.
We will always remember him as a warm and humorous person who
made many friends the world over.

Per Grove Thomsen
Institute For Mathematical Modelling
Danish Technical University


From: Per-Ake Malmquist <>
Date: Sat, 5 Oct 1996 17:59:48 +0200
Subject: Orthogonal Matrices as a Product of Jacobi Rotations

Any SO(n) matrix can be written as the product of
n(n-1)/2 Jacobi rotations. As an example, a 3-dim
rotation matrix can be factored into three rotations
by Euler angles. By extension of this example, a set
of 2x2 rotations that generate a give SO(n) matrix
can easily be found.

Q: What is the oldest known solution(s) to this problem?
Is there a traditional attribution?
Is the problem treated in any generally available textbook
or journal?

Per-Ake Malmqvist (


From: Tony Kearsley <>
Date: Mon, 30 Sep 96 08:32:16 EDT
Subject: Change of Address for Anthony J. Kearsley

Dear Colleagues,
I have accepted a position at the National Institute of
Standards and Technology in Maryland. My new coordinates
are now:
Anthony J. Kearsley
Applied and Computational Mathematics Division
National Institute of Standards and Technology
820 Quince Orchard Road - Room 375
Gaithersburg, MD 20899-0001
Phone: (301) 975-6103
FAX: (301) 990-4127
Warmest wishes, Tony


From: G Meurant <>
Date: Fri, 4 Oct 1996 08:48:37 +0200
Subject: Phone Numbers in France

Dear na.digest,

Starting October 18, all the phone numbers in France are going to change.
Even though most people use e-mail to communicate, I was thinking the
following information could be useful.

Starting Friday October 18, 1996, at 11pm, all the phone numbers in France
will change. We are switching from an 8-digit system to a 10-digit system.
Two new digits will be put in front of the old phone numbers depending in
which region the phone number is located. France will be divided in 5 zones.
To know in which zone a phone number is you just have to look at the 2 first
digits of the 8-digit old number and apply the following rules:

01: Paris and suburbs: 30,34,39,40-49,53,55,60,64,69
02: North-West: 31-33,35,37-41,43,47,48,51,54,96-99
03: North-East: 20-29,44,60,80-89
04: South-East: 42,50,67-79,90-95
05: South-West: 34,45,46,49,53,55-59,61-63,65

So, as an example, my office phone number that is now 45 95 65 39 in a suburb
of Paris will become 01 45 95 65 39.

BUT, be careful if you call from abroad you must not dial the 0 in front of
the number, so you just have to use a 9-digit number. To call me you will dial
33 1 45 95 65 39. Moreover, when you are in France and you want to place an
international call, you won't have to use the 19 as before but 00 as in all
civilzed countries.

Good luck
Gerard Meurant


From: Janos Pinter <>
Date: Fri, 4 Oct 96 09:59:38 -0300
Subject: Global Optimization Software Review

Dear Colleagues:

At the INFORMS meeting in Atlanta (Nov 3-6, 1996), we are giving a tutorial
on global optimization (GO). In this context, we would also like to discuss
software you (may) have developed for solving global optimization problems.
This information will also be used in our forthcoming software review.
Approaches to solve continuous GO problems, as well as various combinatorial
optimization algorithms are of relevance for the review.

Please send us a summary description (no more than two pages), providing the
following information:

- name of the software (including version names, if appropriate)
- GO areas / problem types covered by the software
- the basic algorithmic approach used, mentioning enhancements
- hardware and software environment / platform(s)
- test results and/or 'real world' applications
(experience: problem sizes solved, difficulties?)
- availability and form of user instructions / manual
- availability of software demonstration product(s)
- availability and form of technical support
- public domain or commercial availability information
- addditional comments (as needed)
- contact addresses (postal, phone, fax, e-mail, WWW - as per your wish)
- most important related references (up to five)

Please follow - as much as possible, with modifications as needed - the
format of the software package descriptions in the book "Optimization
Software Guide" (by Jorge More and Stephen Wright, SIAM, 1994).

Please send to both of us this information by e-mail (preferably as a Latex
file, or ASCII, if necessary). Also, please pass along our information
request to interested colleagues - we would like to make the survey as
complete as possible.

We will mail the collected information (possibly in a summarized form) to
every contributor; your comments and suggestions will be most welcome.

Thank you very much for your cooperation.

Panos Pardalos (;
Janos Pinter (;


From: James Weston <>
Date: Fri, 4 Oct 1996 20:14:41 GMT
Subject: Sparse, Symmetric Eigenvalue Problems


I am a member of a group of three which, for the last three years, has been
researching parallel algorithms for the partial eigensolution of large sparse
symmetric matrices. We have recently developed explicit restart techniques for
the Lanczos algorithm some of which are very efficient when implemented on a
massively parallel machine such as the CM-200. One of the techniques seems to
cope very well with the computation of closely clustered eigenvalues.

We have already assessed the performance of this technique using test matrices
from the Harwell Boeing Collection of Sparse Matrices. However, we would like
to test the technique further using other matrices which arise in 'real world'
applications. Does any one have access to a large sparse symmetric matrix of
this type which has closely clustered eigenvalues, particularly at the upper
end of the spectrum?

If so, would it be possible for us to gain access to it, thereby enabling a
more detailed assessment of the performance of the technique to be carried out?

Many thanks in anticipation,


James Weston
School of Information and Software Engineering
University of Ulster
Cromore Road
N. Ireland
BT52 1SA
Telephone: 01265 324582


From: T. Terlaky <>
Date: Fri, 04 Oct 1996 23:36:46 METDST
Subject: New Book on Interior Point Methods

New Book on Interior Point Methods

Interior Point Methods of Mathematical Programming
Tam\'as Terlaky (Ed.)
Kluwer Academic Publishers (September 1996)
ISBN 0-7923-4201-1

This book primarily intends to give an introduction to the theory of
Interior Point Methods (IPMs) in Mathematical Programming.
At the same time we try to give a quick overview of the
impact, of the extensions of IPMs to smooth nonlinear optimization
and to give an impression of the potentials of IPMs in solving
difficult practical problems.

The book is divided into three parts. Part I summarizes the basic
techniques, concepts and algorithmic variants for linear programming.
Part II is devoted to specially structured and smooth convex programming
problems, while Part III illustrates some application areas. The authors
of the different chapters are experts of the specific area and were
asked to give a relatively easy introductory survey.


Part I Linear Programming
B. Jansen, C. Roos, T. Terlaky
2 Affine Scaling Algorithm
T. Tsuchiya
B. Jansen, C. Roos, T. Terlaky
4 Potential Reduction Algorithms
K. M. Anstreicher
5 Infeasible-Interior-Point Algorithms
S. Mizuno
E.D. Andersen, J. Gondzio, Cs. Meszaros, X. Xu

Part II Convex Programming
F. Jarre
A. Yoshise
M.V. Ramana, P.M. Pardalos
10 Implementing Barrier Methods for Nonlinear Programming
D.F. Shanno, M.G. Breitfeld, E.M. Simantiraki

Part III Applications, Extensions
11 Interior Point Methods for Combinatorial Optimization
J. E. Mitchell
P.M. Pardalos, M.G.C. Resende
13 Interior Point Approaches for the VLSI Placement Problem
A. Vannelli, A. Kennings, P. Chin


From: Arnold Neumaier <>
Date: Sat, 5 Oct 1996 14:25:20 +0100
Subject: Protein Folding Survey

I have written a big survey paper on protein folding.
Preprints are available at

An abstract is appended below.

Arnold Neumaier, Vienna

A. Neumaier, Molecular modeling of proteins and mathematical prediction
of protein structure, SIAM Rev., to appear.

This paper discusses the mathematical formulation of
and solution attempts for the so-called protein folding problem.
The static aspect is concerned with how to
predict the folded (native, tertiary) structure of a protein, given its
sequence of amino acids. The dynamic aspect asks about the possible
pathways to folding and unfolding, including the stability of the
folded protein.

>From a mathematical point of view, there are several main sides
to the static problem:

- the selection of an appropriate potential energy function;

- the parameter identification by fitting to experimental data; and

- the global optimization of the potential.

The dynamic problem entails, in addition, the solution of (because of
multiple time scales very stiff) ordinary or
stochastic differential equations (molecular dynamics simulation),
or (in case of constrained
molecular dynamics) of differential-algebraic equations.
A theme connecting the static and dynamic aspect is the determination
and formation of secondary structure motifs.

The present paper gives a self-contained introduction to the
necessary background from physics and chemistry and surveys some of
the literature. It also discusses the various mathematical problems
arising, some deficiencies of the current models and algorithms,
and possible (past and future) attacks to arrive at
solutions to the protein folding problem.


From: Peter Prinz <>
Date: Mon, 30 Sep 1996 17:36:52 +0100 (MET)
Subject: Gabor Digest

The Numerical Harmonic Analysis Group (NUHAG) is proud to announce
the "Gabor Digest"

The Gabor Digest is a private initiative inspired by the success
of Wim Sweldens Wavelet Digest
but not competing with it. It is supposed to be an informal platform
for the exchange of ideas related to Gabor analysis, Weyl-Heisenberg
frames and its applications. In addition it should be a central node
in the WWW for those seeking information in this area.

In the beginning we plan two different subjects:

(1) The Gabor Digest; is a mailing list sent out periodically
(about every month). It contains information about recent
contributions in theory of Gabor analysis and its
applications in signal and image processing, pattern
recognition, vision, time-frequency analysis etc.
Further the Gabor Digest provides information about upcoming
conferences, questions, publications, software and algorithms.

(2) The Gabor Archive; is a place to collect publications
and abstracts resp., concerning Gabor analysis and Weyl-Heisenberg
frames. At the moment only a few abstracts and papers are available,
but authors are encouraged to contribute published papers and preprints.
The Gabor archive contains only the abstracts, and optionally
links to ftp-sites where the papers can be received. If You do not
have the possibility to offer an ftp site where your paper can be found,
we are willing to store it at the Gabor Archive.

Peter Prinz & Hans G. Feichtinger

( ,


From: Gerd Kunert <gerd.kunert@Mathematik.TU-Chemnitz.DE>
Date: Tue, 1 Oct 96 08:35:52 MES
Subject: Trace Theorem

Dear colleagues,

In several textbooks the trace theorem for fractional Sobolev spaces
is dealt with. Basically the trace operator acts from

H^s (Omega) --> H^{s-1/2} (Gamma) if s > 1/2,

with Gamma being the boundary of Omega, and if some additional
assumptions on the domain Omega are satisfied.

However, I did not find a source that stated if the trace theorem is
valid for s = 1/2, or if it is not valid, or if it just can not be
proven. So I wonder if there is a (comparatively) simple counterexample
for s = 1/2, or if the trace theorem can be guaranteed under some more
restrictive assumptions.

Gerd Kunert


From: Wai Sun Don <>
Date: Tue, 1 Oct 96 11:07:19 -0400
Subject: Numerical Differentiation by Pseudospectral Methods

Hello, this message announces the PseudoPack v2 beta.

PseudoPack is a software library for numerical differentiation by
pseudospectral methods.

It is being developed at Brown University by Prof. Wai Sun Don and
Alex Solomonoff.

Three different Collocation Methods are incorporated in the package:
Fourier Method
Chebyshev Method
Legendre Method
For now, they are all based on the Gauss-Lobatto points.

Three type of algorithms are provided:
Matrix-Matrix Multiply Algorithm
Even-Odd Decomposition Algorithm
Transform Algorithm
with the exception of Legendre Method that has no known Transform Algorithm.

The software package is written in Fortran 77 with the C preprocessor. We
have used a few Fortran 90 extensions, namely long function and variable
names, and use of the enddo keyword.

It is freely available, at least to all non-commercial users.

You can read the description and/or download the tarfile from our Web site,

Questions, comments, or suggestions to

Wai Sun Don and Alex Solomonoff


From: Bruce Wade <>
Date: Wed, 2 Oct 1996 11:32:39 -0500 (CDT)
Subject: International Journal of Applied Science and Computation

The International Journal of Applied Science and Computation will be
publishing a special issue edited by David Schultz and Bruce Wade on
computational fluid dynamics. Any topics related to this area will be
considered for publication.

Requests for additional information can be addressed to
Contributors should send three copies of their paper to:

David Schultz, Professor
Department of Mathematical Sciences
University of Wisconsin-- Milwaukee
PO Box 413
Milwaukee, Wisconsin 53201-0413


From: Michael Thune' <>
Date: Mon, 30 Sep 1996 16:00:48 +0200 (MET DST)
Subject: Conference on Computer Science Education

Uppsala, Sweden
June 1 - 5, 1997
Call for Participation

The ACM Special Interest Groups in Computer Science Education (SIGCSE)
and Computer Uses in Education (SIGCUE) announce the second annual
conference and working meeting on the theme of integrating technology
into computer science/informatics education. Submissions are requested
in all related areas.

This includes, but is not limited to, the following topics:

- Closed laboratories/Practical experiences
- Computer science education research
- Computer supported collaborative learning
- Different/new approaches to examination
- Distance learning
- E-mail & bulletin board-based class support
- Effective use of portable computers
- Effective use of the World Wide Web
- Evaluating teaching methods
- Graphics/visualization
- Instructional technology
- Integrating networks and distributed computing
- IT & ethics
- Multimedia/interactive learning
- Need for future changes to courses

Two Distinct, Intermingled Events
This event will consist of both a traditional conference and a set of
working group meetings.

Conference: The conference will consist of invited speakers,
technical sessions (featuring papers, panels, demonstrations,
and posters), and interaction with the working groups.
Conference attendees will have several opportunities to provide
important feedback to the working group members.

Working Groups: Each working group will focus on a particular
aspect of technology in computer science education. Those who
participate in the working groups will meet often during the
conference and work together to produce a document or other
material for publication immediately after the conference.

See the Web site or for more detailed information


From: Barry Koren <>
Date: Thu, 3 Oct 1996 10:52:47 +0100
Subject: Computational Fluid Dynamics Workshop

DNS and LES of Complex Flows
University of Twente, July 9-11, 1997

Call for Papers

Invited speakers:

J. Eggels (Shell Research Amsterdam), M. Germano (Politecnico Torino)
J. Jimenez (Univ. Madrid), L. Kleiser (ETH Z"urich)
M. Kloker (Univ. Stuttgart), M. Lesieur (Univ. Grenoble)
F. Nieuwstadt (TU Delft), Ph. Spalart (Boeing)
P. Voke (Univ. Surrey).

A challenging problem in CFD is the simulation of turbulent flow
in practical applications. At present RaNS and DNS/LES form two
complementary approaches to this problem, each with their own
limitations and complications. This three-day meeting will focus on
recent developments in numerical and physical modelling of
transitional and turbulent flow. The invited speakers will provide the
framework of the meeting in which the present state of the art is
confronted with the grand challenges of industry.

Contributions on developments in RaNS, LES and DNS or industrial
problems in CFD are welcome. Two page abstracts can be submitted
before February 1, 1997 to

Mrs. M. Scholten,
Department of Applied Mathematics,
University of Twente,
P.O.Box 217,
7500 AE Enschede,
The Netherlands.

The costs of attending the meeting, including registration,
accomodation, lunches, informal dinner and a copy of the proceedings are
fl. 600. Special rates are available for (graduate)

More information and pre-registration:,, Mrs. M. Scholten or with

Sponsors: J.M. Burgers Centre, ERCOFTAC, Silicon Graphics


From: Jan Kok <>
Date: Thu, 3 Oct 1996 16:13:18 +0200
Subject: Conference of the Dutch Community of Numerical Mathematicians


1997 Conference of the Dutch Community of Numerical Mathematicians
24 - 26 September 1997, at Woudschoten (NL)

The next conference of the Dutch Community of Numerical Mathematicians
(WNW) has been scheduled for 24, 25 and 26 September 1997, to be held
at the Woudschoten Conference Centre, Zeist, The Netherlands.

Topics of the 1997 conference are:

1. nonlinear boundary-value problems (with special attention to
continuation methods and bifurcation)
2. generalized eigenvalue problems and singular-value decomposition
3. numerical treatment of financial models

For information please apply to the secretary of the organizing committee:
Jan Kok
CWI - Centrum voor Wiskunde en Informatica
Organizing committee Woudschoten Conference
P.O. Box 94079
NL-1090 GB Amsterdam
Mail to:

Conference URL: (English) (Dutch)


From: Chenyi Hu <>
Date: Thu, 3 Oct 1996 12:55:32 -0700
Subject: Workshop on Computational Science and Engineering

Call for Papers
International Workshop on Computational Science and Engineering (IWCSE'97)
Hefei, China, May 27-28, 1997

IWCSE'97 is organized by Anhui University, the University of Science
and Technology of China, Hefei High Performance Computing Center of China,
and the University of Houston-Downtown.

Topics of interest include high performance computing in science and
engineering, key technologies in parallel computing, parallel algorithms for
numerical and non-numerical problems, models, languages, tools and
environments for parallel programming, and applications.

General-chairmen: SHI, Zhongci, Chinese Academy of Sciences

For more details, please see


From: Jack Dongarra <>
Date: Thu, 3 Oct 1996 14:55:58 -0400
Subject: Parallel Benchmark Working Group

Dear Colleague,

The ParkBench (Parallel Benchmark Working Group) will meet
in Knoxville, Tennessee on October 31th, 1996.

The meeting site will be the Knoxville Downtown Hilton Hotel.
We have made arrangements with the Hilton Hotel in Knoxville.

Hilton Hotel
501 W. Church Street
Knoxville, TN
Phone: 423-523-2300

We should plan to start at 9:00 am October 31th and finish about 5:00 pm.
If you will be attending the meeting please send me email so we can better
arrange for the meeting.

The objectives for the group are:
1. To establish a comprehensive set of parallel benchmarks that is generally
accepted by both users and vendors of parallel system.

2. To provide a focus for parallel benchmark activities and avoid
unnecessary duplication of effort and proliferation of benchmarks.

3. To set standards for benchmarking methodology and result-reporting
together with a control database/repository for both the benchmarks and
the results.

The following mailing lists have been set up. Whole committee Low level subcommittee Compact applications subcommittee Methodology subcommittee Kernel subcommittee

Jack Dongarra
Erich Strohmaier


From: Jens Burmeister <>
Date: Sun, 6 Oct 1996 11:06:59 +0200 (MET DST)
Subject: Numerical Treatment of Multi-Scale Problems

Second Announcement

The GAMM Committee "Efficient numerical methods for pde" in cooperation
with the Christian-Albrechts-Universit"at Kiel organizes the

13th GAMM-Seminar Kiel
Numerical Treatment of Multi-Scale Problems

Chairmanship: W. Hackbusch (Kiel), G. Wittum (Stuttgart)

Date: January 24th to 26th, 1997

Location: Mathematisches Seminar und
Institut f"ur Informatik und Praktische Mathematik,
Universit"at Kiel (Germany)

Topics: Numerical Treatment and Implementation Aspects of

- problems
defined on complicated geometries,
with highly varying coefficients,
- coarsening strategies for
multi-level methods,
finite element spaces,
- (discrete) homogenisation techniques,
- multi-scale discretisations.

Abstracts: Please send abstracts (10-20 lines) of your lecture by
Nov. 15, 1996. Notice of acceptance will be given by
Nov. 30. All participants, whether giving a talk or not,
have the possibility of sending an abstract of their
work on the topic of the conference. The collection
of abstracts will be available at the conference.

Conference fee: DM 70,- (to be paid after arrival)

Proceedings: The first twelve GAMM-Seminars were held at Kiel in 1984, 1986-
1996. The corresponding proceedings have been published
in the series "Notes on Numerical Fluid Mechanics" by
Vieweg Verlag, Braunschweig, Germany
(Volumes 10, 16, 21, 23, 30, 31, 33, 41, 46, 49, 51 and 54).

Local organization :
J. Burmeister, Tel. : ++49-431-880-4462, Fax : ++49-431-880-4054,



From: Marsha Berger <berger@BERGER.CIMS.NYU.EDU>
Date: Tue, 1 Oct 96 11:20:49 EDT
Subject: Postdoctoral Position at Courant Institute

The Courant Mathematics and Computing Laboratory (CMCL) of
the Courant Institute expects to have a postdoctoral research
fellowship available starting on or around January 1, 1997.

The CMCL has ongoing research activities in a number of areas
of computational mathematics, including combustion, heat transfer,
computational fluid dynamics, and materials science. Our interests
range from applied analysis to the development of numerical algorithms
and software for a variety of advanced computer architectures.

The fellowship is supported by the DOE Office of Energy Research,
through the division of Mathematical, Information and Computational Sciences.

Interested candidates should send a resume, research statement, preprints
or thesis if available, and three letters of recommendation to:

Peter Lax, Chair
DOE Fellowship Committee
Courant Institute
251 Mercer St.
New York, NY 10012

NYU is an AA/EO employer.


From: Y. F. Hu <>
Date: Wed, 2 Oct 96 09:53:37 BST
Subject: Positions at Daresbury Laboratory, Warrington, UK

Computational Engineers

Computational Engineering Group
Department for Computation and Information
Daresbury Laboratory, Warrington, UK

Vacancies exist at the Daresbury Laboratory for 2 HSO's to work
in the application of high performance parallel computing to problems
in engineering, in particular in the areas of Computational Fluid
Dynamics (CFD) and computational mechanics. The main purpose of
the work is to support university and industrial activities in
the area of computational engineering by:

* collaborative development of parallel implementations
of engineering applications targeted at high performance
computing systems;
* supporting the exploitation of high performance computing facilities
with these applications with on-site facilities including:
a 64 processor Intel iPSC/860; a 16 processor IBM SP2; and
clusters of HP and IBM workstations, and off-site facilities
include a 16 processor J90 and a 512 Cray T3D;
* disseminating information through workshops and newsletters
into the academic and industrial communities.

The post is in the Computational Engineering Group which has
parallel computing interests in: parallel mesh partitioning; parallel
mesh generation; dynamic load balancing; advanced turbulence modelling;
process engineering, parallel algorithms; and internal and external
flows for both compressible and incompressible flows. The Group has
a number of state-of-the-art workstations including Silicon Graphics
and DEC workstations and is an acknowledged centre of excellence in
parallel programming. The main duties of the post will be:

* to stimulate and assist in the development of leading edge parallel
applications in collaboration with industry and academe;
* to assist groups with the exploitation of high performance computing
facilities at Daresbury Laboratory and other national centres;
* to liaise with initiatives supporting related activities e.g. ERCIM,
* to disseminate information via appropriate mechanisms,
such as newsletters, workshops and meetings and to assist in organising
workshops on parallel computing and its exploitation within engineering.

The ideal candidates will have a degree and at least 2 years relevant
experience or a Ph.D. in an appropriate computational subject. The
candidates will play a key role in parallel algorithm development and
supporting the exploitation of parallel computing in high performance
engineering applications. A good knowledge of CFD or computational
mechanics is essential and the ability to work within a team on academic
and industrial projects is necessary. Substantial experience of original
code development in a high level language such as FORTRAN is necessary.
Experience in C or other advanced programming languages would be beneficial
and a willingness to learn and understand a variety of computing platforms
is essential.

Further Information

Both vacancies are fixed term for three years. Starting salary will be
in the range 13,491 - 19,669 pounds, dependent on experience; there
is also a non-contributory superannuation scheme and a generous
leave allowance.

Please send a CV to:

Personnel Division
Rutherford Appleton Laboratory
Oxfordshire OX11 0QX

Please quote reference VN1418/96

The closing date for applications is 11th October 1996


From: Peter Hislop <>
Date: Wed, 2 Oct 1996 12:19:28 -0400 (EDT)
Subject: Position at University of Kentucky

The Department of Mathematics at the University of Kentucky
invites applications for at least one tenure-track Assistant
Professorship to begin in Fall 1997 (subject to budgetary
approval.) We are interested in
applicants in the areas of numerical analysis and
algebra/number theory. However, applications in other areas are
also welcome. We are especially interested in
applications from women and
minority groups. Using the AMS application cover sheet (if
possible), applicants should submit a vita, a description of
research and future plans, evidence of effective teaching, and
arrange to have at least three letters of recommendation sent to:
Chair of the Recruiting Committee, Department of Mathematics, 715
POT, University of Kentucky, Lexington, KY 40506-0027. The
deadline for submission of applications is 31 January 1997. We
expect to begin evaluating applications 1 December 1996.


From: James Sweeney <>
Date: Thu, 03 Oct 1996 22:48:54 -0700
Subject: Position at Stanford University

Assistant Professor Job Opening
Stanford University
Department of Engineering-Economic Systems and Operations Research

The Stanford University Department of Engineering-Economic Systems and
Operations Research (EES&OR) invites applications for a tenure track
position at the Assistant Professor level. Our goal is to fill the
position by September 1, 1997. We will start reviewing applications in
autumn; complete applications received before January 17, 1997 will
receive first priority.

EES&OR addresses operational, strategic, and policy problems, using logical
and mathematical models to provide insight and solutions. EES&OR was
founded in 1996 by a merger of the Department of Engineering-Economic
Systems and the Department of Operations Research. The department offers
degree programs leading to Master of Science, Engineer, and Doctor of
Philosophy. It will be developing an undergraduate program in the near

The department has special interest in theory and application within the
following areas: continuous, discrete, and numerical optimization;
probability and stochastic processes; dynamic systems and simulation;
economics, finance, and investment; decision making, including decision
analysis, dynamic programming, and planning under uncertainty; operations
and services; corporate and individual strategy; and private and public
policy issues. More information is available through the EES&OR WWW

We intend to hire a new faculty member who has an outstanding
methodological foundation and an interest in addressing operational,
strategic, or policy problems. The successful candidate must demonstrate
promise of becoming an exceptional teacher and researcher.

Candidates should send a resume (including research accomplishments,
teaching experience, publications), at least one research paper (published
or unpublished), and names of at least three references. References
should be encouraged to communicate directly, by letter, with the search
committee. Applications should be sent to:

Chair, EES&OR Search Committee
Department of EES&OR
Stanford University
Stanford, CA 94305-4023

Stanford University is an equal opportunity employer, encouraging
applications from women, minority candidates, and disabled persons.


From: Heinz W. Engl <>
Date: Fri, 04 Oct 1996 08:12:04 EDT
Subject: Position at University of Linz


For a joint research and development project between the Department of
Industrial Mathematics at the University of Linz (Austria) and the
affiliated company MathConsult GmbH, we are looking for a mathematician
(with at least a degree comparable to a Master's degree or diploma) to
be employed at MathConsult from January 1,1997, for one year initially
(renewable in case of success). The project involves modelling and numerical
simulation of solid and gas flows and chemical reactions. Programming should
be done in C++. Competitive salary.
For employment law reasons, citizenship of a European Union country or
of Norway or Switzerland required.
For further information contact Prof.Heinz Engl, University, A-4040 Linz,
Austria; fax: +43-732-2468855, E-Mail:


From: E. B. Saff <>
Date: Tue, 1 Oct 1996 17:18:17 -0400
Subject: Contents, Constructive Approximation

Table of Contents: Const. Approx., Vol. 12, No. 4, 1996

443 V. E. Maiorov
Widths and Distributions of Values of the Approximation
Functional on the Sobolev Spaces with Measure

463 S. D. Fisher
Widths and Optimal Sampling in Spaces of Analytic Functions

481 F. Peherstorfer
Minimal Polynomials for Compact Sets of the Complex Plane

489 J. A. Adell and J. de la Cal
Bernstein-Type Operators Diminish the $\phi$-Variation

509 Giorgio Mantica
A Stable Stieltjes Technique for Computing Orthogonal
Polynomials and Jacobi Matrices Associated with a Class
of Singular Measures

531 G. Valent
Co-recursivity and Karlin-McGregor Duality for Indeterminate
Moment Problems


555 G. Nurnberger
Bivariate Segment Approximation and Free Knot Splines:
Research Problems 96-4


From: Art Werschulz <>
Date: Thu, 3 Oct 1996 09:49:57 -0400 (EDT)
Subject: Contents, Journal of Complexity

TABLE OF CONTENTS: Journal of Complexity -- December, 1996
Volume 12, Number 4

A Special Issue for the Foundations of Computational Mathematics Conference
Rio de Janeiro, January 1997

Felipe Cucker and Mike Shub

Greg Wasilkowski, "Average Case Complexity of Multivariate Integration
and Function Approximation"


Homotopy Methods and Real Machines
organized by Felipe Cucker and Herb Keller

Pascal Koiran, "Hilbert's Nullstellansatz is in the Polynomial Hierarchy"

Information-based Complexity
organized by Erich Novak and Henryk Wozniakowski

Karin Frank and Stefan Heinrich, "Computing Discrepancies of Smolyak
Quadrature Rules"

Peter Hertling, "Topological Complexity of Continuous Operations"

Karl-Heinz K|fer, "On the Expected Number of Shadow vertices of the
Convex Hull of Random Points

Erich Novak and Ingo Roschmann, "Numerical Integration of Peak

Erich Novak and Henryk Wozniakowski, "Topological Complexity of

Sergei G. Pereverzev and Sergei G. Solodki, "The Minimal Radius of
Galerkin Information for Fredholm Problem of the First Kind"

Leszek Plaskota, "Worst Case Complexity of Problems with Random
Information Noise"

Arthur G. Werschulz, "The Complexity of Definite Elliptic Problems
with Noisy Data"

organized by Clovis Gonzaga and Michael Todd

Shinji Mizuno, Mimrod Megido and Takashi Tsuchiya, "A Linear
Programming Instance with Many Crossover Events"
Yinyu Ye, "How Partial Knowledge Helps to Solve Linear Problems"

Systems of Algebraic Equations and Computational Algebraic Geometry
organized by Thomas Lickteig and Marie-Frangoise Roy

Dario Bini and Victor Y. Pan, "Graeffe's, Chebyshev-like and
Cardinal's Processes for Splitting a Polynomial into Factors"

John Dalbec, "An Algebraic Proof of Barlet's Join Theorem"

Laureano Gonzalez-Vega and M'hammed El Kahoui, "An Improved Upper
Complexity Bound for the Topology Computation of a Real Algebraic
Plane Curve"

Thomas Lickteig and Marie-Frangoise Roy, "Semi-Algebraic Complexity
of Quotients and Sign Determination of Remainders"

Victor Y. Pan, Myong-hi Kim, Akimou Sadikou, Xiaohan Huang, Ailong
Zheng, "On Isolation of Real and Nearly Real Zeros of a Univariate
Polynomial and Its Splitting into Factors"

Vision and Related Computational Tools
organized by Jean-Michel Morel and David Mumford

Olivier Catoni, "Simulated Annealing and I.E.T. Algorithms: Theory
and Experiments"


End of NA Digest