NA Digest Sunday, May 10, 1992 Volume 92 : Issue 19

Today's Editor:

Cleve Moler
The MathWorks, Inc.

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to


From: Bill Henshaw <>
Date: Fri, 8 May 92 15:11:07 EDT
Subject: Farouk Odeh

It is with great sadness that we must tell you that Farouk Odeh passed
away this week. Farouk, who had been at IBM Research for over thirty
years, was the manager of the differential equations group. He will
be greatly missed by all his friends and colleagues. We knew him not
only as an outstanding and insightful mathematician, but also as a
gracious, friendly, and intuitive man with a gentle sense of humor who
seemed to know more than most of us about the poetry of life.


From: Gene Golub <>
Date: Sun, 3 May 92 20:03:14 CDT
Subject: Congratulations to Alan George

Congratulations to Alan George on his election to the Royal Society of
Canada. Alan's work on sparse matrices has been of great importance to
many areas of application.

Gene Golub


From: Eric Grosse <>
Date: Mon, 4 May 92 10:32 EDT
Subject: Photos from Golub Festivities

A few of Tony Chan's photos from Gene Golub's 15th birthday celebration
in Minnesota are available by anonymous ftp from in
the directory netlib/photo/golub92. These are (binary) GIF-format
image files.

If "ftp" and "GIF" sound foreign to you, now is probably not the time
to explore the mysteries of image file formats. On the other hand, if
you're a graphics enthusiast, download and enjoy!

More generally, it would be possible and perhaps useful to compile
group photos from conferences and photos of individual mathematicians,
both current and historical. Aside from the cost in disk space and
network bandwidth, the only disadvantage that occurs to me is the
risk of abuse by digital image manipulation. I welcome advice from
anyone with insight.

Best wishes, Eric


From: Andrew Stuart <stuart@sccm.Stanford.EDU>
Date: Mon, 4 May 92 08:01:51 -0700
Subject: Address Change for Andrew Stuart

As of April 1st 1992, my new address is:

Andrew Stuart
Program in Scientific Computing and Computational Mathematics
Stanford University
California 94305-4040, USA.

Phone: 415-723-8142
Fax: 415-723-1778


From: Edward A. Celarier <>
Date: Sun, 3 May 1992 21:02 EST
Subject: Info on PC-style Math Coprocessors

I am considering the purchase of a coprocessor for my 80386-based
p.c. There seem to be a number of contenders in the 8087-plug-compatible
league: IIT, Cyrix, Intel, and Weitek, in addition to some other versions.
(My PC has a socket for a special processor from Weitek, for example.)

Can someone point me in the direction of some reliable information about
the numerical aspects of these? Are they all IEEE-compliant? Are
80387-plug-compatible chips really all equivalent, state-for-state, to
the Intel chip?

I would appreciate e-mail to CELARIER@CEBAF2.CEBAF.GOV.

Many thanks,

Edward A. Celarier
Dept of Chemistry
Hampton University
Hampton, VA

[Editor's Note: Other people might be interested in this answer to
these questions, too. Please submit any definitive response to Thanks. -- Cleve]


From: David Billinghurst <>
Date: Wed, 6 May 92 14:28:47 -1000
Subject: Laplace's Equation with Nontrivial Boundary Condtions

We are starting a study on anode consumption in an electro-reduction
process. This is a moving boundary problem. The current flow is determined
by Laplace's equation in multiple domains, and the anode surface ( one of
the internal boundaries ) is consumed at a rate proportional to the
local current density.

Physically the process is self-regulating, since regions of the anode
that are closer to the cathode draw a higher current and burn back
preferentially. In addition there are two very different time scales present:
the current relaxes (almost) instantaneously, while the anode is consumed

We plan to solve for the current flow at each time step, and adjust
the boundary explicitly. This reduces the problem to one of solving
Laplace's equation in multiple irregular regions, with conductivity
differences of two or three orders of magnitude, a couple of hundred
times with small changes to the boundary location each time.

We would appreciate hearing from anybody who

1. Can point us to any relevent literature.
2. Has worked on similar problems analytically or numerically.

Bob Hannaford David Billinghurst

Comalco Research Centre
15 Edgars Rd
Thomastown Vic 3074 Phone: +61 3 469 0777
Australia Fax: +61 3 462 2700


From: Ake Bjorck <>
Date: Thu, 7 May 92 15:39:10 +0200
Subject: Bibtex File on Least Squares

I have compiled a bibtex reference base covering aspects of numerical
solution of least squares problems. It contains about 520 references, and in
particular the references for my chapter on Least Squares Methods in
"Handbook of Numerical Analysis, Volume I, North Holland 1990". I am now
making it available by anonomous ftp at in the directory pub/references

The bibtex base has been modelled after a bibtex reference base created
by Pete Stewart (see NA Digest Vol. 91, No. 9). I am using almost all of
his conventions, with respect to citation keys, keywords, etc. I have also
found Pete's program to search a bib file very useful for browsing or for
creating sub-bibliographies. This program is contained in the file bibsearch.tar
and available by anonymous ftp at in the directory pub/references

Ake Bjorck


From: K.S.Yajnik <cmmacs!>
Date: Tue, 5 May 92 19:37:06 EDT
Subject: Date of Conference in Bangalore, India


Our announcement in last week's Digest of the IMACS Int Symposium on
Math. Modelling and Scientific Computing inadvertently omitted the
date of the conference. I apologise for the incomplete announcement.
The date and place are:

December 7-11, 1992
Bangalore, India

If you have any questions please write/e-mail/fax to:

1) K.S.Yajnik, Head, C-MMACS, National Aeronautical Laboratory
Belur Campus, Bangalore
Tel #: (91)(812)574649
fax: (91)(812)560392

2) S.K.Dey, Dept of Math., Eastern Illinois Univ.
Charleston, IL 61920-3099
Tel # : (217)581-3217
Fax #: (217)581-5188

3) E.J.Kansa, Lawrence Livermore National Lab
P.O. Box 808, L-200
Livermore CA 94551-0808


From: Alan Edelman <>
Date: Fri, 8 May 92 18:14:09 PDT
Subject: Any Large Linear Algebra Problems?


Last month, I initiated my second large dense linear algebra survey.
So far, I have received only ONE definitive response. Is large dense
linear algebra an unimportant subject? Do supercomputers currently need
large dense linear algebra libraries or are more specialized toolboxes
needed for now?

Perhaps my survey comes at a busy time of the year or perhaps I am
targeting the wrong audience. Right now, I have no choice to conclude
based on the data that I have so far received that nobody is working on
solving a dense eigenvalue problem of size bigger than 5000 and almost nobody
is solving large dense linear systems not arising from a boundary element

Is this true? If you are even VAGUELY aware of large dense
eigenvalue problems or large dense linear systems other than boundary
element calculations, I beg you to respond with either a completed
form or by forwarding this form to your colleague.
Thank you
-- Alan Edelman

Alan Edelman, Dept of Mathematics, University of California,
Berkeley, CA 94720

My calendar file tells me that it is now time for the second
annual large dense linear algebra survey. This year, I hope to
cover two subjects:

1. large dense linear systems (like last year)
2. large dense eigenvalue problems

It is not within my resources to print out and mail letters to
every science department at universities and industries, so I beg
readers to spread my survey by word of mouth, and strongly urge
you and your colleagues to participate. Readers interested in
last year's survey are invited to obtain it by anonymous FTP
from in /pub/edelman/survey1991.

In order to confine the topic of discussion, we do not consider
any matrix that can be parameterized by significantly fewer than n^2
elements to be dense. Thus a Toeplitz matrix or a matrix of the form
A*A' where A is sparse are not considered dense in this context.
I will allow matrices generated for Panel Methods and Moment Methods
to be considered dense.

Part 1: Linear Systems

A. Largest LU or QR factorization
Has anyone solved a system of size bigger than 60,000 using
traditional LINPACK or LAPACK style methods? If so, please
tell me the time it took, why you solved the problem,
how accurate the solution was, and how you know.
Have you tried a condition estimator for your problem?
Did you consider a Krylov space based iterative method for
your problem?

B. I an interested in the solutions to any dense matrix
bigger than 20,000 for purposes other than Panel Methods
and Moment Methods.

Part II: Eigenvalue Problems

A. I am interested in all eigenvalues problems for dense square matrices
of order at least 5,000. Please carefully describe
where you are in the range of wanting all eigenvalues and all
eigenvectors to merely wanting one eigenvalue. Do your
eigenvalues fall along a curve or cluster or are they scattered
and well separated? Have you evaluated the conditioning of your problem,
and if so, how?

B. Would you like to solve a large dense eigenvalue problem of
order greater than 50,000 if you had the resources? How large
can you foresee your problem getting at this point?

How big is your matrix?
What kind of matrix? (Symmetric, complex, double precision?)
What is the solution method?
What is the time for solution?
On which machine?
How accurate was your solution? (Explain how you know)
What is your confidence in this accuracy?
Could the newly released LAPACK be used for your problem?
(LAPACK replaces LINPACK and EISPACK as the current best linear algebra
software library. Information is available through netlib.)
Please describe your application area.

1. If appropriate please refer to the publication most closely related
to your particular problem. In most cases this will be an article
authored by you or a member of your group.

2. Please suggest an expository article or book that would be
most accessible to a non-specialist trying to understand your


From: Horst D. Simon <>
Date: Fri, 8 May 92 14:46:47 -0700
Subject: U. S. Legislation on Foreign Research Assistants

Dear na-net readers,

I would like to bring your attention a bill which is
currently going through hearings in the house science
subcommittee. I have received the following information
from a colleague here at NASA Ames.

I believe that the goal of this bill, an attempt of restricting
the ability of foreign graduate students to work as
research assistants, will severely impact the research
in many graduate departments in a negative way, while not addressing
at all the problem of increasing enrollment by U.S.
students in math and science graduate departments.
This should be of concern to many na-net readers.

-- Horst Simon.

TO: NAFSAnet Users
FROM: Lisa Jacobson Treacy, Amy Yenkin, NAFSA Central Office
RE: New Legislation on Foreign R.A.s; Forms Available from
GPO (NAFSA Update No. 178)
DATE: May 7, 1992

1. Bill Introduced on Research Assistantships for Foreign

In an attempt to encourage colleges and universities to
recruit more Americans in doctoral programs, especially in the
fields of math and science, Rep. Paul Henry (R-MI) recently
introduced the "American Math and Science Student Support Act"
(HR 4595). The bill would impose strict reporting requirements
as part of grant applications for federal funds which include:
1. the name and country of origin of each nonimmigrant employed
as a research assistant on a federal grant, 2. each such alien's
intent toward seeking permanent residence status, 3. a
description of efforts made to hire U.S. citizens or permanent
residents, and 4. why no U.S. citizens or permanent residents
were available.
On April 29th, the House Science Subcommittee held a hearing
at the bill. NAFSA was present at the hearing and submitted
testimony for the record. Although all the witnesses agreed that
the low number of Americans pursuing PhDs is a problem, all but
one were critical of the remedy offered by the bill. ON THE
FORWARD. However, based on a May 6 article in the Chronicle of
Higher Education, the underlying issue of the role of foreign
students at U.S. institutions is likely to continue to come under
increased scrutiny.
A detailed story on the bill and NAFSA's testimony will be
featured in the next issue of the Government Affairs Bulletin.
Subscribers will receive the May issue in the next two weeks.


From: Wayne Enright <>
Date: Thu, 7 May 1992 16:05:38 -0400
Subject: Conference: Scientific Computation on Workstations

Scientific Computation on Workstations: Current and Future Environments

Preliminary Announcement:

A symposium on ``Scientific Computation on Workstations: Current and Future
Environments'' will be held in Toronto on September 17, 1992, to be followed by
open sessions of the IFIP Working Group 2.5 (Numerical Software) on September 18
and the morning of September 19.

Recent hardware and software developments have led to powerful low-cost
workstations which are fast enough and have sufficient memory and disk space to
permit the effective solution of many significant problems that arise in
scientific and engineering applications. Problem solving environments that
exploit the advantages such workstations offer are being developed to allow
practitioners to solve their problems without requiring them to be systems or
programming experts. In the symposium an overview will be presented of some of
these environments and of how they are likely to evolve in the next few years.
Feature presentations by Cleve Moler (MATLAB - Mathworks), Keith Geddes (MAPLE
- Waterloo), Wayne Enright (ODEs - Toronto), Eugene Fiume (Graphics - Toronto)
as well as representatives from NAG and IMSL have been planned.

The open sessions of the Working Group will feature presentations, mostly by
members of the Group, of current work on Numerical Software -- including such
topics as programming language facilities, benchmarks and performance,
mathematical methods, and education issues. There is a possibility that some
time will be available for short presentations by others on September 18th.
Anyone interested in contributing such a presentation should contact one of the
organizers by the end of May.

The meetings on September 17 and 18 will be held at the Delta Chelsea Inn.
Room rates will be $76 single and $91 double. Hotel reservations should be made
directly with the hotel:

Delta Chelsea Inn, 33 Gerrard Street West, Toronto, M5G 1Z4, Canada
Phone: (416) 595-1975, Fax: (416) 585-4362, Group Code (or 'Q-name'): GFRORE.

Reservations should be made as soon as possible as only a limited number of
rooms are available at this special rate.

The registration fee for the symposium is $25 and advance registration is
needed to ensure that space is available. To register, or to be placed on our
mailing list to receive more information in the future, please contact one of
the organizers, W.H. Enright or T.E. Hull, at:

Department of Computer Science
University of Toronto
Toronto M5S 1A4


The symposium will be supported jointly by WG 2.5, the Information
Technology Research Centre of Ontario, and the Natural Sciences and Engineering
Research Council of Canada.


From: Carlos de Moura <>
Date: Sat, 9 May 92 15:51:17 +0200
Subject: Brazilian School on High Performance Scientific Computing


Rio de Janeiro, August 3-7, 1992
Support and Organization:
IBM-Brasil and LNCC/CNPq, Rio
Confirmed lecturers

Enrico Clementi
Universite de Strasbourg I - Louis Pasteur
CRS4 - Centro Ricerca e Sviluppo, Studi Superiori in Sardegna
1. Evolution of Computers and of Computer Use in Science
2. The MOTECC Initiative

Craig Douglas
Yale University & IBM Yorktown Heights
Theory and Implementation of Multigrid Methods: Why It Might Work and How
to Make It Run Like a Bat Out of Hell on Serial and Parallel Computers

Peter M. Kogge
IBM Senior Technical Staff (FSC), IEEE Fellow
Declarative Computing

Marianela Lentini
Universidad Simon Bolivar
An implementation on transputers of the Simplex Method

Gerard Meurant
CEA - Centre d'\'Etudes de Limeil-Valenton
Computational Linear Algebra

Ulrich Trottenberg
GMD-Bonn & Universit\"at K\"oln
1. Parallel architetures and programming models
2. Parallel strategies for matrix, grid and multi-grid data

Olof Widlund
Courant Institute of Mathematical Sciences, New York University
Domain decomposition algorithms for elliptic partial differential equations

Paul Woodward
University of Minnesotta

Organizing Committee

M. Annarumma - IBM Brasil
A. Cascon - LNCC & UFF
A. Gadelha Vieira - LNCC & UFRJ
A. Santoro - CBPF & FermiLab Visiting Scholar
J. P. Schiffini - IBM Brasil
R. Shellard - PUC/RJ & CERN Visiting Scholar

C. A. de Moura - LNCC & College de France Visiting Scholar


Date: Fri, 08 May 92 21:39:52 BST
Subject: Parallel Course at CERFACS

(European Centre for Research and Advanced Training in Scientific Computation)
Will Organize

An Introduction to Parallel Computing

A three-day advanced short course
Tuesday 9th June - Thursday 11th June 1992
in Toulouse, France


This course on parallel computing will be held at CERFACS, Toulouse from
June 9th to 11th, 1992 under the auspices of CERFACS.

The course is aimed at scientists in industry and academia who wish to
learn about parallel computing and familiarize themselves through hands-on
experience with a range of parallel computers. These will include

-ALLIANT FX/80 and FX/2800
-IBM 950
-CM2 from Sophia
-BBN TC2000

Throughout the course, there will be a strong emphasis on hands-on
experience and it is the intention that students will finish the course
with a basic knowledge of parallelism and how to appreciate, compare, and
use a range of different parallel architectures.


Tuesday, 9th June 1992

Introduction to vectorization, J. DONGARRA
Introduction to parallel architectures, J. DONGARRA
Shared Memory computers, P. AMESTOY
Distributed Memory computers, R.TUMINARO
Description of machines and access, I. D. LEVINE

Wednesday, 10th June 1992

Programming parallel machines, I. D. LEVINE, S. SMITH, R. TUMINARO
Solution of sparse equations, I. DUFF, P.AMESTOY

Thursday, 11th June 1992

Tools and Parallel Programming, W. JALBY
Block iterative methods D. RUIZ
Multigrid Techniques, R. TUMINARO
Hands-on....... applications
Gala dinner (optional)


Participants are advised to register before May 25 by Email, fax or phone to

Dominique BENNETT,
CERFACS, 42 Ave Gustave Coriolis,
Tel : (33) 61 19 31 31
Fax : (33) 61 19 30 30
EMail :

The enrolment fee is 5,930. FF (of which 930F is VAT) which includes lunch
and all refreshments. For full-time students the fee is reduced to 3,558 FF
(of which 558 F is VAT). A written request for this reduction, together
with verification of student status, must be submitted along with your

Delayed registrations will increase the registration fees by 10%.
The CERFACS' "agrement" number is 7331P003131.

* cheque made out to GIP CERFACS
* or by bank transfer to Barclays Bank, 3 rue Genty Magre, 31000 Toulouse,
Bank Code 30588, Agency Code 12501, Account 03092257101, RIB key 02
* or in cash on the first day of the course


From: Richard Brualdi <>
Date: Mon, 4 May 92 17:49:45 CDT
Subject: Contents: Linear Algebra and its Applications

Contents Volume 171, July 1, 1992

Fuad Kittaneh (Bloomington, Indiana)
A Note on the Arithmetic-Geometric-Mean Inequality for Matrices 1

Charles A. Hall and Xiu Ye (Pittsburgh, Pennsylvania)
Construction of Null Bases for the Divergence Operator Associated With
Incompressible Navier-Stokes Equations 9

Svatopluk Poljak (Praha, Czechoslovakia)
Minimum Spectral Radius of a Weighted Graph 53

Curtis Greene (Haverford, Pennsylvania)
Proof of a Conjecture on Immanants of the Jacobi-Trudi Matrix 65

Jianming Miao and Adi Ben-Israel (New Brunswick, New Jersey)
On Principal Angles Between Subspaces in Rn* 81

Arthur M. DuPre (Newark, New Jersey) and Seymour Kass (Boston, Massachusetts)
Distance and Parallelism Between Flats in Rn 99

Ling Chen (Guangzhou, China) and Chi Song Wong (Windsor, Ontario, Canada)
Inequalities for Singular Values and Traces 109

Jiang Erxiong (Shanghai, China)
A Note on the Double-Shift QL Algorithm 121

J. Pasupathy and R. A. Damodar (Bangalore, India)
The Gaussian Toeplitz Matrix 133

B. Uhrin (Budapest, Hungary)
Touching Points of a Star-Shaped Set With an Affine Subspace 149

Daniel Hershkowitz (Haifa, Israel)
Recent Directions in Matrix Stability 161

Yang Shangjun (Hefei, Anhui, People's Republic of China)
A Note on the Exponent Set of Primitive Minimally Strong Digraphs 187

P. Turkowski (Krakow, Poland)
Structure of Real Lie Algebras 197

Sung Je Cho, Seung-Hyeok Kye, and Sa Ge Lee (Seoul, Korea)
Generalized Choi Maps in Three-Dimensional Matrix Algebra 213

Wilbur N. Dale and Malcolm C. Smith (Columbus, Ohio)
A Note on Eventually Time-Invariant Systems 225

Chin Chang and Tryphon T. Georgiou (Minneapolis, Minnesota)
On a Schur-Algorithm Based Approach to Spectral Factorization: Connection With
the Riccati Equation 233

James V. Burke (Seattle, Washington) and Michael L. Overton (New York, New
Stable Perturbations of Nonsymmetric Matrices 249

Book Review

Stephen G. Nash (Fairfax, Virginia)
Review of Fundamentals of Matrix Computations by David S. Watkins 275

Author Index 279


End of NA Digest