NA Digest Saturday, October 12, 1991 Volume 91 : Issue 41

Today's Editor: Cleve Moler

Today's Topics:

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to


From: June Donato <donato@mira.EPM.ORNL.GOV>
Date: Wed, 9 Oct 91 15:10:38 EDT
Subject: New address for June Donato

Effective immediately my address is

June M. Donato
Oak Ridge National Laboratory
Mathematical Sciences Section, EPM Division
P.O. Box 2008, Building 6012
Oak Ridge, TN 37831-6367


phone: (615) 576-7928
FAX: (615) 574-0680


From: Uri Ascher <>
Date: 8 Oct 91 21:19 +0100
Subject: Phone Numbers at U. British Columbia

As of this month, the first 3 digits of phone
numbers at the University of British Columbia
have changed from 228- to 822-
(or UBC -- with the hope that no one will have
been inconvenienced by this cute idea).
In particular, my office number is (604) 822-4907
Computer Science FAX (604) 822-5485

Uri Ascher


From: Alan Edelman <>
Date: Thu, 10 Oct 91 16:39:29 PDT
Subject: Bay Area NA Day

BAY AREA Numerical Analysis DAY at BERKELEY
November 16, 1991

There will be a Bay Area NA Day in Berkeley on Sat, Nov 16. This day
is an opportunity to meet colleagues and students (and also visitors)
in the Bay area.

We are currently ironing out the details of scheduling, but we
anticipate that the first talk will start around 10 (coffee and
munchies before hand) and the last talk will end around 4:30. The
program will consist of about eight 30 minute talks, most of which
will be given by students, post-docs, and younger faculty.

The evening will end with a Chinese banquet.

Further details will be announced. Stay tuned.
All are welcome. Hope you can come,

-- Alan Edelman


From: M. J. Saltzman <>
Date: Tue, 8 Oct 91 17:05:47 -0400
Subject: LP Codes and Call for Information

In NA Digest Sunday, September 15, 1991 Volume 91 : Issue 37,
Kristjan Jonasson <> writes:
>I would like to find software (preferably FORTRAN subroutines) for
>solving large sparse linear programming problems. Ny intention is
>to use these in a research project on nonlinear minimax problems.
>Can anyone help me?

and Jerzy Wasniewski <> writes
>Does anyone have, or know where I could get, any public domain
>software for Sparse Linear Programming Problem. We need this
>software for the education purpose.
>Thank you in advance.
>Best regards,
> Jerzy

As far as I know there are no PD linear programming codes. There is a
tableau code (i.e. not sparse) in _Numerical_Recipes_, and there may
be some other toy codes published in other textbooks. There are
several commercial packages that offer educational discounts. Some of
these offer either callable libraries or the ability to link to
user-written subroutines. I don't know if the digest is the place to
discuss the merits of individual packages, but I will be happy to
continue the discussion by e-mail. Also, a fairly complete list of
commercial LP packages for PCs was published in _OR/MS_Today_, v17#5
(October 1990).

Questions about optimization software seem to come up frequently,
if not here, certainly on Usenet News. I have offered on Usenet to
compile a list of optimization codes. I would like to arrange to
make the list available for electronic distribution over NA-Net,
Usenet or maybe Netlib. There is some chance that it could be
published in a forum like _OR/MS_Today_ as well.

If you are the author of an optimization code(*) (and did not respond
to my earlier query on Usenet), and you would like to be included on
such a list, please send me the following information:

1) The name of your product.
2) Your name or the name of your company (contact names, addresses,
phone and email).
3) The problem or problems it is designed to solve.
4) The algorithm or algorithms you implement (if they are based on
well-known methods like simplex, branch-and-bound, etc), and the
language your code is written in.
5) The platform(s) your code runs on, and the list price or range of
prices for each machine.
6) Size of problems you can handle.
7) Whether you offer special academic or classroom pricing.
8) Is your code in the public domain?
9) If not, does your product include source code?
10) How can your code be aquired?
11) Any other comments you would like to make (be brief).

(*) Authors only for the moment, please. Once I get a fairly stable
list together, I will ask for corrections and additions.

Thanks for your interest and assistance.

Matthew Saltzman
Clemson Univ Math Sciences

[Editors Note: Terrific project! Thanks for taking it on.
I, for one, would sure like to see the results. Good luck.
-- Cleve]


From: Stefano Foresti <>
Date: Wed, 9 Oct 91 10:16:36 -0600
Subject: Adaptive Methods for PDEs

I am investigating the literature of adaptive methods for partial
differential equations. Particularly, I am interested in parabolic
problems with space-time adaptation.

I would appreciate any information on recent advances in this field,
references, and possibly copies of technical reports or preprints.
I would also welcome any information on publicly available codes.

Stefano Foresti
Utah Supercomputing Institute
and Department of Mathematics
85 SSB
University of Utah
Salt Lake City, Utah 84112, USA
tel: (801)581-3173 Fax: (801)581-5217


From: Stewart Levin <>
Date: Thu, 10 Oct 91 15:14:52 -0500
Subject: Information Sought About "Gold Codes"

I'm attempting to trade off some quality for computational speed
in finite difference modeling. The plan is to inject more than
one emitter (source) into a 2D or 3D medium simultaneously and
distinguish them by supply them with different waveforms which
have relatively small cross-correlations with each other. Early
work with pseudo-random sequences were not too bad, but I've been
told I should be using Gold codes. Does anyone have a computer
subroutine for generating such sequences and/or comments on
whether this is a reasonable thing to do?

Stewart A. Levin
Mobil Research and Developement


From: Henk van der Vorst <>
Date: Mon, 7 Oct 91 12:10:40 +0100
Subject: More on Refereeing

I do not like the idea of punishing unwilling referees, I'd rather
prefer to motivate all possible referees.

Some journals are very nice towards referees, so that at least
you feel acknowledged afterwards, e.g., BIT lists the referees
at the end of each complete volume.
BIT even goes so far to give a small present to the referees for
their work in the past year. I am not suggesting that all journals
would adopt that policy, but I must confess that it gave me a
pleasant feeling at least once a year (with respect to BIT).

Nowadays many people have to specify their scientific activities,
and refereeing is accepted (at least at some Dutch universities) as
a scientific activity (though of lower rank than publications).
Maybe it would help if one would receive a written proof of
having refereed for some journal.

Henk van der Vorst

Henk A. van der Vorst
Department of Mathematics tel: +31-30-533732
University of Utrecht Telefax: +31-30-518394
P.O. Box 80.010
Utrecht, The Netherlands


From: Eduardo Sontag <>
Date: Wed, 9 Oct 91 16:38:31 EDT
Subject: About Journal Refereeing

Let me follow-up on this discussion with a couple of suggestions for
improving the review process and some comments on Gear's letter.

(1) Rewarding people who are particularly good at refereeing.

This is done to some extent automatically, as such people are invited to join
editorial boards ---which is taken by many to be correlated with the
individual's standing in the corresponding field. Also editors may tend to
be more sympathetic to papers by such individuals in "borderline" cases.
No need for a rule here, just let human nature take care of it.

(2) Make sure to ask.

It is extremely effective to first ASK --by email-- a potential reviewer
if (s)he is willing to do the job. I usually include one or two paragraphs
explaining the topic, say paraphrasing the abstract. The whole process takes
a couple of days, even with overseas referees, and once people agree they find
it hard to later say they don't have the time. (The danger here is that most
people might just-say-no, while presumably it is harder to not do a review
once the paper is on your desktop. But I strongly believe that the opposite
is true; the courtesy is appreciated, and one can always include a flattering
statement in the email plus suitable pleas of desperation...)

(3) Make life EASY for the reviewer.

Always insist that there is no need for the referee to send hardcopy. E-mail
reviews are far easier to do --no need to walk to the printer, to find an
envelope, to type an address, to walk to the mailroom. May seem trivial, but
it can make a tremendous difference for busy people.

(4) Divide work.

Sometimes a superficial browsing is very useful when evaluating a paper,
especially if one can in addition find a detailed-but-narrow reviewer. For
instance, one may have a graduate student check all epsilons and deltas, while
asking someone senior in the field to take a quick look at the paper and give
an evaluation of its relevance, apparent novelty, and so forth.

(5) Be tough with authors.

It is hard to read a badly-written paper. So return to the authors,
unreviewed, a paper that is in that category. It is extremely frustrating to
have to first rewrite a paper for oneself before being able to referee it.

Of course, the above will not solve the problem. The REAL problem is that
people publish too much, especially when there is so little to say. With
electronic distribution of preprints, this is a ridiculous situation. The
causes are many, and we are all to blame, but I think that publishers (and
professional organizations) are partly responsible for the proliferation of

Eduardo Sontag


From: Beresford Parlett <>
Date: Fri, 11 Oct 91 17:12:24 PDT
Subject: More on Breakdown of the Refereeing Process

Breakdown of Refereeing Process

In response to the flurry of messages discussing the publication process in
SIAM, and other journals, I would like to suggest a thought experiment that
puts the publication issue in perspective.

A mechanism that permits the publication of worthy papers in recognized
specialized journals is of benefit to the corresponding technical community.
However no one has said that authorship should be either profitable or

Consider the situation in which salary, but not rank, would be inversely
proportional to publication rate (measured in pages per year). In addition
page charges would not be payable either from contracts or by institutions.
Students, postdoctoral scholars, and less affluent academics would seek
sponsors among their colleagues who would be allowed to pay those charges.

In contrast to the current situation which encourages the dissemination of
information the scheme outlined above would provide an interesting trade off
between material and intellectual rewards. Some discoveries would be worth
some immediate sacrifice and a few years of hardship. Authors judgement
would gradually become more refined. Promotion in rank at universities would
proceed in the usual way.

I suspect that members of SIAM would prefer the present situation with its
inconvenient symptoms rather than one that tried to cure the malady.

Information is toxic except in small doses.

Beresford Parlett


From: Robert Kaplan <>
Date: Fri, 11 Oct 91 09:32:41 -0400
Subject: Journal of Computational Physics Editors

Additional Information Regarding Journal of
Computational Physics

While some of the editors of Journal of Computational
Physics have resigned (Drs. Alder, Bell, Hedstrom, Mirin,
and Marx), the majority of the editors will continue (e.g.,
Drs. Brackbill, Chorin, Glowinski, Johnson, Marcus, Norman,
Osher, Roe, etc.). In conjunction with the scientific
community, Academic Press will name a new editor-in-chief in
the very near future.

As indicated in the previous announcement, authors and
referees should send correspondence concerning manuscripts to

Journal of Computational Physics
Editorial Office
1250 Sixth Avenue
San Diego, CA 92101

Telephone: (619) 699-6467
Fax: (619) 699-6859

In addition, all new manuscripts should be submitted
directly to the above address. The manuscripts will then be
routed to the journal's editors for peer review and
publication consideration. The editorial office in San
Diego has been established to provide secretarial and
operational support for the new editor-in-chief.

Academic Press is also pleased to make the following

1) Effective immediately, we will accept electronic
manuscripts, including TeX and LaTeX. Contact the San
Diego office for details.

2) Starting in 1992, the journal will be available at a
reduced subscription rate for individuals subscribing
to the journal for personal use. In 1992, the personal
rate will be $100 for subscribers in North America.
Contact the San Diego office for details.

3) The backlog of accepted papers that the journal has
experienced in recent years will be eliminated in the
coming months. This backlog will be published during
1992, at no extra cost to libraries. In the future,
the journal will review, accept, and publish
manuscripts in a timeframe similar to the other main
journals of physics and computation.

Academic Press appreciates that Journal of Computational
Physics is the premier journal of the field and we are
committed to its continuing excellence. We believe the
editorial transition that is currently underway will result
in a journal that is better able to face the challenges of
the future. As always, it is our aim to continue to publish
the journal in the manner that best serves the scientific

We very much welcome suggestions and advice about the future
direction of the journal. Please contact the Academic Press
Senior Editor for Physical Science, Bob Kaplan at

Academic Press
955 Massachusetts Avenue
Cambridge, MA 02139

Telephone: (617) 876-3907
Fax: (617) 661-3608

with your comments.


Date: Wed, 9 Oct 91 17:21 EDT
Subject: SIAM Conference on Optimization

Fourth SIAM Conference on Optimization
Sponsored by SIAM Activity Group on Optimization

May 11-13, 1992
Hyatt Regency Hotel
Chicago, Illinois

Invited Presentations

Convex Optimization Problems Arising in Controller Design
Stephen P. Boyd, Stanford University

Development of Codes for Large-Scale LP, QP and NLP
Roger Fletcher, University of Dundee, Scotland

Algorithms for Solving Large Nonlinear Optimization Problems
Nicholas I. M. Gould, Rutherford Appleton Laboratory, United Kingdom

Cheap Gradients and Beyond: Algorithmic Differentiation in Optimization
Andreas Griewank, Argonne National Laboratory

Solving Large-Scale Network Problems
Rutgers University

Optimization and Stability in the Scheduling of Manufacturing Systems
P.R. Kumar, University of Illinois, Urbana

Recent Progress in Network Flows
Eva Tardos, Cornell University

Recent Progress on Interior Point Methods
Michael J. Todd, Cornell University

Interior Point Methods for Large-Scale Nonlinear Optimization Problems
Margaret H. Wright, AT & T Bell Laboratories

(Partial list)

Recent Theoretical Advances in Interior Point Methods
Kurt Anstreicher, Yale University

Combinatorial Optimization
Francisco Barahona, IBM Thomas J. Watson Research Center

Optimization Problems in Chemical Engineering
Lorenz Biegler, Carnegie-Mellon University

Constrained Nonlinear Optimization
Richard H. Byrd, University of Colorado, Boulder

Advanced Interfaces for Optimization Software
Robert Fourer, Northwestern University

Large-Scale Nonlinear Optimization
Philip R. Gill, University of California, San Diego

Optimization in Control and Differential Equations
Carl T. Kelley, North Carolina State University

Recent Computational Advances in Interior Point Methods
Sanjay Mehrotra, Northwestern University

Network Flow Algorithms
James B. Orlin, Massachusetts Institute of Technology

Parallel Algorithms in Optimization
Stephen Wright, Argonne National Laboratory

For further information regarding this conference, please contact
SIAM Conference Department, 3600 University City Science Center,
Philadelphia, PA 19104-2688; telephone: 215-382-9800; fax: 215-386-7999;


From: Roy Mathias <>
Date: Thu, 10 Oct 91 14:47:09 CDT
Subject: Workshop in Numerical Ranges and Radii

Preliminary Announcement

A workshop on
``Numerical Ranges and Numerical Radii''

August 10 (Monday) - August 15 (Saturday), 1992.
Department of Mathematics
The College of William and Mary

The aim of the proposed workshop is to bring researchers
on numerical ranges and numerical radii from different
(research and geographic) areas together to exchange
ideas on the subject. In particular, there are four
primary objectives for the workshop.

(a) To study and further explore applications of various
kinds of generalized numerical ranges and numerical
radii in different branches of science.
(b) To discuss existing mathematical tools and techniques
and try to generate new methods to handle problems on
numerical ranges and numerical radii.
(c) To discuss possible research projects or computer projects
on numerical ranges and numerical radii appropriate for the
undergraduate or graduate level.
(d) To exchange research problems, ideas and experience on the

While the main theme of the workshop is on numerical ranges and
numerical radii, discussion will be focused on the relations
and applications of the subject to several specific topics.

Day 1. Operator theory and C*-algebras
Day 2. Norms and Matrix inequalities
Day 3. Decomposable numerical ranges and Quantum physics
Day 4. Systems theory and Computer generation of numerical ranges
Day 5. Location of eigenvalues
Day 6. Other related subjects such as completion problems, linear
preserver problems, etc.

For more information concerning the workshop please contact

Dr. Chi-Kwong Li
Department of Mathematics
The College of William and Mary
Williamsburg, VA 23187 USA

tel: (804) 221-2042

Support for this workshop comes from the The College of William
and Mary. Support funds are being sought for workshop participants
and more details will be known by the end of March, 1992.

Note: Persons from North America planning to attend the second
ILAS meeting at Lisbon, August 3 - 7, 1992, may be able
to include a stop at Williamsburg on their reutrn tickets.


End of NA Digest