NA Digest Sunday, July 28, 1991 Volume 91 : Issue 30

Today's Editor: Cleve Moler

Today's Topics:

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to


From: Sven Hammarling <>
Date: Mon, 22 Jul 91 12:40 GMT
Subject: Wilkinson Prize to Petzold and to Tang

The Wilkinson Prize for Numerical Software is awarded in honour of the
outstanding contributions of James Hardy Wilkinson to the field of numerical
software. The award is presented to the entry that best addresses all phases
of the preparation of high quality numerical software. The prize is sponsored
by Argonne National Laboratory, The National Physical Laboratory and the
Numerical Algorithms Group.

It is a great pleasure to announce that the first Wilkinson Prize has been
awarded to Linda Petzold for the DASSL package, which was judged to be a
significant contribution to numerical software in general and to the solution
of differential/algebraic equations in particular. The prize was presented
to Linda by Heather Wilkinson at the ICIAM meeting in Washington. Many
congratulations to Linda.

The entry by Peter Tang, on software for the elementary functions in Ada, was
also judged to be of outstanding merit and was given a "most honourable
mention". The overall standard of the thirteen entries was extremely high and
was a fitting tribute to Jim Wilkinson.

Potential entrants for the second Wilkinson prize, to be awarded in three
years time, should contact Jorge More' at ANL ( or
Brian Ford at NAG Ltd ( for further details.

Sven Hammarling.


From: Ahmed Sameh <>
Date: Mon, 22 Jul 91 08:36:42 CDT
Subject: Change of Address for Ahmed Sameh

Effective August 16 my new address is:

Ahmed Sameh
Department of Computer Science
University of Minnesota
200 Union Street S.E.
Minneapolis, MN 55455

Telephone: (612) 625-0726
FAX : (612) 625-0572
e-mail :


From: R. J. LeVeque <>
Date: Fri, 26 Jul 1991 16:51:10 +0200
Subject: Restoration of Addresses for Loyce Adams and Randy LeVeque

As of the early August, we will be back in Seattle. Our addresses there are:

Loyce Adams
Applied Mathematics, FS-20
University of Washington
Seattle, WA 98195

phone: (206) 543-5077
department: (206) 543-5493
fax: (206) 685-1440

Randall LeVeque OR
Applied Mathematics, FS-20 Mathematics, GN-50
University of Washington University of Washington
Seattle, WA 98195 Seattle, WA 98195

phone: (206) 685-3037 phone: (206) 543-7898
department: (206) 543-5493 department: (206) 543-1150
fax: (206) 685-1440 fax: (206) 543-9285


From: Kumbukage Pradeep Seneviratne <>
Date: Thu, 25 Jul 91 17:33:14 -0400
Subject: SVD for Banded Matrices

I need to find the singular value decomposition of some banded matrices.
Since the matrices are very large I would like to know whether there are
routines which computes the SVD of a matrix utilizing the fact that the
particular matrix is banded. Please let me know.

Thank you.
Kumbukage Pradeep Seneviratne


From: John Prentice <>
Date: Tue, 23 Jul 91 12:47:16 MDT
Subject: Multigrid and CFD

I am formulating a proposal to model a chemical vapor deposition
reactor. This requires solving the Navier-Stokes equations with
heat conduction, real gas equations of state, and reactive chemistry
in three dimensions on an irregular grid. The goal of the work is to
develop methods for modeling these reactors which are computationally
less expensive than current methods (conventional finite difference
or finite volume techniques). One thought is to speed the solution
of the large linear systems that have to be solved. Another is to
decouple the reactive chemistry and solve that part of the problem
on its own mesh using a different time step than is used for the
transport part of the problem.

Along the lines of the first thought, to speed the solution of the
linear system, I have been wondering to what extent and how successfully
people have been able to solve CFD type problems using multigrid
techniques. Our problem does not involve shocks, so the solutions
are at least continuous functions. However, there are large convective
gradients in the flow and very large thermal gradients. So there is
likely to be significant high frequency components to the solution. On
the face of it, I would assume this makes multigrid a poor choice.
However, I have had some private conversations with other local
researchers who are working on CFD problems and who have indicated
that this may not be the case. I would be interested in finding out
if anyone has looked at such problems and could point me at some
literature or at people to talk to. I would also entertain the idea of
collaborating on this research if anyone has some clever ideas.

John Prentice
Quetzl Consulting
Albuquerque, New Mexico


From: John Prentice <>
Date: Wed, 24 Jul 91 01:38:04 MDT
Subject: Need Access to a 64k processor CM2

I am interested in running some quantum scattering calculations on
the CM2. I have access to the one at NCSA and the one at the University
of Maryland. These machines are fine for my smaller calculations and
for development, but ultimately I need access to a CM2 with the full
complement of processors and memory. The computationally intensive
part of my problem is solving a complex valued dense linear system.
I would appreciate any information of where and how to get access to
such a system.


John K. Prentice (Internet)
Department of Physics and Astronomy
University of New Mexico


From: American Mathematical Society <>
Date: Tue, 23 Jul 91 10:45:19 EDT
Subject: e-MATH Available from AMS


e-MATH is a node on the Internet providing mathematicians with the
ability to electronically communicate with a central information
source. Partially funded by a three-year National Science Foundation
grant, e-MATH consists of a dedicated computer facility and associated
software maintained and supported by American Mathematical Society
staff members. It is intended that e-MATH become an
electronically-based clearing house for timely professional and
research information in the mathematical sciences.

e-MATH currently offers a searchable, on-line Combined Membership
List, a professional register for employment and postdoctoral
opportunities, a repository of AMS-supported TeX software, the
Mathematical Reviews classification scheme, and more. Planned future
services include the on-line electronic distribution of the AMS
Bulletin, two conferencing systems that would allow moderated and
unmoderated conferences, and a prototype electronic journal in the
mathematical sciences.

To successfully connect to e-MATH, you need:

o a connection to an INTERNET host;
o VT100 connectivity in communications software and host operating system;
o terminal tabs set at every eight columns.

To access e-MATH, enter the command:


or telnet Login and password are e-math. For further
information or help, send e-mail to:


Date: Thu, 25 Jul 91 11:57 EDT
Subject: SIAM Conference on Applied Linear Algebra

Fourth SIAM Conference on Applied Linear Algebra

September 11-14, 1991

Radisson Hotel Metrodome
Minneapolis, Minnesota

Sponsored by SIAM Activity Group on Linear Algebra

Conference Themes

Parallel Matrix Computations
Direct Methods for Large and Sparse Systems
Interior Point Methods in Mathematical Programming
Rational Matrix Functions in Systems Theory
Core Linear Algebra
Numerical Methods for Markov Chains
Signal Processing and Wavelets
Linear Algebra in Statistics

Deadline Dates

Hotel Reservations
August 23, 1991

Advance Conference Registration
September 4, 1991

Organizing Committee
Richard A. Brualdi, Chair
University of Wisconsin, Madison
David H. Carlson
San Diego State University
George Cybenko
University of Illinois, Urbana-Champaign
J. Alan George
University of Waterloo, Canada
Gene H. Golub
Stanford University
Roger A. Horn
The Johns Hopkins University
John G. Lewis
Boeing Computer Services
Mitchell B. Luskin
University of Minnesota, Minneapolis
Salvatore D. Morgera
McGill University, Canada
George P.H. Styan
McGill University, Canada
Paul Van Dooren
Philips Research Laboratory, Belgium

For a complete preliminary program, please contact SIAM,
( or write: SIAM Conference
Department, 3600 University City Science Center, Philadelphia, PA
19104-2688; telephone: 215-382-9800; fax: 215-386-7999.


Date: Wed, 24 Jul 91 17:31:19 MEZ
Subject: Course in Germany on Simulation

CCG-Course Announcement

The Carl-Cranz Gesellschaft, a nonprofit society for engineering and
scientific education is organizing a course on

Numerical Integration Methods for the Simulation of Constrained and
Unconstrained Mechanical Systems

in Oberpfaffenhofen (near Munich), Germany, Sept. 30 - Oct. 2, 1991.

The course outline:
G.Bock, Augsburg - Concept of Modern Simulation Tools
- Introduction to Numerical Integration Methods
H.Pesch, Munich - One Step Methods
P.Rentrop, Munich - Modern Implementations of Runge-Kutta and
Multistep Methods
- Stiff ODEs
E.Eich, Augsburg - Numerical Treatment of ODEs with Discontinuities
(with special emphasis on typical discontinuities
occuring in mechanical systems)
P.Rentrop, Munich - Differential-Algebraic Systems (DAEs)
C.Fuehrer, Wessling - Mechanical Systems in DAE Formulation and their
Numerical Treatment
E.Eich/C.Fuehrer - Numerical Exercises with DASSL and ODASSL

The course addresses engineers and mathematicians from research and
development departments in industry and research institutions involved
in the dynamic analysis of complex mechanical systems such as vehicles
of all kinds, robots, mechanisms.
It is also designed for faculty members, research
associates, and graduate students specializing in the analysis and
simulation of dynamic systems.

Participation Fee: DM 1.260,- including detailed manuscripts
(Members of German Universities may get a reduction on request)

For more information, please contact:
Claus Fuehrer, Inst. Dynamik der Flugsysteme, DLR, D-8031 Wessling



From: Richard Sincovec <>
Date: Tue, 23 Jul 91 09:14:57 -0400
Subject: Ada on Super-Computers Workshop

TRI-ADA '91, San Jose, CA, October 21, 1991

Call for Participation and Contribution

The top priority of users of supercomputers is performance. This
workshop focuses on the ways that ADA can address this priority.

Super-computers are becoming more wide spread because of inexpensive
microprocessor technology and innovative computer architectures. These
architectures rely extensively on parallelism in many forms and from
fine grain to coarse grain. Ada with its numeric processing, tasking
and generic abstraction features has the potential to address the
growing problem of programming super-computers effectively.

The aim of this workshop is to bring together interested
super-computer users, Ada technologists, compiler vendors, computer
manufacturers, and Ada 9X Mapping/Revision participants to address the
following issues, each in a 1 and 1/2 hour session of panel and
audience debate and discussion:

1) Users Needs

Currently, FORTRAN is the lingua franca of super-computer users. What
steps are needed to make Ada the language of choice for users of

2) Compiler Technology

Super-computer architectures require sophisticated compilers to
achieve effective performance. Ada, for performance reasons, also
requires a sophisticated compiler. How can both be accommodated?

3) Language Issues

In some instances, the Ada Language Reference Manual restricts the use
of parallel processing. Also, scientific users have specialized needs
in handling floating pointing representation, storage and operation.
What are the critical issues and what steps can be made to accommodate
such issues now and in the Ada 9X revision process?

4) Wrap Up

The results of the workshop (lists of issues, proposed solutions and
recommendations) will be summarized and published in Ada Letters.

If you are doing work related to any of these areas and would like to
bring your expertise and point of view to bear on the above questions
as a panelist, then we invite you to submit a short description (1-3
pages) of what your interests are and in which session you are
interested in participating. Short position statements from invited
panelists will start each session and then the floor will be opened by
a moderator for debate and discussion with the panel and the audience.
The dead-line for submitting a position brief as a speaker/panelist is
August 8 1991. Selection of panelists will be made by August 25,
1991. The audience attendance is limited to about 75 persons so
please contact me if you are interested in attending.

Richard Sincovec
P.O. Box 2008, Bldg 6012
Oak Ridge, TN 37831-6367
FAX: (615)574-0680


Date: Wed, 24 Jul 91 09:43 EDT
Subject: Contents: SIAM J. Math. Anal.

SIAM J. Math. Anal.
November 1991, Vol. 22, Number 6

Ca( ) Solutions of a Class of Nonlinear Degenerate Elliptic Systems Arising
in the Thermistor Problem
Hong Xie and W. Allegretto

Existence and Multiplicity of Positive Radial Solutions for Semilinear
Elliptic Equations in Annular Domains
Song-Sun Lin and Feng-Ming Pai

Partial Regularity in Problems Motivated by Nonlinear Elasticity
Nicola Fusco and and J. Hutchinson

Period Doubling with Higher-Order Degeneracies
Bruce B. Peckham and Ioannis G. Kevrekidis

Variational Principles Without Definiteness Conditions
Paul Binding and Qiang Ye

Persistence of Invariant Tori in Systems of Coupled Oscillators.
II. Degenerate Problems
Masaji Watanabe and Hans G. Othmer

The Existence of Infinitely Many Traveling Front and Back Waves in the FitzHugh-Nagumo Equations
Bo Deng

Stability Analysis of Stationary Solutions of Bistable Reaction-Variable
Diffusion Systems
Hideo Ikeda and Masayasu Mimura

Integral Equation Methods in a Quasi-Periodic Diffraction Problem for the
Time-Harmonic Maxwell's Equations
J. C. Nedelec and F. Starling

Diffusion of Fluid in a Fissured Medium with Microstructure
R. E. Showalter and N. J. Walkington

On Parabolic Volterra Equations in Several Space Dimensions
Hong-Ming Yin

Analyticity of Solutions of the Koreteweg-De Vries Equation
Nakao Hayashi

Geometry of Rational Functions and Nonlinear Integrable Systems
Yoshimasa Nakamura

The Interior Transmission Problem and Inverse Scattering from
Inhomogeneous Media
B. P. Rynne and B. D. Sleeman

On the Inverse of the Discrete Two-Dimensional Wave Operator at CFL
Robert Glassey, Ernst Horst, Andrew Lenard, and Jack Schaeffer

Approximation by Piecewise Exponentials
Junjiang Lei and Rong-Qing Jia

On the Determination of Ziglin Monodromy Groups
Richard C. Churchill and David L. Rod


End of NA Digest