NA Digest Saturday, February 26, 1994 Volume 94 : Issue 9

Today's Editor:
Jack Dongarra
University of Tennessee / Oak Ridge National Labaotory
Stan Green
University of Tennessee

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to


From: Joseph Traub <>
Date: Sun, 20 Feb 94 14:13:17 EST
Subject: Contents, Volume 10, #1, Journal Of Complexity


Table of Contents
March 1994

Regular Papers

1. Is it Possible to Know a Problem Instance is Ill-Posed?
Author: J. Renegar

2. Bounds on the number of units for computing arbitrary
dichotomies by multilayer perceptions
Author: M. Cosnard, P. Koiran, and H. Paugam-Moisy

3. Parallel Speed-Up of Monte Carlo Methods for Global
Author: R. Shonkwiler and E. Van Vleck

4. Tractability and Strong Tractability of Linear Multivariate
Author: H. Wozniakowski

Festschrift Papers

1. Average Saving Effects in Enumerative Methods for Solving
Author: K. Borgwardt and J. Brzank

2. A Multi-step Algorithm for Hankel Matrices
Author: A. W. Bojanczyk and G. Heinig

3. Learning from Hints
Author: Y. S. Abu-Mostafa


From: Xiaozhuo Yang <>
Date: Sun, 20 Feb 94 14:45:42 EST
Subject: Numerical Analysis Question

I am looking for an answer to the following question.
Let p(x) = x^{n} + a_{n-1} x^{n-1} + ... + a_{0} be a nomic polynomial
in the complex plan. Let r_{1},r_{2}, ..., r_{m} are all the distinct
roots of p(x). Let d be the smallest distance between the m roots. The
question is if it is possible to estimate d or give a "good" lower bound
ONLY by the coefficients of p(x), a_{n-1},...., a_{0}.

Thank you very much for response.

Yang Xiaozhuo


From: Tom Scavo <>
Date: Sun, 20 Feb 94 11:53:03 -0800
Subject: Re: A Meta-answer

In NA-digest Vol 94, Issue 7, Arne Raeithel writes:
Arne> What kind of "digest" is this, ... na.digest is nothing but a medium
Arne> to post problems, ...

In NA Digest, V. 94, # 8, Bill Gear <> writes:
Bill> Good -- I like it that way. The one question I posed to this audience
Bill> brought in several answers, some of which were very useful to me. One
Bill> person even mailed me a collection of (paper) papers. I doubt that
Bill> many other NA-digest readers would have been interested in these
Bill> details, (but a couple who were did email me asking to pass on any
Bill> good answers I got to my question).

I agree with Arne. An interactive forum along the lines of the
Usenet newsgroup sci.math.num-analysis would be more productive.
In fact, I think NA Digest and sci.math.num-analysis would
benefit greatly from each other. I know that some mailing lists
are automatically gated into the appropriate newsgroup. Maybe
that could be done in the case of NA Digest. I, for one, would
welcome the change.

Just look how long it's taken to get these three related messages
to the list. In a newsgroup, it's not uncommon for a thread to
generate the same reponse in a matter of hours.

Tom Scavo


Date: Sun, 20 Feb 94 22:17:43 +0100
Subject: FTP site at UTIA

The ftp address of the Institute of Information Theory and Automation
(UTIA) of the Academy of Sciences of the Czech Republic, Prague has
been changed to:

This site contains data and documents from the Institute and from the
Academy of Sciences. Some of the directories that may be found at this
site are:

Technical reports of the Institute. Included in this directory is a
postscript file which contains all of the abstracts of this report

Additional technical reports of the Adaptive Control group

Documents and data of the Subcommission for Computer Networks of the
Academy of Sciences

Data offered to the public by the Institute staff

Archives of public mailing lists served by the

* Institute of Information Theory and Automation (UTIA) *
* Academy of Sciences of the Czech Republic, Prague *
* Mail: *
* UTIA AV CR Tel: +42 2 6605-2422 *
* Pod vodarenskou vezi 4 Fax: +42 2 6641-4903 *
* 182 08 Praha 8 E-mail: *
* Czech Republic *


From: Simon Chamlian <>
Date: Sun, 20 Feb 94 17:40:39 EST
Subject: Eigenvalues of a Toeplitz matrix

I have a 2000x2000 symmetric Toeplitz matrix that needs to be inverted
and the eigenvalues and eigenvectors calculated.

Does anyone know the existence of subroutines that calculate the inverse,
eigenvalues and eigenvectors by taking advantage of the Toeplitz
structure of the matrix?
Do Linpack and/or Lapack have such routines?

Any comments will be appreciated.
Thanks in advance.



From: Alan J. Laub <>
Date: Sun, 20 Feb 94 15:01:41 PST
Subject: Faculty Opening in Scientific Computation at UCSB

University of California, Santa Barbara
College of Engineering

Faculty Opening in Scientific Computation

The College of Engineering at the University of
California, Santa Barbara is in the process of
developing an interdisciplinary graduate program
aimed at education and research at the intersection
between applied mathematics, numerical analysis,
computer science, and the application of computational
methods to the solution of problems in science and
engineering. The College is seeking a faculty member
to develop and lead the program. The level of the
position is open and multi-departmental appointments
are possible. Senior candidates should have
established an international reputation in an area
within the broad framework of the program. Junior
candidates should have outstanding research potential.

Applicants should send their resumes and the names and
addresses of at least four professional references to:
Search Committee in Scientific Computation
Dean's Office, College of Engineering
Engineering I, Rm. 1016
University of California
Santa Barbara, CA 93106
Applications and nominations will be received until
the position is filled. UCSB is an equal opportunity
affirmative action employer.


From: Christian Brechbuehler <>
Date: Mon, 21 Feb 1994 12:35:37 --100
Subject: A Matrix Eigenvalue Problem Re: NA Digest, V. 94, # 8

On Wed, 16 Feb 94 09:25 CST, Richard Luczak <> asked
in NA Digest, V. 94, # 8:

> Let A be a tridiagonal matrix of the form
> -2 2
> 1 -2 1
> . . .
> 1 -2 1
> 2 -2
> What is the formula for all eigenvalues of matrix A?

Let A be n by n with n > 1.
The following is a conjecture, verified only for n=2, n=5 and n=32.

Eigenvector j (not normalized) has the components
cos(i j Pi/(n-1)),
where 0 <= j < n
0 <= i < n.

Arbitrarily picking the first row of A, we have the Eigenvalues
2(cos(j Pi/(n-1)) - 1).
They are in descending order for increasing j. The largest Eigenvalue
is 0, the smallest is -4, independent of n.

Christian Brechbuehler

Communication Technology Lab, Image Science Group
Federal Institute of Technology (ETH) Zurich, Switzerland


From: Iain Duff <>
Date: Mon, 21 Feb 94 12:15:32 GMT
Subject: Contents, IMAJ Numerical Analysis Volume 14 Number 2. April 1994

Allgower E, Bohmer K, and Zhen M
Branch switching at a corank-4 bifurcation point of semi-linear
elliptic problems with symmetry.

van Dorsselaer J L M and Spijker M N
The error committed by stopping the Newton iteration in the numerical
solution of stiff initial value problems.

Chiang Y-L F
Properties of optimal schemes for linear 1D PDE initial-value
hyperbolic problems with variable coefficients.

Lopez-Marcos M A
Numerical analysis of pseudospectral methods for the
Kuramoto-Sivashinsky equation.

Chen Z and Hoffmann K-H
An error estimate for a finite-element scheme for a phase field model.

Miller J J H and Wang S
A new non-conforming Petrov-Galerkin finite-element method with
triangular elements for a singularly perturbed advection-diffusion

Priestley A
The positive and nearly conservative Lagrange-Galerkin method.

Potra F A and Venturino E
Low-order methods for Cauchy principal value integrals with endpoint


Date: 21 Feb 1994 11:25:03 -0700 (PDT)
Subject: Mathematics Modeling Workshop for Graduate Students at Claremont

Mathematics Modeling Workshop
for Graduate Students
June 6-14,1994

The Claremont Colleges will host this Workshop as a continuation of the IMA
Summer Program in Mathematical Modeling held August, 1992, and the
Workshop held at Claremont in 1993. See SIAM News, Nov., 1993,
for a description of the latter Workshop. Funding has been
provided by the National Security Agency. The goals are the same as

(1) To expose 30 mathematics and statistics graduate students
to the excitement of approaching real world problems.

(2) To introduce students to the team approach to problem solving.

In the Workshop the students will be divided into six teams to work on
problems brought by experienced modelers. These problems are derived
from industry or applied science, and require new a fresh insight for
their definition and solution. The problem presenters are being
recruited and their names will be released subsequently.

Note: During June 1-4 an International Conference on Differential
Equations and Applications to Biology and Industry will be held in
Claremont. This Conference pays tribute to Stavros Busenberg who died
last year. Main speakers in Industrial Mathematics are Avner Friedman
and John Ockendon. Students accepted to the Workshop may be interested
in attending this Conference (the Industrial Mathematics portion of the
Conference will be June 3-4), and applications to attend both will
be accepted.

Application Procedure. Graduate students in mathematics, applied mathematics,
statistics or operations research can be nominated for this program by a
faculty member by sending a letter of recommendation. The student must
provide a copy of a recent transcript.

The deadline for applications is May 2.

Students will be expected to finance their travel, although some travel
funds may become available for the students attending the Busenberg
conference. The workshop will cover local living expenses for U.S. citizens
and permanent residents.

All correspondence should be sent to:
MATHEMATICS DEPT., CLAREMONT, CA 91711-3988. Phone 909-621-8080,
FAX 909-621-8390, E-Mail:


From: Mike Boucher <>
Date: Mon, 21 Feb 1994 18:51:33 -0800
Subject: Parallel matrix multiplication

In the last issue of this newsletter, Steve Huss-Lederman mentioned a number
of parallel matrix multiplication subroutines available in the public
domain. A library of parallel linear algebra subprograms that includes
parallel BLAS, LINPACK, and LAPACK is available commercially for SPARC from
DSS, Inc. It is tuned for SPARC and gets close to peak speeds for SPARC,
SuperSPARC, and microSPARC. More information is available from:


From: William B. Sawyer <>
Date: Tue, 22 Feb 1994 10:14:23 --100
Subject: Summer Student Internship Program At CSCS


This year, the Swiss Scientific Computing Center CSCS-ETH, in
collaboration with other Swiss institutions, will organize a summer
course for undergraduate students in Manno (TI), Switzerland. CSCS'
Summer Student Internship Program (SSIP) is embedded in the Parallel
Computer Architectures Education Laboratory (ParEdLab), part of CSCS'
Section of Research and Development (SeRD).

During the summer break, for a duration of thirteen weeks between July
and October, up to eight students will be taught basic skills in
programming Distributed Memory Parallel Processors (DMPPs). Each
student is also expected to complete a small research project. The
topics for these projects will deal with today's well-known
programming problems of DMPPs, e.g. development of programming tools,
new algorithms, and scientific applications for such machines.

Student profile

Students participating in the summer student internship program should
have good programming knowledge in either C or Fortran, but no advance
knowledge in parallel systems and parallel programming is required.
The internship program is mainly targeted for Computer Science and
Electrical Engineering students, but also students from other, more
application-oriented fields, e.g. Mechanical Engineering or Material
Sciences, as well as from Mathematics and Physics, can and should
apply. Candidates should have completed their Bachelor of Science
degree, or, in Europe, have passed their intermediate exams at the
university level.


The course will commence on July 18, and end on October 14, 1994. It
will start with a two week introduction to parallel distributed
systems, covering both hardware/system issues and programming of
DMPPs. The lessons will also include exercises on the platform to be
used by the students later on for their own ten-week project.

Each student will give two oral presentations on her/his project --
one intermediate presentation after 5 weeks, and a final presentation
after the project's completion. Each student must summarize her/his
achievements in a final technical report before the SSIP completes.

Advisory scheme

Each student will be assigned an advisor who will assist throughout
her/his stay at CSCS. Each advisor will supervise not more than two


CSCS will provide all technical infrastructure, e.g. office space,
workstations and parallel distributed development platforms, such as a
Cenju system supplied by NEC as part of the Joint CSCS-ETH/NEC
Collaboration in Parallel Processing, a Meiko CS-1/860, as well as a a
Hewlett-Packard and Sun workstation clusters.

Financial considerations

CSCS will cover local expenses of all students participating in the
summer student internship program. Support for travel costs is


A wide range of project topics, from the implementation of development
tool components to the development of parallel applications using High
Performance Fortran (HPF) and other modern programming paradigms, is
available. A partial list of the project topics will be supplied with
the application materials.

Application process

You can request application materials by sending an e-mail to Will
Sawyer,, with subject line "SSIP'94 application
request". Complete instructions will be supplied. The deadline for
requesting application materials is March 31, 1994. Completed
applications must be received by April 15, 1994 for consideration.


From: K. Borovkov <>
Date: Tue, 22 Feb MSK 12:33:18 +0300
Subject: Rus-Eng Eng-Rus Dict. on Probability, Statistics, and Combinatorics Available

I have recently published a book with SIAM,
entitled Russian-English English-Russian Dictionary on
Probability, Statistics, and Combinatorics, published in
January 1994. It consists of approximately 216 pages,
and is a softcover volume. Those of you who are interested
in this field may find my book very helpful. Prepayment is
required; shipping charge will apply.

Please feel free to contact SIAM for further information:
SIAM, 3600 University City Science Center,
Philadelphia, PA 19104-2688; telephone:
215-382-9800; fax: 215-386-7999;



From: Ramon Harps <>
Date: Tue, 22 Feb 1994 15:28:32 +0100
Subject: Announcement of ICIAM95


July 3 - 7, 1995


INVITED SPEAKERS: H. Akaike, Japan; V.I. Arnold, Russia;
F. Bacelli, France; M.V. Berry, UK; H. Brezis, France;
C. Cercignani, France; I. Daubechies, USA; P. Degond, France;
H. Foellmer, Germany; K. Hasselmann, Germany; E.J. Hinch, UK;
R. James, USA; W. Kahan, USA; J.B. Keller, USA; J.-L. Lions, France;
J. Marsden, USA; G. Meyer, USA; D. Mumford, USA;
K. Murota, Japan; H. Neunzert, Germany; L.G. Nilsson, Sweden;
A. Perelson, USA; C. Peskin, USA; F. Pfeiffer, Germany;
A. Quarteroni, Italy; R. Rannacher, Germany; G. Sivashinski, Israel;
A. Tayler, UK; L. Trefethen, USA; P. Van Dooren, USA; M. Ward, Canada.

CONFERENCE TOPICS: The programme consists of invited and
contributed lectures, minisymposia, poster presentations,
and an exhibition.
The presentations are solicited in all areas of applied
mathematics, computer science, applied probability and
statistics, scientific computing, and applications in science,
medicine, engineering, economics and other related fields.

to submit a paper, which you may present in lecture or poster
format. Authors will have approximately 15 minutes for the lecture,
with additional 5 minutes for discussion. Alternatively they may
elect poster presentations, which allow interactive discussions
with individuals interested in their work.

MINISYMPOSIA: A minisymposium is a session of speakers focusing on
a single topic. The organizer of a minisymposium invites the
speakers and decides on the topic they are to deliver.

Persons who would like to present a paper (lecture or poster format)
or would like to organize a minisymposium are encouraged to submit
a proposal on an appropriate ICIAM 95 form.

AUGUST 31, 1994

INFORMATION: Forms, guidelines and further information are
available at

Univ. Regensburg, NWF I - Mathematik
D-93053 Regensburg, Germany

Phone: +49-941-943-4918, Fax: +49-941-943-4005



From: Niels Houbak <>
Date: Tue, 22 Feb 94 15:33:14 DNT
Subject: Call for papers for CISS'94, ETH, Zurich

C A L L F O R P A P E R S .

CISS - First Joint Conference of International Simulation Societies
August 22 - 25,1994 , ETH Zurich, Zurich , Switzerland


1: Simulation Methodology and Tools
2: Simulation of Energy and Environmental Systems
3: Simulation of Multibody Systems
4: Simulators, Methods, and Applications of Real Time Simulation
5: Simulation by Means of Vector-, Parallel, and Neurological Computers
6: Simulation in Design and Manufacturing
7: Simulation in Biology, Ecology and Medicine
8: Simulation using Bond-Graphs
9: Building Simulation

Jurgen Halin, Walter Karplus,
ETH Zurich, Switzerland University of California
Institute of Energy Technology 3732 Boelter Hall, Los Angeles
Clausiusstrasse 33, CH-8092 Zurich CA 90024, USA
Tel.: +41-1-632-4608, Tel.: 001-310-8252929
Fax.: +41-1-262-2158 Fax: 001-310-8252273
E-mail:Halin@czheth5a.bitnet E-mail:

I. Bausch-Gall (D); L. Birta (CDN); Bo Hu Li (TJ); P. Born (F);
F. Breitenecker (A) F.E. Cellier(USA) R. Crosbie (USA); L. Dekker (NL);
J.M.Giron Sierra(E) F. Hauser (CS); R. Huntsinger(USA) G. Iazeolla (I);
A. Javor (H); K. Juslin (SF); E. Kerckhoffs(NL); P. Kropf (CH);
W. Krug (D); A. Lehmann (D); F. Lorenz (B); P. Lorenz (D);
H.J. Luthi (CH); F. Maceri (I); I. Molnar (H); E. Mosekilde (DK);
A. Pave (F); B. Quatember (A); E. Rothauser (CH); M. Rusnak (SR);
S.J. Sapirie (CH); M. Savastano (I); B. Schmidt (D); G. Vansteenkiste (B);

Extended abstracts of the papers (two pages typewritten without drawings
and tables) are due to arrive IN QUADRUPLICATE at the office of the
General-Conference-Chairman, Jurgen Halin, before March 15, 1994.
Please mention in your abstract the originality of your contribution.

Only original papers, written in English, which have not previously been
published elsewhere will be accepted. In case you want to organize a panel
discussion, please contact the General-Conference-Chairman.

Each abstract will be reviewed by at least three members of the International
Program Committee. The notification of acceptance or rejection will be
sent by March 31st, 1994.

The abstract must also contain the following information:
1. The title of the paper, the topics, and the number of the conference theme.
2. Name, occupation and/or title, and complete mailing address.
3. Telephone and/or fax number.
4. Eventually E-mail address.

Authors Member of Soc. Other Particip.
Before June 15, 94 sfr 675,00 sfr 675,00 sfr 750,00
After June 15,94 sfr 725,00 sfr 850,00
On request a limited number of participants from Eastern European countries
and students can get a reduction of the registration fee.


Date: Tue, 22 Feb 94 14:01:11 EST
Subject: SIAM Journal on Applied Math - Table of Contents 54-3


The Dynamics of a Tippe Top
A. C. Or

Blow-Up in a System of Partial Differential Equations with Conserved First
Integral. Part II: Problems with Convection
C. J. Budd, J. W. Dold, and A. M. Stuart

Large-Scale Averaging Analysis of Single Phase Flow in Fractured Reservoirs
Zhangxin Chen

Shortcomings of Existing Finite Element Formulations for Subsurface Water
Pollution Modeling and Its Rectification: One-Dimensional Case
M. G. Rabbani and James W. Warner

Direct-Formulation Finite Element (DFFE) Method for Groundwater Flow Modeling:
Two-Dimensional Case
M. Ghulam Rabbani

Stability of Nonlinear Periodic Internal Waves in a Deep Stratified Fluid
M. D. Spector and T. Miloh

The Effect of Microstructure on Elastic-Plastic Models
Lianjun An and Anthony Peirce

Estimations for Nonsymmetric Effective Coefficients
Gelu I. Pasa

The Computation of One-Parameter Families of Bifurcating Elastic Surfaces
Frank E. Baginski

Reflection of Localized Beams From a Nonlinear Absorbing Interface
J. A. Powell, E. M. Wright, and J. V. Moloney

Stable Solutions for a Catalytic Converter
Helen Byrne and John Norbury

Fast Subsystem Bifurcations in a Slowly Varying Lienard System Exhibiting
M. Pernarowski

Fixed-Point Smoothing of Scalar Diffusions I: An Asymptotically Optimal Smoother
Y. Steinberg, B. Z. Bobrovsky, and Z. Schuss

A Processor-Shared Queue That Models Switching Times: Heavy Usage Asymptotics
Charles Knessl and Charles Tier


From: John R. Gilbert <>
Date: Tue, 22 Feb 1994 12:33:27 PST
Subject: SIAM Applied Linear Algebra Conference Deadline



Sponsored by the SIAM Activity Group on Linear Algebra
June 15-18, 1994
Snowbird Ski and Summer Resort
Snowbird, Utah

The deadline for CAMERA-READY copy for the conference proceedings is
March 7, 1994. SIAM has mailed packets to the authors who returned the
reply card last year; call SIAM at 215-382-9800 or email
if you need a packet. The preliminary program is also available by
anonymous ftp to machine (IP number as file


March 7, 1994: Deadline for camera-ready papers for conference proceedings
March 21, 1994: Advance registration deadline for contributed participants
June 15-18, 1994: SIAM Conference on Applied Linear Algebra

This meeting will experiment with a new format for contributed papers, in
response to concerns about conflicts between parallel sessions. The new
format is described in the preliminary program.


* Geometry and Eigenvalues
Persi Diaconis, Harvard University
* Parallel Matrix Computations
Robert S. Schreiber, RIACS-NASA Ames Research Center
* Linear Algebraic Duality for Discrete Optimization
Leslie E. Trotter, Cornell University
* Control Theory and Linear Algebra
Israel Gohberg, Tel Aviv University, Israel
* Nonnegative Matrices: Can the Next Century Top This One?
Charles R. Johnson, College of William and Mary
* Recent Advances in Iterative Methods Solving Linear Systems
Anne Greenbaum, Courant Institute, New York University
* Fast Parallel Algorithms for Eigenproblems
James Demmel, University of California, Berkeley
* The SIAM Activity Group on Linear Algebra Prize Lecture
Prize to be awarded at the conference


* William M. Kahan, University of California, Berkeley


* Algebraic Ricatti Equations and Their Applications
Peter Lancaster, Calgary
* Iterative Methods for Large Sparse Systems
Roland Freund, AT&T Bell Labs
* Teaching Linear Algebra
Gilbert Strang, MIT
* Parallel Multisplittings and Applications of Domain Decomposition
Dianne P. O'Leary, Maryland; Rosemary Renaut, Arizona State;
and Barry F. Smith, UCLA
* Eigenvalues, Geometry, and Graph Theory
Alex Pothen, Old Dominion and ICASE
* Accuracy Issues for Eigenvalue and Singular Value Problems
Jesse L. Barlow, Penn State; and Kresimir Veselic, Hagen
* Matrix Function Developments
Ralph Byers, Kansas
* Iterative Methods for Image Restoration
Lothar Reichel, Kent State
* Numerical Methods for Structured Matrices
Angelika Bunse-Gerstner, Bremen
* Linear Algebra in Optimization
Thomas F. Coleman, Cornell
* Direct Methods for Large Sparse Systems
Iain S. Duff, Appleton Lab and CERFACS
* Topics in Matrix Theory
Richard Brualdi, Wisconsin


As an experiment, this meeting has been organized so that the
presentation of contributed papers will be a dialogue rather than
a monologue. Each paper may be presented in three forms: in a
proceedings volume, as a poster display, and as part of a 2-hour
"common interest" discussion session. We expect most of the
contributed papers to be presented in all three forms, though this
is not required. The program contains details. The deadline for
proceedings papers is MARCH 7, 1994.


Beresford N. Parlett (Chair), University of California, Berkeley
Harm Bart, Erasmus University, Rotterdam
Richard A. Brualdi, University of Wisconsin
John R. Gilbert, Xerox Palo Alto Research Center
Sven Hammarling, Numerical Algorithms Group
John G. Lewis, Boeing Computer Services
Paul Van Dooren, University of Illinois


From: Michael Olesen <>
Date: Tue, 22 Feb 1994 15:00:34 -0600 (CST)
Subject: Workshop on Domain-based Parallelism And Problem Decomposition Methods


On April 25 and 26, 1994, the University of Minnesota
Supercomputer Institute will host a Workshop on Domain-Based
Parallelism and Problem Decomposition Methods in Computational
Science and Engineering, organized in cooperation with the SIAM
Activity Group on Supercomputing and the Army High Performance
Computing Research Center. The workshop will bring together
researchers from many scientific and engineering disciplines who
exploit `problem decomposition techniques" in various continuous
or discrete senses to develop parallel algorithms. The invited
speakers will take a broad perspective suitable for an
interdisciplinary audience of engineers, physical scientists, and
applied mathematicians. There will be no contributed papers.
However, a poster session will be organized for those participants
who wish to present their work. Attendence to the workshop is
limited, so registration is required. Early registration is

For information or to obtain a registration form, contact one of
the organizers, David Keyes (ICASE) or Youcef Saad (University of
Minnesota, CS) or Donald Truhlar (University of Minnesota,
Chemistry), or contact Michael Olesen, Symposium Administrator,
Supercomputer Institute, 1200 Washington Avenue South,
Minneapolis, MN 55415, phone:(612) 624-1356, fax: (612) 624-8861,


ZLATKO BACIC (New York University): "Bound States of Strongly
Coupled Multidimensional Molecular Hamiltonians by the Discrete
Variable Representation Approach".

PETTER BJORSTAD (Institute for Informatics in Bergen, Norway):
"Domain Decomposition and Parallel Computing with Applications to
Oil Reservoir Simulation".

XIAO-CHUAN CAI (University of Kentucky): "Domain Decomposition
Methods for Nonsymmetric and Indefinite Systems of Equations".

CHARBEL FARHAT (University of Colorado at Boulder): "Tailoring DD
Methods for Fast Parallel Coarse Grid Solution, Multiple Right
Hand Sides, and Adaptive Refinement".

ROLAND GLOWINSKI (University of Houston): "Fictitious
Domain/Domain Decomposition Methods for Partial Differential

WILLIAM GROPP (Argonne National Laboratories): "Parallel Software
for Domain Decomposition Algorithms".

GEORGES JOLICARD (Universite de Franche-Comte): "The Wave Operator
Theory as a Tool to Define Active Spaces and Simplified Dynamics
in Large Size Quantum Spaces".

DANA KNOLL (Idaho National Engineering Laboratory): "Newton-
Krylov-Schwarz Methods Applied to the Tokamak Edge Plasma Fluid

ANDREW LUMSDAINE (Notre Dame University): "Domain Decomposition in
Space and Time: Accelerated Waveform Methods for Semiconductor
Device Simulation".

ALFIO QUARTERONI (Universita degli Studi di Milano in Italy):
"Domain Decomposition Methods for Wave Propagation Problems".

FRANCOIS-XAVIER ROUX (Office National d"Etudes et de Recherches
Aerospatiales in Chatillon, France): "Parallel Implementation of a
Domain Decomposition Method for Non-Linear Elasticity Problems".

DAVID SCHWENKE (NASA Ames Research Center): "New Ideas for Problem
Decomposition Methods in Quantum Mechanical Reactive Scattering".

ELLEN STECHEL (Sandia National Laboratories): "Linear Scaling
Algorithms for Large-Scale Electronic Structure Calculations".

TAYFUN TEZDUYAR (University of Minnesota): "Parallel Finite
Element Computation of Flow Problems".

V. VENKATAKRISHNAN (Institute of Computer Applications in Science
and Engineering in Virginia): "Parallel Implicit Methods for
Aerodynamic Applications on Unstructured Grids".

H.-J. WERNER (Universitat Bielefeld): "Electron Correlation in

JACOB WHITE (Massachusetts Institute of Technology): "Aspects of
Multipole and FFT Accelerators for the Rapid Solution of Three-
Dimensional Potential Integral Equations, with Applications".


From: T.R.Hopkins <>
Date: Wed, 23 Feb 1994 15:44:34 +0000
Subject: Interpolation problem

Suppose you were given, for each point of a square grid, the value
of a function f(x, y), together with {\partial f}/{\partial x} and
{\partial f}/{\partial y}, but no higher order derivatives. How would
you interpolate in this data? (I note that the data overdetermines a
cubic element, but leaves a quartic element underdetermined.)

Any pointers very welcome.

Tim Hopkins


Date: Thu, 24 Feb 1994 10:50:20 -0305
Subject: First Kind Integral Equations

I am interested in efficient numerical methods for a linear first kind integral
equations in Augmented Galerkin method. I need an algorithm or an C or Fortran
program to solves an overdetermind set of linear equations in the L1 norm or in
Lp norm,i.e., an program for L1 approximation. Thanks.


From: Barak Galanti <>
Date: Thu, 24 Feb 1994 11:26:16 +0200
Subject: Looking for n-dimensional Fast Chebyshev Transform software

I am searching for n-dimensional Fast Chebyshev Transform software.

Any help would be greatly appreciated.
Many thanks in advance,

Barak Galanti


Date: Thu, 24 Feb 94 13:34:11 EST
Subject: SIAM J. on Matrix Analysis & Applications Table of Contents


Strong Hall Matrices
Richard A. Brualdi and Bryan L. Shader

Computing the PSVD of Two 2 x 2 Triangular Matrices
Gary E. Adams, Adam W. Bojanczyk, and Franklin T. Luk

Nonlocal Perturbation Analysis of the Schur System of a Matrix
M. M. Konstantinov, P. Hr. Petkov, and N. D. Christov

An Algorithm for the Single-Input Pole Assignment Problem
Rafael Bru, Jose Mas, and Ana M. Urbano

Backward Error Estimates for Toeplitz Systems
J. M. Varah

Moore-Penrose Inversion of Square Toeplitz Matrices
Georg Heinig and Frank Hellinger

Fast Triangular Factorization and Inversion of Hankel and Related
Matrices with
Arbitrary Rank Profile
Debajyoti Pal and Thomas Kailath

The Arnoldi Method for Normal Matrices
Thomas Huckle

Unicity of Biproportion
Louis de Mesnard

Characterizations of Scaling Functions I. Continuous Solutions
David Colella and Christopher Heil

Linear Operators Preserving Complex Orthogonal Equivalence on
Roger A. Horn, Chi-Kwong Li, and Dennis I. Merino

A Parallel Algorithm for Computing the Singular Value Decomposition
of a Matrix
E. R. Jessup and D. C. Sorensen

Accurate Downdating of Least Squares Solutions
A. Bjorck, H. Park, and L. Elden

Dynamical Systems that Compute Balanced Realizations and the
Singular Value
U. Helmke, J. B. Moore, and J. E. Perkins

Iterative Consistency: A Concept for the Solution of Singular
Systems of Linear
M. Hanke

Algorithms for Computing Bases for the Perron Eigenspace with
Nonnegativity and Combinatorial Properties
Michael Neumann and Hans Schneider

On Rank-Revealing Factorisations
Shivkumar Chandrasekaran and Ilse Ipsen

Symmetric Toeplitz Matrices with Two Prescribed Eigenpairs
Moody T. Chu and Melissa A. Erbrecht

Perturbation Analysis of a Condition Number for Linear Systems
Zhi-Quan Luo and Paul Tseng

A Second-Order Perturbation Expansion for the SVD
Richard J. Vaccaro

Probabilistic Bounds on the Extremal Eigenvalues and Condition
Number by the
Lanczos Algorithm
J. Kuczynski and H. Wozniakowski

An Error Model for Swarztrauber's Parallel Tridiagonal Equation
Nai-Kuan Tsao

Trust Region Problems and Nonsymmetric Eigenvalue Perturbations
Ronald J. Stern and Henry Wolkowicz


From: Prof L.M. Delves <>
Date: Fri, 25 Feb 1994 21:45:28 +0000 (GMT)
Subject: Research Postdoctoral Position at University of Liverpool

Three Year Research Postdoctoral Position

Institute of Advanced Scientific Computation
University of Liverpool
Liverpool UK

Applications are invited for this SERC-funded post
to work on techniques for automatic recognition
of small objects in SAR (Synthetic Aperture Radar) images.
The ideal candidate will have, or shortly expect to receive,
a PhD in Computational Mathematics or an allied subject. Experience
with SAR and/or parallelism would be helpful but is not essential

The post is available for filling immediately; salary will be up to
approximately 15,500 Pounds Stg. The work will be carried out
in collaboration with the SAR groups at
DRA Malvern and NA Software Liverpool.
Applications can be sent by mail, fax or email to:

Prof L.M Delves
Tel +44 51 794 4752
Fax: +44 51 794 4754
University of Liverpool
PO Box 169
Liverpool Merseyside L63 9BX


From: Rudnei Dias da Cunha <>
Date: Sat, 26 Feb 94 12:25:43 +0000
Subject: PIM version 1.1 available

We would like to announce the availability of version 1.1 of PIM.

This version includes support for complex systems of linear equations,
in both COMPLEX and DOUBLE COMPLEX data formats.

The user's guide has been revised to reflect this new feature and we
have fixed some bugs in the examples.

The new version is available via anonymous ftp from (

in the file


Rudnei Dias da Cunha
Computing Laboratory
University of Kent at Canterbury
Canterbury, Kent CT2 7NF, United Kingdom


From: Lawrence F. Shampine <H5SR1001@VM.CIS.SMU.EDU>
Date: Sat, 26 Feb 94 10:17:39 CST
Subject: Book on Numerical Solution of Ordinary Differential Equations Available

The book

Lawrence F. Shampine
ISBN 0-412-05151-6

is now available from the publisher, Chapman & Hall, 29 West
35th Street, New York, NY 10001, (212) 244-3336.

From the preface: "This book is an introduction to the numerical
solution of the initial value problem for a system of ordinary
differential equations (ODEs). ... Chapter 1 provides a sense of
what kinds of problems can be solved numerically and describes how
typical problems can be formulated in a way that permits their
solution with standard codes. ... Chapter 2 explains some of the
limitations of numerical methods. ... Chapter 3 is devoted to the
computational problem and associated software issues. Chapters 4
through 8 derive the basic numerical methods, prove their convergence,
study their stability, and consider how to implement them effectively.
By narrowing the focus of the book to the most important methods, it
is possible to consider them rather carefully and still make minimal
demands on the mathematical and computational background of the reader.
... The book presents a perspective of the numerical solution of
ordinary differential equations that results from some two decades of
research and development by the author in both universities and
scientific laboratories. ... "


End of NA Digest