## NA Digest Tuesday, November 25, 1992 Volume 92 : Issue 44

### Today's Editor:

Cleve Moler
The MathWorks, Inc.
moler@mathworks.com

### Submissions for NA Digest:

Mail to na.digest@na-net.ornl.gov.

Mail to na.help@na-net.ornl.gov.

-------------------------------------------------------

Date: Tue, 17 Nov 92 16:07:17 CST
Subject: Polynomials Orthogonal Over an Equilateral Triangle

I wonder if some kind soul would direct me to literature that deals
with polynomials that are ORTHOGONAL on an EQUILATERAL triangle.
The Appell functions are orthogonal over a region covered by a
right angled triangle, but I do not know if these can be modified
to obtain polynomials that would be orthogonal over an equilateral
triangle. Also, any literature that deals with rational functions
orthogonal over an equilateral triangle.

I would be very grateful for any and all information on this interesting
topic (it affects both finite elements and mechanical CAD).

Thank you,

------------------------------

From: Brigitte Verdonk <verdonk@wins.uia.ac.be>
Date: Wed, 18 Nov 92 11:15:31 +0100
Subject: Block Tridiagonal Linear Systems

We are looking for recent papers on parallel techniques for the solution
of block tridiagonal linear systems. Our last reference is a paper which
appeared in 1987 in Linear Algebra and Its Applications and we would like
to get up to date with the techniques developed since then.

Brigitte Verdonk
verdonk@wins.uia.ac.be

------------------------------

From: Joe Crcar <sepp@snll-arpagw.llnl.gov>
Date: Thu, 19 Nov 92 18:53:06 PST
Subject: L-1 Solution of Overdetermined Systems

Hi Net!

I'm looking for (fortran) software that will solve
overdetermined systems of linear equations in the
L-1 (least absolute sum) sense. Couldn't find it
in netlib. The systems I have are small (50x20)
and dense.

Regards, Joe Grcar (na.grcar@...)

------------------------------

From: Franklin Luk <luk@cs.rpi.edu>
Date: Fri, 20 Nov 1992 08:04:47 -0500
Subject: Resnick Center for Physics Education

Rensselaer Polytechnic Institute will honor Professor Robert Resnick
(of the Halliday-Resnick Physics text fame) by creating a Robert Resnick
Center for Physics Education. The Center will be directed by an endowed
faculty chair, who will direct support staff, visiting professors
and graduate students in researching and developing innovations
in physics education. I submitted this news bit to NA-Net to show
how highly education is valued at major American research institutions.

------------------------------

From: Gene Golub <golub@sccm.Stanford.EDU>
Date: Fri, 20 Nov 92 14:14:21 PST

Several persons have written me, saying they would like to buy Dover
publications and other books published in the US. You can order books
directly from the Stanford bookstore.

Send an e-mail msg to bookstore@sierra.stanford.edu, giving the name of the
book and a credit card number. Don't expect an acknowledgment; their computer
skills are limited. If it is easier, send a FAX to 415/ 322-1936.

The bookstore is excellent and they sell, books, t-shirts, and Apples.
Try this; you may like it.

Gene

------------------------------

From: Michael E.Hosea <mhosea@sun.cis.smu.edu>
Date: Mon, 23 Nov 92 11:28:31 CST
Subject: Calculating Runge-Kutta TEC's

I have developed an item of mathematical software that may be of interest
to those working with Runge-Kutta methods. It is an ANSI C program that
calculates the truncation error coefficients of Runge-Kutta methods. The
code is called RKTEC and is available from NETLIB in the MISC directory.

RKTEC is based on the approach of Albrecht in

P. Albrecht, "A New Theoretical Approach to Runge-Kutta Methods",
SIAM J. Numer. Anal., 24 (1987), 2, pp. 391-406.

Details of the extensions of Albrecht's work required, the algorithms used,
and illustrative examples of the performance of the code are provided in

M. E. Hosea, "Rapid Calculation of Runge-Kutta Truncation Error
Coefficients", Technical Report #92-7 (1992), Mathematics Dept.,
Southern Methodist University, Dallas, TX 75275-0156.

In principle RKTEC can calculate truncation error coefficients of
arbitrarily high orders for any Runge-Kutta formula. (In practice the
amount of memory required increases as the order is increased and as the
number of stages in the formula is increased.)

Regards,
Mike Hosea (mhosea@sun.cis.smu.edu)

------------------------------

From: Roy Mathias <mathias@imafs.ima.umn.edu>
Date: Mon, 23 Nov 92 14:29:58 CST
Subject: Report on Summer Institute at Leuven

NATO ADVANCED STUDY INSTITUTE examines linear algebra for large scale
and real time applications

By Alan Edelman and Roy Mathias

Continuing on the success of the 1988 NATO ASI on "Numerical Linear
Algebra, Digital Signal Processing and Parallel Algorithms," a second
NATO ASI was held in Leuven, Belgium from August 3--14, 1992 at the
Katholieke Universiteit. The institute brought together researchers
in linear algebra along with those who study real time computations in
a forum that discussed the state of the art and the needs for the
future.

Though the locals may know and love their Katholieke Universiteit,
researchers throughout the world know the site as
"esat.kuleuven.be.ac" thanks to the tireless efforts of the chief
organizer, Mark Moonen who must have broken some kind of record in
numbers of electronic mail messages sent on organizing a conference.
For his efforts he has become the second recipient of the "order of
the shiny red hat." The conference owes its success also to
the other committee members: Angelika Bunse-Gerstner, Bart De Moor,
Gene Golub, Sven Hammarling, Jean Meinguet, and Joos Vandewalle; we
are also indebted to Marc Moonen, Bart Motmans, Geert Schelfhout,
Peter Van Overschee, Fillep Van Poucke, and Paul Vanvuchelen for their
efforts with the local arrangements and their generous hospitality.

Not only were we enriched scientifically, we also became experts on
saying "thank-you" and "please" in Dutch. Also of note is that the
organizers had the forethought to provide individual computer accounts
for each participant, a necessity in this community right up there
with lodging and meals. There was a sports afternoon and a trip to
Brugges to take our minds from mathematics.

Thanks to the efficiency of electronic communication, and the
standardization of LaTeX, the conference proceedings will be published
in record time. They will be published by Kluwer in 1993. We
refer readers to this book for a fuller synopsis of the material
covered during the meeting, and here we report (these authors biases
of) some memorable events.

George Cybenko provided a tutorial on
wavelets generalizing the usual formulation into the context of
unitary operators and multiresolution analysis. He also listed
applications of supercomputers in signal processing including seismic
processing in which, he pointed, out varies with the price of
oil. Cybenko left us with sobering thoughts regarding high
performance supercomputing, by asking whether there has been sometimes
"too much politics and promotion" with promises that have
mislead the public into too high an expectation that something
remarkable will happen in the next five years. (This question has
become a hot topic of debate in quite a number of circles.)

Margaret Wright reminisced about her first class with Gene Golub,
where she learned the most important question of numerical analysis:
"What is zero?"in the context of discussing large scale optimization.
In particular, the protein folding problem, whose solution will all but
guarantees the solver a Nobel Prize, leads to very interesting linear
algebra issues. This problem requires collaborative efforts
of many specialists.

Simon Haykin described applications of Adaptive Noise Cancellation via
interference canceling such as echo cancellation in telephony and
radar. His approach is the minimization of a measure of
"information" content between outputs. He further illuminated the
applications of self-organized learning to principal components
analysis.

A recurring theme of the conference was the question, "What is
Large?" Ph. Toint reminded us that linear programs with five million
variables can be solved, while a non-linear program of size 500 can be
difficult. Toint pointed out that non-linear optimization problems
are of importance because nature optimizes and the universe is
non-linear. He reported a quote (I did not catch the source) of
common wisdom in optimization technology that says that "I'd rather
work on today's algorithms on yesterday's computers rather than the
other way around." In particular, Toint discussed the Lancelot
project that has solved a variety of problems including the growth
factor problem for complete pivoting in Gaussian Elimination.

Linda Kaufman also addressed the issue of what is large by considering
the image reconstruction problem of Pet Scans. She believes that
doctors would be reluctant to use supercomputers because it would
force them to cooperate with each other, when they are more used to
creating their own "fiefdoms."

Morven Gentlemen said that he was not aware of any success stories of
large scale linear algebra in robotics. The basic kinds of
computation in the field of robotics are the control algorithms and
sensor data interpretation for robotics. Though there may not be any
large scale success stories, linear algebra plays a major role both
conceptually and computationally in the relationships between
coordinate systems, transitions between segments, and path
planning. He described his first talk as describing how to make
a robot do what you want assuming you already know what it should do.
His second talk described how to find out enough about the robot's
environment so as to decide what it should do. He pointed out that
we'd be surprised as to how many things can go wrong with a robot
trying to cut its own birthday cake.

Bo Kagstrom manipulated the Swedish flag the same way
matrix pencils are manipulated to obtain their Kronecker structure.
Kagstrom told us why the singular case of matrix
pencils is harder than the regular case, and what possible Kronecker
forms appear as perturbations of other forms through the algorithm.
His second talk considered a direct method for reordering eigenvalues
in the GUPTRI form of regular matrix pencils.

Frank Luk illustrated some parallel computing efforts on the CM-5
including difficulties with the initial documentation of the
underlying partial fat-tree network.

Beresford Parlett who was immediately followed by Vince Fernando
described their recent simplifications of the prize winning work of
Demmel and Kahan by finding an accurate algorithm for the bidiagonal
singular value problem by using an algorithm that was essentially
known to Rutishausser. Parlett also noted that there was a lot of
money holding down the slides that were prone to sliding off.
The sliding slides were a recurring source of humor during the two
weeks.

Roland Freund surveyed the latest in iterative method technology for
nonsymmetric matrices. He described the look-ahead Lanczos method,
and pointed out how appropriate it was that this was all worked out
before Lanczos' one hundredth birthday. Freund also wanted us to know
that he was not opposed to the idea of preconditioning, contrary to
popular belief. Noel Nachtigal discussed the implementation of
the QMR method. Lanczos was considered probablistically by
Jacek Kuczynski. A Faber Polynomial approach to iterative methods
was considered by Gerhard Starke.

Jim Demmel considered tradeoffs between parallelism and stability.
He discussed LAPACK war stories providing suggestions for the future
of numerical computing. David Stewart had to fight a war with one
participant. His LAPACK style library is the results of
one researcher, rather than the international collaborative effort of
the LAPACK project. Nevertheless, he dazzled us with his matrix
computations software library written in C.

Charlie van Loan discussed material from his new book on matrix
factorizations and the Fast Fourier Transform. Somehow less of a
surprise to the audience than one might expect, he pointed out that
knew about his famous series. (After all, Gauss may have done it all?)
He also introduced a knew kind of structured approximation to a dense
matrix, the Kronecker product approximation which might potentially be
used as a preconditioner. It turns out that the Kronecker product
approximation has an SVD solution.

Pete Stewart considered the numerical treatment of Markov Chains.
A Markov chain is a unique kind of eigenvalue problem, because the largest
eigenvalue is already known. Many methods have been proposed for the
solution. Stewart also told us about the rank determination problem
in the presence of error. Rank determination is often performed via
rank revealing algorithms. In the presence of error it becomes even
more important to test elements in the decomposition to see if they
may be regarded as 0.

Steve Boyd discussed convex optimization problems that can be stated
as the optimization of a linear function subject to a positive semidefinite
matrix constraints. He presented some 'five line codes' based on interior
point methods to solve these problems. The theme of his two talks was that
once an engineering problem is reduced to a convex optimization problem
it should be regarded as solved. Florian Jarre talked about similar results.

Petter Bjorstad discussed the implementation of the BLAS on the Maspar
supercomputer and the solution of large linear systems arising in
structural analysis using parallel computers. He impressed us with a
description of the high speed network connecting the major Norwegian
sites, when in the past young boys (Petter many years ago) helped in
the communication of information by grabbing the appropriate disks and
running across the hall to be placed on the computer at the right
point in the calculation.

Alan Edelman turned down the opportunity to take a random walk to
the bars on the last night (and an even more random walk home)
so that he could tell us about the game of eigenvalue roulette.

There were many other interesting talks. We apologize for not summarizing
them here, and encourage you to see the forthcoming proceedings.
In summary, everyone felt this was a wonderful conference, and we
greatly express our appreciation to the organizers.

------------------------------

From: Ingrid Tokka <tokka@esat.kuleuven.ac.be>
Date: Tue, 17 Nov 92 08:47:59-0100
Soubject: Workshop In Leuven, Belgium

Workshop on
Numerical Techniques for Scientific Computing
and Mathematical Engineering

Date : Wednesday, December 2, 1992

Location : Arenberg Castle, Katholieke Universiteit Leuven,
Kardinaal Mercierlaan 94, B-3001 Leuven (Heverlee), Belgium

Information can be obtained from

Prof. Dr. Bart De Moor
Ms. Ingrid Tokka
ESAT - Department of Electrical Engineering
Kardinaal Mercierlaan 94
B-3001 Leuven
Belgium
tel: 32/16/220931
fax: 32/16/221855
email: demoor@esat.kuleuven.ac.be

Speakers:

Prof. Dr. Gene Golub,
Department of Computer Science, Stanford University.
Solving Large, Structured Systems

Department of Computer Science, K.U.Leuven.
From Euclid to Lanczos

Prof. Dr. Bart De Moor,
Department of Electrical Engineering, K.U.Leuven.
The singular value decomposition in signal processing and
control engineering

Prof. Dr. Paul Dierckx,
Department of Computer Science, K.U.Leuven.
Smoothing with splines

Prof. Dr. Carl De Boor,
Department of Mathematics, University of Bielefeld.
Computational aspects of multivariable polynomial interpolation

We would also like to invite you to:

The ceremony for the Honorary Degree of
Prof. Dr. H. Czichos, Berlin
Prof. Dr. G. Golub, Stanford

Date : Tuesday, December 1, 1992
Time : 4.00 pm
Location : Central University Building,
De Hallen, Naamsestraat 22, B-3000 Leuven

Welcome and introduction by the Dean of the \\Faculty of Engineering,
Prof. Dr. E. Aernoudt

- Lecture by Prof. Dr. H. Czichos: "Industrial and Materials
Technologies, Policy and R & D Trends in Europe."

- Laudatio for Prof. Dr. Czichos by Prof. Dr. J. Roos

- Laudatio for Prof. Dr. Golub by Prof. Dr. B. De Moor

- Words of thanks by Prof. Dr. G. Golub

------------------------------

Date: Wed, 18 Nov 92 06:58:55 -0700
Subject: Copper Mountain Conference on Multigrid Methods

!!! REMINDER !!!

ANNOUNCEMENT
and
CALL FOR PAPERS

COPPER MOUNTAIN CONFERENCE
ON
MULTIGRID METHODS

April 4-9, 1993

ORGANIZING INSTITUTIONS

The University of Colorado at Denver
Front Range Scientific Computations, Inc.
The Society for Industrial and Applied Mathematics

Department of Energy
National Science Foundation

CONFERENCE CHAIRMEN

Tom Manteuffel and Steve McCormick,

PROGRAM COMMITTEE

Joel Dendy, Los Alamos National Laboratory
Jan Mandel, University of Colorado at Denver
Seymour Parter, University of Wisconsin
Joseph Pasciak, Brookhaven National Laboratory
Boris Rozovski, University of Southern California
John Ruge, University of Colorado at Denver
Klaus Stueben, Gesellschaft f. Math. u. Datenverarbeitung
James Thomas, NASA Langley
Pieter Wesseling, Delft University
Olof Widlund, Courant Institute

WORKSHOP CHAIRMAN

Paul Frederickson, RIACS

CIRCUS CHAIRMAN

Craig Douglas, IBM and Yale University

SPECIAL FEATURES

Circus
Preliminary Proceedings
Student Paper Competition
Special Journal Publication of Proceedings
Workshops

TOPICS OF INTEREST

Domain Decomposition Methods
General Iterative Methods
Multigrid, Multilevel, Multiscale, Multiresolution Methods

Abstracts Dec. 15, 1992
Student Papers Dec. 15, 1992
Papers for Prelim. Procs. Feb. 15, 1993
Lodging Reservations March 1, 1993
Early Registration March 1, 1993

------------------------------

From: Annie Cuyt <cuyt@wins.uia.ac.be>
Date: Fri, 20 Nov 92 15:13:50 +0100
Subject: Nonlinear Numerical Methods and Rational Approximation

CONFERENCE ANNOUNCEMENT

``Nonlinear Numerical Methods and Rational Approximation''

Organiser: A. Cuyt (UIA, Antwerp).
Scientific Committee: A. Bultheel (KUL, Leuven), A. Cuyt (UIA, Antwerp),
J. Meinguet (UCL, Louvain-la-Neuve), J.-P. Thiran (FUN, Namur).

Invited Speakers: A. Gonchar (Moscow), M. Gutknecht (Zurich),
W. B. Jones (Boulder, USA), D. Lubinsky (Witwatersrand, S-Africa),
E. Saff (Tampa, USA).

In 1993 another conference will be organised on the use of rational functions
in different fields of numerical analysis. It will take place at the
University of Antwerp (UIA) from September 5 to 11. Registration will cover
participation, housing and a copy of the conference proceedings.

Each of the following sessions will be introduced by a one-hour survey
lecture. All other participants are invited to present a 20-minute
research talk. Proceedings will be published similar to the ones from
1987. Since Antwerp is in 1993 Europe's cultural capital, special
attention will be paid to housing and social events for the participants.

Sessions: The emphasis will be on Pade approximation,
Rational interpolation, Rational approximation, Continued fractions
and Orthogonal polynomials. For each of those topics we also welcome
all Multivariate and Multidimensional problems, Applications, Error
analysis and Software development.

Further Information: Annie Cuyt, Dept of Mathematics and
Computer Science, University of Antwerp (UIA), Universiteitsplein 1,
B--2610 Wilrijk-Antwerp, Belgium, tel: (32) 3/820.24.07, fax: (32)
3/820.22.44, secr: (32)3/820.24.01, e-mail: cuyt@wins.uia.ac.be

------------------------------

From: Karen Hahn <khahn@cs.rutgers.edu>
Date: Mon, 23 Nov 92 16:14:41 EST
Subject: IMACS Symposium on Symbolic Computation '93

INTERNATIONAL IMACS SYMPOSIUM - 1993
SYMBOLIC COMPUTATION
New Trends and Developments

WHERE: Lille, France
WHEN: Monday 14th - Thursday 17th of June 1993

Chairman: G. Jacob, France (jacob@lifl.fr)
Co-Chairman: Stan Steinberg, U.S.A. (stanly@math.unm.edu)

There are sixteen proposed sessions:

1. Differential Equations (Ramis and MacCallum)
2. Visualisation of Combinatorial Structures (Maylis Delest)
3. Programmation with Constraints and High Level Compilation (Christian Lair)
4. Complexity (M. Giusti)
5. Robotics and Motion Planning (G. Jacob)
6. Discrete Event Processes (G.Jacob)
7. Computer algebra and solving polynomial equations (V. Gerdt)
8. Theory and Practice of Algebraic Algorithms (V. Trevisan)
9. Real Algebraic Numbers in Symbolic Computation (T. Recio)
10. Geometry and Algebraic Topology (Sergeraert)
11. The Impact of Computer Algebra on the Use and Understanding of
Fundamental Interactions (H. Caprasse)
12. Applications of Symbolic Computation in Engineering and Applied Mathematics
(M. Crespo da Silva)
13. Application of Symbolic Computation to Numerical Solving of PDEs (V. Ganzha)
14. Symbolic Computation Interfaces to Numeric Computation (R.L. Peskin)
15. Solving Differential Equations by Computer Algebra (F. Schwarz)
16. Computer Algebra as a Tool in Teaching Calculus and Differential Equations
(V. Ganzha, W. Strampp)

Professor Steinberg or Professor Jacob.

MAILING ADDRESS: Symposium IMACS SC 93
Bat. M3-Informatique
Universite Lille I
59655 Villeneuve d'Ascq Cedex France

REGISTRATION: \$250 (1250 FF), including proceedings and banquet
IMACS members: \$235 (1175 FF)
Students \$60 (300 FF), banquet and proceedings not included.

LOCATION OF CONFERENCE: M.A.C.C.
Bd. Paul Langevin
59650 Villeneuve d'Ascq, France.

HOW TO GET THERE: By plane to Lille-Lesquin airport, then
minutes by taxi.
By train - Paris to Lille, 2 hrs 30 mins,
\$35 (160 FF), then metro from Lille
station to ``Cite Scientifique'', 8 FF.

HOUSING: Participants should make their own hotel reservations.
You may contact the conference secretariat for a list of hotels.

------------------------------

From: Dennis Gannon <gannon@moose.cs.indiana.edu>
Date: Sun, 15 Nov 1992 21:54:53 -0500
Subject: Computaional Sciences at Indiana Univ.

Position in Computational Science
Indiana University, Bloomington

Indiana University is expanding its program in Computational
Science and invites applications for a faculty position in
the program. We are particularly interested in applicants
with an interdisciplinary orientation in computational
science and a background in Astronomy, Chemistry, Computer
Science, Mathematics or Physics. Applicants with other
backgrounds who can make a strong contribution to this
interdisciplinary program will also be considered. The
position is tenure-track in one of the above departments
(depending on background) and teaching and the development
of a vigorous research program in this department is
expected.

Of particular interest are those candidates with experience
in supercomputing and massively parallel systems.

Applicants should send a curriculum vitae, publication list,
documentation of teaching experience and accomplishments, a
one page statement of research goals, and the names of three
references. Applications should be sent to: The
Computational Sciences Program, Office of the Dean of Arts
and Sciences, Indiana University, Bloomington, IN 47405.
Indiana University is an equal opportunity employer.

Informal contacts by email to gannon@cs.indiana.edu are also
welcome.

------------------------------

Date: Mon, 16 Nov 1992 14:15:16 -0700
Subject: Position at University of Colorado, Boulder

Department of Computer Science

Applications are invited for faculty positions in
numerical and parallel computation. Preference will be given
to candidates at the assistant professor level. Applicants
should show strong promise in both research and teaching.

The Computer Science Department at the University of Colorado
The Department is the recipient of two consecutive five-year
Institutional Infrastructure (previously CER) grants from the
National Science Foundation that emphasize parallel and
numerical computation. It is a member of the recently formed
DARPA National Consortium in High Performance Computing
and a major participant in recently announced NSF and NASA
Grand Challenge Applications Group grants. The computing
environment in the department includes a multitude of computer
workstations and a large variety of parallel computers.

Applicants should send a current curriculum vita and the
names of four references to Professor Robert Schnabel, Chair,
Department of Computer Science, Campus Box 430, University of Colorado,
Boulder, CO 80309-0430. One-page statements of research and teaching
interests would also be appreciated. Review of applications will
begin on January 1, 1993, although all applications postmarked
before March 1, 1993 are eligible for consideration.
Earlier applications will receive first consideration.
Appointment can begin as early as August 1993.

The University of Colorado at Boulder has a strong institutional
commitment to the principle of diversity in all areas. In that
spirit, we are particularly interested in receiving applications
from a broad spectrum of people, including women, members of
ethnic minorities, and disabled individuals.

------------------------------

From: George Sell <sell@a1.arc.umn.edu>
Date: Fri, 20 Nov 92 14:25:57 CST
Subject: Postdoctoral Positions at the AHPCRC

Postdoctoral Positions

Army High Performance Computing Research Center (AHPCRC)

UNIVERSITY OF MINNESOTA

The AHPCRC runs an interdisciplinary research program that focuses on the high
performance computational aspects of engineering, physical sciences, computer science
and mathematics. The primary computational system is a 32K Connection Machine,
CM-5 with 2,176 Vector Floating Point (64 bit) units and 17.6 gigabytes of main memory.
Other computational resources include a Graphics and Visualization Laboratory and
access to several Crays. The Center is currently emphasizing research activities in the
following areas:

Computational Continuum Mechanics, including computational fluid dynamics
with finite volume and finite element techniques;

High Performance Computing Issues in Nonlinear Dynamics, with special
emphasis on long-time dynamical issues arising in fluid flows, convective
flows, and reacting flows;

Computer Science, including parallel algorithms for matrix computations,
multibody dynamics and optimization, systems environment tools,
heterogeneous computing over networks, and performance evaluations;

Emerging Areas, covering computational biomedical sciences,
computational chemistry, and modern materials.

We are inviting applications for postdoctoral fellowships (research associate
positions) beginning in September 1993, but alternative starting dates may be
requested. All requirements for a doctorate must be completed prior to the
starting date. The materials listed below must be received by January 15, 1993:

1. Description (maximum two pages) of research background and plans for
participating in one of the AHPCRC above areas.

2. Curriculum vitae including a list of publications.

3. Three letters of recommendation, to be sent directly to the AHPCRC.

All correspondence should be directed to:
Tayfun E. Tezduyar, Director of Fellows Program
AHPCRC/University of Minnesota
1100 Washington Avenue South
Minneapolis, Minnesota 55415

The University of Minnesota is an equal opportunity educator and employer.

------------------------------

From: Stephen Wright <wright@antares.mcs.anl.gov>
Date: Fri, 20 Nov 92 15:03:03 CST
Subject: Postdoc at Argonne/Northwestern

We have recently received funding for a postdoctoral position in high
performance computing for large scale optimization. The successful
applicant will be appointed by Northwestern University but will spend most
of their time at Argonne National Laboratory.

If you are interested, contact Steve Wright, wright@mcs.anl.gov, (708) 252-7847,

Jorge Nocedal (Northwestern University)
Steve Wright (Argonne National Laboratory)

Northwestern University is an AA/EO employer.
All applicants must be able to provide proof
of eligibility to work.

------------------------------

From: John Mason <mason@rmcs.cranfield.ac.uk>
Date: Fri, 20 NOV 92 10:12:31 BST
Subject: New Journal: Advances in Computational Mathematics

ANNOUNCEMENT OF NEW JOURNAL - AND CALL FOR PAPERS :

ADVANCES IN COMPUTATIONAL MATHEMATICS (First issue: Spring 1993)

Editors-in-Chief:

Charles A. Micchelli - Academic Editor
Mathematical Sciences Department
IBM Research Center
PO Box 218
Yorktown Heights NY 10598
USA

e-mail: cam@YKTVMZ (Bitnet)

John C. Mason - Managing Editor
Applied & Computational Mathematics Group
Royal Military College of Science
Shrivenham
Swindon SN6 8LA
England

e-mail: mason@rmcs.cran.ac.uk

Publisher:

J.C. Baltzer AG
Scientific Publishing Company
Wettsteinplatz 10
CH-4058 Basel
Switzerland

EDITORIAL BOARD

D. Arnold (USA), C.T.H. Baker (England), D. Braess (Germany),
J.H. Bramble (USA), C. Brezinski (France), K. Burrage (Australia),
C.K. Chui (USA), M.G. Cox (England), G. Cybenko (USA), W. Dahmen (Germany),
R.A. De Vore (USA), D. Donoho (USA), C. Douglas (USA),
S.W. Ellacott (England), W.H. Enright (Canada), R. Fletcher (Scotland),
W. Freeden (Germany), T.L. Freeman (England), M. Gasca (Spain),
K.O Geddes (Canada), R.N. Goldman (USA), G.H. Golub (USA),
T.N.T. Goodman (Scotland), J.A. Gregory (England), E. Grosse (USA),
S.J. Hammarling (England), C. Hoffman (USA), A. Iserles (England),
R.-Q. Jia (Canada), S.L. Lee (Singapore), T. Lyche (Norway), S. Mallat (USA),
J.C. Mason (England), C.A. Micchelli (USA), T. Poggio (USA),
A. Quarteroni (Italy), L. Reichel (USA), J.M. Sanz-Serna (Spain),
R. Schaback (Germany), L.L. Schumaker (USA), S. Seatzu (Italy),
T.W. Sederburg (USA), I.H. Sloan (Australia), E. Tadmoor (Israel),
G. Wahba (USA), W.L. Wendland (Germany).

Software Editor: R. Reuter, IBM Deutschland GmbH, Heidelberg, Germany.
E-mail: reuter@dhdibm1.bitnet

SCOPE
Advances in Computational Mathematics is an interdisciplinary journal of
high quality, driven by the computational revolution and emphasising
innovation, application and practicality. This journal will be of
interest to a wide audience of mathematicians, scientists and engineers
concerned with the development of mathematical principles and practical
issues in computational mathematics.

Publication areas include computational aspects of algebraic, differential
and integral equations, statistics, optimization, approximation, spline
functions and wavelet analysis. Submissions are especially encouraged in
modern computing aspects such as parallel processing and symbolic computation
and application areas such as neural networks and geometric modelling.

All contributions should involve novel research. Expository papers are
also welcomed provided they are informative, well-written and shed new
light on existing knowledge. The journal will consider the publication
of lengthy articles of quality and importance. From time to time
special issues of the journal devoted to meetings or topics of
particular interest to the readers will be published with the guidance
of a guest editor. Ideas for such special issues can be communicated to
the Editors-in-Chief.

Software of accepted papers will be tested and be made available to
readers. Short announcements, a problem section and letters to the
Editor will also appear in the journal at regular intervals. Advances
in Computational Mathematics will be published quarterly.

CALL FOR PAPERS

Authors are cordially invited to submit their manuscripts in triplicate
to the Managing Editor .

All the manuscripts will be refereed. The decision for publication will
be communicated by the Managing Editor. After acceptance of their
paper, authors who can should send a diskette with the TEX (or LATEX or
AMS-TEX) source of their paper together with a hard copy of it and the
letter of acceptance to the Managing Editor. For papers concerning
software an ASCII diskette is needed.

------------------------------

From: SIAM <helfrich@siam.org>
Date: Tue, 17 Nov 92 09:56:30 EST
Message-Id: <9211170956.A18975@siam.org>
Subject: Contents: SIAM Control and Optimization

SIAM Journal on Control and Optimization
Vol. 31, No. 2, April 1993

The papers in this issue are dedicated to Wendell Fleming on the
occasion of his 65th birthday.

A Tribute to Wendell H. Fleming

Discrete-Time Controlled Markov Processes with Average Cost Criterion:
A Survey
Aristotle Arapostathis, Vivek S. Borkar, Emmanuel Fernandez-Gaucherand,
Mrinal K . Ghosh, and Steven I. Marcus

Fleming-Voit Processes in Population Genetics
S. N. Ethier and Thomas G. Kurtz

Curvature-Driven Flows: A Variational Approach
Fred Almgren, Jean E. Taylor, and Lihe Wang

Front Propagation and Phase Field Theory
G. Barles, H. M. Soner, and P. E. Sougandis

European Option Pricing with Transaction Costs
Mark H. A. Davis, Vassilios G. Panas, and Thaleia Zariphopoulou

Estimation of the Quadratic Variation of Nearly Observed
Semimartingales with Application to Filtering
Jean Picard

Convex Duality and Nonlinear Optimal Control
Richard Vinter

------------------------------

From: SIAM <gallaghe@siam.org>
Date: Thu, 19 Nov 92 20:42:04 EST
Subject: Contents: SIAM Scientific Computing

SIAM Journal on Scientific Computing
March 1993 Volume 14, Number 2

CONTENTS

Analysis of the Implicit Euler Local Uniform Grid Refinement Method
R. A. Trompert and J. G. Verwer

Vector-Supercomputer Experiments with the Primal Affine Linear
Programming Scaling Algorithm
K. O. Kortanek

Numerical Computation of the Scattering Frequencies for a
Cylindrically Symmetric Potential
Musheng Wei and George Majda

Problem: Application to Spectral Element Discretization
Yvon Maday, Dan Meiron, Anthony T. Patera, and Einar M. Ronquist

Explicit Runge--Kutta Pairs with One More Derivative Evaluation
than the Minimum
P. W. Sharp and E. Smart

Superparallel FFTs
Hans Munthe-Kaas

Numerical Experience with a Class of Algorithms for Nonlinear
Optimization Using Inexact Function and Gradient Information
Richard G. Carter

A Multiresolution Method for Distributed Parameter Estimation
Jun Liu

A Parallel Implementation of an Iterative Substructuring Algorithm
for Problems in Three Dimensions
Barry F. Smith

Modified Cholesky Factorizations for Sparse Preconditioners
Tamar Schlick

Optimal Parallel Solution of Sparse Triangular Systems
Fernando L. Alvarado and Robert Schreiber

A Flexible Inner-Outer Preconditioned GMRES Algorithm

A Transpose-Free Quasi-Minimal Residual Algorithm for Non-Hermitian
Linear Systems
Roland W. Freund

Preserving Symmetries in the Proper Orthogonal Decomposition
Nadine Aubry, Wen-Yu Lian, and Edriss S. Titi

Timely Communication

Wavelets and Multigrid
William L. Briggs and Van Emden Hensen

------------------------------

End of NA Digest

**************************
-------