NA Digest Sunday, September 6, 1987 Volume 87 : Issue 69

This weeks Editor: Cleve Moler

Today's Topics:

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


To: na@score.stanford.edu
Subject: Positions at RIACS
Date: Wed, 02 Sep 87 09:25:37 -0700
From: raugh@riacs.edu

POSITIONS AT RIACS: in Advanced Algorithms and Architectures Project

The Research Institute for Advanced Computer Science (RIACS), located forty
miles southeast of San Francisco at NASA Ames Research Center, is seeking
two scientists to join the Advanced Algorithms and Architectures
(AAA) project.

The AAA project is a collaborative effort between NASA and RIACS to explore
the use of new parallel computer architectures for scientific applications,
primarily computational fluid dynamics and computational chemistry. The
research is being conducted jointly with NASA's Numerical Aerodynamic
Simulation Program (a national pathfinder for supercomputing in fluid-
dynamics research) and with the Ames Research Center's Computer Systems and
Research Division, which provides supercomputing resources for many of
NASA's scientific programs.

Scientific objectives of the AAA project include development and
implementation of kernel parallel algorithms for scientific applications,
comparative performance analysis of different architectures for NASA
applications, and study of languages for parallel computing. Information
acquired through our studies will be used in the design of future machines.

The AAA staff currently consists of the project leader, 4 visiting
scientists, and 3 Stanford University graduate students. NASA physicists,
chemists, and computer scientists are collaborating with this group.
Current research topics include parallel multigrid techniques for solving
Navier-Stokes equations, parallel kernel algorithms using methods from
domain decomposition, parallel solutions of incompressible flows, and
parallel particle-method tools for hydrodynamics.

If you are interested in joining this project, send us your resume, reprints
of significant papers or reports, and names of references. Both visiting
and long-term positions are available. Preference will be given to
scientists who have a Ph.D. in computer science or a related field and who
have demonstrated success in working in a multidisciplinary research
environment. Address applications and requests for further information to

Dr. Michael R. Raugh
AAA Project Leader
NASA Ames Research Center, MS 230-5
Moffett Field, CA 94035
(e-mail address: raugh@riacs.edu)

RIACS is an institute of Universities Space Research Association (USRA).
RIACS' charter is to conduct computer-science research in collaboration with
NASA scientists.

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

Date: 3 Sep 87 14:57:00 EDT
From: "SIAM" <siam@wharton.upenn.edu>
Subject: SIAM meeting, Denver, October 12-15
To: "na" <na@score.stanford.edu>
Reply-To: "SIAM" <siam@wharton.upenn.edu>

Computational Sciences to be featured at the
SIAM 35th Anniversary Meeting


Computational science will be a major theme of SIAM's annual meeting in l987,
according to Hans Weinberger, program chair for the meeting from the
University of Minnesota, Minneapolis. The meeting, which will be held in
Denver, October 12-15, marks SIAM's 35th anniversary. Weinberger and his
committee have assembled an impressive group of speakers who will focus on
problems in the computational sciences as well as on other areas of current
broad interest to applied mathematicians. Weinberger and others observe that
computational science seems to mean different things to different people, but
in the aggregate, it can be said to include scientific computing, computer
science, discrete mathematics, computational mathematics, numerical analysis
and algorithm development all of which will be covered at the Denver meeting.

In addition to the regular sessions, there will be two special presentations.

The John von Neumann Lecture will be presented by Richard M. Karp from the
University of California at Berkeley, who will speak on the complexity of
parallel computation. The George Polya Prize recipient, Andrew Yao from
Princeton University, will speak on boolean circuits and computational
complexity. Additional topics to be covered are robust control, and inverse
scattering.

Invited Presentations

Modeling the Earth System Robust Control Theory and Matrix
Francis P. Bretherton, National Perturbation Problems
Center for Atmospheric Research John Doyle, California Institute
of Research
Computational Hypersonic Aerodynamics
Douglas L. Dwoyer, NASA Langley Research Liquid Crystal Theory
Center Jerry L. Ericksen, University
of Minnesota
Viscoelastic Properties of Rodlike
Polymers in Solution Inverse Scattering and Guy C.
Berry, Carnegie Mellon University Applications to Nonlinear
Evolution Equations
Recent Progress in Theoretical Reflection Mark J. Ablowitz, Clarkson
Seismology: Identifying Partial University
Differential Equations from Attributes of
Their Solutions Graph Minor Algorithms: Path
William W. Symes, Rice University Routing and Coloring
Paul D. Seymour, Bell
Communications Research

Corporation

Representatives of ICEMAP, the Interagency Committee for Extramural
Mathematics Programs of the federal government, will give an overview of
federal funding for mathematical research, with a discussion of the funding
patterns that are evolving in their agencies.








Minisymposia

Scientific Computing

Computational Advances at the NSF Supercomputer Centers
John Connolly, National Science Foundation

Supercomputing Research: Scientific Results on the Cornell National
Supercomputing Facility
Lawrence Lee, Cornell University, and Linda Morris, Cornell University

Science, Supercomputing and Graphics/Imaging
Robert B. Wilhelmson, National Center for Supercomputing Applications,
University of Illinois, Urbana-Champaign

Optimization and Supercomputing
Michael J. Healy, Boeing Computer Services

Numerical Methods for Viscous, Incompressible Flow
John B. Bell, Lawrence Livermore National Laboratory and Alan E. Berger,
Naval Surface Weapons Center

Ship Wave and Floating Body Problems
Ralph Kleinman, University of Delaware

Methods for Compressible Fluid Computations
Phillip Colella, Lawrence Livermore National Laboratory; James G. Glimm,
Courant Institute of Mathematical Sciences; and David Sharp, Los Alamos
National Laboratory

Solitons and Coherent Structures
David K. Campbell, Los Alamos National Laboratory

Algorithm Development

Dense Matrix Computation on Vector and Parallel Computers
Danny Sorensen, Argonne National Laboratory

Sparse Matrix Computation on Vector and Parallel Computers
J. Alan George, Oak Ridge National Laboratory and the University of Tennessee,

Knoxville

Integrated Environments for Scientific Computing
Guy William Cherry, Tektronix, Inc.

New Vector Algorithms for the IBM 3090 Vector Facility
Fred G. Gustavson, IBM T.J. Watson Research Center

Multigrid and Related Methods

Domain Decomposition Methods for Partial Differential Equations
Roland Glowinski, University of Houston



Adaptive Mesh Refinements in Finite Element Methods
James H. Bramble, Cornell University

Finite Element Method in Meteorological and Oceanographic Flows
Beny Neta, Naval Postgraduate School

Multigrid Methods
A special four-part session organized by Stephen S. McCormick, University of
Colorado, Denver

Discrete Mathematics/Computer Science

Computational Geometry
Bernard Chazelle, Princeton University

Combinatorial Optimization and Applications
Bernhard Korte, Universitat Bonn

Intersection Graphs and Their Applications
F. R. McMorris, Office of Naval Research, and William T. Trotter, Jr., Arizona

State University

Complexity of Parallel and Distributed Computation
Eli Upfal, IBM Almaden Research Center

Control Theory and Robotics

Research Issues in Robotics
John E. Hopcroft, Cornell University

Mathematics of Robust Control Theory
John Doyle, California Institute of Technology

Applications of Robust Control Theory
John Doyle, California Institute of Technology

Inverse Scattering

Inverse Scattering Problems
Roger G. Newton, Indiana University

Math Education

The l987 Mathematical Contest in Modeling
Ben Fusaro, Salisbury State College



If you are interested in attending the meeting and would like further
information, you can respond through e-mail: SIAM@Wharton.Upenn.edu
or you may fill out and return the attached form to: Conference Coordinator,
SIAM ll7 South l7th Street, l4th Floor, Philadelphia, PA l9l03-5052.
Telephone 215-564-2929.







SIAM 35th Anniversary Meeting
October 12-l5, l987
Marriott Center City, Denver, Colorado

I am interested in attending the meeting please send me the following
materials:

[] Advance registration material
[] Preliminary meeting program
[] Information on the short course

[] Please send me information about SIAM

[] I am interested in exhibiting software/hardware products

Name_________________________________________________________________________
First Middle Initial Last
Organization_________________________________________________________________

Department___________________________________________________________________

Address______________________________________________________________________

City_______________________________State________________Zip___________________


Phone ( )

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

Date: 4 Sep 87 00:18:04 GMT
From: lbl-csam.arpa!baden@ucbvax.Berkeley.EDU (Scott B. Baden)
Subject: Re: Random Vortex Method
To: na@score.stanford.edu

In reply to Arun's note:

Erik Tiemroth wrote a Thesis in Naval Architecture,
that Chorin has told me he thinks gives a reasonably
good overview. I have read parts of this myself,
and thought it quite good. Here is the reference:

Erik C. Tiemroth
The Simulation of the Viscous Flow Around a Cylinder
by the Random Vortex Method
Ph. D. Dissertation
Department of Naval Architecture and Offshore Engineering
University of California
Berkeley, CA
MAY 1986

A few years ago Tony Leonard wrote a survey for JCP, and
you may want to look that, too:

A. Leonard
Vortex Methods for Flow Simulation
Journal of Computational Physics
Vol. 37, pp 289-335, 1980


Scott Baden
Lawrence Berkeley Lab, Mathematics

[Editor's Note: The contributor, Scott Baden, has also written
a Ph. D. thesis about the implementation of the random vortex
method on parallel computers.]

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

Date: Fri, 04 Sep 87 11:10:44 CET
From: Rolf Jeltsch <JELTSCH%DACTH51.BITNET@forsythe.stanford.edu>
To: NA@score.stanford.edu
Subject: Junior visiting 1 year position in Aachen

The Institue for Geometry and Practical Mathematics of the
RWTH Aachen will have a junior level 1 year visiting position.
We are interested in a young scientist who has finished his Ph.D.
in numerical analysis and who would like to work on numerical
problems arising in climate modelling. We investigate energy
balance models in one and two space dimensions. Hence it
would be advantagous if the candidate has experience in one or more of
the fields: boundary value problems in ODEs, parabolic problems , bifur-
cation problems. Teaching duties are 2 hours of an excercise session each
week in numerical analysis.
Starting date is somewhere in October or early November, 1987.
The Institute has approximately 20 scientists. Most are working
in numerical analysis and only a few are working in
geometry and computer aided geometry.
For more information call Rolf Jeltsch (49) 241 803950
or write to Rolf Jeltsch
Institut fuer Geometrie und Praktische Mathematik
Templergraben 55
RWTH Aachen
D-5100 Aachen, Fed. Rep. of Germany

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

Date: Fri, 04 Sep 87 13:33:33 SET
From: Toint Philippe <PHTOINT%BNANDP10.BITNET@forsythe.stanford.edu>
Subject: Ron Dembo's address
To: People on NaNet <NA@score.stanford.edu>

Does anyone know an address (postal or e-mail) where Ron Dembo can
EFFECTIVELY be reached ? I would be delighted if you could forward
it to me...

Many thanks

Philippe Toint
Department of Mathematics
Facultes Universitaires ND de la Paix
61, rue de Bruxelles
B-5000 Namur (BELGIUM)
e-mail : PHTOINT@BNANDP10.BITNET
pto!fun-cs.uucp

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

Date: 3 Sep 87 21:31:43 GMT
From: ecsvax!urjlew@mcnc.org (Rostyk Lewyckyj)
Subject: Help interpolating function of three variables
To: na@score.stanford.edu

I am looking for code, preferably in FORTRAN, or references to code,
or a cookbook description for an algorithm, to do interpolation for
a function F(x,y,z) of three variables, given F tabulated on a regular,
i.e. xl,(dx),xh; yl,(dy),yh; zl,(dz),zh, grid.
The F that I am immediately concerned with is nicely continuous and
smooth, but I may want to handle Fs that are less well behaved.
The IMSL subroutine package has routines IBCCCU,IBCEVL and IBCIEU
to do BiCubic spline interpolation for functions G(x,y)
of two variables, but there is nothing for three variables.

Thank you in advance
Rostyk
--
Reply-To: Rostyslaw Jarema Lewyckyj
urjlew@ecsvax.UUCP , urjlew@tucc.bitnet
or urjlew@tucc.tucc.edu (internet)

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

Mail-From: GOLUB created at 4-Sep-87 13:32:45
From: prlb2!prlb.DECNET!courtois@seismo.CSS.GOV
Date: Fri, 4 Sep 87 19:54:49 +0100
To: seismo!score.stanford.edu!golub@seismo.CSS.GOV
Subject: New phone and Fax numbers for Philips Research

Since September 3, the telephone and fax numbers of the Philips Research
Laboratory in Brussels are changed:

New phone numbers: Central: 32 2 674 22 11
My direct line: 32 2 674 22 69

Fax number: 32 2 674 22 99.

P.-J. Courtois

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

Date: Fri, 4 Sep 1987 22:22 EDT
From: Richard M. Heiberger <V5807E%TEMPLEVM.BITNET@forsythe.stanford.edu>
To: <sadler@BUCKNER-EMH.ARPA>, <NA@score.stanford.edu>
Subject: Response to Sadler

Regarding the query by Sadler in v87, #67, let me recommend:
William G. Cochran and Gertrude M. Cox, Experimental Designs, Second Edition.
Wiley, 1957.

In addition to being an excellent text on the subject, they also include a
very comprehensive collection of designs. Plan 10.11 provides 3 weeks of the
12 in your example. Reassigning numbers to individuals and replicating the
entire plan four times will give you 12 weeks. It won't come out even in the
example, however, because person number one must be paired with three others
each week and the remaining 19 are not divisible by three. If you have some
freedom in the choice of number of treatments (persons), and block sizes
(number at each table), you can get a balanced design.

Richard M. Heiberger
Temple University

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

End of NA Digest

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