NA Digest Saturday, March 5, 1988 Volume 88 : Issue 10

Today's Editor: Cleve Moler

Today's Topics:


>From: Xiao-He Zhang <xiaohe@Tybalt.Caltech.Edu>
Date: Mon, 29 Feb 88 04:48:15 PST
Subject: Finding Degenerate Eigenvectors

Hi, netters,

I am looking for an algorithm, or better a subroutine I can call,
to find all DEGENERATE eigenvectors of a general real or complex
matrix of size < 10 x 10. I have tried those subroutines for GENERAL
real or complex matrices in LINPACK, EISPACK and IMSL. They give me
correct eigenvalues and both left and right eigenvectors when the
matrix is not degenerate, i.e when all eigenvectors are different.
However, they give me at best identical eigenvectors for the degenerated
eigenvalue, which is incomplete for my problem; or something does not
make sense at all. I searched LINPACK and EISPACK for the keyword
"degenerate". Nothing appropriate was found.
If any of you have some suggestion, please e-mail it to me as I do not
read netnews as often as I SHOULD. Thank you for your help.


Xiao-He Zhang || | xiaohe@caltech.bitnet


>From: Kris Stewart <>
Date: Tue, 01 Mar 88 05:00:04 PST
Subject: Numerical Analysis Course Numbering

I belong to a Math Sciences Department which includes as separate
groups: Math, Applied Math, Statistics, Math Education and Computer
Science. I teach Numerical Analysis and am a member of the computer
science group (as a developer of mathematical software this is where
I felt most comfortable). In response to a department review from last
year, we are renaming computer science courses within the department
as CS ###, primarily to aid students in CS whose transcripts don't
immediately reflect CS. I have always felt Numerical Analysis should
be placed in between Applied Math and Computer Science and would like
the course double numbered, i.e. the same course would appear in the
catalog as CS ### and MATH ###. The computer science group is in favor
of this, some in the math groups are vehemently opposed, offerring
the justification that 'other schools' put Numerical Analysis in the
Math Department.

I would really like to hear the opinion of other practicing Numerical
Analysts on how things are done at your University. Perhaps also, how
you wish things were done at your University.

Kris Stewart
Department of Mathematical Sciences
San Diego State University
San Diego, CA 92182 (619-942-1012)
( or q300058@calstate.bitnet)


>From: R. G. Bartle <RGB@MATH.AMS.COM>
Date: Tue 1 Mar 88 11:23:54-EST
Subject: Gateaux

For some time I have felt uneasy about the circumflex in Gateaux' name,
since I notice that it does not appear on (at least some of) his papers.
I checked with the grand dragon, J. Dieudonne', who reports that
Gateaux' name appears on the official roster of all alumni of the Ecole Normale
Sup. who were killed in WW I, and that his name appears WITHOUT a circumflex.
Dieudonne' writes: "I think this clinches the matter." So do I.
I never thought that Gateaux was a piece of cake.


>From: Adrian Baddeley <munnari!natmlab!dmsim!>
Date: 1 Mar 88 23:47:16 GMT
Organization: CSIRO, Div. of Maths and Stats, Lindfield, Aust.
Subject: Re: Triangulation of 2-d domains

In article Volume 88, #8, ashcraft@yale.UUCP (Cleve Ashcraft) writes:

>I am interested in ordering sparse matrices arising from 2-d finite
>element problems which use linear triangular elements. The density
>of the grid points in the domain should be roughly equal throughout
>the domain, and the triangularization should be "good". The
>triangularization should exhibit no apparent regularity, this last
>property is very important.

Try using the Delaunay triangulation, e.g.
Lee D.T. and Schachter B J, Two algorithms for
constructing a Delaunay triangulation,
Int. J. Comput. Inform. Sci. 9 (1980) 219-242.

Green P J and Sibson R , Computing Dirichlet tessellations in
the plane. Computer J. 21 (1978) 168-173.

Given a finite set of points in the plane, this produces a triangulation
(of the convex hull of the points) which uses the given points as
the triangle vertices. So you can produce the type of regularity you want
by strewing points in some quasi-regular manner over your plane region.

There is an engineer/programmer called S.W.Sloan from the University of
Newcastle (Australia) who has written and used Delaunay triangulation
algorithms for finite element problems. The algorithms are published but,
sorry, I don't have the reference to hand.
Adrian Baddeley, CSIRO Division of Mathematics & Statistics, Sydney, Australia.
PO Box 218, Lindfield NSW 2070, Australia. Phone: +61 2 467 6062 (24 hrs)


>From: Y. F. Chang <>
Date: Fri, 4 Mar 88 11:11 PST
Subject: Request for DAE test problems

This is a request for multiple-constraint problems, or DAE's with
more than one algebraic equation.

I have just finished writing and testing a program to solve
multiple-constraint problems. It was tested on a compound
pendulum, where a second pendulum hangs from the first, which
in turn hangs from the ceiling. I shall next test the program
on a coupled pendulum, where the energy transfers back and forth
between a pair of pendula.

I seek additional examples for test of robustness.

Since my BITNET address is new to the system, some nodes may not
recognize the 'CLAREMONT.EDU'. In that case, try


If you prefer the USMail, try

Y. F. Chang
976 W. Foothill Blvd
# 298
Claremont, CA, 91711


>From: George Adams <>
Date: Thu, 3 Mar 88 22:23:51 EST
Subject: Supercomputing '88 Conference


November 14-18, 1988
Hyatt Orlando
Kissimmee, Florida, USA

Sponsored by:
Computer Society of the IEEE and ACM SIGARCH

In Cooperation with:
Argonne National Laboratory, Lawrence Livermore National Laboratory,
Los Alamos National Laboratory, NASA Ames Research Center,
National Center for Atmospheric Research,
National Science Foundation, SIAM, Supercomputing Research Center

Papers submitted by: March 14, 1988
Research exhibit proposals due: April 4, 1988
Poster proposals due: August 2, 1988

Conference Highlights-
Keynote Speaker: Seymour Cray, Cray Research
Banquet Speaker: Carl Conti, IBM Enterprise Systems
ACM Annual Computer Chess Tournament

Supercomputing '88 is a new conference that will bring together
supercomputing system researchers, designers, and users to
report new advances and experiences, state needs, suggest future
directions, and contribute to discussions. It will include
tutorials, a high quality technical program, on-line and video
taped demonstrations, informal poster sessions, vendor and
university exhibits, and product briefings.

TOPICS OF INTEREST. Examples include, but are not limited to,
the following:
Science and Supercomputing
The Impact of New Technology on the Future of Supercomputing
Supercomputing Execution Environment
Supercomputing Development Environment
Supercomputing Application Environment
Supercomputing System Evaluation
Supercomputing Management Issues
Mass Storage and Supercomputers
Technical Aspects of Products
User Experience

PAPERS. Authors are invited to submit papers which report
concrete results and experience. Papers reporting important
negative results are also encouraged. Selection criteria will
include originality, clarity, and relevance.

Requirements: Papers must be original material not previously
published. Papers must be submitted without conditions; authors
must obtain any necessary approvals and/or clearances prior to
submission. Copyright release will be required. Authors of
accepted papers will be responsible for retyping corrected
papers on special forms to be provided and for preparing visual
material for their presentations using guidelines to be
provided. Camera-ready copy is due July 18, 1988. Presentation
visual material is due for quality review October 4, 1988.

Instructions: Submit five copies to the Program Chairman by
March 14, 1988. Papers must be in English, be typed double-
spaced, and not exceed 25 pages (about 5000 words). Papers must
have: (1) a title page that lists the name, mailing and
electronic address, and telephone number for each author; (2) an
abstract; (3) keywords; (4) and the presentation media
requirement. For multiple author papers, identify the
corresponding author and the presenting author.

RESEARCH EXHIBITS. Some space will be available for researchers
with demonstration-oriented exhibits of their research.

Instructions: Contact the Program Chairman.

POSTERS. In addition to informal evening poster sessions, an
on-line poster session will be scheduled where people who have
developed interesting applications will demonstrate them using
exhibitor equipment.

Instructions: Contact the Program Chairman. Proposals for on-
line posters should be made jointly with the collaborating

sessions will be organized so that supercomputing center
managers can share recent progress, discuss common problems, and
consider opportunities for collaboration.

Supercomputing '88 Organizing Committee
General Chairman George Michael, LLNL
Program Chairman Stephen Lundstrom, Stanford
University and PARSA
Deputy Chairman Robert Voigt, ICASE
Exhibits Chairman Roger Anderson, LLNL
Finance Chairman Sidney Fernbach, consultant
Local Arrangements Chairman Dennis Duke, Florida State Univ.
Publication Chairman Harlow Freitag, SRC
Publicity Chairman George Adams, Purdue University

Supercomputing '88 Advisory Committee
Robert Borchers Lawrence Livermore National Laboratory
Bill Buzbee National Center for Atmospheric Research
F. Ron Bailey NASA Ames Research Center
Melvyn Ciment National Science Foundation
Jack Dongarra Argonne National Lab
Doug DeGroot ACM-Sigarch; Texas Instruments
Joanne Martin IEEE Computer Society, TC on Supercomputing; IBM
Norman R. Morse Los Alamos National Laboratory
Paul Schneck Supercomputing Research Center
Daniel Sorenson Argonne National Laboratory

For information on the conference, program, or exhibits contact
one of the following:

General Chairman Program Chairman Exhibits Chairman
George Michael, L-306 Stephen F. Lundstrom Roger Anderson, L-306
P. O. Box 808 Stanford University P. O. Box 808
Livermore, CA 94550 Stanford, CA 94305 Livermore, CA 94550
(415) 422-4239 (415) 723-0140 (415) 422-8572

For registration information contact the Computer Society of the IEEE,
1730 Massachusetts Ave., N.W., Washington, DC 20036-1903. (202) 371-1013


End of NA Digest