NA Digest Sunday, December 9, 1990 Volume 90 : Issue 43

Today's Editor: Cleve Moler

Today's Topics:


From: NA-Net Organizers -- Gene, Cleve, Jack, Mark and Bill
Date: Sun Dec. 9 19:40:59 PST 1990
Subject: NA-NET is Moving from Stanford to Oak Ridge

NA-NET is Moving from Stanford to Oak Ridge

Over the last few months we have rewritten much of the NA-NET
system and moved it to Oak Ridge National Laboratory. It is
running on the same machine that serves We have
added a number of features, such as auto-joining and auto-address
changes. The message below describes the current system in place at You should switch to using the Oak Ridge NA-NET, will be turned off soon.

The NA-NET originated at Stanford University by Gene Golub. Much of
the software we've used up to now was written by Mark Kent. The
Oak Ridge NA-NET operation will be under the direction of Jack Dongarra.
Bill Rosener of the University of Tennessee has rewritten the software.
Cleve Moler will continue to edit the NA-NET News Digest.

Beginning January 1, 1991 mail to will be forwarded
to for reforwarding.

This mail facility was created to allow numerical analysts (na) an
easy method of communicating with one another. The main advantage
of the NA-NET is uniformity of addressing. All mail is addressed
to the Internet host "" at Oak Ridge National
Laboratory. Hence, members of the NA-NET do not need to remember
complicated addresses or even where a member is currently located.
As long as moving members change their e-mail address in the NA-NET
(see feature 5 below) everything works smoothly.


1). Individual (unicast) messages
This is the most frequently used feature. Each member of the
NA-NET has a unique na-net name. This nanet name is usually
the same as the members last name. However, if there is more
than one member with the same last name then the first initial
is usually prepended to their last name to form their na-net name.
For example, if John Smith and Mark Smith both wanted to join the
na-net then mail addressed to "" would
be used to send mail to John Smith and ""
would be used to send mail to Mark Smith.

2). NA-NET News Digest
Any mail sent to "" will be considered
for distribution to all members of the NA-NET. About once a week
the Editor of the NA-NET News Digest will go over the messages which
have queued up, pick out the ones that are thought to be of general
interest to the numerical analysis and mathematical software
community, combine them in a News Digest format, and mail the
Digest to everyone on the mailing list.

3). Joining the NA-NET
To join the NA-NET send mail to "".
In the message body specify the following three fields.


The values can be specified in any order. The program which parses
your mail message is case sensitive. The subject line of your
message is ignored. Zero or more spaces can separate the key words
from there actual values. If there is a member of NA-NET
with the same lastname, then a message will be sent back to you.
If this happens try prepending your first initial to your last name.
In this case the "key" which identifies you will not be your lastname.
See Appendix A for example on how to join the na-net.

4). Removing membership
To remove your membership from the NA-NET send mail to
"". In the message body specify
the following three fields.


The values can be specified in any order. Again the subject line
of your message is ignored. Zero or more spaces can separate the
key words from there actual values. A message will be sent to both
the deleted address and the address making the request informing
you that your name has been removed. See Appendix A for example on
how to remove your membership.

5). Changing e-mail address
To change your e-mail address send mail to "".
In the message body specify the following four fields.


The values can be specified in any order. Again the subject line
of your message is ignored. Zero or more spaces can separate the
key words from there actual values. A message will be sent to both
the old-address as well as the new-address informing you that the
change has taken place. See Appendix A for example on how to change
your e-mail address.

6). Help
Questions and comments about the NA-NET should be addressed to:

A person will read all mail messages within reason to this

Mail sent to:

will return the message you are currently reading.

7). Current list of all members
Mail sent to the following address will result in a mail message
being sent back to you containing all members of the NA-NET.

Below are some real examples to give you an idea of how the
NA-NET works.

1). For example, to mail to Gene Golub.

mail to:

2). Mail sent to a nonexistent na-net name. For example,

mail sent to:

will result in the following message being returned to you.

NA-NET key not found. Message returned.

3). Mail sent to a non-unique na-net name. For example,

mail sent to:

will result in the following message being returned to you.

Ambiguous key: try one of the following

fox, david = na.dfox
fox, phyl = na.pfox
fox, x = na.cfox

4). An example of how Gene Golub would join the na-net.

Subject: anything

Lastname: golub
Firstname: gene

5). An example of how Mike Foobar would remove his membership
in the na-net.

Subject: anything

Lastname: foobar
Firstname: mike
key: foobar

6). An example of how Mike Fox would change his e-mail address.

Subject: anything

Lastname: fox
Firstname: mike
New-address: mfox@new.address
key: mfox

Jack Dongarra
University of Tennessee and
Oak Ridge National Laboratory


From: Dave Carlson <>
Date: Mon, 3 Dec 90 13:52:29 PST
Subject: Minisymposium on Teaching Issues in Linear Algebra


A minisymposium on Teaching Issues in Linear Algebra is being organized
for the SIAM Conference on Applied Linear Algebra September 11-14, 1991 in
Minneapolis. The session will include a report with recommendations from
the 1990 Workshop on the Undergraduate Linear Algebra Curriculum, and
discussion from the floor. In addition, there will be several presentations
on curricular and pedigogical issues in linear algebra teaching. (This
session is intended to complement the minisymposium "Teaching Linear Algebra
with Software Tools", also being organized for the meeting.)
If you would like to be considered as a presentor, please send, by January 5,
1991, an abstract or brief description of your proposed presentation to the
David Carlson
Mathematical Sciences Department
San Diego State University
San Diego, CA 92182-0314


From: Michael Mascagni <>
Date: Tue, 4 Dec 90 08:38:44 EST
Subject: Parallel Random Number Generation

I am interested in recent work on parallel random number generators. I am
most interested in SIMD/vector type generators, but recent MIMD work is also
of use. I am most interested in generators where at least something is known
about them theoretically, but any reference is appreciated.

I also would like to obtain software for statistical testing of RNGs for
quality in Monte Carlo. Either software or references to testing methods
would be appreciated (especially parallel testing methods)



From: Klaus Peters <kpeters@cdp.uucp>
Date: Mon, 3 Dec 90 13:39:13 -0800
Subject: Change of Address for Alice and Klaus Peters

Folowing up on earlier notice, we would like to announce that we
have joined Jones & Bartlett, Boston, as publishing partners. Our
new address is: Jones & Bartlett, 20 Park Plaza, Boston, MA 02116
(617) 482-3900. E-mail: cdp!

-- Alice and Klaus Peters


From: Tim Kelley <>
Date: Tue, 4 Dec 90 09:28:30 EST
Subject: Optimization Methods in Differential Equations and Control

A conference on ``Numerical Optimization Methods in Differential
Equations and Control'' will be held in Raleigh, NC, from July 15, 1991
through July 17, 1991 as a follow on meeting to the ICAIM. Over 30
speakers have agreed to attend and a poster session will be held for
contributed presentations. For additional information contact

C. T. Kelley,
Dept. of Mathematics, Box 8205
North Carolina State University
Raleigh, NC 27695-8205


From: David E. Bernholdt <uflorida!ufqtp!>
Date: 4 Dec 90 20:35:03 GMT
Subject: Sparse BLAS1

About 10 days ago, I sent out a request for information on sparse
BLAS-1 implementations. Here is a summary of the response...

There is a paper by Dodson, Grimes and Lewis (DGL) describing a set of
sparse BLAS-1 routines. The paper and model routines are available from
netlib. I understand it will be appearing in the ACM Transactions on
Mathematical Software as well. This is the most "popular" such
definition of which I am aware.

These routines have been implemented in the NAG Mark 14 Fortran

The IBM ESSL library implements 10 of the routines -- the single and
double precision, but not the comple or double complex.

Cray's SCILIB has several sparse BLAS-1 type operations. They have
different names and arguments from the Dodson, Grimes and Lewis

I have heard two different rumors about Cray implementing the DGL
definitions: From one source that they have been implemented,
but from another that they are not scheduled for release to the public.

In any event, it looks like the DGL proposal is begining to catch on
with the vendors.

As I suspected, there is little concensus on what "sparse blas" should
be. I think it is fairly clear for BLAS-1, but much less so for
higher levels. I get the feeling that most people working with sparse
problems are "rolling their own" basic routines. Some people
expressed doubt that sparse versions of the higher-level BLAS would ever
catch on.

That being said, I should note that the paper "Are there iterative
BLAS?" by Oppe and Kincaid includes a slightly different approach
sparse level-1 BLAS (as part of what they call "iterative BLAS") --
more general and somewhat more flexible than the DGL proposal, but
with basically the same operations available.

The Oppe & Kincaid paper is the only other proposal which received any
mention. They are aiming at a fundamental set of routines for the
development of sparse iterative solvers, so they include numerous
routines aimed at different common storage/problem structures.

Thanks to all who replied. I hope this summary is useful.

David Bernholdt
Quantum Theory Project bernhold@ufpine.bitnet
University of Florida
Gainesville, FL 32611 904/392 6365


From: David Brown <dlb%guarneri.c3@LANL.GOV>
Date: Mon, 3 Dec 90 04:53:51 MST
Subject: DOE Computational Science Fellowships Available

The following is an announcement for a new DOE Computataional
Science Graduate Fellowship program. This is a four-year all-expenses-paid
fellowship that includes summer employment at a DOE lab, a personal
workstation for the fellow and supercomputing time on NERSC computers.
Please let promising graduating seniors and first year grad students
at your departments know about this. U.S. citizenship is required.

Universities must obtain DOE accredidation to be eligible
to host these graduate fellows. You may wish to pursue this as



The program is designed to support highly capable sci-
ence and engineering students interested in pursuing doc-
toral study in applied sciences or engineering disciplines
with applications in high-performance computing. The pro-
gram will provide a stipend, tuition and fees, institutional
allowance, and funding for a work station for each of the
Fellows. The program is sponsored by the US Dept. of Energy
Scientific Computing Staff.


The program is open to U.S. citizens who hold a B.S.
degree in either the life or physical sciences, engineering
or mathematics. Applicants may be entering or current gradu-
ate students. The program will not accept students who have
already received department (faculty) approval on a Ph.D.
thesis topic.


For the 1991-92 awards, applications must be received
at the Science/Engineering Education Division, Oak Ridge
Associated Universities, by 4:30 p.m., January 28, 1991.
Fellowship selection will be announced by mid-April 1991.


Student applications consist of undergraduate and gra-
duate transcripts, faculty references, an academic and
career goal statement and a listing of work experiences and
publications. After the 1991-1992 cycle, GRE scores will be
required. Application forms should be requested from the
address given below.


To accept DOE Computational Science Fellows, universi-
ties will have to be approved for participation in the pro-
gram. Acceptance will be based on material submitted in the
university application. These materials include a descrip-
tion of the curriculum, enrollment data, previous and ongo-
ing research, postgraduate employment records, and faculty


Fellows will receive annuals stipends at $18,000 (first
year); $19,200 (second year); $20,400 (third year); and
$21,600 (fourth year). The program will also pay full tui-
tion and fees, an institutional allowance of $1,000 a year,
some travel expenses and matching funds for a workstation.


All fellows will be required to participate in a
research assignment at a DOE or DOE-approved facility in a
research assignment related to ongoing high-performance com-
puting activities. The required length of this assignment is
12 weeks. Under certain circumstances, Fellows may also per-
form their thesis research at DOE facilities.


Computational Science Graduate Fellowship Program
Science/Engineering Education Division
Oak Ridge Associated Universities
P.O. Box 117
Oak Ridge, Tennessee 37830-0117
Telephone: (615) 576-0128
Telefax: (615) 576-0202


From: Michael Osborne <>
Date: Fri, 7 Dec 1990 09:11:08 EST
Subject: Positions at Australian National University

at the Australian National University

The School of Mathematical Sciences seeks to make at least one new
appointment to augment its existing research programs in
areas of Advanced Computation. These include active programs in the solution
of partial differential and integral equations, optimization, computer
intensive statistical methods, and symbolic and exact computation, which
stress the development both of algorithms appropriate to innovative computer
architectures and of data visualisation techniques. These activities are
well supported by the University with computing facilities which include a
Fujitsu VP100 supercomputer (shortly to be upgraded to a VP2000) and a 16000
processor Connection Machine in addition to medium scale machines and
The University through its Centre for Information Sciences Research is an
affiliate of the Argonne ACRF and is receiving a prototype Fujitsu CAP
machine for joint development work.
These positions will be full time research positions at the level of
Postdoctoral, Research, and Senior Research Fellow. Part time appointments
are also possible, as are extended visiting appointments during leave of
absence from a home institution.
Enquiries and requests for further information can be addressed to Dr. M.R.
Osborne (email or, FAX
Closing date: 25 January 1991

Salary: Senior Research Fellow; $A45,729-$A54,255 p.a.
Research Fellow; $A33,163-$A43,096 p.a.
Postdoctoral Fellow Grade 2; $A33,163-$A43,096 p.a.
Postdoctoral Fellow Grade 1; $A28,792-$A32,762 p.a.
Salaries are presently under review.

Appointment: Senior Research Fellow /Research Fellow up to three years with
the possibility of extension to five years; Postdoctoral Fellow normally two
years with the possibility of extension to three years.

Applications: These should be sumitted in duplicate to the Registrar, The
Australian National University, GPO Box 4, Canberra, ACT 2601, Australia,
quoting the reference number, and including a curriculum vitae, and
the names of three academic referees. The University reserves the right not
to make an appointment or to appoint by invitation at any time. Further
information is also available from the Registrar.



From: Willard Miller <>
Date: Thu, 6 Dec 90 08:27:22 CST
Subject: IMA Postdoctorates in Industrial Mathematics



IMA announces 2 new one-to-two year positions in
Industrial Mathematics, effective September 1, 1991.
These appointments are in addition to the 4 IMA
Postdoctorates in Industrial Mathematics positions
announced earlier. The new positions are are funded
jointly by the IMA and by Cray Research. Industrial
Mathematics Postdoctorates are designed to prepare
mathematicians for research careers involving
industrial interaction. Applicants should
have received their Ph.D. in Mathematics or Applied
Mathematics by September 1, 1991. Background in
applied linear algebra and/or numerical analysis is
desired, but no knowledge in molecular biology is
required. Postdoctorates will spend 50% effort
working with scientists from Cray Research on one
of the following topics:

1) Protein folding for drug research

2) Searching algorithms for genetic simulation

and 50% effort in the regular IMA program.

IMA also announces a new one-to-two year position in Industrial
Mathematics effective mid-July, 1991. This appointment is
in addition to the 6 IMA Postdoctorates in Industrial
Mathematics positions announced earlier. The new position
is funded jointly by the IMA and by Siemens A G of Munich.
Industrial Mathematics Postdoctorates are designed to prepare
mathematicians for research careers involving industrial
interaction. Applicants should have received their Ph.D.
in Mathematics or Applied Mathematics by September 1, 1991.
Background in applied linear algebra and/or partial
differential equations and/or numerical analysis is desired,
but no knowledge in semiconductors is required. The
Postdoctorate will spend 50% effort in Munich working with
scientists from Siemens A G on the topic

Semiconductors: Processing Device and Circuit Simulation

and 50% effort in Minneapolis with the regular IMA program.
While in Munich the Postdoctorate will have contact with
researchers at the University of Augsburg. The position
will start in mid-July 1991 with the IMA summer program on

The following materials must be submitted (all material
should arrive by January 15, 1991):

(1) Personal statement of scientific interests,
research plans, and reasons for wishing to participate
in this program. (This is an essential part of the

(2) Curriculum vitae and a list of publications.

(3) Three letters of recommendation, to be sent
directly to the IMA.

All correspondence should be sent to

MINNEAPOLIS, MN 55455-0436

The University of Minnesota is an equal opportunity
educator and employer, and specifically invites and
encourages applications from women and minorities.

of Technology, Indiana University, Iowa State University,
Michigan State University, Northern Illinois
University, Northwestern University, Ohio State
University, Pennsylvania State University, Purdue
University, University of Chicago, University of
Cincinnati, University of Houston, University of
Illinois (Chicago), University of Illinois (Urbana),
University of Iowa, University of Manitoba,
University of Michigan, University of Minnesota,
University of Notre Dame, University of Pittsburgh,
Wayne State University.

Bellcore, Cray Research, Eastman Kodak, Hitachi,
General Motors, Honeywell, IBM, Motorola, 3M, UNISYS


Date: Wed, 5 Dec 90 09:28 EDT
Subject: Contents, SIAM Journal on Computing

April 1991 Volume 20, Number 2


Minimum Weighted Coloring of Triangulated Graphs, with Application to
Maximum Weight Vertex Packing and Clique Finding in Arbitrary Graphs
Egon Balas and Jue Xue

Approximate Levels in Line Arrangements
Jiri Matousek

Parallel Algorithms for Channel Routing in the Knock-Knee Model
Joseph JaJa and Shing-Chong Chang

Some Observations on Separating Complexity Classes
Ronald V. Book

An O(nlog2h) Time Algorithm for the Three-Dimensional Convex Hull Problem
Herbert Edelsbrunner and Weiping Shi

A General Sequential Time-Space Tradeoff for Finding Unique Elements
Paul Beame

The Power of Alternating One-Reversal Counters and Stacks
Oscar H. Ibarra and Tao Jiang

Interpolation and Approximation of Sparse Multivariate Polynomials
over GF(2)
Ron M. Roth and Gyora M. Benedek

Lower Bounds for Computations with the Floor Operations
Yishay Mansour, Baruch Schieber, and Prasoon Tiwari

Probably Approximate Learning of Sets and Functions
B. K. Natarajan

Efficient Parallel Algorithms for Testing k-Connectivity and Finding
Disjoint s-t Paths in Graphs
Samir Khuller and Baruch Schieber

Time and Message Bounds for Election in Synchronous and Asynchronous
Complete Networks
Yehuda Afek and Eli Gafni

Good and Bad Radii of Convex Polygons
Peter Gritzmann, Laurent Habsieger, and Victor Klee

For additional information please contact Vickie Kearn,
Publisher, Society for Industrial and Applied Mathematics (SIAM),
3600 University City Science Center, Philadelphia, PA 19104-
2688; (215) 382-9800; FAX: (215) 386-7999; E-mail:


Date: Fri, 7 Dec 90 13:01 EDT
Subject: Contents, SIAM Journal on Numerical Analysis

SIAM Journal on Numerical Analysis
APRIL 1991 Volume 28, Number 2


Convergence of the Grid-Free Point Vortex Method for the Three-Dimensional
Euler Equations
Georges-Henri Cottet, Jonathan Goodman, and Thomas Y. Hou

Convergence of a Point Vortex Method for Vortex Sheets
Thomas Y. Hou, John Lowengrub, and Robert Krasny

Numerical Approximation of the Solution of a Variational Problem with a
Double Well Potential
Charles Collins, David Kinderlehrer, and Mitchell Luskin

Spectral Approximations of the Stokes Equations with Boundary Conditions
on the Pressure
Christine Bernardi, Claudio Canuto, and Yvon Maday

Quasi-Optimal Pointwise Error Estimates for the Reissner-Mindlin Plate
Lucia Gastaldi and Ricardo H. Nochetto

Geometry Related Convergence Results for Domain Decomposition Algorithms
Tony F. Chan, Thomas Y. Hou, and P. L. Lions

The Finite Volume Element Method for Diffusion Equations on General
Zhiqiang Cai, Jan Mendel, and Steve McCormick

A Finite-Element Approximation Theory for the Drift Diffusion
Semiconductor Model
Joseph W. Jerome and Thomas Kerkhoven

A Spectral Method of Characteristics for Hyperbolic Problems
Endre Suli and Antony Ware

On Nonconforming Combinations of Various Finite Element Methods for
Solving Elliptic Boundary Value Problems
Zi-Cai Li

Convergence of Trust Region Algorithms for Optimization with Bounds
when Strict Complementarity Does Not Hold
M. Lescrenier

Some Runge-Kutta Formula Pairs
J. H. Verner

Estimation of the Error in the Reduced Basis Method Solution of the
Differential Algebraic Equation Systems
Meeei-Yow Lin Lee

A Method of Analytic Centers for Quadratically Constrained Convex
Quadratic Programs
Sanjay Mehrotra and Jie Sun

A Globally Convergent Augmented Lagrangian Algorithm for Optimization
with General Constraints and Simple Bounds
Andrew R. Conn, Nicholas I. M. Gould, and Philippe L. Toint

Polynomial Extrapolation from [-1,1] to the Unit Disc
L. Brutman

For additional information, contact Vickie Kearn, Publisher,
Society for Industrial and Applied Mathematics (SIAM), 3600
University City Science Center, Philadelphia, PA 19104-2688;
telephone (215) 382-9800; FAX: (215) 386-7999; e-mail:


End of NA Digest