NA Digest Tuesday, September 21, 1993 Volume 93 : Issue 36

Today's Editor:

Cleve Moler
The MathWorks, Inc.

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to


From: Tony Chan <>
Date: Fri, 17 Sep 1993 16:28:30 -0700 (PDT)
Subject: Message from Heidi Householder

I'd like to relate the following message from Heidi Householder,
the widow of Alston Householder, which I summarized from a
recent phone conversation with her:

1. She would like to thank everybody who has sent letters of
condolence to her after Alston's death. She plans to
personally write everybody back but right now she is not
in good enough shape to do that yet.
2. She wants to make a correction to a mistake in some public
reports of Alston's death: Alston didn't die at home but at
the Santa Monica Hospital.
3. Several people took photos of the Householders at the
Householder Symposium at Lake Arrowhead and had promised
to send copies to them. She'd like to remind
people that she'd be very happy to receive those photos.
Her address is: 6235 Tapia Dr., Malibu, CA 90265.

- Tony Chan


From: Arnold Neumaier <>
Date: Fri, 17 Sep 93 08:33:32 +0200
Subject: Change of Address for Arnold Neumaier

New address, valid starting September 20, 1993:

Arnold Neumaier
c/o James E. Schmidt
30 Johnson Drive
Murray Hill, NJ 07974 USA

phone: 908-464-5694

work address:

Arnold Neumaier
AT&T Bell Laboratories
600 Mountain Avenue, Room 2C-463
Murray Hill, NJ 07974-0636


Arnold Neumaier


From: Richard Brankin <>
Date: Wed, 15 Sep 1993 09:08:13 +0100 (BST)
Subject: Help for IEEE Floating Point Woes

I imagine that there are many who experience the same difficulties as
Joe Grcar's colleague which were described in last week's NA Digest.
Also, I imagine that the non-hackers never get to grips with what they
could do to overcome the difficulties. The following may be of help,
for SGI Indigo users anyway,


and don't forget to link with "-lfpe" when building an executable. This
has the effect "underflow to zero; bomb out with a traceback when either
overflow, or divide by zero, or an invalid operand is detected", with
which I expect most people would be happy. And, for what it's worth, the
f77 compile flags "-g -C -u -automatic -trapuv" help to cut out many
programming errors.

As usual, the gory details are in the man pages, somewhere! :-)

Richard Brankin
NAG Ltd.


From: Simon Chamlain <>
Date: Wed, 15 Sep 93 18:45:24 EDT
Subject: Circuit Drawings

A colleague and I are writing a manual in which we have to include
some circuit pictures. Does anyone knows a software (shareware or
commercial) that does circuit drawing (without any simulation)
and saves it in a well known graphic format (TIFF, CGM etc) ?

Any information on such a software and the way I can get it (or
order it) will be greatly appreciated.



From: William B. Sawyer <>
Date: Wed, 15 Sep 93 09:46:50 +0200
Subject: HPF codes sought

Could anyone give me a pointer to High Performance Fortran codes in the public
domain? Our research institute is involved in a project to test a HPF
translator (which takes HPF code and generates F77 with message passing
primitives). Basically any type of codes will do, but preferably numerical
applications which are matrix operation intensive.

Thanks for any suggestions,
Will Sawyer
Centro Svizzero di Calcolo Scientifico


From: Takeshi Nanri <>
Date: Sat, 18 Sep 93 19:17:05 JST
Subject: Availability of SIMPLE

Could anyone kindly tell us availability of the program
called SIMPLE, a two-dimensional Lagrangian hydrodynamics code,
which was reported in SIGPLAN.
We would like to use the code for benchmarking of a parallel system.
The code is said to be developed at LLNL.

Takeshi Nanri


From: Jack Dongarra <>
Date: Tue, 14 Sep 93 10:58:06 -0400
Subject: Position at University of Tennessee

The Department of Computer Science at The University of Tennessee,
is seeking applicants for the position of Senior Research Associate.

Applicants should have a Master's degree in Computer Science, Electrical
Engineering or have the equivalent background. Experience with computer
networks, TCP/IP, Unix operating system, the design of computer systems,
X-window applications, parallel architectures, scientific computing,
interfacing peripherals with computer systems and use of protocol analyzers
would be pertinent.

Responsibilities include designing, prototyping, testing, maintaining, and
documenting computer systems related to network computing.

The position involves developing software related to the PVM project.
Familiarity with parallel architectures and algorithms is also desired.
Additional benefits of the position include a competitive salary,
travel opportunities, access to state-of-the-art computational facilities
(including both parallel architectures and high-performance workstations),
and collaborative research opportunities in a very active research
program in advanced scientific computing.

Inquiries should be directed to:
Jack Dongarra
Computer Science Department
University of Tennessee
Knoxville TN 37996-1301

Phone: (615) 974-8295, Fax: (615) 974-8296


From: Stig Skelboe <>
Date: Tue, 14 Sep 93 23:13:13 +0200
Subject: IFIP WG 2.5 Symposium

Copenhagen, 21-22 October, 1993

Symposium organized by IFIP WG 2.5 and
UNI-C, The Danish Computing Centre for Research and Education

Technical Session I. (Thursday, 21st October 1993, 9.00-12.00)

Opening of the 20th Meeting of IFIP WG 2.5 (Fosdick)

Opening of the Technical Meeting on "Numerical Software" (Skelboe)

Peter Jensen: Evaluation of ship hull seakeeping and wave resistance
Sven Mattisson: CONCISE - a parallel electrical circuit simulator
based on waveform relaxation
Zahari Zlatev: Large-scale air pollution modelling

Methods, Algorithms, and Implementations
Neil D. Jones: When does specialization pay off on practical

Technical Session II. (Thursday, 21st October 1993, 14.00-17.00)

Methods, Algorithms, and Implementations
Ulrich Kulisch: Computer arithmetic - Recent developments in
hardware and software
Kaj Madsen: A new continuation algorithm for linear programming
John Reid: The new Harwell sparse matrix codes MA47 and MA48
Kjell Gustafsson: A feedback control view of Runge Kutta methods
and their implementation

Wine and cheese (Thursday, 21st October 1993, 17.00-19.00)

Technical Session III. (Friday, 22nd October 1993, 9.00-12.00)

Libraries, Problem Solving Environments and Tools
Per Christian Hansen: Matlab regularization tools
Siegfried Rump: PROFIL - A fast interval library
Per Grove Thomsen: SIMPLE 2000 - an ODE solving environment

Technical Session IV. (Friday, 22nd October 1993, 14.00-17.00)

Proposals and Standards
Ulrich Kulisch: Proposal for accurate floating-point vector
John Reid and Richard Hanson: Condition handling in Fortran
John Reid: Supporting the IEEE standard in Fortran
M. Vouk: Numerical features in C - Status report on the
standardization effort

The workshop will be held in "Lille Auditorium", Department of
Computer Science, University of Copenhagen, Universitetsparken 1,
DK-2100 Copenhagen, Denmark.

Please register by sending name, address and E-mail address to:
Stig Skelboe, UNI-C, Building 305, DTH, DK-2800 Lyngby, Denmark,
E-mail: fax: +45 45 93 02 20
Please indicate if you need travel and/or hotel information.


From: H.J.J. te Riele <>
Date: Wed, 15 Sep 1993 12:02:28 +0200
Subject: Symposium on Massively Parallel Computing and Applications


In 1993-1994, the Centre for Mathematics and Computer Science Amsterdam (CWI)
and the University of Utrecht (RUU) are organising a series of symposia
on massively parallel computing and applications. The first meeting, held
on June 4, 1993, was devoted to: Topics in Environmental Mathematics.
The second meeting is focused on the theme:


Date: September 24, 1993
Location: CWI, Kruislaan 413, 1098 SJ Amsterdam
Room: Z009


Patrick Dewilde and Alle-Jan van der Veen (TU Delft)
Approximation of Matrices with Matrices of low
(computational) complexity
Arnolf Reusken (TU Eindhoven)
A Robust and easy parallelizable multigrid method for
2D convection-diffusion
L.M. Freeman and J.M. Bull (Univ. of Manchester, UK)
Parallel Algorithms for Numerical Integration
Stevan Vandewalle, Graham Horton (KU Leuven)
On the Massively Parallel Solution of Parabolic PDEs

NOTE: The lecture by Margreet Louter-Nool which was announced earlier
by ordinary mail unfortunately has to be canceled.

The organisers:

H.J.J. te Riele (CWI, tel. 020-5924106)
(send your email address to if you wish to receive a LaTeX-file
of the abstracts of the lectures)
H.A. van der Vorst (RUU en CWI)


From: John Tucker <>
Date: Wed, 15 Sep 93 14:05:18 EST
Subject: Statistical Methods in Software Engineering


[Please note there is no registration fee to attend, and you
may register either by email to John Tucker, or by registering at the
registration desk in the National Academy of Sciences building, 2100
C Street, N.W. on the day of the forum. You may either make hotel
reservations through National Academies Travel (to obtain a
discount for attendees, 1-800-367-2038, or 202-334-3768) at the One
Washington Circle Hotel, One Washington Circle, N.W.,
Washington, D.C., 20037, or you may make your own arrangements
at any hotel or motel of your choice. Please also note that, although
the forum itself ends at Noon on Oct. 12, the afternoon of 10/12 is
devoted to a meeting of the Panel on Statistical
Methods in Software Engineering in Room 250 of the same NAS
building. The open session of that panel meeting can be attended
by interested guests, but seating is limited there; please indicate on
the registration form if you want to be a guest at the panel meeting
open session.]

(No Registration Fee)
October 11-12, 1993
Auditorium, National Academy of Sciences,
2100 C Street, N.W., Washington, DC

Monday October 11

8:00 - 9:45 Software Process
Chair: Gloria Davis (NASA)
Speakers: Ted Keller (IBM)
David Card (CSC)

10:15 - 12:00 Software Metrics
Chair: Bill Curtis (SEI)
Speakers: Victor Basilli (U Maryland)
John Munson (U Florida)

1:00 - 2:45 Software Dependability & Testing
Chair: Rich DeMillo (U Purdue)
Speakers: John Knight (U Virginia)
Dick Lipton (Princeton U)

3:15 - 5:00 Case Studies
Chair: Daryl Pregibon (AT&T)
Speakers: Tsuneo Yamaura (Hitachi)
Stuart Zweben (Ohio State U)

Tuesday October 12

8:00 - 9:45 Nonstandard Methods
Chair: Sid Dalal (Bellcore)
Speakers: Madhav Phadke (Phadke Associates)
Eric Sumner Jr (AT&T BL)

10:15 - 12:00 Software Visualization
Chair: Steve Eick (AT&T)
Speakers: Will Hill (Bellcore)
John Stasko (Georgia Tech)

1:00 - 1:30 Closed Session Panel Meeting

1:30 - 5:00 Open Working Session of Panelists


From: Iain Duff <>
Date: Thu, 16 Sep 93 15:57:14 +0100 (BST)
Subject: Report on Fox Memorial and Prize Meeting

Symposium in Honour of Leslie Fox and the Sixth Leslie Fox Prize

The Sixth Leslie Fox Prize Competition was the first to be held since
Leslie's death and was combined with a Symposium in his honour. Both were
held in the Nuclear Physics Lecture Theatre in Oxford, the Prize meeting on
Thursday June 24th with the Symposium starting that evening and continuing
on the Friday.

The attendance on both days of around eighty was very good given the fact
that many British Universities were still involved with examiner's meetings
and other end-of-session delights. Even more gratifying was the number of
participants from overseas, several of whom were visitors to Oxford when
Leslie was there.

The audience was treated to two excellent days of numerical analysis, the
first in the main serious, the second with many reminiscences mixed
into some very entertaining and thought provoking presentations.
Speakers, chairmen, and attendees all combined to make a very fitting
tribute to Leslie Fox.

The Adjudicating Committee for the Fox Prize comprised Nancy Nichols
(chairman from Reading), Charlie Elliott (Sussex), and Christopher
Baker (Manchester). They had the difficult task, first of selecting
finalists from nineteen very high quality entrants, then the even more
unenviable task of choosing prizewinners from the finalists. The finalists
and their subjects are given below and, since I do not feel I could
do justice to their excellent papers and presentations by summarizing them
here, I leave it to readers to enquire further directly with the
authors. The cast (in order of appearance) was:

D.J. Higham (University of Dundee)
"The Dynamics of Variable Stepsize Runge-Kutta Algorithms"

Z. Jia (University of Bielefeld)
"Generalized Block Lanczos Methods for Large Unsymmetric Eigenproblems"

R. Mathias (University of Minnesota)
"The Stability of Parallel Prefix Matrix Multiplication with
Applications to Tridiagonal Matrices"

A. Edelman (University of California, Berkeley)
"Eigenvalue Roulette and Random Test Matrices"

P. Lin (Oxford University)
"Characteristic Galerkin Schemes for Scalar Conservation Laws in
Two Space Dimensions"

Y. Li (Cornell University)
"On the Convergence of Reflective Newton Methods for Large-scale
Nonlinear Minimization Subject to Bounds"

The standard was so high that the Committee decided to award all finalists
a prize with a First Prize being awarded to Yuying Li from Cornell
University and Second Prizes to Alan Edelman, Des Higham, Zhongxiao Jia,
Peixiong Lin, and Roy Mathias. Clemency Fox, who attended all
the sessions of the meeting, presented the prizes which included book
donations from IMA, OUP, and Chapman and Hall in addition to a monetary

A feature of the day was that the chairmen were invited to reminisce prior
to fulfilling their primary role. To this end Nancy Nichols, Iain Duff, and
John Reid all made comments about the origins and history of the
Fox Prize Competition and their memories of Oxford in the golden days of
Leslie. The art of chairmanship was even more displayed in the evening
when Mike Powell, on the Thursday evening session of the Symposium,
gave an entertaining description of sporting exchanges between the
Cambridge and Oxford golf teams using the actual trophy as a visual
aid. The speaker for the session was Gene Golub which was doubly
appropriate because not only was he a friend and visitor to Oxford but
he was also a prime instigator in the foundation of the Leslie Fox Prize.
Gene spoke on modified eigenvalue problems with reference to a very early
paper by Leslie on the computation of latent roots, prompting a lively
debate on nomenclature. The tone of the whole meeting was well set by the
excellent atmosphere established by Mike and Gene in this first session
of the Symposium.

The setting of the Fellows Garden at Balliol for the aperitifs was a
marvellous prelude to a memorable dinner at Balliol which I am sure would
have been greatly enjoyed by Leslie himself. Indeed several speakers
remarked that Leslie would have very much enjoyed the proceedings although
he might have found the alternative fixture of the visiting Australian
cricket team at the Parks a little tempting. The dinner, made even
more memorable by the quality of the wine sponsored by NAG Ltd, was a time
for more reminiscing led by entertaining and informative speeches by
Bill Morton who introduced Brian Ford who then, after some remarks of his
own, introduced Charles (E.T.) Goodwin the after dinner speaker.
Charles spoke mainly about Leslie's pre-Oxford days at the National
Physical Laboratory. The banter and comraderie continued well into
the evening in the Senior Common Room and, after what for some seemed a
rather short night, at breakfast on the Friday morning.

Suitably refreshed, we reconvened in the Nuclear Physics Lecture
Theatre for a wide range of contributions from speakers, chairmen, and
floor. The mathematical range was testimony to the great breadth and
influence that Leslie brought to Numerical Analysis and the range in mode
of presentation reflected the admiration and esteem in which all
participants held him. After some opening remarks by Bill Morton, Geoff
Hayes took the chair. Geoff shares a distinction with the first speaker,
Donald Kershaw, of having been to the same school as Leslie in Yorkshire.
Donald spoke about some work of Leslie Fox with Linda Hayes at Oxford
related to work of Wronski that has generated recent interest in the
numerical linear algebra community. Frank Olver then gave a highly
entertaining and informative talk on "Superasymptotics". The second
morning session was chaired by Sean McKee, the first UCINA coordinator and
golfing colleague of Leslie, and included talks by Alan Taylor on
differential equations arising from the Study Groups with Industry and by
an ex-Oxford student Nick Gould on linear algebra issues in optimization.

After a most pleasant lunch in Balliol, the Symposium recommenced with a
talk by Andrew Stuart, another ex-Oxford student and possibly the last
to be directly influenced by Leslie. He talked about "Analysis and
computations for a model of phase transitions". The chairman of the
session was Charles Clenshaw who was a colleague of Leslie's at the
National Physical Laboratory. Hans Stetter, who had visited Oxford several
times as a guest of Leslie gave the second talk on "Defect correction from
Gauss to the present day" and was followed by Leslie's Oxford colleague
and co-author David Mayers on the subject of "Relaxation" .. an appropriate
title for the last talk and a topic on which Leslie had been involved
since his early work with Southwell. The meeting was brought to a
close with characteristically witty remarks by John Mason who irreverently
reviewed the talks at the Symposium as if they were contenders for a
senior citizen's Leslie Fox Prize.

At the end of the meeting, presentations were made to the lassies from
the Oxford Computing Laboratory who had worked hard to ensure the smooth
running of both the meeting and the arrangement with Balliol College.
At the final afternoon tea, a very lively conversation amongst all
participants, friends of Leslie and thence friends together, concluded
a most successful and enjoyable event and one which did much to uphold the
memory of our recently departed dear friend and colleague.

Although registration costs were kept low, the generosity of several
sponsors (Chapman and Hall, ICL, IMA, NAG Limited, Nuclear Electric, and
OUP) enabled a healthy surplus to be donated to the Leslie Fox Prize Fund
which will be used to support further Leslie Fox Prize Competitions.

A memorial booklet is planned which will include, in addition to abstracts
of the presentations, some biographical notes and a bibliography, reports
on the Memorial Service held at Balliol in January, a list of students from
Oxford during Leslie's reign, and a history of the Leslie Fox Prize. This
booklet will be sent to all participants at the meeting but extra copies
will be available from the Oxford University Computing Laboratory.

Iain Duff
1 September 1993


From: Lester Ingber <>
Date: Fri, 17 Sep 1993 09:22:05 -0700
Subject: Adaptive Simulated Annealing (ASA)

Adaptive Simulated Annealing (ASA) version 1.43

To get on or off the ASA email list, just send an email to with your request.

Significant CHANGES since 1.34

Remarks were added in the NOTES for HP support, and for Turbo C, Turbo
C++, and MS Quick C on PCs.

Another option was added, the ASA_PRINT_MORE Printing Option, to give
more intermediate printout, i.e., the values of the best parameters and
cost function at each new acceptance state.

Wall Street Journal

I have been told that the WSJ will mention the world-wide use of the
ASA code, in an article to appear soon. I gave examples of some
projects using ASA, but I had to insist that the relevant people would
have to be contacted previous to citing them; see the related comment
in General Information below. Of course the press has the last word on
what they will publish/interpret.


I regularly receive requests to be able to run ASA with FORTRAN. I
cannot maintain both a C and a FORTRAN code, but there does seem to be
a genuine need to interface ASA with FORTRAN. In the NOTES are a
couple of suggestions: (1) Use f2c on FORTRAN programs; I have done
this and it works very well. (2) Try CFORTRAN to interface the C and
FORTRAN codes: (a) call FORTRAN from ASA, e.g., from cost_function(),
call a FORTRAN function that performs the actual calculation of the
cost function; (b) call ASA from FORTRAN, e.g., using the ADAPTING
section in the NOTES as a guide, call asa_main() from a FORTRAN
function. Can someone prepare templates for (a) and/or (b)? This
probably isn't easy to prepare for public release; about half a dozen
people started, but didn't complete such a project.

The new reference for this preprint in is
%A L. Ingber
%T Simulated annealing: Practice versus theory
%J Mathl. Comput. Modelling
%D 1993
%P (to be published)
As announced previously, this is a much expanded version of the
original draft, e.g., including new ideas and calculations regarding
"quenching." In the acknowledgements, I give a sincere thanks to the
many users who read parts of previous drafts and who sent me their own
(p)reprints on simulated annealing. I'd be interested in hearing about
any systems that find the QUENCHing options as useful (lucky?) as the
ASA test problem did in this paper.

General Information

The latest Adaptive Simulated Annealing (ASA) code and some related
(p)reprints in compressed PostScript format can be retrieved via
anonymous ftp from [] in the pub/ingber

Interactively: ftp, [Name:] anonymous, [Password:]
your_email_address, cd pub/ingber, binary, ls or dir, get
file_of_interest, quit. The latest version of ASA is asa-x.y.Z (x and
y are version numbers), linked to asa.Z. For the convenience of users
who do not have any uncompress utility, there is a file asa which is an
uncompressed copy of asa-x.y.Z/asa.Z; if you do not have sh or unshar,
you still can delete the first-column X's and separate the files at the
END_OF_FILE locations. There are patches asa-diff-x1.y1-x2.y2.Z up to
the present version; these may be concatenated as required before
applying. The INDEX file contains an index of the other files.

If you do not have ftp access, get information on the FTPmail service
by: mail, and send only the word "help" in the
body of the message.

If any of the above are not possible, and if your mailer can handle
large files (please test this first), the code or papers you require
can be sent as uuencoded compressed files via electronic mail. If you
have gzip, resulting in smaller files, please state this.

Sorry, I cannot assume the task of mailing out hardcopies of code or

People willing to be contacted by others who might be interested in
their ASA projects could keep me informed on (1) the title and/or short
description of their project, and (2) whether I have permission to
release their names as well as the description of their projects.


Prof. Lester Ingber
Lester Ingber Research
P.O. Box 857 EMail:
McLean, VA 22101 Archive:


From: Carlos Moura <>
Date: Tue, 14 Sep 93 14:48:12 EST
Subject: Contents: Computational and Applied Mathematics

Table of Contents for
(Matematica Aplicada e Computacional)
V.12, n.1, '93

Methods for Comparison in Functional Differential Equations

Nonlinear Plane Waves in the Klein-Gordon-Maxwell Field Theory

J. A. M. Felippe-de-SOUZA and A. J. PRITCHARD
A note on Overall Observability: the joint Problem of State Estimation
and Parameter Identification

Local Controllability of Nonlinear Systems on Surfaces

S. M. S. GODOY and J. G. dos REIS
The ``Square Wave'' type Functions as a Limit of Solutions of
three Mathematical Models

Exponential Decay for a Class of Initial Boundary Value
Problems in Thermoelasticity


End of NA Digest