NA Digest Sunday, March 20, 1994 Volume 94 : Issue 12

Today's Editor:
Cleve Moler
The MathWorks, Inc.

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to


From: Bharat Kumar <>
Date: Mon, 14 Mar 1994 14:53:48 -0500 (EST)
Subject: Ordering for Sparse Cholesky Factorization

I'm interested in obtaining the code for determining an ordering of
a sparse matrix that minimizes the number of parallel elimination
steps among the class of perfect orderings (orderings with no fill),
based on the approach by Jess and Kees. I would appreciate any
pointers about where I could obtain the above source code.

Thanks in advance,

Bharat Kumar
Dept. of CIS, The Ohio State University


From: Andrei Knyazev <>
Date: Mon, 14 Mar 94 14:17:03 -0500
Subject: Complex Eigenvectors from MATLAB

I was surpised by the following results from MATLAB.

| 10 5 6 | | 14 -4 8 |
A = | 5 10 6 |, B = | -4 4 -3 |
| 6 6 10 | | 8 -3 6 |

Both A and B are symmetric, real, positive definite, well conditioned
matrices. Now, try the standard MATLAB function eig for finding a matrix
V of generalized eigenvectors of A\B (or B\A).

>> [V,D] = eig(A,B)

V =
2.4303e-01+ 2.6897e-03i -5.6521e-01+ 1.6569e-01i 5.9361e-01+ 1.6691e-01i
-5.7729e-01- 6.3891e-03i 6.8994e-01- 2.0226e-01i 3.7123e-01+ 1.0438e-01i
-7.7946e-01- 8.6267e-03i -3.5409e-01+ 1.0380e-01i -6.6071e-01- 1.8578e-01i

The result is complex! However, interchanging A and B gives real vectors.

>> [V,D] = eig(B,A)

V =
5.8900e-01 6.1663e-01 -2.4305e-01
-7.1898e-01 3.8563e-01 5.7732e-01
3.6899e-01 -6.8634e-01 7.7951e-01

Both answers are correct. The two V's have the eigenvectors in different
orders and with different complex scalar factors. But I was surprised
to get the complex results from the first calculation.

"Everything is real in my mind.." - J. Lennon.

-- Andrei Knyazev

[Editor's Note. Andrei is not the first to be surprised by this.
For eig(A), MATLAB already checks to see if A is real symmetric and,
if it is, uses the symmetric QR algorithm, which gives real results.
I am now thinking that for eig(A,B), we should check if both A and B
are real symmetric and if B is also positive definite and well
conditioned. If so, instead of the general QZ algorithm, we should use

R = chol(B);
[V,D] = eig(R'\A/R);
V = R\V;

This will give real eigenvalues and eigenvectors and the eigenvectors
will automatically be normalized so that V'*B*V = I. The cost of the
check is small compared to the cost of the general algorithm. -- Cleve]


From: Randy Bank <>
Date: Mon, 14 Mar 94 15:06:05 PST
Subject: PLTMG Edition 7.0

I am happy to announce that a new version of the PLTMG package is now available.

PLTMG edition 7.0 is a package for solving linear and nonlinear elliptic partial
differential equations in general regions of the plane. It is based on
continuous piecewise linear triangle finite element approximation, and features
a damped Newton iteration, adaptive local mesh refinement, hierarchical basis
multigrid iteration, and pseudo-arclength continuation for parameter
dependencies. The package also includes an initial mesh generator, a skeleton
generator, and several graphics routines. The source code is available by
anonymous ftp from Netlib and Mgnet.

Full documentation can be obtained in the PLTMG User's Guide, available from
SIAM publications, 1400 Architects Building, 117 South 17th Street,
Philadelphia PA 19103-5052. They can also be reached by
email at The title is: PLTMG: A Software Package for Solving
Elliptic Partial Differential Equations (ISBN 0-89871-330-7).

Randy Bank


From: V. I. Kostin <>
Date: Tue, 15 Mar 94 18:01:41 +0600
Subject: New Book on Numerical Methods in Linear Algebra

This is the second announcement about appearance of our book. The
first one was made 1 year before and, maybe, was too early. The
book has realy appeared in the last summer.

New Book on Numerical Methods in Linear Algebra

"Guaranteed Accuracy in Numerical Linear Algebra"
by S.K.Godunov, A.G.Antonov, O.P.Kiriljuk and V.I.Kostin
ISBN 0-7923-2352-1

(Updated and revised translation of the second Russian edition of
" The Guaranteed Precision of Linear Equations Solutions in
Euclidean Spaces")

Audience: Researchers whose work involves numerical
methods of linear algebra.

Exposition is closed and a reader need not additional sources of
information. The first chapter "Singular Value Decomposition" is
devoted to introduce abstract linear algebraic base such as
singular values and vectors, special orthogonal transformations,
bidiagonalization and tridiagonalization and so on. In the more
concrete next chapter "Systems of Linear Equations" notions of
condition number, characteristic of inconsistency and gap are
introduced. The main goal here is development of perturbations
theory for different cases of full rank and rank deficient
systems of equations. Simplification of bidiagonal and tridia-
gonal matrices is described in the chapter 3 "Deflation Algo-
rithms for Band Matrices". There is used new techniques based on
advanced Sturm theory and completely different from classical
SVD- and QR- algorithms. Chapter 4 is entitled "Sturm Sequences
of Tridiagonal Matrices" and is the clue chapter in the book.
Such interesting pecularity of Sturm sequences as their monoto-
nicity is used for calculation of so called two-sided Sturm se-
quences. Elements of these sequences determine parameters for
deflation algorithms. Chapter 5 "Pecularities of Computer Compu-
tations" is intended for detail description of round-off errors
which may occur in real computations. All kinds of errors (rela-
tive and absolute ones) are taken into account in the estimates
and provide guaranteed accuracy of results.

Bibliography 47 items. Index. 535 pp.

Published by Kluwer Academic Publishers,
P.O.Box 17, 3300 AA Dordrecht,
The Netherlands.


From: John E. Osborn <jeo@emmy.UMD.EDU>
Date: Tue, 15 Mar 1994 09:24:34 -0500
Subject: Ivo Babuska Receives Birkoff Prize

Professor Ivo Babuska of the University of Maryland has been selected as
co-recipient of the the 1994 Birkhoff Prize. He shares this prize with S.R.S
Varadhan of the Courant Institute.

This prize is awarded by the American Mathematical Society and the Society
for Industrial and Applied Mathematics. The selection committee particularly
cited Babuska's work on the reliability of finite elements methods; the
development of a general framework for finite element error estimation; and
the development of p and h-p finite element methods.

Bruce Kellogg
John Osborn


From: Toby Driscoll <>
Date: Tue, 15 Mar 1994 10:26:42 -0500
Subject: Schwarz-Christoffel Toolbox for MATLAB 4.x

The Schwarz-Christoffel Toolbox for Matlab 4 is a collection of
M-files for the interactive computation and visualization of
Schwarz-Christoffel conformal maps. The Schwarz-Christoffel
transformation is a formula for a conformal map from the upper
half-plane to the interior of a polygon, which may have slits
or vertices at infinity. The transformation can also be modified to
produce maps from the unit disk onto either the interior or the
exterior of the target polygon.

Major features of the toolbox include:

Graphical input of polygons

Solution of the parameter problem for half-plane, disk, and
exterior maps

Computation of forward and inverse maps

Adaptive plotting of images of orthogonal grids for

Command-line and graphical user interfaces

Online and Postscript documentation

The toolbox requires Matlab 4.1 or later, except under MS-Windows,
where version 4.0 is sufficient. Little Matlab expertise is required.

The Schwarz-Christoffel Toolbox can be obtained by anonymous ftp to, in the directory pub/driscoll/SC-Toolbox, or at
the MathWorks ftp site,, in the directory
pub/contrib/misc. Version 1.0 of the toolbox is available in the shar
file, and the Postscript documentation is in the file Other archive formats may be available at the Cornell
site. If you do not have the Optimization Toolbox, you also need the
file pub/contrib/optim/ at the MathWorks ftp site.

Toby Driscoll
Center for Applied Mathematics
Cornell University
Ithaca, NY 14853

Phone: (607) 255-8272


From: Tao Lin <>
Date: Tue, 15 Mar 1994 16:57:18 -0500
Subject: Fredholm Integro-differential Equation

Dear friends:

I will be very grateful if you can let know any references
about numerical methods of boundary value problems of a
one dimensional Fredholm type integro-differential equation. The
the integro-differential equation consists of an usual one dimensional
second order self-adjoint differential operator and
a Fredholm integral operator with a weakly singular kernel.
The boundary value problem is posed in an interval [-L, L], with
Neumann boundary condition. After spending several days in our
library during the spring break, I found only two papers
discussing this type of problems, and I will certainly be happy to
share them with anyone who has interest in them.

Thanks a lot in advance,

Tao Lin
Department of Mathematics
Virginia Tech.
Blacksburg, VA 24061


From: Roland Freund <>
Date: Tue, 15 Mar 94 23:51 EST
Subject: Professor Josef Stoer's 60th birthday


In celebration of Professor Josef Stoer's 60th birthday
on June 21, 1994, a colloquium will be held at the University
of Wuerzburg in Wuerzburg, Germany.

The colloquium will be on Friday June 24, 1994, starting
at 1:00 pm, and is expected to close the same evening at
6:30 pm. In the evening all participants are invited to
join us for dinner at a local restaurant.

The following speakers have agreed to give a talk:

Prof. R. Bulirsch (Technical University Muenchen)
Dr. R.W. Freund (AT&T Bell Laboratories)
Dr. M. Stoer (Konrad-Zuse-Zentrum Berlin)
Prof. R.S. Varga (Kent State University) (tentatively)
Prof. J. Zowe (University of Jena)

We would like to invite everybody to participate.

A brief email reply of those who plan to participate is most
welcome. Please send your reply by May 1st 1994 to any of
the following email addresses:


R.W. Freund, F. Jarre, C. Lubich


From: Ken Jackson <>
Date: Wed, 16 Mar 1994 09:05:49 -0500
Subject: Southern Ontario NA Day

15 April 1994
Computer Science Department
University of Toronto

The Twelfth Annual Southern Ontario Numerical Analysis Day will be held
in the Computer Science Department, University of Toronto, on Friday,
April 15, 1994. The distinguished speaker for the day is Professor
Larry F. Shampine, Mathematics Department, Southern Methodist
University, Dallas, Texas. In addition to the distinguished speaker,
there will be contributed talks of 20 to 30 minutes, including 5
minutes for questions. Any interested persons are invited to submit an
abstract (at most one page long) for consideration. In keeping with
our tradition, we encourage graduate students to give contributed
talks. Talks in all areas of numerical analysis will be considered.
Anyone wishing to present a talk should send electronic mail to: or

or contact any of the following:

Christina Christara
Tom Fairgrieve
Ken Jackson

Dept. of Computer Science
University of Toronto
Toronto, Ontario
Canada M5S 1A4
Tel. (416) 978-7360, 978-7075, 978-7816
Fax (416) 978-1931

Please let us know if you plan to attend, whether or not you wish to
give a talk. We will send directions and suggestions for accommodation
when we hear from you.

Please submit abstracts before March 28. Indicate your preference for a
20 or 30 minute talk. Speakers will be notified of acceptance by April 7.


From: Herman te Riele <>
Date: Wed, 16 Mar 1994 15:13:27 +0100
Subject: Symposium in Amsterdam March 25


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.

This is to announce the fifth meeting which centres around the theme:


Date: Friday, March 25, 1994
Location: CWI, Kruislaan 413, Amsterdam
Room: Z009

11.00 - 11.05: Welcome
11.05 - 11.50: Ivan Graham (speaker) and R.K. Coomer
(School of Mathematical Sciences, Univ. of Bath, UK)
Massively parallel methods for semiconductor device modelling
12.00 - 12.45: Rudnei Dias da Cunha (speaker) and Tom Hopkins
(Computing Laboratory, Univ. of Kent at Canterbury, UK)
Designing a portable numerical package for parallel architectures

Next meeting and theme:
June 3, 1994: Parallel Numerical Algorithms and CFD - Applications
(this meeting will take place at Delft University of Technology,
following a three-day Summerschool on Parallel Computing in Fluid Dynamics,
organised by P. Wesseling)

Dates and themes of the previous meetings:
Febr. 4, 1994: Parallel Numerical Algorithms and Software
Nov. 26, 1993: Computational Number Theory and Cryptography
Sept. 24, 1993: Parallel Numerical Algorithms
June 4, 1993: Topics in Environmental Mathematics

For further information, e.g., about how to reach CWI, contact H.J.J. te Riele
(CWI, tel. 020-5924106).
If you wish to receive a LaTeX-file of the abstracts of the lectures,
please send a request to .

Herman te Riele (CWI)
Henk van der Vorst (RUU en CWI)


From: Ernest Rothman <>
Date: Thu, 17 Mar 1994 09:15:55 EST
Subject: MS in Applied Mathematics at Salve Regina University


Salve Regina University's Master of Science in Applied Mathematics,
an interdisciplinary program with a strong computational component,
is designed to prepare students from a wide variety of educational
backgrounds for productive careers in industry, entry into Ph.D.
programs at other institutions, or careers in education. The program
combines the personalized attention one would expect from a small
college with the advantages of a larger research university.

Since Salve Regina University is connected to the Internet, our
graduate students and faculty easily communicate with scientists
all over the world, and have access to state-of-the-art supercomputers
available at national centers. At Salve one can take classes with
professors who maintain active research programs (in numerical
analysis, scientific computing, and differential equations), and
have applied mathematics to real-world situations in industry and
government agencies. Although active in research and consulting, Salve
faculty regard teaching as their primary commitment and are easily
accessible to students.

Salve Regina University is located on a picturesque campus in the
historic seaport of Newport, Rhode Island.

For more information please contact:

Dr. Ernest E. Rothman
Department of Mathematical Sciences
Salve Regina University
100 Ochre Point Avenue
Newport, RI 02840-4192

Phone: 1 401 847 6650 Ext. 3437
FAX: 1 401 847 0732


From: Douglas Arnold <>
Date: Fri, 18 Mar 1994 09:36:32 -0500
Subject: Course on the MODULEF finite element library

A course on the MODULEF Finite Element Library
will be held at Penn State University, July 11-15, 1994

The Modulef finite element library is an extensive finite element
system developed under the leadership of the French Institut National
de Recherche en Informatique et en Automatique (I.N.R.I.A.). Through
the Club Modulef academic and industrial researchers from all over the
world have benefited from and contributed to the library. Some of the
features of Modulef include:

* modular organization, designed for adaptability and extendability
* full access to source code
* rapid incorporation of current theoretical developments
* portability: runs on workstations, mainframes, and supercomputers
* complete English documentation
* extensive 2-D and 3-D mesh generation and visualization facilities
* many algebraic solution methods including direct and iterative methods
* domain decomposition techniques
* interactive mode

The 5 day course will consist of lectures by several of the leading
Modulef developers as well as hands-on training sessions. It will
explain the design concepts behind the code, the use of the library both
through the interactive drivers and a calling program, and the extension
of the library through the addition of new modules.

Tuition and Modulef licensing offer

Course tuition is $300 for full-time academic participants and $600 for
industrial participants. Tuition includes a copy of the book "Modulef: A
Modular Library of Finite Elements." Participation will be limited to
the first 20 people to register.

Purchase of the Modulef library involves the payment of a one time Club
Modulef entry fee and an annual license fee. The entry fee is 2,000 FF
for academic institutions and 8,000 FF for industrial corporations, and
the license fee is 4,000 FF for academic institutions and 12,000 FF for
corporations. (The current exchange rate is approximately 6 FF to the
dollar.) The entry fee will be waived for academic participants of the
course who purchase the code within six months of the course and reduced
by 50% for industrial participants.

To register or to obtain more information about the Modulef library or the
course, contact

Marina Vidrascu
BP 105
78153 Le Chesnay Cedex

Marina's phone number is +33 1 39 63 54 20 and her e-mail address is

Some information is also available on the World-Wide Web. Point
your browser at the URL
(A browser with good image capability, like mosaic, is recommended.)


From: Jiming Liu <>
Date: Fri, 18 Mar 1994 23:27:17 -0500 (EST)
Subject: Survey of Applications of Nonlinear Programming

Dear Colleagues:

We are conducting a survey of practical applications of nonlinear
programming. Please take a couple of minutes to fill out the
following questionnaire and send it to by e-
mail. We will send you a summary once the survey is completed.
Any comments are quite welcome. Thank you.


Anthony V. Fiacco
Jiming Liu
Department of Operations Research
The George Washington University
Washington DC 20052

Current Applications of Nonlinear Programming (NLP)

1. Name and Complete Mailing Address (Including E-mail if Possible)

2. Affiliation and Position

3. Nature and Specific Area of Your Personal Interest

4. List of Your Relevant Works (attach your main reference list,
if possible)

5. Most Important Potential Applications of NLP?

6. Most Important Applications and Results of NLP in Practice?

7. What is Needed to Stimulate Application of NLP?

8. Most Useful NLP Software Available or Needed and Expected Use
and Impact? Computer Implementations?

9. Major Impediments to Widespread Practical and Commercial Use of NLP?

10. Is Sensitivity Analysis of Practical Interest in NLP? Is This
Information Often Needed or Actually Calculated? Is Software
Available? Major Impediments? Implementations?

11. What Important Applications Require Global Solutions?

12. Comments, Question, Options? Please Provide Important Current
References to NLP Applications, if Possible.


From: Tony Skjellum <tony@Aurora.CS.MsState.Edu>
Date: Mon, 14 Mar 94 21:36:54 CST
Subject: Position at Mississippi State

Research Assistant Position
(One-half Time)
in Computational Engineering

Mississippi State University invites applicants for a Research Assistant I
or II, non-tenure half-time position in Computational Engineering in its
NSF Engineering Research Center for Computational Field Simulation. A Master
of Science degree is required in applied mathematics, computer science or a
closely related discipline, with experience in parallel processing, numerical
methods, and scientific software. Salary is competitive and commensurate with
degree and experience.

Real knowledge of the Unix system as well as C and Fortran programming are
essential. Previous experience with parallel machines such as Intel, nCUBE,
IBM, or cluster computing are strong plusses. Knowledge of theory and/or
numerical methods for partial differential equations is important. Experience
solving real problems on parallel systems is a big plus. The successful
candidate will interact with a dynamic group of researchers at the MSU ERC and
at national laboratories (including periodic visits to national laboratories),
through our collaborations and contracts. Interested persons should submit a
complete resume with names and addresses of at least three references to:

Dr. Joe F. Thompson, Director
Engineering Research Center for
Computational Field Simulation
P.O. Box 6176
Mississippi State, MS 39762

Applications will be accepted through March 31, 1994 or until the position
is filled. MSU is an affirmative action, equal opportunity employer. Women
and minorities are encouraged to apply.


From: Bob Robey <>
Date: Tue, 15 Mar 94 13:50 MST
Subject: Postdoc Position at Syracuse

The Alex G. Nason Prize
Two-year Computational Science Postdoctoral Fellowship
at Syracuse University

This fellowship encourages talented post doctoral researchers to
participate in the research of the Northeast Parallel Architectures
Center (NPAC) at Syracuse University. NPAC conducts an
interdisciplinary program to use high-performance computing
technologies in scientific and industrial applications. NPAC offers a
full range of the most up-to-date hardware and software technologies.
The Nason Prize encourages researchers to apply these or other
innovative techniques to scientific and/or industrial applications.

Computational Science is a relatively new field that is emerging at
the interface of computer science and application disciplines,
including engineering, physics, chemistry, and information science and
technology. Recognizing Computational Science as an important new
field, Syracuse University began a major initiative in this area in
1990. The Nason Prize was established by the Nason Foundation in
recognition of alumnus Alex G. Nason's commitment to the advancement
of Syracuse University and the furtherance of knowledge and useful
applications in Computational Science.

In 1994-95, the Nason Prize will include a salary of $53,000 plus
fringe benefits; and $10,000 of research support. A new two-year
Fellow will be named each year through 1996.

Application Guidelines
Send a cover letter, curriculum vitae, and three letters of reference
to: Donna McCammon, Personnel Administrator, NPAC, 111 College Place,
Syracuse University, Syracuse, NY 13244-4100. [email:]

Materials must be postmarked on or before April 22, 1994.


From: Bill Harrod <>
Date: Sat, 19 Mar 94 17:40:22 CST
Subject: Position at Cray Research

Position Announcement: Senior Programmer/Analyst
Mathematical Software Group
Cray Research, Inc.
Eagan, Minnesota

The Cray Research Mathematical Software Group invites qualified
individuals to apply for a position as Senior Programmer/Analyst.

The Mathematical Software Group supports Cray Research's math and
scientific libraries. Members of the group do research on algorithm
development, develop, optimize, and support numerical software, and
work with applications analysts to provide highly optimized
computational kernels for use as building blocks in application
programs. Particular areas of interest within the group include direct
and iterative methods for solving sparse linear systems, in-core and
out-of-core solvers for dense linear systems and eigenvalue problems,
and signal processing. Each individual participates in several
different projects, developing and supporting software for both
the scalable CRAY T3D and the vector/parallel Cray C90/YMP computing

The minimum requirement is a Ph.D. or equivalent experience in
computer science or applied mathematics. Preference will be given to
candidates with the following additional qualifications:

1. Experience in developing and optimizing software for
high-performance computer architectures, especially scalable
parallel computers.

2. Expertise in numerical methods for signal processing, including
FFT routines.

3. Demonstrated ability to work independently and as a member of a

This position offers a competitve salary and benefits package, an
excellent working environment, and unmatched access to state-of-the-art
high-performance computing power.

Inquiries may be directed to

Dr. William J. Harrod
Cray Research, Inc.
Mathematical Software Group
655F Lone Oak Drive
Eagan, MN 55121
(612) 683-5249

Resumes and other supporting documents should be sent by April 15 to:

Cray Research, Inc-MSG
Resume Processing Center
1620 Olson Drive
Chippewa Falls, WI 54729


From: Dinesh <>
Date: Sun, 20 Mar 94 17:42:44 EST
Subject: Position at University of North Carolina

Department of Computer Science
Tenure Track Position

We invite applications for an assistant professor positions to begin
August 1994. Candidates must hold (or expect to hold) a Ph.D. We will
give highest priority to those who have strong research credentials
in parallel and scientific computing.

Our department supports a world-class effort in very-high-performance
real-time graphics and virtual environments research. Increasingly this
revolves around high-speed general purpose parallel computers, not just
the special machines we build here. The application domain for these
graphics systems is becoming increasingly sophisticated, typically
revolving around scientific applications and visualization. We have
extensive collaborations with scientists at UNC, Duke, the NC
Supercomputing Center,and other places in the area of parallel scientific
computation and visualization. There are also ongoing efforts in the
specification and development of parallel applications in collaboration
with John Reif at Duke and researchers at Kestrel Institute in Palo Alto.

For all of these reasons the department is particularly keen to expand and
strengthen our parallel computing faculty. If you can run mosaic, you can
have a look at a department brochure and a couple of the research projects
via the URL

The area around here is hopping (in activities, that is, and not in the
geophysical sense). It's an excellent place to live. The cost of living
is not too high, yet the flavor and style is cosmopolitan.

Apply by electronic mail to or postal mail to Faculty
Search Committee, Campus Box 3175, Sitterson Hall, Chapel Hill, NC 27599-3175.
The University of North Carolina is an equal opportunity, affirmative
action employer.


From: Stratis Gallopoulos <>
Date: Thu, 17 Mar 94 10:19:46 CST
Subject: Academic Programs in Computational Science


Summary of a Conference Report
Randall Bramley (
Department of Computer Science, Indiana University, Bloomington
E. Gallopoulos (
CSRD, University of Illinois at Urbana-Champaign

This conference was held February 10-12 1994, in Albuquerque, NM,
and was co-sponsored by the U.S. Department of Energy (represented
by Gary Johnson) and by the High Performance Computing
Engineering Research Program of the University of New Mexico (UNM).
The conference was chaired by Brian Smith (
and Frank Gilfeather ( of UNM.
In the next few paragraphs we summarize some conference
highlights. An extensive report on the presentations, opinions,
and discussion will appear in the Summer 1994 issue of the
new IEEE Computational Science & Engineering magazine
(call (714) 821-8380 for more information).
An extended abstract is also available from the
IEEE Computer Society gopher (Info.Computer.Org).

The goals of the conference were "to examine different
models for CSE academic programs, compare their features, learn from
their experience, and describe course offerings available in
high performance scientific and engineering computing."
The conference drew approximately 80 participants that included
K-12 program developers, university faculty, and representatives from
national labs, industry, and government. Several members from the
NA-net community were present, including Gene Golub, Elias Houstis,
Liz Jessup, Syvert Norsett, Dan Sorensen, Steve Vavasis, Kris Stewart,
and Mike Heath. The conference was organized around panel discussions,
with plenty of audience participation and a poster session.
The first session on "K-12 Programs and Activities"
pointed to major problems faced by developers of K-12 programs,
such as the difficulty of making curricular changes and
establishing uniform goals and criteria, the shrinking
of budgets, and the weak state of mathematics science and
mathematics. More issues were raised in the
"Undergraduate Programs" session. Coordination and support by
the home institutions, turf wars, and limited budgets were seen
as typical problems. The programs that seem successful are the
ones that were created "bottom-up," rather than by administrative fiat.
A side-effect was that frequently CSE courses sat in a no-man's
land between the usual academic departments.
Issues included the source for continuing funding
to sustain the seed curricular development supported
by agencies such as NSF. Most participants seemed to agree
that although there should be undergraduate CSE courses,
there should not be an undergraduate major in CSE.

Panelists at the "Graduate Programs and Curricula"
session described programs at Rice, Arizona, Louisiana State,
Michigan, Stanford, Syracuse, and UC Davis.
They spoke of CSE programs, not departments, and described
their basic structures and characteristics and difficulties.
One major difficulty was persuading other departments that
they were not already producing "virtual" CSE graduates
and having these departments replace some of their core courses
in favor of CSE-oriented ones. Because many programs relied
on the initiative of a few faculty members, many differences
between the programs can be traced to political realities
within each university rather than substantial differences
in philosophy. Panelists felt that graduates would benefit
more by having their advanced degree labeled from a traditional
department and augmented by a phrase showing a CSE emphasis.
Regarding computer science, it was suggested that non-CS
departments should offer "immigration courses," which
would allow computer science students a view of applications,
without forcing them to follow the long list of prerequisites designed
for majors in that discipline. The "Industrial Reality Check" session
addressed the question, "What does industry expect from CSE graduates?".
The panelists' comments (from Martin Marietta, ORNL, Dupont, and
Exxon) reflected the downsizing in research-oriented projects and the
greater computational sophistication demanded from their new hires.
The fourth session presented the "Academic Response and
Evaluation." Panelists spoke of the role of software as
the vehicle for computational science, the importance of virtual
laboratories, and the need to distinguish real knowledge from
technical detail. Representatives of the national labs and
supercomputer centers presented some of their programs and experiences
in the session
"Laboratory and Supercomputing Center Educational and Training Efforts."
The final session had the theme
"Industrial and Government Response and Evaluation."
Gary Johnson, who was its organizer, said that the essence of CSE is a set
of cultural attributes and attitudes to problem solving.
He stressed that universities, government, and industry
should be equal partners in defining and participating in
CSE activities, stressing the role of the
National Information Infrastructure for making
CSE possible The industrial viewpoint was reviewed by
panelists from Cray Research and Amoco.
Equipment donations, mentoring, internships, fellowships, and
conference sponsorships were regarded as the standard
ways industry is being asked to contribute to education.
It was felt that there is a long-term market for CSE graduates
but that industry should be more active in defining and communicating its
requirements to academia.

This conference was an opportunity for many people to get
together and see other viewpoints on just what CSE is, what
it should consist of, and how it should be taught.
Although there were diverse ideas and sometimes fractious
discussion, a surprisingly large number reached consensus
on several issues. For example, in all attempts to define
CSE (and there were many), there was strong agreement that
CSE is interdisciplinary, it is application and problem-solving oriented,
and it uses computations as an essential component.
Another consensus item was the power of using computational science
to motivate and interest students at all levels in their
study of science, mathematics, and computing.
This conference also made clear that creators of courses in
computational science should first explore the many resources
already available, prepared through support by NSF, DOE,
and other agencies. A final item of general agreement was
that the conference should be held again next year: indeed,
this very interesting conference set the stage for future
discussion regarding the future directions of CSE education.


From: Richard Brualdi <>
Date: Wed, 16 Mar 1994 13:57:28 -0600 (CST)
Subject: Contents, Linear Algebra and its Applications

Contents Volume 201

Robert Grone (San Diego, California)
A Biography of Marvin Marcus 1

Chi-Kwong Li (Williamsburg, Virginia)
Linear Operators Preserving the (p, q) Numerical Radius 21

William Watkins (Northridge, California)
Unimodular Congruence of the Laplacian Matrix of a Graph 43

Morris Newman (Santa Barbara, California)
Tridiagonal Matrices 51

Russell Merris (Hayward, California)
A Note on Unimodular Congruence of Graphs 57

Raphael Loewy (Haifa, Israel) and
Stephen Pierce (San Diego, California)
Linear Preservers of Balanced Singular Inertia Classes 61

E. R. Barnes (Atlanta, Georgia) and A. J. Hoffman
(Yorktown Heights, New York)
Bounds for the Spectrum of Normal Matrices 79

Alexander Kovacec (Coimbra, Portugal)
On a Conjecture of Marcus and de Oliveira 91

N. Bebiano and M. E. Miranda (Coimbra, Portugal)
On a Recent Determinantal Inequality 99

Richard A. Brualdi (Madison, Wisconsin) and
Bryan L. Shader (Laramie, Wyoming)
Minimum Permanents on Special Faces of the Polytope of Doubly
Stochastic Matrices 103

Mao-Ting Chien (Taipei, Taiwan) and Bit-Shun Tam (Tamsui, Taiwan)
Circularity of the Numerical Range 113

Henryk Minc (Santa Barbara, California)
Minimum Permanents of Doubly Stochastic Matrices With Prescribed
Zero Entries on the Main Diagonal 135

Richard Arens (Los Angeles, California) and
Moshe Goldberg (Haifa, Israel)
Weighted I*be Norms for Matrices 155

Morris Newman and Robert C. Thompson (Santa Barbara, California)
A Counterexample Connected With Gersgorin's Theorem 165

LeRoy B. Beasley (Logan, Utah) and Daniel J. Scully
(St. Cloud, Minnesota)
Linear Operators Which Preserve Combinatorial Orthogonality 171

Robert Grone, Stephen Pierce, James Ross (San Diego, California),
and Chi-Kwong Li (Williamsburg, Virginia)
Spectral Bounds Derived From Quadratic Forms on Decomposable
Tensors 181

Frank Uhlig (Auburn, Alabama)
Computing the Inertias in Symmetric Matrix Pencils 199

George W. Soules (Princeton, New Jersey)
An Approach to the Permanental-Dominance Conjecture 211

Author Index 231


From: Daniel Baltzer <>
Date: Thu, 17 Mar 1994 15:26:13 +0100
Subject: Contents, Numerical Algorithms

Numerical Algorithms, Volume 6, No. 3 - 4, 1994, ISSN 1017 1398
Editor-in-Chief: Claude Brezinski

pp 205-227: S. Elhay and J. Kautsky,
Jacobi matrices for measures modified by a rational factor

pp 229-244: A. Sidi,
Convergence of intermediate rows of minimal polynomial and reduced rank
extrapolation tables

pp 245-273: J.C. Dodu, T. Eve and M. Minoux,
Implementation of a proximal algorithm for linearly constrained nonsmooth
optimalization problems and computational results

pp 275-296: D. Amitai, A. Averbuch, S. Itzikowitz and E. Turkel,
Asynchronous and corrected-asynchronous finite differences solutions of
PDEs on MIMD multiprocessors

pp 297-316: G. Plonka,
Optimal shift parameters for periodic spline interpolation

pp 317-351: A.-M. Bellido,
Construction of iteration functions for the simultaneous computation of the
solutions of equations and algebraic systems

pp 353-378: M.K. Ng and R.H. Chan,
Fast iterative methods for least squares estimations

pp 379-418: J.-C. Fiorot, P. Jeannin and S. Taleb, New control massic
polygon of a B-rational curve resulting from a homographic change of

pp 419-423: Book reviews

Submissions of articles and proposals for special issues are to be
addressed to the Editor-in-Chief:

Claude Brezinski
Laboratoire d'Analyse Numerique et d'Optimisation
Universite des Sciences et Technologies de Lille
59655 Villeneuve d'Ascq Cedex

postal address:
Paris Drouot BP 18
75433 Paris Cedex 09

Requests for FREE SPECIMEN copies and orders for Numerical Algorithms are
to be sent to: E-mail:

J.C. Baltzer AG, Science Publishers
Asterweg 1A
1031 HL Amsterdam
The Netherlands
tel. +31-20-637 0061
fax. +31-20-632 3651


From: SIAM <>
Date: Thu, 17 Mar 94 15:20:41 EST
Subject: Contents, SIAM Control and Optimization

SIAM Journal on Control and Optimization
Volume 32, Number 4, July 1994

An Adaptive Servomechanism for a Class of Infinite-Dimensional Systems
Hartmut Logemann and Achim Ilchmann

Minimax-Optimal Strategies for the Best-Choice Problem When a Bound
is Known for the Expected Number of Objects
T. P. Hill and D. P. Kennedy

The H-infinity-Problem with Control Constraints
Viorel Barbu

Positive Dependence of a Class of Multivariate Exponential Distributions
Ingram Olkin and Y. L. Tong

Observability and Observers for Nonlinear Systems
J. P. Gauthier and I. A. K. Kupka

Decomposition and Parametrization of Semidefinite Solutions of the
Continuous-Time Algebraic Riccati Equation
Harald K. Wimmer

A Strong Separation Principle for Stochastic Control Systems Driven
by a Hidden Markov Model
Raymond Rishel

Optimal Switching in an Economic Activity Under Uncertainty
Kjell Arne Brekke and Bernt Oksendal

L-infinity-Exact Observability of the Heat Equation with Scanning
Pointwise Sensor
Alexander Khapalov

Boundary Control of a One-Dimensional Linear Thermoelastic Rod
Scott W. Hansen

Control of Infinite Behavior of Finite Automata
J. G. Thistle and W. M. Wonham

Supervision of Infinite Behavior of Discrete-Event Systems
J. G. Thistle and W. M. Wonham

A Version of Olech's Lemma in a Problem of the Calculus of Variations
Arrigo Cellina and Sandro Zagatti

Characterization of the L2-Induced Norm for Linear Systems with
Jumps with Applications to Sampled-Data Systems
N. Sivashankar and Pramod P. Khargonekar

The Equivalence of Extremals in Different Representations of
Unbounded Control Problems
J. Warga and Q. J. Zhu

Controllability of a System of Two Symmetric Rigid Bodies in Three Space
Michael J. Enos

Optimal Angular Velocity Tracking with Fixed-Endpoint Rigid Body Motions
Michael J. Enos

Erratum: On the Optimal Tracking Problem
Ofer Zeitouni and Moshe Zakai


From: E. B. Saff <>
Date: Thu, 17 Mar 94 15:54:16 EST
Subject: Contents, Constructive Approximation

Volume 10 Numbers 1 1994

1 Constructive Approximations to the Invariant Densities of
Higher-Dimensional Transformations
A. Boyarsky, P. Gora, and Y. S. Lou

15 On the Darling-Mandelbrot Probability Density and the Zeros
of Some Incomplete Gamma Functions
John S. Lew

31 Convex Polynomial and Spline Approximation in C[-1,1]
Yingkang Hu, Dany Leviatan, and Xiang Ming Yu

65 Fourier Series of Functions Whose Hankel Transform is
Supported on [0,1]
Juan L. Varona

77 Best Uniform Approximation by Harmonic Functions on
Subsets of Riemannian Manifolds
P. M. Gauthier and D. Zwick

87 Simultaneous Lagrange Interpolating Approximation Need
Not Always Be Convergent
S. P. Zhou

95 Strong Converse Inequality for Kantorovich Polynomials
W. Chen and Z. Ditzian

107 Lehmer Pairs of Zeros, the de Bruijn-Newman Constant \Lambda,
and the Riemann Hypothesis
George Csordas, Wayne Smith, and Richard S. Varga

131 Maximal Polynomial Subordination to Univalent Functions
in the Unit Disk
Vladimir V. Andrievskii and Stephan Ruscheweyh


End of NA Digest