NA Digest Sunday, February 14, 1993 Volume 93 : Issue 7

Today's Editor:

Cleve Moler
The MathWorks, Inc.

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to


From: Klaus Peters <>
Date: Fri, 12 Feb 93 08:58:25 EST
Subject: Change of Address for Klaus and Alice Peters

Change of address:

Klaus and Alice Peters, publishers of books in mathematics and computer science since 1964, have started their own publishing company and can now be reached at

A K Peters, Ltd., 289 Linden Street, Wellesley, MA 02181
Phone: 617-235-2210, FAX: 617-235-2404

Klaus Peters


From: Tamir Tassa <>
Date: Sun, 14 Feb 93 16:02:29 +0200
Subject: A Degenerate Parabolic Equation

A degenerate parabolic equation:

Let u=u(x,t) be a solution of the degenerate parabolic equation
u_t + f(u)_x = (Q(u)*u_x)_x (the * denotes multiplication), subject to
a bounded and compactly supported initial data, where the viscosity
coefficient, Q(u), is always positive except for u=0 where Q(0)=0.

I would like to have a reference, where the question of smoothness of
solutions to such equations is addressed.

Sincerely yours,
Tamir Tassa
Tel-Aviv University


From: H. Murakami <>
Date: Fri, 12 Feb 93 23:05:05 JST
Subject: Complexity of Tridiagonal Eigenvalue Problem

If anyone knows the answer of the folloing question, please let me know.

Assume, I have a unsymmetric tri-diagonal matrix of order n.
The 3n-2 nonzero entries are sometimes all real or somtimes all complex numbers.

What is the currently know fastest method to calculate all the
eigenvalues of this matrix, for a given requirement of m-bits accuracy,

If it were real symmetric or (complex) hermitian tri-diagonal matrix,
using let's say, the Sturm bisection method, the computational order
for solving all the eigenvalues is, about O(n^2 log(m)) I think.

Thank you.

H. Murakami


From: Christian Hoffmann <>
Date: Mon, 8 Feb 93 07:38:18 GMT
Subject: Elements of Space Probes

As an owner of the dynamical solar system simulator program "Dance of the
Planets" for PC, I am interested in current and past osculating elements of
space probes like Pioneer, Voyager, Galileo and the like. It would make
tremendous fun to simulate these objects through their encounters with
the planets.

Does anyone know of electronic mails in astronomy? Or amateur astronomy?

Christian Hoffmann
Sternwarte Muelimatt
CH-8915 Hausen am Albis


From: Cleve Moler <>
Date: Sun Feb 14 11:21:21 EST 1993
Subject: 1993 MATLAB Conference

1993 MATLAB Conference
Cambridge, Massachusetts
October 18 - 20

The first MATLAB Conference will be held October 18-20, 1993, at the
Hyatt Regency in Cambridge, MA. The conference will provide a chance
for users of MATLAB, SIMULINK, and the MATLAB toolboxes to meet each
other and members of The MathWorks staff.

The conference will include:
* Invited talks by experts in fields of interest to MATLAB users.
* Contributed papers from the MATLAB and SIMULINK user community.
* Technical presentations by the developers of MATLAB and SIMULINK.
* Seminars by toolbox authors.
* Social activities.

Call For Papers
The MathWorks welcomes abstracts of papers for possible presentation
at conference. Abstracts should be approximately 300 words in length.
The deadline for submitting abstracts is March 31, 1993. Abstracts
will be reviewed and authors notified of their acceptance by May 15.
Authors who have submitted an abstract by the deadline date and are
selected to present papers will have their registration fee waived.
Please submit abstracts to:

MATLAB Conference
The MathWorks, Inc.
24 Prime Park Way
Natick, MA 01760

or to:

For More Information
The Winter 1993 issue of the MathWorks Newsletter contains more
information about the conference.

If you would like to be on the mailing list to receive additional
information on the conference, please send your name, postal mailing
address and phone number to: You can also
put your name on the mailing list by phoning (617) 439-9962 or by
sending a FAX to (508) 653-2997.


From: Albert Greenberg <>
Date: Mon, 8 Feb 93 13:32 EST
Subject: DIMACS Workshop on Parallel Algorithms

DIMACS Workshop on
Parallel Algorithms for Unstructured and Dynamic Problems
June 2-4, 1993

Parallel computing has been quite successful solving large problems
having very regular structure, because the structure naturally leads
to a balanced allocation of data and computations across the
processors, and to efficient communications between processors.
Examples of such problems can be found in matrix computation, in
signal/image processing, and in natural sciences. However, in many
important mathematical, scientific and industrial problems data
dependencies are highly irregular and/or evolving at run time.
Outstanding examples include discrete event simulations, branch and
bound techniques, unstructured sparse graph and matrix problems,
adaptive grid methods, as well as many others. It is desirable to
develop parallel algorithms, compiler techniques, and hardware for
efficiently solving large, irregular problems. The purpose of the
workshop is to bring together researchers from a wide variety of
fields in order to explore the existing algorithms, heuristics, and
systems, to suggest new methods, and to identify some common


The three day workshop (Wednesday, June 2 - Friday, June 4) will be
held at DIMACS at Rutgers University, Piscataway, New Jersey. DIMACS
is the National Science Foundation science and technology center for
discrete mathematics and computer science. The workshop will include
invited and contributed talks, and possibly a poster session.
Contributions are solicited in appropriate applications (e.g.,
discrete event simulations, sparse matrix computations), systems
(e.g., compilers, hardware), and theory. For those that wish to
present a talk, send a paper or abstract to the organizers by February
22, 1993.

By the end of March 1993, authors will be notified of acceptance, and
the program will be distributed. The organizers plan to arrange
informal gatherings on each day of the Workshop to promote
discussions. Short abstracts will be published in the workshop proceedings.
This will not preclude publication of the full paper elsewhere.

The DIMACS Conference Center can accommodate about 100 participants.
Subject to this capacity constraint, the Workshop is open to all researchers.
The Workshop will not have a registration fee, and a limited amount of travel
support will be available.

Albert G. Greenberg Andrew T. Ogielski
AT&T Bell Laboratories Bell Communications Research
600 Mountain Avenue 445 South Street
Murray Hill, NJ 07974 Morristown, NJ 07960
(908) 582-3395 (201) 829-4192
(908) 582-2379 (FAX) (201) 829-4391 (FAX)

If you are interested in attending or contributing, please send email to either
of the organizers, stating your name/affiliation/area of interest/intentions.


From: Paul Messina <>
Date: Thu, 11 Feb 1993 22:13:39 -0800
Subject: CSCC Supercomputing Workshop

The Second CSCC Delta Applications Workshop

Sponsored by the Concurrent Supercomputing Consortium

March 25-26, 1993
Norfolk Waterside Marriott, Norfolk, Virginia

Note: The CSCC Workshop will follow the SIAM Parallel Processing
Conference which is also being held at the Norfolk Waterside Marriott
on March 22-24, 1993. For information on the SIAM conference, send
email to or call (215) 382-9800.


- To provide a forum for presenting applications that are being pursued on
the Intel Touchstone Delta,

- To promote exchange of information on techniques and software for
massively parallel computing, with emphasis on those that have worked well
on the Delta.

The program will feature:

- Delta applications that have produced important new scientific results,

- Programming tools and methods,

- Software clinic: an opportunity to get advice on programming techniques and
debugging help from experts from Intel SSD and Concurrent Supercomputing
Consortium sites,

- Presentations of work in progress will also be included.


The workshop will be open to the general research community, including
members of the Concurrent Supercomputing Consortium.


"Solving Grand Challenge Problems on Parallel Computers at Oak Ridge
National Laboratory"
Rich Sincovec (Oak Ridge National Laboratory)

"Phase Refinement and Extension for Macromolecular Structure Determination
Using Distributed Memory MIMD Machines"
Dan Marinescu, John R. Rice, Marius A. Cornea-Hasegan, Zhonguyn Zhang
(Purdue University)

"Quenched QCD Spectrum on the Delta"
Seyong Kim (Argonne National Laboratory)

"Future Directions for CSCC"
Paul Messina (Caltech)

"Computing Unsteady 3-D Incompressible Flows at High Reynolds Number Using a
Fast Implementation of the Vortex Particle Method"
Gregoire S. Winckelmans, Anthony Leonard, John Salmon (Caltech), Michael S.
Warren (Los Alamos national Laboratory)

"Implementation of an Explicit Navier-Stokes Algorithm on a Distributed
Memory Multiprocessor"
Steve Scherr (Wright-Patterson Air Force Base)

"Multi-Purpose Tree-Codes on the Delta"
John Salmon (Caltech)

"The Full Wave Analysis Miniature Telecommunication Components on the Intel
Faiza Lansing (Jet Propulsion Laboratory), Stephen Gedney, University of

"A Parallel/Distributed Memory Programming Strategy for a Periodic
Tridiagonal Solver"
Thomas Eidson (NASA Langley Research Center)

"Bond-Stretch Isomerism in Strained Organosilicon Compounds: An Application
of Ab Initio Electronic Structure Theory"
Jerry Boatz (Edwards Air Force Base)

"Polygon Graphics for Interactive Scientific Visualization on the Delta"
David Ellsworth (University of North Carolina)

"Hybrid Spectral Element Solvers: Performance on the Intel Delta and other
Parallel Supercomputers"
Ron Henderson (Princeton University)

"Scalability Studies of an Incompressible Flow Solver Based on Unstructured
Grids on the Delta"
Ravi Ramamurti, Rainald Lohner (Naval Research Lab)

"Towards a Seamless Parallel Finite Element Analysis Package for the
Automated Solution of Large Scale 3D Electromagnetic Scattering Problems on
the Touchstone Delta"
J.C. McComb, S. Araki, R. D. Ferraro, J. Parker, K. Tembekjian, J. E.
Patterson (Jet Propulsion Laboratory)

"DIME++: A Language for Parallel PDE Solvers"
Roy Williams (Caltech)

*This is a sampling of talks to be presented at the workshop. These talks
are confirmed as of early February. Between now and the workshop, we
expect to add more talks to the workshop program to include late-breaking
Delta results.


If you would like to present results you have obtained on the Delta, or
would like to give a talk related to your experiences in using the Delta,
please contact Paul Messina at Caltech (818-356-3907), email: Provide a title and brief abstract and indicate
how much time you need to present the work. Indicate if you would like a time
slot of 15 or 30 minutes. Talks can be informal. Presentations will be
accepted at the last minute, to allow the newest scientific results obtained
on the Delta to be reported.


The registration fee is $100. To register, contact:
Mary Maloney (818) 356-3732
Peggy Olsen (818) 356-6739


An informal proceedings, containing the slides presented by each workshop
speaker, will be prepared following the workshop and sent to the

Most of the workshop speakers are actively involved in high-performance
computing at one of the institutions participating in the Concurrent
Supercomputing Consortium. The consortium, which acquired the Intel Delta
as their first major computational resource, was formed in November 1990 by
a group of individuals interested in making progress in massively parallel
supercomputing. The consortium members are:

Argonne National Laboratory
Pacific Northwest Laboratory
California Institute of Technology
Jet Propulsion Laboratory
The Center for Research on Parallel Computation
(an NSF Science and Technology Center)
Defense Advanced Research Projects Agency
Intel's Supercomputer Systems Division
National Aeronautics and Space Administration (NASA)
National Science Foundation
Lawrence Livermore National Laboratory
Los Alamos National Laboratory
Purdue University
Sandia National Laboratories


From: Dugald Duncan <>
Date: Fri, 12 Feb 93 08:26:04 GMT
Subject: Mathematical Methods in Petroleum Science

Novel Mathematical Methods in Petroleum Science

Wednesday 28th April 1993
At the Royal Society of Edinburgh,
Edinburgh, Scotland

There is a range of new mathematical techniques which are potentially
capable of solving many complicated problems in petroleum science whose
solution has proved intractable by traditional means. Neural network
techniques provide a way of unravelling correlations in complex systems
with applications, for example, to rock recognition in subsurface
reservoir mapping. Applications of cellular automata include modelling
the flow and hydrodynamical properties of complex fluids such as
structured polymer-containing fluids and cement slurries, simple fluids
in random media, such as the porous rock encountered in oil reservoirs
and lubricating fluids. The workshop will consider both fundamental
aspects of these novel methods and examples of petroleum science based

The meeting is being organised by the International Centre for
Mathematical Sciences in collaboration with the Petroleum Science and
Technology Institute.

Organizing committee:
J Carr (Department of Mathematics, Heriot-Watt University), Peter
Coveney (Schlumberger Cambridge Research) and Ken Sorbie (Department of
Petroleum Engineering, Heriot-Watt University ).

For further information please contact:
Frank Donald, Project Development Director, ICMS,
Heriot Watt University, Edinburgh EH14 4AS, UK.
Tel UK 031 451 3256, Fax UK 031 451 3249,


R Ball (Cavendish Laboratory, Cambridge)
Renormalisation of two-phase flow

P Coveney (Schlumberger Cambridge Research)
Cellular Automaton Simulations of Cement
Hydration and Microstructure Development

P Fletcher (Schlumberger Cambridge Research)
Analysis of time and frequency domain data
using neural networks: some oilfield applications

D Harris (Department of Physics, University of Edinburgh)
Application of neural nets to lithofacies identification
in reservoir description

A Schlijper (Shell Research, Thornton Research Centre, Chester)
Simulation techniques for determination of
microrheological properties

D Sherrington (Department of Physics, University of Oxford)
How neural networks work

K Sorbie (Department of Petroleum Engineering, Heriot-Watt University)
An overview of mathematical problems in Petroleum Science


From: R. Baker Kearfott <>
Date: Wed, 10 Feb 93 14:44:22 CST
Subject: Interval Analysis Info -- Change in FTP Server

To minimize expenses, the main mail server at our university is changing.
My email correspondents should still be able to reach me at:,
just as before. However the area for anonymous FTP is now accessed
differently. BEFORE, anonymous FTP was accessed by:


NOW, anonymous FTP will be accessed by:


The directory structure of interest to my colleagues should be as before.
In particular, I remind you that a bibliography of works on interval
computations by researchers in the former Soviet Union, along with ordering
information, is found in the directory:


Information on the conference on Numerical Analysis with Automatic
Result Verification (Feb. 25 to Mar. 1, 1993) is found in:


The latter directory will continue to carry information, such as
instructions for submitting to and ordering the proceedings, after
the conference.

The FTP server may not be operational after March 1.

Please inform me if there are any questions or problems.

R. Baker Kearfott (Internet)


From: Ed Wegman <>
Date: Tue, 9 Feb 93 23:17:29 -0500
Subject: Position at George Mason University

We are looking for a Director for our new Institute for Computational
Sciences and Informatics. The attached announcement describes the position
and the expectations of the Institute. I have recently received funding
for an Intel Paragon/S (56 nodes). We have 18 SG machines and 15 NeXT
plus a lot of other miscellaneous UNIX hardware. The Institute is home
for a Ph.D. Program in Computational Sciences. I'd be happy to respond
to direct enquiries about the details of the program (provided they're not
too overwhelming in number). Thanks.

George Mason University


The George Mason University of the Commonwealth of Virginia is seeking
candidates for the position of Director of the Institute for Computational
Sciences and Informatics (CSI). The Director, who reports to the Provost,
is responsible for leadership of the Institute and fulfillment of its research,
educational and outreach mission. The Institute's mission includes the
development of new approaches to scientific research based on rapidly
developing computational methodology across a broad range of disciplines
such as computational physics, mathematics and statistics, space sciences,
environmental sciences, global change, the interface of computation and
biological science, including bioinformatics, and others. Research in the
Institute is organized both through multi-disciplinary teams and through
disciplinary centers in some of the fields. The Institute is developing
and implementing a new Ph.D. program in Computational Science with current
enrollment of 68 students and with an anticipated steady enrollment of 150
students. Institute's current computational facilities include a network of
workstations, and servers, with the possible addition of a highly parallel
machine. A significant emphasis in the Institute's mission is given to the
interaction with local high-technology industry and government laboratories
in the Washington Metropolitan area.

The successful candidate should be a recognized researcher in a field with
a significant computational component, with a strong record of individual
research and scholarship. The candidate should have demonstrated the ability
to provide leadership and management of research teams, and should have
experience with a variety of research funding sources. The ability to foster
interactions with high technology industry and laboratories, and the ability
to provide leadership to an interdisciplinary doctoral program will be

Those wishing to be considered for the position or wanting to nominate a
candidate should write to:

Chair, CSI Director Search Committee
George Mason University
116 Science and Technology Building I
Fairfax VA 22030-4444

Applications should include a letter, a curriculum vitae, and the names of at
least three references. Review of the applications will begin on February 1,
1993, and the position is expected to be available July 1, 1993. Women and
minorities are encouraged to apply.

George Mason University, a State University of the Commonwealth of Virginia, is
an affirmative action/equal opportunity employer.


From: Des Higham <>
Date: Fri, 12 Feb 93 14:32:49 GMT
Subject: NA Post-Doc at Dundee

Numerical Analysis Post-Doc at Dundee

The Science and Engineering Research Council will fund a three
year Research Assistantship for a project entitled
``The Dynamics of Time-Stepping in the
Numerical Analysis of Differential Equations''

The researcher will join a very active Numerical Analysis group at the
University of Dundee and will work closely with grantholders
Dr. D.J. Higham and Dr. D.F. Griffiths.
The main aim of the project is to investigate the long-term behaviour of
numerical methods on nonlinear, functional ODEs and semi-discretised PDEs.

The appointment is to be made at the RA 1A level, with a current
starting salary of 12,638 pounds sterling per annum.

The start date is no more than six months after 1st April 1993.
Applicants should hold, or expect to complete before the start date,
a PhD, and should have a strong background in numerical analysis
and differential equations.

Applications and requests for further information should be directed to

Dr D.J. Higham
Department of Mathematics and Computer Science
University of Dundee
Dundee DD1 4HN, Scotland, UK
Tel: 0382 23181 Extension 4474

Applicants should send a CV and the names and addresses of two people
from whom references can be obtained.


From: Richard Brualdi <>
Date: Tue, 9 Feb 93 07:04:19 CST
Subject: Contents: Linear Algebra and its Applications

Contents Volume 180, February 1993

R. McEachin (Fort Wayne, Indiana)
Closing the Gap in a Subspace Perturbation Bound 7

L. Elsner (Bielefeld, Germany) and M. Neumann (Storrs, Connecticut)
Monotonic Sequences and Rates of Convergence of Asynchronized
Iterative Methods 17

Erich Bohl (Konstanz, Germany) and Peter Lancaster (Calgary, Alberta, Canada)
Perturbation of Spectral Inverses Applied to a Boundary Layer
Phenomenon Arising in Chemical Networks 35

Andre C. M. Ran (Amsterdam, The Netherlands), Lieba Rodman, and
Jonathan E. Rubin (Williamsburg, Virginia)
Direct Complements of Invariant Lagrangian Subspaces and Minimal
Factorizations of Skew-Symmetric Rational Matrix Functions 61

Jean-Pierre Mongeau, Gilles Deslauriers, and Serge Dubuc
(Montreal, Quebec, Canada)
Continuous and Differentiable Multidimensional Iterative
Interpolation 95

Shu-An Hu, James F. Hurley (Storrs, Connecticut),
and Tin-Yau Tam (Auburn, Alabama)
Nonconvexity of the Permanental Numerical Range 121

L. Yu. Kolotilina (St. Petersburg, Russia)
Lower Bounds for the Perron Root of a Nonnegative Matrix 133

Carolyn Eschenbach (Atlanta, Georgia)
Idempotence for Sign-Pattern Matrices 153

Michael Eiermann (Heidelberg, Germany)
Fields of Values and Iterative Methods 167

Gerhard Starke (Karlsruhe, Germany)
Fields of Values and the ADI Method for Non-normal Matrices 199

R. Baeza-Vega and R. Benavides (Temuco, Chile)
Associative Bilinear Forms in Some Baric Algebras 219

Guorong Wang and Yuhua Lin (Shanghai, People's Republic of China)
A New Extension of Leverrier's Algorithm 227

Wenxin Ma (Ottawa, Ontario, Canada)
The Weak Identity T2n for nxn Symmetric Matrices 239

Charles R. Johnson (Williamsburg, Virginia), D. D. Olesky,
Michael Tsatsomeros, and P. van den Driessche
(Victoria, British Columbia, Canada)
Spectra With Positive Elementary Symmetric Functions 247

Referees, Volumes 161-180 263

Author Index, Volumes 161-180 267


Contents Volume 181, March 1, 1993

William Watkins (Northridge, California)
The Cone of Positive Generalized Matrix Functions 1

Daniel J. Scully (Saint Cloud, Minnesota)
Maximal Rank-One Spaces of Matrices Over Chain Semirings.
II. (u, i)-Spaces 29

Helena Albuquerque (Coimbra, Portugal),
and Alberto Elduque (Zaragoza, Spain)
On the Generators of Lie Superalgebras 45

R. Gow (Dublin, Ireland) and M. C. Tamburini (Brescia, Italy)
Generation of SL(n, Z) by a Jordan Unipotent Matrix
and Its Transpose 63

Alan C. Wilde (Ann Arbor, Michigan)
Nearly Commuting Projections 73

Wang Yan (Hebei, People's Republic of China)
Two Enumeration Theorems in Singular Symplectic Geometry
and a Class of PBIB Designs 85

Francisco Marcellan and Gabriela Sansigre (Madrid, Spain)
On a Class of Matrix Orthogonal Polynomials on the Real Line 97

Ton Geerts (Tilburg, The Netherlands)
Solvability Conditions, Consistency, and Weak Consistency for Linear
Differential-Algebraic Equations and Time-Invariant Singular Systems:
The General Case 111

Charles R. Johnson (Williamsburg, Virginia)
and Peter M. Nylen (Auburn, Alabama)
The Sprinkling Problem 131

Fumio Hiai (Ibaraki, Japan) and Denes Petz (Budapest, Hungary)
The Golden-Thompson Trace Inequality is Complemented 153

Zevi Miller and Dan Pritikin (Oxford, Ohio)
Eigenvalues and Separation in Graphs 187

J. Kastner and Ch. Zylka (Leipzig, Germany)
On Generalized Uniformly Tapered Matrices 221

G. Rousseau (Leicester, England)
An Expansion for Certain Symmetric Determinants 233

Paul Yiu (Boca Raton, Florida)
Maximal Normal Sets of n-Planes in R2n 241

D. W. Richardson (University Park, Pennsylvania),
B. D. O, Anderson (Canberra, Australia),
and N. K. Bose (University Park, Pennsylvania)
Matrix-Fraction Description From Frequency Samples 251

Richard Arens (Los Angeles, California) and Moshe Goldberg
(Haifa, Israel)
Quadrative Seminorms and Jordan Structures on Algebras 269

Author Index 279


End of NA Digest