NA Digest Sunday, April 7, 1996 Volume 96 : Issue 14

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

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to

URL for the World Wide Web: -------------------------------------------------------

From: John Tucker <>
Date: Mon, 01 Apr 96 11:21:00 EST
Subject: Query: Finding All Relative Maxima

This naive query originates from a remark made by a chemist that it would
be of considerable value to him to be able to determine ALL the relative
maxima, not just the global maximum, for a given objective function of
several variables (arising, I believe, in super-computer modeling of
molecular dynamics). Being removed from what is now state-of-the-art in
optimization, I ask for information: does there exist theory and software
that (under suitable conditions, naturally) provides an efficient means
to accomplish this? Please email responses to Thank you
in advance for any replies.

John R. Tucker
Board on Mathematical Sciences
Room NAS 315
National Academy of Sciences
2101 Constitution Avenue, N.W.
Washington, D. C. 20418
(202) 334-2421
Fax: (202) 334-1597


From: Indraneel Das <>
Date: Mon, 1 Apr 1996 11:27:56 -0600 (CST)
Subject: Condition of Random Matrices

Consider a random nxn real, matrix A such that:
1. A(i,i) is distributed normally with mean 1 and
sigma^2 for i = 1, ... , n

2. A(i,j) is distributed normally with mean mu and variance
sigma^2 (same sigma as above) for all i not equal to j

Find the probability that A has rcond < r_crit, where say,
r_crit = 1e-10 when
a.> mu (mean of each off-diagonal elt) = 0
b.> mu = 1 .
[ You can fix a value of the variance if that simplifies things]

-- Indraneel Das
[ Graduate Student ]
Department of Computational and Applied Mathematics
Rice University
Houston, TX 77251 -1892
Ph. (713) 527 8101 x3810 e-mail:


From: Saleet Jafri <>
Date: Tue, 2 Apr 1996 10:54:22 -0500
Subject: Fast ODE solvers

I am working on large coupled systems of ODE's and am looking for a
fast solver. I decompose my grid into pieces and solve the ODE
system on the subgrids. The application is cardiac network models.
Any suggestions would be appriciated. Thanks

Saleet Jafri
Department of Biomedical Engineering
The Johns Hopkins University School of Medicine
Rm 411 Traylor Research Building
720 Rutland Ave
Baltimore, MD 21205



From: Hairer Ernst <>
Date: Tue, 2 Apr 1996 10:17:39 +0200
Subject: Request for Papers on Optimization

My former student Valerie Brasey, Red Cross delegate in Tuzla/Bosnia,
sent me the following letter from one of her field officers:

... we work on a similar areas: on research field of
optimization. She refers me to you, hoping that you could
help University Tuzla
Your aid can be reflected in books from area of my interest:
techniques of optimization and simulation and that include
the follows:
1. Where is it possible to get those books
2. if it was possible to obtain some reprints from that area
3. if it is possible to abtain some programme packages for
PC from that area
Thank You in advance for Your reply, I send You best regards.
Sincerely Yours,
Mr. Suad Halilcevic, dipl. el. ing.
Fakultet Elektrotehnike Masinstva
Univerzitet u Tuzla
Bosnia and Herzegovina

If you are doing research on optimization, Mr. Suad Halilcevic
would be very grateful, if you send him some of your papers and/or
information on excellent recent books and programs.
Thanks in advance for all of your help. Ernst Hairer


From: E. B. Saff <>
Date: Tue, 2 Apr 1996 17:22:11 -0500
Subject: Software to Accompany Nagle and Saff ODE Book

A differential equations instructional utility package (DOS based)
to accompany the text "Fundamentals of Differential Equations" by
Nagle and Saff (Addison-Wesley Pub) is now available via the ftp site
The software was developed by E. Moskona, S. Tashev and K. Ivanov.


From: Chris Jones <>
Date: Thu, 4 Apr 1996 08:22:49 -0800 (PST)
Subject: New Book on Visualization and Optimization

I am pleased to announce the publication of my book,
Visualization and Optimization, published by Kluwer Academic
Publishers, ISBN0-7923-9672-3. The illustrated book discusses
how techniques from computer graphics can be usefully applied
to optimization practice and research; it also discusses
how optimization has been applied to visualization itself.

I have negotiated a limited time, 25% discount with the publisher.
This discount is available by using an order form published on
the Web. This order form is available at URL:


Chris Jones
Associate Professor
Management Science Department
School of Business Administration
353 Mackenzie Hall
University of Washington
Seattle, WA 98195
206-685-9392 (fax)


From: Takeo Ojika <>
Date: Mon, 01 Apr 1996 12:55:35 +0900
Subject: Conference on Virtual Systems and Multimedia

International Conference on Virtual Systems and Multimedia '96
(VSMM '96) in Gifu

September 18-20, 1996
Gifu Convention Center, Gifu, Japan

The second International Conference on Virtual Systems and Multimedia
will be held in Gifu on 18-20 September 1996. This conference is
organized by the International Society on Virtual Systems and
Multimedia(VSMM), and Executive Committee VSMM'96 in Gifu.

The theme of VSMM'96 is advanced technologies for seamless systems
between virtual worlds and the real world. The conference will
provide a forum for virtual reality and multimedia, and for academic
researchers to exchange up-to-date technological knowledge and
experience. In addition to the technical sessions, there will be
plenary and invited sessions.

Papers describing original work in, but not limited to, the following
technical areas are invited:


Prof. Takeo Kanade, CMU., USA
Prof. Hiroshi Harashima , Univ. of Tokyo
Dr. Myron Krueger, Artificial Reality Corp., USA,
Prof. Peifa Jia, Univ. of Tsinghua, R.P.China,
Prof. Robert J. Stone, InSys Limited(VR Division)&Univ. Salford, UK,
Prof. Hal Thwaites, Concordia Univ., Canada,
Prof. David Zeltzer, MIT,USA

For more details, please e-mail to


From: Ken Jackson <>
Date: Mon, 1 Apr 1996 16:55:48 -0500
Subject: Computational Finance Workshop

Computational Finance Workshop

Thursday, May 2, 1996
Sandford Fleming Bldg., Room 1105
10 King's College Rd.
University of Toronto

Sponsored by: ITRC - Information Technology Research Centre
IIPR - Institute of Insurance and Pension Research
The Departments of Computer Science at the Universities
of Waterloo and Toronto

Organizers: Peter Forsyth, Computer Science, University of Waterloo
Ken Jackson, Computer Science, University of Toronto

Modern models in finance rarely admit analytical solutions. This
workshop will focus on computational methods for obtaining approximate
solutions to financial models with application to such areas as:
option pricing, portfolio selection, and risk management. Ten
speakers from banks, consulting companies, software companies and
universities will present their work on new models and numerical
methods, including schemes based on Monte Carlo, Lattices, and Partial
Differential Equations.

The workshop is intended to increase awareness of advances in
computational finance and to discover opportunities to improve
practices or undertake further research.

There is no registration fee, but we would appreciate receiving your
registration before April 26.


Prof. Phelim P Boyle, School of Accounting, University of Waterloo,
"New Methods in Monte Carlo"

Prof. Ken Vetzal, School of Accounting, University of Waterloo,
"PDE Models of the Term Structure: Matching Yield and Volatility Curves"

Mr. Baoyan Ding, Combinatorics and Optimization, University of
Waterloo, "Portfolio Selection with Skewness"

Dr. Michel Crouhy, Vice President, CIBC, Global Analytics,
Market Risk Management Division, "An Overview of Numerical
Problems in Derivative Products and Risk Management"

Mr. Rob Zvan, Computer Science Dept., University of Waterloo,
"Robust Numerical Methods for PDE Models of Asian Options"

Dr. Izzy Nelken, Consultant and Principal, Harris Investment
Management, Chicago, "Modeling Convertible Bonds"

Mr. Yidong Liu, Computer Science Dept., University of Toronto,
"Modeling Path Dependent Options"

Dr. Ron Dembo, Algorithmics Inc., Toronto, Title: TBA

Prof. Sheldon Lin, Statistics Dept., University of Toronto,
"Bounds on Multiple Contingent Claims"

Mr. Hailiang Yang, Statistics Dept., University of Waterloo,
"Asset Allocation with Transaction Costs"

More information: Rosanna Reid, email or fax: 416-978-7207


From: Geza Seriani <>
Date: Tue, 2 Apr 1996 13:34:48 +0200 (METDST)
Subject: Conference on Theoretical and Computational Acoustics


PLACE & DATE: Columbia University, New York, USA, July 14-18, 1997.
Prof. Yih-Hsing Pao (National Taiwan University, R.O.C.)
Prof. Allan D. Pierce (Boston University, USA)
Prof. Martin H. Schultz (Yale University, USA)
Prof. Er-Chang Shang (University of Colorado, USA)
Dr. Yu-Chiung Teng (Columbia University, USA)

SPONSORS: U.S.Naval Undersea Warfare Center (NUWC), NATO SACLANT (Undersea
Research Center, ITALY), Yale University, Columbia University, Boston
University, University of Colorado, National Taiwan University (REPUBLIC
OF CHINA), Osservatorio Geofisco Sperimentale (OGS, Trieste, ITALY).

OBJECTIVES: Multidisciplinary forum for researchers in theoretical
and computational acoustics and related topics, such as:
- Computational aero-, seismo-, and ocean-acoustics; hydroacoustics,
bio-engineering acoustics, nonlinear acoustics, structural acoustics;
- Multidimensional wave modeling: grid methods (spectral,FD,FEM,FV,BIE,etc),
ray & symbolic & ODE methods, supercomputing & parallel methods;
- Wave propagation in boreholes, rocks, fluid-solid interfaces,
anisotropic media, viscoelastic, porous, fractured & random media;
- Ultrasonic wave propagation and nondestructive testing;
- Computational and experimental seismology;
- Shallow water acoustics and environmental/bottom parameter extraction;
- Global acoustics and large-scale monitoring;
- Acoustic tomography and full-field inversion;
- Real applications, experimental results, and data compression.

ADDRESS & INFO: Dr. Yu-Chiung Teng, Aldridge Lab. of Appl. Geophysics,
Columbia University, New York, NY 10027 U.S.A.
Tel: (212) 854-2910 /Fax: (212) 854-6508 /


From: Anne-Marie De Meyer <Anne-Marie.DeMeyer@CC.KULEUVEN.AC.BE>
Date: Thu, 04 Apr 96 11:35:53 UTC+0200
Subject: High Performance Computing in Europe on IBM Platforms

High Performance Computing in Europe on IBM Platforms
Krakow, September 8 - 11, 1996

Organizing institutions

The 8th annual meeting of SUP'EUR will be held in Krakow, Poland.

The Conference is organized jointly by:
- Academic Computer Center CYFRONET - KRAKOW,
- Institute of Computer Science, AGH,
- SUP'EUR Steering Committee.

SUP'EUR 96 will focus on the following topics:

- Parallel and distributed computing,
- IBM trends in high performance computing,
- New IBM products,
- Experience with SP2,
- High performance storage systems,
- Environments, languages and tools,
- Applications,
- Graphics and visualization,
- Education and training.

The Sup'Prize contest

Sup'Prize is a contest (sponsored by IBM) in the development of parallel
applications on IBM platforms. The result of the Sup'Prize contest
will be presented at the SUP'EUR 96 Conference. You can find more about
the Sup'Prize contest from:

Anne-Marie De Meyer
coordinator of the Sup'Prize 1996
Computing Centre
KU Leuven
de Croylaan 52A
3001 Leuven
Fax: 32 16 322999

For more information about SUP'EUR 96 please contact:
ftp: //

Prof. Jacek Moscinski
SUP'EUR Steering Committee Chairman
Academic Computer Centre CYFRONET - Krakow
P.O.Box 386
Nawojki 11
30-950 Krakow 61
Phone: (48 12) 341 766; 173 964
Fax: (48 12) 341 084; 338 054


From: Christian Heimann <>
Date: Thu, 4 Apr 1996 12:11:31 +0100 (MESZ)
Subject: Workshop Paralleles und Verteiltes Rechnen

4. Workshop "uber Wissenschaftliches Rechnen
Termin: In der Woche vom 9.-11.10.1996
Ort : Technische Universit"at

Die Kombination preiswerter Rechner mit Bauteilen aus Massenproduktionen
"ubersteigt h"aufig die Leistung von Supercomputern, die in teurer
Einzelfertigung hergestellt werden. Dies f"uhrte zum Aufschwung von
Parallelrechnern und Workstationfarmen mit RISC Prozessoren und
erkl"art ihren sprunghaften Anstieg am Marktanteil im High Performance

Verbunden mit dieser Entwicklung sind Fragen zur optimalen Verteilung von
Aufgaben auf verschiedene Rechner. Wesentlich ist hierbei die Wechselwirkung
zwischen parallelen Algorithmen, dem Netzwerk bzw. der Hardware, und der
Kommunikationsart. Ebenso stellt sich die Frage der Organisation dieser
Rechner, die als Arbeitsplatzrechner oder dedizierte Produktionsrechner
eingesetzt werden k"onnen. Diese Fragen reichen von geeigneten parallelen
Algorithmen "uber Batchsysteme, von Fehlertoleranz und Lastbalanzierung in
verteilten Systemen bis hin zu Netztechnologien und Vernetzungsfragen.
Nur eine ausgewogene Balance der Elemente Algorithmus, Hardware/Netzwerk
und Kommunikation wird effektive Ergebnisse zeigen.

Dieser Workshop spannt eine Br"ucke zwischen diesen Fragen und verbindet
Parallelrechnen, d.h. Softwaretools, Programmiermodelle und Algorithmen
f"ur Parallelrechner, mit verteiltem Rechnen, d.h. Tools zur
Organisation von Arbeitsplatzrechnen, und dazu notwendiger Netztechnologie.

Beitr"age zu folgenden Schwerpunkten sind m"oglich:

- Programmiermodelle
- Softwaretools
- Algorithmen
- Organisation, Lastbalanzierung und Fehlertoleranz
- Metacomputing
- Anwendungen
- Kommunikation
- neue Netztechnologien.

Besonders erw"unscht sind Beitr"age, die sich mit der Wechselwirkung
dieser einzelnen Elemente und deren Auswirkung auf die Effektivit"at

Als Beitr"age sind Voll- und Kurzbeitr"age m"oglich.
Die Beitr"age sollten in Deutsch oder Englisch sein. Sie sollten in
Papierform oder elektronisch (postscript level 2) an
geschickt werden. Akzeptierte Beitr"age werden in einem Proceedingsband

Die Beitr"age sollten in LaTeX mit dem Springer Lecture Notes Style
llncs.sty, verf"ugbar via


geschrieben sein. Die Vollbeitr"age sollten ein Maximum von 12 Seiten
und die Kurzbeitr"age einen Umfang von 4 Seiten nicht "uberschreiten.

Elektronische Anmeldung:

Weitere Informationen finden Sie unter

Dort k"onnen Sie sich auch elektronisch zur Tagung anmelden und einen
Vortrag einreichen.

Anmeldung per e-Mail:

Senden Sie bitte ihre Antwort per e-Mail an


From: Robert Corless <>
Date: Thu, 4 Apr 1996 14:25:40 -0500
Subject: Southern Ontario NA-Day '96


Saturday, April 27, 1996, 8:30am
Room 240, Western Science Centre
University of Western Ontario

The final schedule for SONAD 96 is given below, along with abstracts for the
invited presentations. The abstracts for the graduate student presentations
are available on our Web site,

Please note that one of our invited speakers, George Corliss, will be giving
another presentation on the day before the conference. We encourage you to
arrive early on Friday so that you can attend George's talk, as well as the
wine and cheese reception in the evening. The abstract for this talk is
given below.

We have had a terrific response regarding the student presentations. As a
result, we have filled all the available slots in the schedule. However, we
will also be displaying posters of graduate student work outside the
conference hall. If you are interested in displaying a poster, please send
a title to Rob Corless at by April 19.


Presentation: George Corliss, "Operator Overloading as an Enabling
Technology for Automatic Differentiation"
Friday, April 26, 4:00pm
Room 240, Western Science Centre

Conference: Saturday, April 27, 8:30am-5:30pm
Room 240, Western Science Centre


Michel Pettigrew, TBA
Mohamed Hamed, "Numerical Simulation of Unsteady Thermocapillary
Convection with Large Scale Interfacial Deformations"
Peter Varlagas, "Anisotropic FEM Mesh Generation Using Rivara Refinement"
Sultan Sial, "A Finite Volume Method for Simulating Phase Transitions"
Saar Friedman, "Particle Tracking for Multiphase Flow Visualization"
Reza Sadri, "Entrance Effects in Channel Flows"
Marzban Kermani, "A Preview of Automeshing: Its Status and
Future in the Industry"
Serge Kruk, "A Semidefinite Approach to Nonlinear Optimization Problems"
Robert Zvan, "Robust Numerical Methods for PDE Models of Asian Options"
Hiroshi Hayashi, "A Delay Differential Equation Solver by a
Continuous Runge-Kutta Method with Defect Control"
George Corliss, "War Stories from Industrial Mathematics"


Operator Overloading as an Enabling Technology
for Automatic Differentiation
Dr. George Corliss
Marquette University
Friday, Apr. 26, 4:00pm, Rm 240, Western Science Centre

We present an example of the science that is enabled by object-oriented
programming techniques. Scientific computation often needs derivatives for
solving nonlinear systems such as those arising in many PDE algorithms,
optimization, parameter identification, stiff ordinary differential
equations, or sensitivity analysis. Automatic differentiation computes
derivatives accurately and efficiently by applying the chain rule to each
arithmetic operation or elementary function. Operator overloading enables the
techniques of either the forward or the reverse mode of automatic
differentiation to be applied to real-world scientific problems. We
illustrate automatic differentiation with an example drawn from a model of
unsaturated flow in a porous medium. The problem arises from planning for
the long-term storage of radioactive waste.

War Stories from Industrial Mathematics
Dr. George Corliss
Marquette University

In 20+ years of consulting and teaching industrial and applied
mathematics, my students and I have been involved in some
very interesting, and some very frustrating, projects. We have
helped win (and lose) the America's Cup yacht races, navigate
commercial airliners, select stock portfolios, prescribe and monitor
drug treatment, recover oil, diagnose hypertension, design packing
crates for plastic containers, monitor electrocardiogram signals,
recommend which day to harvest alfalfa, design a toilet for the
German ICE high-speed trains, control the Laudering process in beer
brewing, select a fragrance for Pledge, and many others. This talk
will discuss some of the more interesting projects, the tools
applied, and the lessons learned. The intent is to convey the
flavor of the field of practicing industrial mathematics. A central
precept is ``Listen to your client'' so that you solve the problem
they have, rather than the problem you wish they had. You may apply
anything you know, but whatever you know is never enough.

A Preview of Automeshing: Its Status and Future in the Industry
Dr. Marzban Kermani
Cray Research

Meshing and pre-processing of geometry data, for the purpose of analysis,
has always been a very time-consuming matter. As the confidence level of
managers in CAE increases, more people are relying on CAE. However, as
more CAE is done and as the models get bigger, the problem of dealing
with the geometry becomes significantly larger, to the point that some
people decide it does not make economic sense to do a thorough analysis.
With faster processors today, it is possible to automate this process and
quite a bit of activity both in academia and industry is being focused on
this subject. In general, automatic tetra meshing has matured quite a
bit and both Delauney and Octree approaches offer almost everything an
analyst wants. However in the area of hexahedral meshing, there is more
work that needs to be done. A lattice method approach has been developed
by Cray Research, which is implemented in a code called Hexar. There are
some shortcomings in the approach. If these issues shortcomings are
addressed, faster and more reliable CAE can be performed. Some of these
issues are discussed in order to encourage curious minds to take up some
of these challenges for advancement towards virtual engineering.


From: Basem Attili <>
Date: Wed, 3 Apr 1996 00:05:08 +0300 (SAUST)
Subject: Theme Issue of the Arabian Journal for Science & Engineering

The editorial board of the Arabian Journal for Science and Engineering
(AJSE) plans to publish a theme issue in January of 1998 in Numerical
Methods for Differential Equations and Applications. The AJSE hopes to
bring together in a single issue research papers which represents the
state of the art in this vast and rapidly growing area.

- Numerical methods for ODEs and for PDEs
- Numerical functional analysis of DEs
- Numerical methods for linear and nonlinear eigenvalue problems.
- Numerical solution of DEs in science and engineering
- Computational aspects of numerical methods for DEs


- Deadline for submission of manuscripts: January 31, 1997
- Notification of acceptance of papers: June 30, 1997


Three copies of manuscripts, including illustrations and references
should be submitted to

Managing Editor
Arabian Journal For Science and Engineering
KFUPM Box # 8
King Fahd University of Petroleum and Minerals
Dhahran 31261, Saudi Arabia

More details may be obtained by contacting the managing editor at the
above address, telex 801060 KFUPM SJ, Fax (966-3) 860-3306, e-mail


From: Yuefan Deng <>
Date: Wed, 3 Apr 96 08:44:12 EST
Subject: Postdoctoral Position at Cornell Medical School

POSITION OPEN: Postdctoral Position in Supercomputer Simulation of Kidney.

INSTITUTION: Cornell University Medical College

ADDRESS; Department of Physiology and Biophysics
Cornell University Medical College
1300 York Avenue
New York, NY 10021

CONTACT PERSON: John L. Stephenson, M.D., Professor of Biomathematics
in Physiology

PHONE: (212) 746-6376
FAX: (212) 746-8690

DATE OF AVAILABILITY: Current. Hope to fill position no later
than July 1.

DESCRIPTION, JOB REQUIREMENTS: The overall goal of our work
is to synthesize experimental data at the membrane and molelcular level into
predictive mathematical models of the mammalian kidney that are useful in
understanding both its normal and diseased function.

Depending on species, the mammalian kidney consists of a
50,000 to one million similar but not identical units, the nephrons,
operating in parallel. Each nephron is a tube approximately 1 cm long and
10\u-3\d cm in diameter. The closed end is wrapped around a specialized
knot of capillaries to form the glomerulus; the open ends merge to empty
into the ureter and thence the bladder.

A protein and cell free filtrate of blood expressed by the glomerular
capillaries is modified as it flows down the nephron by the selective
reabsorption of most of the solutes and water and the selective secretion
of other solutes to form the final urine needed to maintain the composition
of the interstitial fluid bathing the cells of the body within the narrow
limits compatible with life.

Our current research is directed along two parallel paths:
1) Incorporation of the details of transcellular transport
into models of individual tubular segments.
2) Incorporation of the tubular models into architecturally detailed models
of single nephrons and of the whole kidney.
Simulation of these extended models lends itself naturally to the
parallel processing capabilities of supercomputers, and we have ongoing
projects at both the Cornell National Supercomputer Facility and at SUNY,
Stony Brook.

The present position is concerned with the development and coding of the
extended models. Requirements for the position are a background in numerical
analysis, ability to program in both Fortran and C, and some experience in
parallel computation.

APPLICATION REQUIREMENTS: Applications, which should include a resume, a
statement of research interest, a publication list, and the names of three
references should be submitted directly to Professor John L. Stephenson.


From: Dag Wedelin <>
Date: Thu, 4 Apr 1996 12:17:35 +0200 (MET DST)
Subject: Positions at Chalmers University of Technology, Gothenburg


The Department of Computing Science, Chalmers University of
Technology, Gothenburg, Sweden, offers job positions for
work within a three year ESPRIT-project within the HPCN
subprogram (High Performance Computing and Networking). The
project is a cooperation between Chalmers Univ of Tech, Univ
of Patras, Carmen Systems AB and Lufthansa.

The project is concerned with developing a new generation of
fully automatic scheduling methods, especially for airline
crew scheduling. For Chalmers the focus will be on
development of the main discrete optimization algorithms.
Modelling methods and parallel algoritms will also be

One ASSISTANT PROFESSOR (ref no 47/96). The formal
requirement for this postion is a PhD in optimization or
other field relevant to the project. The position also
involves teaching up to 25%. The position is limited to four

One RESEARCH ENGINEER (ref no 48/96). The position includes
technical development, implementation and some documentation
within the project. The formal requirement is a MSc or
similar degree in Mathematics or Computing Science.

One PHD STUDENT (ref no 49/96). The position is intended for
work towards a PhD in an area relevant to the project. Some
participation in underaduate education will be required. The
formal requirement is a MSc or similar degree in Mathematics
or Computing Science.

More information can be given by Researcher Dag Wedelin
(+46-31-7721022, or Head of Department
Christer Carlsson (+46-31-7721038).

Each applicant shall give a short description of his or her
qualifications. For the position as Assistant Professor a
short description of previous research and teaching
experience should also be given. The application should be
sent in two copies to the Personnel Office, Chalmers Univ of
Tech, S-412 96 Gothenburg, Sweden. The DEADLINE for
application is APRIL 19, 1996.


From: Iain Duff <>
Date: Thu, 4 Apr 96 21:57:29 BST
Subject: Contents, IMA Numerical Analysis


Pena J M
Pivoting strategies leading to small bounds of the errors for certain
linear systems.

Dai Y H and Yuan Y
Convergence properties of the Fletcher-Reeves method.

Bellettini G and Paolini M
Numerical simulations of measurements of capillary contact angles.

Coleman J P and Ixaru L Gr
P-stability and exponential-fitting methods for $y''=f(x,y)$.

Schroll H J and Winther R
Finite-difference schemes for scalar conservation laws with source terms.

Rathsfeld A
Error estimates and extrapolation for the numerical solution of
Mellin convolution equations.

Barrett J W and Blowey J F
An error bound for the finite-element approximation of a model for
phase separation of a multi-component alloy.

Hill A T and S\"uli E
Set convergence for discretizations of the attractor.


From: IMACS Administration <>
Date: Wed, 3 Apr 1996 13:58:50 -0500
Subject: Contents, IMACS Applied Numerical Mathematics


Z. Bai
A class of hybrid algebraic multilevel preconditioning methods 389

B. Cano and B. Garcia-Archilla
A generalization to variable stepsizes of Stormer methods
for second-order differential equations 401

C. de Nicola, G. Pinto and R. Tognaccini
A normal mode stability analysis of multiblock algorithms for
the solution of fluid-dynamics equations 419

P.M. de Zeeuw
Development of semi-coarsening techniques 433

S. Garcia
Higher-order incremental unknowns, hierarchical basis, and
nonlinear dissipative evolutionary equations 467

C.-H. Guo
Incomplete block factorization preconditioning for linear
systems arising in the numerical solution of the
Helmholtz equation 495

S.V. Tsynkov
Errata to "An effective numerical technique for solving a
special class of ordinary difference equations".
Applied Numerical Mathematics 18 (1995) 489-501 509


From: Richard Brualdi <>
Date: Thu, 4 Apr 1996 10:52:00 -0600 (CST)
Subject: Contents, Linear Algebra and its Applications

Volume 239

Florence Corpet (Castanet-Tolosan, France) and Artibano Micali
(Montpellier, France)
Algebres Genetiques et Mutation Quantique 1

Chi Fai Chan (Hong Kong)
Refinements of the Schur Inequality for Principal Characters 15

Po-Fang Hsieh (Kalamazoo, Michigan), Mitsuhiko Kohno
(Kumamoto, Japan), and Yasutaka Sibuya (Minneapolis, Minnesota)
Construction of a Fundamental Matrix Solution at a Singular
Point of the First Kind by Means of the SN Decomposition of
Matrices 29

M. N. Spijker and F. A. J. Straetemans (Leiden, The Netherlands)
Stability Estimates for Families of Matrices of Nonuniformly
Bounded Order 77

Hannu Valiaho (Helsinki, Finland)
P*-Matrices Are Just Sufficient 103

Thomas L. Moeller and Jaime Milstein (El Segundo, California)
Algebraic Representations for Finite-State Machines. I.
Monoid-Ring Formulation 109

Akio Arimoto and Takashi Ito (Tokyo, Japan)
Singularly Positive Definite Sequences and Parametrization of
Extreme Points 127

Shu-Lin Liu (Hohhot, People's Republic of China)
Bounds for the Greatest Characteristic Root of a Nonnegative
Matrix 151

Vladimir Matsaev and Vadim Olshevsky (Ramat Aviv, Israel)
Cyclic Dimensions, Kernel Multiplicities, and Gohberg-Kaashoek
Numbers 161

Katsutoshi Takahashi (Sapporo, Japan)
Eigenvalues of Matrices With Given Block Upper
Triangular Part 175

Ismor Fischer (Madison, Wisconsin) and Suk-Geun Hwang
(Taegu, Republic of Korea)
Certain Nonbarycentric Cohesive Matrices 185

K. R. Fuller (Iowa City, Iowa), W. K. Nicholson (Calgary,
Alberta, Canada), and J. F. Watters (Leister, England)
Direct Sums of Reflexive Modules 201

Alessandra Giovagnoli (Bologna, Italy) and Henry P. Wynn
(Warwick, United Kingdom)
Cyclic Majorization and Smoothing Operators 215

Irving S. Reed and Wolfgang F. Kraske (Los Angeles, California)
Reflections, Spinors, and Projections on a Minkowski Space
Underlie Dirac's Equation 227

Author Index 263


End of NA Digest