NA Digest Saturday, September 9, 1995 Volume 95 : Issue 36

Today's Editor:
Cleve Moler
The MathWorks, Inc.
moler@mathworks.com

Submissions for NA Digest:

Mail to na.digest@na-net.ornl.gov.

Information about NA-NET:

Mail to na.help@na-net.ornl.gov.

URL for the World Wide Web: http://www.netlib.org/na-net/na_home.html -------------------------------------------------------

From: Shi Zhong-ci <shi@indigo19.cc.ac.cn>
Date: Fri, 8 Sep 1995 11:01:11 +0000
Subject: Feng Kang Prize Awarded

Announcement

The first Feng Kang Prize of Scientific Computing will be honored to
Professor Chi-Wang SHU, Brown University US, Professor Jinchao XU, Pennsylvania
State University US and Professor Ya-xiang Yuan, the Institute of Computational
Mathematics and Scientific/Engineering Computing of the Chinese Academy of
Sciences for their significant contributions in Computational Fluid Dynamics,
multilevel and domain decomposition methods, and Optimization respectively.

The ceremony of announcement and reception of the prizes will be given at the
General Assembly of the Chinese Society of Computational Mathematics to be held
in Zheng Zhou, China, October 31--November 3, 1995.

The Foundation of Feng Kang Prize of Scientific Computing is founded in
memory of the contribution of late Professor Feng Kang to China's
Computational Mathematics. The Prize is honored to young Chinese
scientists in China or abroad, working in the fields of Scientific and
Engineering Computing whose age should not exceed 45.


Professor Zhong-ci Shi
Chairman
The Committee of the Feng Kang Prize of Scientific Computing
August 22, 1995


------------------------------

From: Jose Manuel Badia Contelles <badia@vents.uji.es>
Date: Wed, 6 Sep 1995 11:03:58 +0200 (METDST)
Subject: Gregory and Karney Eigenvalue Problem

Hi,

I am testing an algorithm to compute the eigenvalues of symmetric
tridiagonal matrices. I am using some matrices from
the book by Gregory and Karney entitled "A Collection of
Matrices for Testing Computational Algorithms". Unfortunately, I
do not have access to this book and I would like to know the formula
for the exact eigenvalues of the following matrix:

A=[bi-1, ai, bi]

ai = a for odd i
ai = b for even i
bi=1

I know by a reference that this matrix is the Example 7.4, pag. 137
of the above cited book.

Thank you

Jose M. Badia e-mail:badia@inf.uji.es
Departamento de Informatica jmbadia@dsic.upv.es
Universidad Jaume I Tel.(+34) 64 345771 ext (4720)
Castellon (SPAIN) (+34) 64 345769 ext (4720)


------------------------------

From: Jeff Cash <j.cash@ic.ac.uk>
Date: Thu, 7 Sep 1995 15:33:24 +0100
Subject: New Code for Stiff ODEs Available

NEW CODE

This is to announce a new code dmebdf.f for the solution of
stiff initial value problems. The code is available from NETLIB
by getting dmebdf.f from ode. The underlying algorithm is
based on modified extended backward differentiation formulae.
These formulae have considerably better stability than BDF and
are A-Stable up to order 4 and A(alpha)-Stable up to order 9.
Results for dmebdf.f on the stiff DETEST test set together with
the corresponding results for STRIDE, RADAU5, LSODE and DESI
(which may be used as comparisons), as well as results from the 'ring'
problem and for problems resulting from the MOL solution of PDEs
are available by anonymous FTP and on the WEB. These can be obtained
as follows:
Anonymous FTP: to cato.ma.ic.ac.uk (155.198.192.82). Codes are
in /pub/jeff and the user should first get readme.
WWW: http://www.ma.ic.ac.uk/~baxter/jeff/readme.html
Any comments (good or bad) will be gratefully received by the
author on
j.cash@ic.ac.uk

-- Jeff Cash


------------------------------

From: Xiaoye Li <xiaoye@CS.Berkeley.EDU>
Date: Thu, 7 Sep 1995 09:53:46 -0700
Subject: New High Performance Sparse Linear System Solver Available

We are pleased to announce the availability of the SuperLU package,
a set of subroutines to solve a sparse linear system A*X=B. It implements
Gaussian elimination with partial pivoting, using supernodes and BLAS
to optimize performance. It has achieved over 120 Mflops on realistic
examples on an IBM RS 6000/590; it is currently the fastest available
algorithm for a variety of problems.

SuperLU is implemented in ANSI C, and must be compiled with standard
ANSI C compilers. We supply the BLAS in C, but for highest performance
optimized BLAS should be used. Currently only the real single-precision and
double-precision versions are provided; the complex versions are still
under construction. The calling sequence is modeled on LAPACK. There
is also a Matlab mex-file interface, test and installation code,
a user guide and detailed technical report.

This package can be obtained via anonymous ftp from

ftp.cs.berkeley.edu : pub/lapack/SuperLU/superlu.tar.gz

SuperLU is joint work of James Demmel, Stan Eisenstat, John Gilbert, Xiaoye Li,
Joseph Liu, and Jinqchong Teo. It will also soon appear as part of the
LAPACK/ScaLAPACK library on netlib, at http://www.netlib.org/clapack.
Please direct all comments and suggestions to the the following people:

Xiaoye Li xiaoye@cs.berkeley.edu
John Gilbert gilbert@parc.xerox.com
James Demmel demmel@cs.berkeley.edu
Jinqchong Teo teo@pasteur.eecs.berkeley.edu


------------------------------

From: Bruce Hendrickson <bahendr@cs.sandia.gov>
Date: Fri, 8 Sep 95 12:53:54 MDT
Subject: Chaco-2.0: Graph Partitioning Software

ANNOUNCING CHACO-2.0
Software for Partitioning and Ordering Graphs

Many problems which arise in the course of scientific computing can be
conveniently described in terms of graph partitioning. A prominent
example is the problem of decomposing a large, unstructured grid across
the processors of a parallel computer. Other applications include generating
nested disection orderings for sparse matrix factorizations and devising
efficient circuit layouts.

Version 1 of our graph partitioning code "Chaco" has been licensed to over
100 sites around the world. We are now releasing version 2.0 with greatly
enhanced performance, ease of use and functionality. Chaco contains a
variety of partitioning algorithms including spectral bisection, quadrisection
and octasection, the inertial method, the Kernighan-Lin/Fiduccia-Mattheyses
algorithm and multilevel partitioners. Advanced techniques that are new to
version 2.0 include terminal propagation (a method for improving data locality
adapted from the circuit community), the ability to map partitions
intelligently to hypercube and mesh architectures, and easy access to the
Fiedler vector to assist the development of new applications of spectral
graph algorithms. This capability has already been used in applications
ranging from gene sequencing to database design.

A user's guide and papers describing the algorithms are available by
anonymous ftp to cs.sandia.gov in the directory pub/tech_reports/bahendr.
Academic user's can obtain the code at no charge under a research license
agreement and it may also be licensed for commercial application. Interested
parties should contact the first author at the address given below.

Bruce Hendrickson (bah@cs.sandia.gov)
Rob Leland (leland@cs.sandia.gov)
Sandia National Laboratories
Albuquerque, NM 87185


------------------------------

From: Mortaza Jamshidian <mori@gayley.math.ucla.edu>
Date: 8 Sep 1995 17:04:10 GMT
Subject: Journal of Statistical Software, A Proposal

Journal of Statistical Software
(proposals and reactions)

On

ftp://ftp.stat.ucla.edu/pub/JSS

and

http://www.stat.ucla.edu/develop/JSS/

you will find a proposal to establish an electronic journal that
documents, distributes, demonstrates, and describes computerized
statistical techniques. The journal will be freely available on
the WWW, and, wherever appropriate, the source code will be both
available for downloading and available for executing interactively
using WWW forms and scripts.

There are four files in the directories mentioned above
= jss.txt (the original proposal)
= jss.mail (email reactions from sci.stat.xxx and allstat)
= jss.core (core group working on the plans)
= jss.disc (extended proposal)

Reactions, suggestions, donations all welcome.


------------------------------

From: Jinchao Xu <xu@math.psu.edu>
Date: Thu, 7 Sep 1995 16:51:11 -0400
Subject: Temporary Change of Address for Jinchao Xu

Dear Colleagues and Friends:

I am on sabbatical leave during this academic year.
After 9/23/95 to summer/96, my working address is
Department of Mathematics
6363 Mathematical Sciences
U.C.L.A
Los Angeles, CA 90095-1555

I am currently visiting ICASE (till 9/22/95):
ICASE, Mail Stop 132C
Building 1192c, W. Taylor St.
NASA Langley Research Center
Hampton, VA 23681-0001

My email and URL addresses remain unchanged:
Email: xu@math.psu.edu
URL: http://www.math.psu.edu/xu/

Thanks for your attention.

Jinchao Xu


------------------------------

From: Rob Stevenson <stevenso@sci.kun.nl>
Date: Tue, 5 Sep 1995 11:54:10 +0200
Subject: Change of Address for Rob Stevenson

My new address is:

Rob Stevenson
Department of Mathematics
University of Nijmegen
P.O. Box 9010
NL-6500 GL Nijmegen
The Netherlands
Phone: +31 (0)80 653232 (After October 10, +31 (0)24 3653232)
Fax: +31 (0)80 652140 (After October 10, +31 (0)24 3652140)
email: stevenso@sci.kun.nl


------------------------------

From: Mohammad Reza Haghighat <mohammad@csrd.uiuc.edu>
Date: Fri, 8 Sep 95 14:20:32 CDT
Subject: New Book on Symbolic Analysis for Parallelizing Compilers

I'd like to announce the availability of the following new book:

Title : SYMBOLIC ANALYSIS FOR PARALLELIZING COMPILERS
Author : Mohammad R. Haghighat
Foreword : Constantine D. Polychronopoulos
Publisher : Kluwer Academic Publishers, 1995
ISBN : 0-7923-9623-5
Pages : 168, Hard Cover

SYMBOLIC ANALYSIS FOR PARALLELIZING COMPILERS is the first complete account of
the state-of-the-art symbolic analysis techniques for parallelizing compilers.
The book makes a compelling case for the potential of symbolic analysis
techniques by applying them with remarkable results to such important program
optimization problems as dependence analysis, static timing or size analysis,
and loop scheduling. Although automatic program parallelization is the focus of
the book, the presented techniques should prove useful in other realms where
understanding of programs is essential in achieving efficiency, proving
correctness, and reengineering the programs.

SYMBOLIC ANALYSIS FOR PARALLELIZING COMPILERS should be of interest to people
in high-performance computing, especially compiler practitioners and researchers
of high-performance compilers in industry and academia. In particular, readers
of the books of Dr. Utpal Banerjee, Prof. Constantine Polychronopoulos, and
Prof. Michael Wolfe may enjoy reading this book. Researchers in abstract
interpretation, semantics-based analysis, and automated reasoning may also find
the material useful and stimulating.

CONTENTS

1 INTRODUCTION
2 RELATED WORK
3 SYMBOLIC ANALYSIS
4 INDUCTION VARIABLES
5 INTERPROCEDURAL SYMBOLIC ANALYSIS
6 TIMING ANALYSIS AND SCHEDULING
7 IMPLEMENTATION AND EXPERIMENTS
8 CONCLUSIONS
A INTERVAL ANALYSIS

Mohammad R. Haghighat
mohammad@csrd.uiuc.edu


------------------------------

From: C Chiu <macchiu@sobolev.cityu.edu.hk>
Date: Mon, 4 Sep 1995 15:38:24 --800
Subject: Conference on Hyperbolic Problems in Hong Kong

Sixth International Conference on Hyperbolic Problems
Theory, Numerics, Applications

June 15-19, 1996, Hong Kong

Objective
This conference is to provide a high-standard forum for mathematical
scientists to present their latest progress on nonlinear hyperbolic
problems. It aims at bringing together senior scientists and young
researchers for academic interaction.

Scientific Committee

Josef Ballmann, RWTH-Aachen
Guiqiang Chen, Northwestern University
Constantine M. Dafermos, Brown University
Xiaqi Ding (Cochair), Shantou University
Bjorn Engquist, UCLA
James G. Glimm, SUNY Stony Brook
S.K. Godunov, Moscow State University
Chaohao Gu, Fudan University
Rolf Jeltsch, Eidgen Tech Hochschule
Peter D. Lax, New York University
Tai-Ping Liu (Cochair), Standford University
Piero Marcati, U Degli Studi Dell'Aquila
Takaaki Nishida, Kyoto University
Benoit Perthame, University of Paris IV
Longan Ying, Beijing University
Tong Zhang, Academia Sinica

Organizing Committee

Roderick Wong (Cochair), Dept of Math, City U of HK
Tong Yang, Dept of Math, City U of HK
Shing-Tung Yau (Cochair), Institute of Mathematical Sciences, Chinese U of Hong Kong
Jun Zou, Dept of Math, Chinese U of Hong Kong

Sponsors

The Chinese University of Hong Kong
The City University of Hong Kong>

Call for Papers
Titles and abstracts of contributed papers must be received by
JANUARY 31, 1996. The abstracts should be typed by Latex not
to exceed one page, and sent to Dr. Tong Yang by email or floppy disk.

For more information, please contact:
Dr. Tong Yang
Department of Mathematics
City University of Hong Kong>
Tel: (852) 2788-9819, Fax: (852) 2788-8561
Email: mago@cityu.edu.hk


------------------------------

From: Fred Hickernell <fred@fisher.math.hkbc.hk>
Date: Thu, 7 Sep 1995 15:06:51 +0800 (EAT)
Subject: Workshop on Quasi-Monte Carlo Methods

WORKSHOP ON QUASI-MONTE CARLO METHODS AND THEIR APPLICATIONS
HONG KONG, DECEMBER 11-13, 1995

The efficiency of of Monte Carlo methods can often be improved by using
uniformly scattered deterministic points instead of pseudo-random points.
These quasi-Monte Carlo methods (also known as number-theoretic methods)
have been applied to a variety of numerical and statistical problems,
including design of experiments, statistical inference, multidimensional
quadrature, optimization and simulation.

This fall a workshop will be held at Hong Kong Baptist University to
discuss the latest developments in quasi-Monte Carlo methods. The invited
speakers include:

Russel Caflisch, University of California at Los Angeles
Kai-Tai Fang, Hong Kong Baptist University
Fred Hickernell, Hong Kong Baptist University
Harald Niederreiter, Austrian Academy of Sciences
Art Owen, Stanford University
Ian Sloan, University of New South Wales
Yuan Wang, Chinese Academy of Sciences, Beijing
Jeff Wu, University of Michigan
Henry Wynn, University of Warwick

The registration fee for this workshop is HK$600 (US$78), which includes
conference materials, lunches and a banquet. Those wishing to attend
and/or contribute a paper should contact either of the organizers:

Kai-Tai Fang (ktfang@hkbu.edu.hk)
Fred Hickernell (fred@hkbu.edu.hk)
Department of Mathematics
Hong Kong Baptist University
Kowloon Tong, HONG KONG
Fax: +852 2336 1505
Phone: +852 2339 7015
http://www.math.hkbu.edu.hk


------------------------------

From: John Dennis <dennis@masc6.rice.edu>
Date: Wed, 6 Sep 95 11:32:46 CDT
Subject: Postdoctoral Postion at Rice University

Part-time Postdoctoral Position at Rice University

The Optimization Project of the Center for Research on
Parallel Computation has a half-time postdoctoral position
in applications and algorithm development for numerical
nonlinear programming. In addition to a research interest
in this area, it is crucial that the applicant be a good
writer and speaker and have an interest in modern software
design and implementation. This position will be responsible
for tracking and managing research budgets and editing reports
for the members of the group. The position will be under
the direction of John Dennis and will be an appointment in
the Department of Computational and Applied Mathematics at
Rice University in Houston, Texas. Contact John Dennis,
dennis@caam.rice.edu, immediately if you are interested in
at least one year in this position.


------------------------------

From: R. Baker Kearfott <rbk5287@interval.usl.edu>
Date: Fri, 8 Sep 1995 16:17:34 -0500
Subject: Contents: Reliable Computing

Reliable Computing. - 1995. - 1 (2) - 102 p.
(Special Issue: Parallel Algorithms for Interval Computations)

CONTENTS

Parallel interval-based reasoning in medical knowledge-based
system CLINAID
Ladislav J. Kohout, Isabel Stabile, Hasan
Kalantar, Maria F. San-Andres, and John Anderson 109

Applications of interval computations to earthquake-resistant
engineering: How to compute derivatives of interval functions
fast
Vladik Kreinovich, David Nemir, and Efren Gutierrez 141

A reliable linear algebra library for transputer networks
Christian P. Ullrich and Roman Reith 173

Parallel accurate linear algebra subroutines
J\"urgen Wolff von Gudenberg 189

Addresses of the Editorial Board Members 201

Information for Authors 203

Contents 205

====================================================================

Reliable Computing. - 1995. - 1 (3) - 162 p.
(Special Student Issue)

CONTENTS
Editors' Introduction 209

Mathematical research

Application of interval Newton's method to chemical engineering
problems
Gopalan V. Balaji and J. D. Seader 215

Ockham's razor in interval identification
Bo H. Friesen and Vladik Kreinovich 225

An automatic and guaranteed determination of the number of roots
of an analytic function interior to a simple closed curve in the
complex plane
Jonathan Herlocker and Jeffrey Ely 239

A general iterative sparse linear solver and its parallelization
for interval Newton methods
Chenyi Hu, Anna Frolov, R. Baker Kearfott, and Qing Yang 251

Optimizing INTBIS on the CRAY Y-MP
Chenyi Hu, Joe Sheldon, R. Baker Kearfott, and Qing Yang 265

Study on sampling techniques with CMMs
Thomas J. McLean and David H. Xu 275

A new characterization of the set of all intervals, based on the
necessity to check consistency easily
Driss Misane and Vladik Kreinovich 285

Interpolation that leads to the narrowest intervals and its
application to expert systems and intelligent control
Hung T. Nguyen, Vladik Kreinovich, Bob Lea, and Dana Tolbert 299

A parallel complex zero finder
Mark J. Schaefer and Tilmann Bubeck 317

A software interface and hardware design for variable-precision
interval arithmetic
Michael J. Schulte and Earl E. Swartzlander, Jr. 325

A bright side of NP-hardness of interval computations: interval
heuristics applied to NP-problems
Bonnie Traylor and Vladik Kreinovich 343

Information
Special Issue 361

1994 Lotfi A. Zadeh Best Paper Award in the field of fuzzy logic 362

Addresses of the Editorial Board Members 363

Information for Authors 365

Contents 367


------------------------------

From: Iain Duff <I.Duff@letterbox.rl.ac.uk>
Date: Mon, 4 Sep 95 10:19:21 BST
Subject: Contents, IMA Numerical Analysis

IMA JOURNAL OF NUMERICAL ANALYSIS --- Volume 15, Number 4.

Martins M M and Trigo M E
An error bound for the modified successive overrelaxation method.

Miettinen M and Haslinger J
Approximation of non-monotone multivalued differential inclusions.

Carr J, Duncan D B, and Walshaw C H
Numerical approximation of a metastable system.

Malek A and Phillips T N
Pseudospectral coollocation methods for fourth-order differential equations.

Lubich Ch and Ostermann A
Linearly implicit time dicretization of non-linear parabolic equations.

Elster C and Neumaier A
A grid algorithm for bound constrained optimization of noisy functions.

List of referees

Index to Volume 15



------------------------------

From: Marilyn Radcliff <radcliff@math.ohio-state.edu>
Date: Wed, 6 Sep 1995 11:14:01 -0400 (EDT)
Subject: Contents, Approximation Theory

Table of Contents: J. Approx. Theory, Volume 82, Number 1, July 1995

Franc\c ois Dubeau and Jean Savoie
Explicit error bounds for spline interpolation on a uniform partition
1--14

V. I. Filippov and P. Oswald
Representation in $L_p$ by series of translates and dilates of one
function
15--29

J\"orgen L\"ofstr\"om
Real interpolation with constraints
30--53

Siqing Xie
Regularity of $(0,1,\dots,r-2,r)$ and $(0,1,\dots,r-2,r)*$ interpola-
lations on some sets of the unit circle
54--59

Kenneth I. Gross and Donald St. P. Richards
Total positivity, finite reflection groups, and a formula of Harish--
Chandra
60--87

Uwe Franken
Extension of functions with $\omega$-rapid polynomial approximation
88--98

G\"unther Meinardus, Hennie ter Morsche, and Guido Walz
On the Chebyshev norm of polynomial $B$-splines
99--122

Erich Novak
The adaption problem for nonsymmetric convex sets
123--134

K. Yu. Osipenko
On $n$-widths of holomorphic functions of several variables
135--155

K. Yu. Osipenko
Exact values of $n$-widths and optimal quadratures on classes of
bounded analytic and harmonic functions
156--175



------------------------------

End of NA Digest

**************************
-------