NA Digest Sunday, June 22, 1997 Volume 97 : Issue 25

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: Nicolas Robidoux <>
Date: Mon, 16 Jun 1997 09:29:54 -0600 (MDT)
Subject: Stopping Criterion for Conjugate Gradients

My first idea was to try the following:

When using conjugates gradient to solve S x = b (S is symmetric
positive definite), since the norm of the residual in exact arythmetic
is monotonically non-increasing, one can assume that roundoff is being
felt when the residual increases.

But this doesn't work. For one reason, the residual whose norm decreases
is not the readily computed one.

The problem is that if the system is nice enough, this actually will
happen only when one reaches exponent overflow since everything will
scale nicely.

Nonetheless, has someone used this criterion as an ADDITIONAL stopping
criterion? Warnings? Caveats?

This brings me to my main questions, which I was hoping and failed to
bypass with the above idea.


Suppose that you are solving Ax = b with a preconditioner M, so that
it is required to solve systems of the form Mx = b. You have decided
that the most practical way to solve the latter system is with an
iterative solver; consequently, you have nested iterative solves.

Does anyone know of a good way of choosing a stopping criterion for
the inner iterative solve so that the outer iterative method is not
too contaminated (which may cause breakdown) by the inner? Methods
which are satisfied with very rough matrix norm estimates preferred.


The second version is essentially identical to the first save that it
relates to contamination from an iterative solve used to effect a
matrix multiply by a FACTOR of the overall coefficient matrix.

Rather than present it in general fashion, I will do it in the context
in which I am working.

Suppose that you are solving the following with conjugate gradients

delta^T star^{-1} delta u + D u = f

where delta is a simple difference operator, star is local and
symmetric positive definite (in 1D, banded), and D is diagonal. The
overall system is symmetric positive definite.

First of all, a local approximation is used as an overall
preconditioner by using a local approximation to star^{-1}, this local
system being solved with conjugate gradients so that the issue raised
in VERSION 1 applies.

Moreover, in order to effect multiplication by
delta^T star^{-1} delta, conjugate gradients is used to solve systems
of the form star v = w. How should one set the stopping criterion on
this latter solve so that the outer iterative processes produce good

In situations where star is not exactly symmetric, I intend to use a
splitting method or a "generalized" conjugate gradients method in
which the preconditioner is the symmetric part, but the same issues

Nicolas Robidoux
Albuquerque Resource Center of the High Performance Computing,
Education and Research Center (starting late August: Massey
University, New Zealand)


From: Medovikov Alexei <>
Date: Tue, 17 Jun 97 15:49:23 +0300
Subject: Fortran Codes for Explicit Methods for Stiff ODE

New Fortran code of explicit method order 3 and order 4 for stiff and
non-stiff ordinary differential equations you can get in .
Programs DUMKA3.for (order 3) and DUMKA4.for (order 4)
don't use linear algebra procedure and give square root acceleration
for stiff problems.

email: for Medovikov Alexei.


From: Rob Bisseling <>
Date: Wed, 18 Jun 1997 10:15:10 +0200
Subject: Oxford BSP Toolset

The Oxford BSP Toolset and Profiling system -- version 1.1

We are pleased to announce the first official release of the Oxford
BSP Toolset which conforms to the BSPlib programming library
definition. BSPlib is an SPMD communications library, and is based on
the Bulk Synchronous Parallel (BSP) model of parallel computation. The
main feature of BSPlib is two modes of communication, one capturing a
BSP oriented message passing approach, and the other reflecting a
(one-sided) direct remote memory access paradigm. The core BSP library
consists of just 20 primitives. Sources are freely available by
anonymous FTP. For more information on the toolset distribution see:


From: Martin Peters <>
Date: Mon, 16 Jun 1997 14:22:24 +0100
Subject: New Book on Scientific Computing with Maple and MATLAB


Solving Problems in Scientific Computing Using Maple and MATLAB

Walter Gander and Jiri Hrebicek (eds.)

3rd expanded and revised edition, Springer, ISBN 3-540-61793-0,
due July 1997, 432 pp., US-$ 49.95, DM 78.-,pounds sterling 34.00

This book consists of 27 chapters with topics ranging from the
calculation of orbits and sun dials to Gauss quadrature and
Runge-Kutta formulae and being situated in domains such as
physics, celestial mechanics, thermodynamics, statistical physics,
quadrature, penetration theory ... The subjects deal with
optimization, initial and boundary value problems, integration,
least-squares problems reflection, coordinate transformations,
conformal mapping, calibration, heat flow, penetration phenomena,
boson particles, compression in metal forming,... Basic elements
are explained (e.g. properties of orthogonal polynomials, the
Newtonian equations of motion, line and curve reflection, etc.).
Formulas whenever presented are made easily understandable.

All chapters have been adapted to the newest versions of Maple
(Version 5 Release 4) and MATLAB 5. All Maple and MATLAB programs
can be obtained from

Contact Person at Springer-Verlag:

Dr. Martin Peters
Mathematics Editor phone: *49-6221-487 409
Springer-Verlag fax: *49-6221-487 355
Tiergartenstr. 17 e-mail: peters
69121 Heidelberg


From: Vladan Babovic <>
Date: Mon, 16 Jun 1997 12:48:59 +0000
Subject: Hydroinformatics 98 Call for Papers

Hydroinformatics '98
Copenhagen, Denmark, August 24-26, 1998

This is the third international conference on Hydroinformatics organized by
the International Association for Hydraulic Research (IAHR) and it is on this
occasion hosted by Danish Hydraulic Institute.

Conference topics encompass, but are not restricted to:

* Data Acquisition and Analysis (SCADA, remote sensing, data modelling,
data management and data-base technology)
* Advances in Numerical Methods and Techniques (advances in 1-, 2-, and 3-D
computational hydraulics, water quality and ecological modelling, data
assimilation, parameter estimation and process identification)
* Control Techniques and Decision Support (development and application of
control techniques, model based control, uncertainty handling, decision
support systems, distributed impact assessment and decision making:
Internet and intranet)
* Experiences with standard software (case studies in coastal and estuarine
processes, water resources/basin management, urban drainage systems,
educational software - computer-aided learning)
* Emergent technologies (evolutionary algorithms, neural networks, fuzzy logic,
distributed and decentralised simulation, object- and agent-orientation,
data mining)

A tutorial day will be held in conjunction to the Conference on Sunday,
August 23, 1998.

* Evolutionary Algorithms * Artificial Neural Networks
* Geographic Information Systems * Computational Hydraulics
* Internet and Intranet * Environmental Impact Assessment
* Data mining * Ecological Simulations

Hydroinformatics '98
Danish Hydraulic Institute
Agern Alle 5, DK-2970 Horsholm
Phone: +45 - 45 76 95 55
Fax: +45 - 45 76 25 67

Authors are invited to submit 500 word abstracts of their papers in the
English language to the Organising Committee of the Conference as soon as
possible and not later than September 15, 1997. The authors of high-quality
abstracts will be invited to submit their full contributions before
February 01, 1998.

September 15, 1997 Submission of Abstracts
October 15, 1997 Acceptance of Abstracts
February 01, 1998 Submission of Papers
August 23, 1998 Tutorial Day and Welcoming of delegates
August 24, 1998 Conference


From: Manuel Salas <>
Date: Mon, 16 Jun 1997 10:02:19 -0500
Subject: Computational Aerosciences Workshop


January 21-23, 1998
H.J.E. Reid Conference Center, NASA Langley Research Center
Hampton, Virginia


Over the last decade, the role of computational simulations in all
aspects of design has steadily increased. Today, large scale
computations are prevalent in all areas of aeronautical, atmospheric,
and space research. However, despite the many advances in computational
modelling over the last decade, the time required for computations
necessary to significantly advance the state-of-the art is far too long.
The impact is that significant improvements in designs, proof of concept
for revolutionary designs, and a basic understanding of important
physics is severely deterred. In fact, it could be argued that
effective/efficient progress in most areas of aeronautic, atmospheric
and much of space access research is currently paced by, and indeed
hostage to, the cost and speed of carrying out parametric computations.
In many cases, further progress is severely impeded or nearly impossible
without a dramatic change in the status quo. The purpose of this
workshop is to share new and emerging ideas that enable complex,
multidisciplinary, numerical simulations in the aerospace sciences in
time periods of minutes to hours instead of weeks or months. Time to set
up the simulations, including surface definition or alteration and
gridding (if required) are included in simulation time as is any
necessary post-processing. It should be emphasized that the workshop is
aimed at examining new ideas and methodologies that may not be fruitful
or possible at the current time, but may be so in the next twenty years.

Session I: Computer Hardware/Networking/Architecture
Keynote Speaker: Burton Smith, Tera Computer Company
Contributed Papers

Session II: Human/Computer Interface
Keynote Speaker: Rick Stevens, Argonne National Laboratory
Contributed Papers

Session III: Geometry Modelling and Grid Generation
Keynote Speaker: Jamshid Samareh, NASA Langley Research Center
Contributed Papers

Session IV: Algorithms
Keynote Speaker: David Keyes, Old Dominion University and ICASE
Contributed Papers

There will be a panel discussion following each session. In addition, a
poster session/reception will be held Wednesday evening, January 21, 1998.

The deadline for submitting abstracts (please limit to one page) is
August 15, 1997. Notification of acceptances will be sent out by October
1, 1997. Final papers/viewgraphs will be published in a NASA Conference
Proceeding. Please submit to:

Emily Todd, Conference Manager
Mail Stop 403
NASA Langley Research Center
Hampton, VA 23681-0001

Additional information about this meeting will be posted on the World
Wide Web at:

* This workshop is open only to U. S. citizens and permanent residents.


From: Jean Marie Chesneaux <>
Date: Tue, 17 Jun 1997 14:12:57 +0200
Subject: Conference on Real Numbers and Computers


Pierre et Marie Curie University
April 27-28-29 1998

The conference continues the series of RNC-symposia which have been held
in St-Etienne (France, 1994) and Marseille (France, 1995). These conferences
traditionally cover all the aspects in relationship with an efficient
handling of real numbers on computers : program portability problems, "exact
arithmetics", numerical reliability, number systems, architectures etc.

Many members of the scientific community are concerned by these problems,
they could share their knowledge and come up with solutions. But they do not
have the opportunity to meet, they do not belong to the same scientific fields
(computer science, number theory, numerical analysis, computer algebra) and
they have a different vocabulary. One of the aims is to put them together
during this meeting.

The topics covered by the conference are

- Algorithms and architectures for "serial" and "on line" arithmetic.
- Relations between number theory, automata theory and computer arithmetic.
- Number systems
- Floating point arithmetic
- Calculability
- Symbolic manipulation of numbers
- Algorithms for "exact" computing
- Multi-precision, interval arithmetic, stochastic arithmetic ...
- Accuracy problems in various fields (geometry, physics,...), and
proposed solutions.

January 1st 1998

You can submit a full paper (not an abstract) to:
preferably in LaTeX. Please request a pattern LaTeX file from the above
address. If you cannot use LaTeX, send 4 copies of a printed version to

Jean-Marie Chesneaux
Laboratoire LIP6
Universite Pierre et Marie Curie
4 place Jussieu, 75252 Paris Cedex 05, FRANCE.

All information can be found at the URL address :
Questions can also be sent at the e-mail adress :

Local committee
(LIP6, UPMC, Paris, France)

Program committee

Jean-Paul ALLOUCHE (France),Rene ALT (France),Jean-Claude BAJARD (France),
Jean-Claude BERGES (France),Vasco BRATTKA (Germany),Jean-Marie CHESNEAUX
(France), Martin Hotzel ESCARDO (United Kingdom),Christiane FROUGNY (France),
Peter KORNERUP (Danmark),David LESTER (United Kingdom),Pierre LIARDET (France),
Dominique MICHELUCCI (France),Jean-Michel MULLER (France),Nathalie REVOL


From: Aurel Galantai and Gyorgy Szeidl <>
Date: Tue, 17 Jun 1997 15:01:55 +0000
Subject: Numerical Methods and Computational Mechanics 98

24-27th August 1998
Miskolc, Hungary
Organized by
Central European Association for Computational Mechanics
Janos Bolyai Mathematical Society Hungarian Academy of Sciences
University of Miskolc

The aim of the conference is to bring together numerical
analysts of classical topics, specialists in computational mechanics
who are interested in the development; implementation and application
of advanced methods for reliable mathematical simulation of
structural and mechanical systems. Special attention will be given
to the finite element method and related techniques. The venue
was selected so as to make it possible for researchers from
central Europe and the former Soviet Union to participate.

-- Numerical algebra (sparse and dense linear systems, eigenvalue
problems, nonlinear systems, parallel algorithms, etc.)
-- Numerical solution of differential equations (IVP and BVP,
for ODEs and DAE, difference equations, FEM and BEM,
multigrid, difference methods, spectral methods, parallel
algorithms, etc.)
-- Computational mechanics (FEM, BEM, Optimization, Parallel
algorithms etc)
Forty five minute plenary lectures, thirty minute minisymposium
lectures and twenty minute long contributed talks are planned.
Honorary chairman:
P. ROZSA, Technical University of Budapest, Hungary

Chairman: B.SZABO, Washington University in St. Louis, USA
R.P.AGARWAL, National University of Singapore, Singapore
I. BABUSKA, The University of Texas, USA
C. BROYDEN, University of Bologna, Italy
M. GRIEBEL, University of Bonn, Germany
M. KLEIBER, Polish Academy of Sciences, Poland
H. MANG, Vienna Technical University, Austria
P. NEITTAANMAKI, University of Jyvaskyla, Finland
E. STEIN, University of Hannover, Germany
W. WENDLAND, Universitat Stuttgart, Germany
T. CZIBERE, University of Miskolc, Hungary
I. KOZAK, University of Miskolc, Hungary
Z. GASPAR, Technical University of Budapest, Hungary
I. PACZELT, University of Miskolc, Hungary

Authors who wish to present a 20-minute contributed lecture should
submit a two-page abstract in English and LATEX format with
article 12pt style. Page size is A4. Authors are kindly asked to
submit their abstracts via e-mail. Deadline: 30th of April, 1998.

Please visit our WWW pages which are regularly updated and provide
you with more information on formal requirements, accommodation
travelling, schedule etc.

If you are interested in attending the conference and wish to be on
the mailing list, please SEND NOW, a message to
--Institute of Mathematics or Department of Mechanics
--University of Miskolc
--3515 Miskolc-Egyetemvaros
--Hungary with your personal data (mail-address, phone number, fax
and e-mail address). The whole text of the First Announcement will
have been sent to persons on our mailing list as of the middle of
Phone: 36-46-365-111 Fax: 36-46-365-174


From: Gareth Shaw <>
Date: Thu, 19 Jun 1997 14:31:59 +0000 (BST)
Subject: NAG Fastflo Courses

NAG Fastflo Courses

To introduce its new Fastflo finite element package NAG (in
collaboration with the developers CSIRO) will be holding two
free one-day courses:

July 8th at the NAG Ltd offices in Oxford, England.

July 17th at Stanford University (immediately after the
SIAM conference, where Fastflo will be exhibited).

The courses will be largely an extended introduction, but will
also give a flavour of some of the more complex problems that
can be tackled using Fastflo.

Fastflo is a flexible 3D finite element software package designed
for simulation and rapid prototyping applications in engineering,
manufacturing, research and teaching, with a particular emphasis
on CFD. It can be used to solve problems ranging from the simple
classical PDEs (Poisson, Helmholtz, Heat equation, etc) to three-
dimensional turbulent flow simulations in complex geometries.
Fastflo is currently available for PCs under Windows 95 and
Windows NT, with UNIX implementations to be announced shortly.
For further information about Fastflo see:

If you are interested in attending one of the introductory courses,
or would like more information about Fastflo, please contact one of
the NAG Response Centres:

UK: Tel: +44 (0)1865 311744
Fax: +44 (0)1865 311755

North America and Canada: Tel: +1 630 971 2345
Fax: +1 630 971 2346


From: Dugald Duncan <>
Date: Thu, 19 Jun 1997 16:56:23 +0100
Subject: Scottish Computational Mathematics Symposium 97

10am-5pm Monday 22nd September
Hume Tower, University of Edinburgh, Scotland.

organised by D.B. Duncan (Heriot-Watt University) and D.M. Sloan
(University of Strathclyde) with the aim of bringing together
mathematicians and others who develop and/or use computer algorithms
to solve mathematical problems.

The meetings are open to everyone interested.


* M J Baines (University of Reading)

* D S Broomhead (UMIST, Mathematics)

* G Stewart (Heriot-Watt University, Petroleum Engineering
and EPS Ltd)

* A Trefethen (NAG)

* L N Trefethen (University of Oxford, Computing Laboratory)


Fees: 20 pounds, (10 pounds students)

To register (by September 10th) , send a cheque payable to Heriot-Watt
University and include your address so that we can contact you. The fee
includes tea, coffee and lunch.

To register or to obtain further information please contact:

SCMS97, Department of Mathematics,
Heriot-Watt University, Edinburgh, EH14 4AS.

WWW Information:


From: Oleg Burdakov <>
Date: Sat, 21 Jun 1997 18:06:44 +0200 (MET DST)
Subject: Sample Copy of Optimization Methods and Software

I would like to let you know that it is possible to order a free sample
copy of the journal "Optimization Methods and Software" (OMS) on the
WWW page:

where there is an item "Send me a free sample copy of the journal:"
The table of Contents for the journal OMS is available on:

Oleg Burdakov,
Managin Editor of the journal OMS


From: Stephen Wright <>
Date: Fri, 20 Jun 1997 11:08:39 -0500
Subject: Postdoctoral Position at Argonne National Laboratory

Postdoctoral Position

Mathematics and Computer Science Division
Argonne National Laboratory

Argonne National Laboratory has an opening for a postdoctoral research
position in the Mathematics and Computer Science Division and invites
outstanding candidates to apply. Candidates should have a Ph.D. in applied
mathematics or computer science and comprehensive knowledge of optimization
algorithms and/or numerical linear algebra. Familiarity with parallel
computing and with scientific computing applications in such areas as
biochemistry, chemistry, materials science, or geophysics is desirable but
not required.

The successful candidate will participate in the development of algorithms
and software for computational crystallography under the umbrella of a
DOE-sponsored project titled "Supercomputer Solution of Massive
Crystallographic and Microtomographic Structural Problems." The candidate
will interact with researchers at Argonne's Advanced Photon Source and at
the Consortium for Advanced Radiation Sources at the University of Chicago.
Please see the following URL for further details:

The Mathematics and Computer Science Division supports an excellent
computational environment that includes access to high-performance
scientific workstations, a scientific visualization and virtual reality
laboratory, and state-of-the-art parallel computers.

Argonne is located in the southwestern Chicago suburbs, offering the
advantages of affordable housing and good schools, as well as easy access to
the cultural attractions of the city.

Applicants must have received their Ph.D. not more than three years prior to
the beginning of the appointment. The appointment is available immediately
and for a one-year term (renewable). Applications should be addressed to
Walter McFall, Box mcs-xray, Employment and Placement, Argonne National
Laboratory, 9700 S. Cass Avenue, Argonne, IL 60439, and must include a
resume and the names and addresses of three references. For additional
information, contact Steve Wright (

Argonne is an affirmative action/equal opportunity employer.


From: Chris Diesch <>
Date: Fri, 20 Jun 97 13:39:04 PST
Subject: Position at NERSC


TITLE: Staff Scientist/Engineer
WORKING TITLE: Computer Scientist/Department Head
SALARY RANGE: $3,500 - $10,200
TYPE: Full time, career

DIVISION: Computing Sciences/NERSC
DEPARTMENT: High Performance Computing Research Dept.
GROUPS: Future Technologies, Scientific Computing,

REPORTS TO: Horst Simon
DATE: Open until filled

Interested parties may forward a resume or curriculum vitae via our
Mailing Address:

One Cyclotron Road, MS 938A Box-NERSC, Berkeley, CA 94720
(Reference "NERS/205" in the cover letter)


via E-Mail to:

1. Send as plain text.
2. Reference "NERS/205" in the SUBJECT of the message.

Visit our web sites at and for more
information about NERSC and the Lawrence Berkeley National Laboratory.


Full time, career position as Department Head of the High Performance
Computing Research Department (Future Technologies, Scientific
Computing, Visualization) within NERSC. Reports to Horst Simon,
Division Director.



Reporting to the NERSC (National Energy Research Scientific Computing)
Division Director, the Department Head of the High Performance
Computing Research Department is expected to provide intellectual and
technical leadership and management to several computer science and
scientific computing research groups, which develop and implement
technologies and strategies for the efficient use of high-performance
computing in a large scale production supercomputer center. With input
from stake holders, the Department Head is expected to develop short-
and long-term research and development plans and propose new
technology directions for NERSC. The Department Head will be directly
responsible for the management of the Future Technologies, Scientific
Computing, and Visualization Groups in NERSC. He/she will create
partnerships with other similar research groups to carry out program
goals. The successful candidate is expected to provide planning and
management support to DOE , as well as to develop new initiatives and
budgets, and obtain funding. The Department Head will maintain a work
environment the embraces diversity and fosters creativity and
innovation, and ensure that DOE and Laboratory rules and policies are



Nationally-known individual with a distinguished record of scientific
accomplishment and program management. Demonstrated experience in
leading and managing computer science/scientific computing research
and development teams. Understanding of the requirements of
large-scale computational science, and be able to project a vision for
scientific computing environments of the future. Successful record of
developing partnerships and collaborative relationships with
universities, research laboratories, and others to carry out research
and to deploy high-performance computing technologies. Strong
management and administrative skills, creativity in decision making
and problem-solving, and experience in human resource management.
Ability to interact with all levels of management at the DOE, the
University of California, and other national laboratories. Excellent
communication skills, and a demonstrated ability to facilitate
communications within institutions and with the research community,
government agencies, and industry.


Ph.D. or equivalent experience in Computing Sciences or related
scientific field.


From: Kelly Thomas <>
Date: Mon, 16 Jun 97 11:08:40 EST
Subject: Contents, SIAM Journal on Applied Mathematics

SIAM Journal on Applied Mathematics
Volume 57, Number 4, AUGUST 1997

Phase Transition in van der Waals Fluid
Din-Yu Hsieh and Xiao-Ping Wang

Hyperbolicity and Optimal Coordinates for the Three-Dimensional Supersonic
Euler Equations
W. H. Hui and Yuanping He

On the Modulational Instability of O(1) Amplitude Waves in Supersonic Boundary
Philip Hall and Demetrios T. Papageorgiou

Shear Band Formulation Due to a Thermal Flux Inhomogeneity
J. A. DiLellio and W. E. Olmstead

Vortical Flow Outside a Sphere and Sound Generation
Omar M. Knio and Lu Ting

Viscous Fingering: An Optimal Bound on the Growth Rate of the Mixing Zone
Felix Otto

Dynamic and Steady-State Behavior of Continuous Sedimentation
Stefan Diehl

Global Asymptotic Behavior of a Chemostat Model with Discrete Delays
Gail S. K. Wolkowicz and Huaxing Xia

Aggregation, Blowup, and Collapse: The ABC's of Taxis in Reinforced Random
Hans G. Othmer and Angela Stevens

Behavior Changes in SIS STD Models with Selective Mixing
James M. Hyman and Jia Li

Local Tomography II
Adel Faridani, David V. Finch, Erik L. Ritman, and Kennan T. Smith

Local Tomography for the Generalized Radon Transform
A. I. Katsevich

Dynamical Hysteresis Without Static Hysteresis: Scaling Laws and Asymptotic
Guillermo H. Goldsztein, Fernando Broner, and Steven H. Strogatz


From: Lisa Dougherty <>
Date: Thu, 19 Jun 97 08:44:00 EST
Subject: Contents, SIAM Journal on Computing

SIAM Journal on Computing
Volume 26, Number 4, AUGUST 1997

An Optimal Probabilistic Protocol for Synchronous Byzantine Agreement
Pesech Feldman and Silvio Micali

Error-Resilient Optimal Data Compression
James A. Storer and John H. Reif

Constant-Time Randomized Parallel String Matching
Maxime Crochemore, Zvi Galil, Leszek Gasieniec, Kunsoo Park, and
Wojciech Rytter

Learning from Multiple Sources of Inaccurate Data
Ganesh Baliga, Sanjay Jain, and Arun Sharma

Singular and Plural Nondeterministic Parameters
Michal Walicki and Sigurd Meldal

Data Structures' Maxima
G. Louchard, Claire Kenyon, and R. Schott

Oracles that Compute Values
Stephen Fenner, Steven Homer, Mitsunori Ogihara, and Alan Selman

Fast Discrete Polynomial Transforms with Applications to Data Analysis
for Distance Transitive Graphs
J. R. Driscoll, D. M. Healy, Jr., and D. N. Rockmore

Doubly Logarithmic Communication Algorithms for Optical-Communication
Parallel Computers
Leslie Ann Goldberg, Mark Jerrum, Tom Leighton, and Satish Rao

The Robot Localization Problem
Leonidas J. Guibas, Rajeev Motwani, and Prabhakar Raghavan

A Fast Algorithm for Optimally Increasing the Edge Connectivity
Dalit Naor, Dan Gusfield, and Charles Martel

Graph Decomposition is NP-Complete: A Complete Proof of Holyer's
Dorit Dor and Michael Tarsi

The Fourth Moment Method
Bonnie Berger

Testing Shared Memories
Phillip B. Gibbons and Ephraim Korach

Polynomial Methods for Separable Convex Optimization in Unimodular
Linear Spaces with Applications
Alexander V. Karzanov and S. Thomas McCormick


End of NA Digest