## NA Digest Sunday, February 14, 1999 Volume 99 : Issue 07

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: Bruce Hendrickson <bahendr@trocious.cs.sandia.gov>
Date: Wed, 10 Feb 99 15:12:37 MST
Subject: Geometric Intersection Problem

While working on the coupling of Lagrangian and Eulerian codes,
I have come across the following simple geometric problem.
This kernel is consuming a significant amount of time for the
large problems of interest. I would be interested to hear
from others who may have addressed the same problem or anyone
who has a suggestion for an efficient approach.

Given:
A tetrahedron and a rectilinear 3D grid.
Find:
The volume of intersection of the tet with each grid cell.

I'm aware that the general problem of intersecting convex polyhedra
is well studied. I'm looking for an approach that is fast in practice
for this special instance.

- Bruce

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

From: Daoqi Yang <yang@math.wayne.edu>
Date: Thu, 11 Feb 1999 09:31:35 -0500 (EST)
Subject: Strong Discontinuities

Dear NA Netters:

I have some questions for people who are dealing with
problems with strongly discontinuous coefficients.

My implementation of finite element methods show that
the solution gets less accurate when the coefficients
become more discontinuous. When the jump in discontinuity
increases from 10 to 10^{5} (10 to the power of 5), the
accuracy decreases from 10^{-3} to about 1. My first
question is that anyone has implemented FEM (or FDM, FVM etc)
carefully such that the accuracy does not deteriorate
even if the jump is 10^{50} or 10^{100}?

I recently implemented FEM with Domain Decomposition and found
that certain domain decomposition methods do not lose accuracy
when the jump increases from 10 to 10^{100} and the conditioning
of the discrete problem gets much better when the jump
becomes large. For example, when the jump is 10^{100}, one iteration
will be enough to get accuracy of order 10^{-3}.
My second question is that anyone has had similar observations?

My third question is that anyone has had encountered real world
problems or know some references that have coefficient
jumps in the order 10^{10}, 10^{50} or 10^{100}?

Thanks. I appreciate your suggestions and references.

Daoqi Yang
yang@math.wayne.edu

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

From: Yair Censor <yair@mathcs2.haifa.ac.il>
Date: Fri, 12 Feb 1999 11:21:41 +0200 (IST)
Subject: Errata, Parallel Optimization by Censor and Zenios.

Errare humanum est...

We proudly announce that our tireless efforts to publish the "first ever
published book without errors" have failed....

For the benefit of the readers of our book (see full publication details
below) we installed an Errata list at the following internet site:
http://www.ucy.ac.cy/ucy/pba/zenios/bookinfo.html

We would be grateful to anyone bringing to our attention further errors,
typos, or omissions of credits and references. We will gladly post those
on the page.

Yair Censor and Stavros Zenios

P.S. If you wish a .ps file of the Errata list e-mailed to you please send
a request to: yair@mathcs2.haifa.ac.il

PARALLEL OPTIMIZATION : THEORY, ALGORITHMS, AND APPLICATIONS
By Yair Censor and Stavros A. Zenios,
A volume in the series: "Numerical Mathematics and Scientific Computation",
Oxford University Press, New York, 1997.
Hardcover, 576 pages, ISBN 0-19-510062-X.
For information on the book, including Title Page, Foreword, Preface,
Organization of the Book, Suggested Course Outlines, Acknowledgements,
and Ordering Instructions, please visit the publisher on the internet at
http://www.oup-usa.org/gcdocs/gc_019510062X.html

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

From: Cristian Oara ER <oara@df.op.dlr.de>
Date: Tue, 09 Feb 1999 12:30:22 +0100
Subject: New Book, Generalized Riccati Theory and Robust Control

NEW BOOK ANNOUNCEMENT
GENERALIZED RICCATI THEORY AND ROBUST CONTROL
A Popov Function Approach
by
Vlad IONESCU, Cristian OARA, Martin WEISS

(410 pages, hard-cover, ISBN: 0-471-97147-2, 60,- GBP)
John Wiley and Sons
1999

FROM THE BACKCOVER:

This comprehensive monograph is unique in providing a generalization
of the algebraic Riccati theory to the indefinite sign and singular
cases from the perspective of the Popov function approach. The whole
development is oriented towards applications in robust systems theory.

MAIN FEATURES:

** All the results are treated in parallel, in continuous and discrete-time
settings, and at the same level of generality and in-depth details
** Necessary and sufficient existence conditions and computable formulas
for Riccati equations and systems under the most general assumptions available
so far in the literature
** Characterizations of the solutions to the Riccati equation by using various
representations of the underlying Hamiltonian system: time domain (input-output
operator), frequency domain (transfer function matrix), and state-space
(transmission matrix pencil)
** Extension of the celebrated Popov positivity theory to the game theoretic
cases
** Unified solutions and extensions of bounded real lemma, small gain theorem,
normalized coprime factorization, spectral and inner-outer factorization,
Redheffer theorem, linear quadratic control
** A signature condition based approach which unifies the solutions to various
mathematical problems from robust control theory, including the maxmin, Nehari,
H-2 and H-infinity control, and robust stabilization problems
** The most general formulas available so far for the suboptimal H-infinity
controller (continuous and discrete-time)
** Optimal solutions to the robust stabilization problems under normalized
coprime factors uncertainties
** Particular attention paid to the peculiarities of discrete-time systems
such as to eliminate the unnatural restrictive assumptions which are usually
present in the literature and which rule out the ``interesting'' cases
** Self-contained monograph, giving all necessary basic results used in the
book

Graduate students in control engineering and those working on the
mathematics of system and control theory will find this book essential
reading. For researchers, this is an indispensable work, when seeking
ever more robust control laws which take into account uncertainties.

CONTENTS:

Preface

Part I. General Matrix Theory and Linear Dynamical Systems

Chapter 1. Matrices, Matrix Pencils, and Rational Matrix Functions.
Chapter 2. Linear Dynamical Systems.

Part II Generalized Riccati Theory

Chapter 3. Popov Triplets.
Chapter 4. Riccati Theory: An Operator--based Approach.
Chapter 5. Riccati Equations and Matrix Pencils: the Regular Case.
Chapter 6. Riccati Systems and Matrix Pencils: the General Case.

Part III Applications to Systems Theory and Robust Control

Chapter 7. Applications to Systems Theory.
Chapter 8. The Four Block Nehari Problem.
Chapter 9. The Optimal H-2 Control Problem.
Chapter 10. The H-Infinity Control Problem
Chapter 11. Robust Stabilization.
References
Index

ORDERING INFORMATION:
Internet: http://www.wiley.com/ordering
http://catalog.wiley.com
Email: customer@wiley.co.uk
Fax: + 44 (0) 1243 770460
Phone: 0800 243407 (UK Free Phone)/
+ 44 1243 843206 (Other countries)
Post: Sarah Hinton, John Wiley and Sons Ltd, Baffins Lane,
Chichester, West Sussex, PO19 1UD, UK

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

From: Lev Krukier <krukier@npt.rnd.runnet.ru>
Date: Mon, 8 Feb 1999 18:06:26 +0300
Subject: Conference on Environmental Mathematical Modeling

SECOND ANNOUNCEMENT and CALL for PAPERS
International Conference on
Environmental Mathematical Modeling
and Numerical Analysis
(EMMNA'99)
May 24 - 31, 1999, Rostov-on-Don, RUSSIA

Institute for Mathematical Modeling, Russian Academy of Science and Rostov
State University are organizing International Conference on Environmental
Mathematical Modeling and Numerical Analysis (EMMNA'99)

The scope of the Conference includes but is not limited to the following
topics:

1. Mathematical modeling of hydrophysical processes in a water basin and
transport of contaminants
2. Groundwater pollution
3. Mathematical modeling of hydrophysical processes in air and air pollution
4. Numerical methods for advection-diffusion equation
5. Numerical methods for the solution of unsteady PDE problems
6. Iterative methods and preconditioners in environmental modeling
7. Efficient time stepping for stiff ODEs in environmental modeling
8. High performance computing in environmental modeling

Chairman: A.A. Samarskii (Russia)

INVITED SPEAKERS (confirmed): Richard Ewing, USA ; A. Hasanov, Turkey ;
A.N. Konovalov, Russia; V.V. Penenko, Russia; V.V. Shaidurov, Russia; A.
Sydow, Germany; P.N. Vabishchevich, Russia; I.I. Vorovich, Russia;
Junping Wang, USA; Ruediger Weiss, Germany; Z. Zlatev, Denmark.

Submission of abstracts: March 15, 1999
Notification of acceptance of abstracts: March 30, 1999
Notification of acceptance of papers: June 30, 1999
Registration and submission of accommodation forms: May 1, 1999

For further information contact us by e-mail : emmna99@uic.rnd.runnet.ru
or check our WWW-page: http://www.uic.rnd.runnet.ru/emmna99/

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

From: Wenyu Sun <wysun@pine.njnu.edu.cn>
Date: Tue, 9 Feb 1999 11:42:17 +0800
Subject: Nanjing Conference on Optimization and Numerical Algebra

International Conference on Optimization and Numerical Algebra
Nanjing Normal University, Nanjing, China
September 27-30, 1999
FIRST ANNOUNCEMENT
Nanjing is a big, beautiful and historical city with about thirty universities
and colleages. The International Conference aims to review and discuss recent
advances and promising research trends in some areas of Optimization and Nume-
rical Algebra. The topics include: Linear Programming and Nonlinear Program-
ming, Convex Programming and Nonconvex Programming, Nonsmooth Optimization,
Global Optimization, Stochastic Programming, Multiobjective Optimization,
Variational Inequalities, Linear and Nonlinear Systems of Equations, Least-
Squares Problems, Computation of Eigenvalue Problems, Matrix Computation and
Generalized Inverses, Applications of Optimization and Numerical Algebra.

Invited speakers include:
O. Axelsson (Nijmegen University, The Netherlands)
J.R. Birge (University of Michigan, USA)
T.F. Coleman (Cornell University, USA)
D.Y. Cai (Tsinghua University, Beijing)
S.C. Fang (North Carolina State University, USA)
M. Ferris (University of Wisconsin, USA)
M. Fukushima (Kyoto University, Japan)
W. Gander (ETH, Switzerland)
J.Y. Han (Chinese Academy of Sciences, Beijing)
E.X. Jiang (Fudan University, Shanghai, Shanghai)
P. Kall (University of Zurich, Switzerland)
W.W. Lin (National Tsinghua University, Taiwan)
W. Niethammer (Karlsruhe University, Germany)
L. Qi (University of New South Wales, Australia)
D. Ralph (University of Melbourne, Australia)
Z. Shen (Nanjing University, Nanjing)
E. Spedicato (University of Bergamo, Italy)
J. Sun (National University of Singapore, Singapore)
J.G. Sun (University of Umea, Sweden)
K. Tanabe (The Institute of Statistical Mathematics, Japan)
Ph.L. Toint (University of Namur, Belgium)
K.L. Teo (The Hong Kong Polytechnic University, Hong Kong)
J.-Ph. Vial (University of Geneva, Switzerland)
A.J. Wathen (Oxford University, UK)
T. Yamamoto (Ehime University, Japan)
Y. Yuan (Chinese Academy of Sciences, Beijing)
J. Zhang (City University of Hong Kong, Hong Kong)
X. Zhang (Chinese Academy of Sciences, Beijing)
Y. Zhang (Rice University, USA)

CALL FOR PAPERS: Titles and abstracts of invited and contributed papers must
be received by July 10, 1999. The abstracts should be typed in Latex, not ex-
ceed one page, and be sent to niqfs@dns.nuaa.edu.cn or wysun@pine.njnu.edu.cn
by e-mail. There is also 1999 Nanjing Award in Optimization and Numerical
Algebra for Young Researcher.
Conference proceedings, special issues of some journals, tours and
accommodations arrangements will be indicated in the Second Announcement.
Further information can be checked in WWW websites of
http://www.cc.ac.cn, http://www.njnu.edu.cn, http://www.nuaa.edu.cn,
http://www.pax.st.usm.edu/~jding/meeting.html or be obtained by contacting conference organizers:
Z.Bai (Chinese Academy of Sciences, bzz@lsec.cc.ac.cn),
L.Qi (University of New South Wales, Sydney, l.qi@unsw.edu.au),
W.Sun (Nanjing Normal University, wysun@pine.njnu.edu.cn).

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

From: David Keyes <keyes@icase.edu>
Date: Tue, 9 Feb 1999 16:58:26 -0500 (EST)
Subject: Parallel CFD Conference

Second Announcement
PARALLEL CFD'99
May 23-26, 1999
Williamsburg, Virginia, USA
http://www.parcfd.org

Parallel CFD'99, the eleventh in an international series of meetings
featuring computational fluid dynamics research on parallel computers,
will be held May 23-26, 1999 in Williamsburg, Virginia.

The technical program features invited speakers, three tracks of
contributed presentations, a poster session, and will be preceded for the
convenience of conferees by an optional half-day tutorial introducing the
PETSc parallel software system, including case studies from fluid
dynamics. Early experiences on teraflops-class ASCI computers and a
report on the ASCI alliance program will also be featured at the meeting.

All realms of phenomena in which fluid dynamical simulations play a
critical role --- from the microscopic scales of high tech manufacturing,
through human and planetary scales, and even to galactic scales --- and
all types of parallel models and architectures are featured. However, it
is the intersection of CFD and parallelism that is of interest, not the
union. Contributions generally focus on implementations and results.
Novel tools and methods supporting the software/hardware environment of
parallel CFD are also of interest.

Special emphases of the 1999 meeting will be parallel methods in
optimization, validation of parallel codes, non-PDE-based formulations of
CFD, and the influence of deep memory hierarchies and high interprocessor
latencies on the design of algorithms and data structures for CFD
applications. Abstracts targeting these themes are especially welcome.

The following international experts from industry, government, and
academia have agreed to make plenary presentations at the conference:

Antoine Desideri, INRIA, France
Harald Echtle, Daimler-Chrysler, Germany
Paul Fischer, Argonne National Lab, USA
Marc Garbey, University of Lyon, France
Trond Kvamsdahl, SINTEF, Norway
Dimitri Mavriplis, ICASE, USA
Kevin McManus, University of Greenwich, UK
John Salmon, California Institute of Technology, USA
John Shadid, Sandia National Lab, Albuquerque, USA
James Taft, NASA Ames Research Center, USA
Shinichi Kawai, National Space Development, Japan

The deadline for contributed abstracts has been extended to February 16,
1999. Authors will be notified of acceptance on or about March 12, 1999.
A refereed conference proceedings will be published by North Holland and
selected papers will appear in a special issue of the Int. J. of CFD.

The early registration deadline is April 15, 1999. For details on
conference venue, abstract preparation and submission, and registration,
please see the conference website, www.parcfd.org.

Local Organizing Committee of Parallel CFD'99:

D. E. Keyes, Old Dominion University and ICASE (Chair)
M. D. Salas, ICASE E. N. Todd, ICASE
V. N. Vatsa, NASA Langley R. G. Voigt, College of William & Mary

Scientific and Organizing Committee of the Parallel CFD Conferences:

R. K. Agarwal, Wichita State U. B. Chetveruskin, Russian Acad. Sci.
A. Ecer, UIPUI D. R Emerson, CLRC
P. Fox, IUPUI M. Garbey, U. of Lyon
A. Geiger, RUS C. Jenssen, SINTEF
D. Keyes, Old Dominion U. C. A. Lin, Tsing-Hua U.
I. Lopez, NASA Lewis D. McCarthy, Boeing
J. McDonough, U. of Kentucky R. Pelz, Rutgers
J. Periaux, Dassault Aviation N. Satofuka, Kyoto Inst. of Tech.
P. Schiano, CIRA A. Sugavanam, IBM
M. E. S. Vogels, NLR D. Weaver, Phillips Laboratory

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

From: Jack Dongarra <dongarra@cs.utk.edu>
Date: Thu, 11 Feb 1999 11:40:23 -0500
Subject: BLAS Technical Forum

BLAS TECHNICAL FORUM, March 16-18, 1998, Oak Ridge, TN

We are planning a BLAS Technical Forum meeting on March 16-18, 1999.
The meeting will be hosted by the University of Tennessee and will be
held at the Ramada Inn & Suites in Oak Ridge, TN:

Ramada Inn & Suites
420 S. Illinois Avenue
Oak Ridge, TN 37830
(423) 483-4371

The Forum has been established to consider expanding the Basic Linear
Algebra Subprograms (BLAS) in a number of directions in the light of
modern software, language, and hardware developments. The first meeting
of the Forum was held in Nashville on February 19-20, 1996.

We strongly urge and encourage attendance at the meeting
so that we can make tangible progress towards much needed standards.
Wide input is needed to help ensure that emerging proposals are useful
and acceptable to the community.

It is appreciated that it is not easy for everyone to attend the
meetings of the Forum, but we would nevertheless welcome your input
since we wish the discussion to be as open as possible, and the results
to reflect consensus from the community at large.

The tentative schedule for the meeting will be:
9am-5pm: Tues, March 16
9am-5pm: Wed, March 17
9am-1pm: Thurs, March 18

Morning and afternoon snacks will be provided, but you're on-your-own
for lunches.

A rough agenda for the meeting follows:

* Second/third reading of the following chapters:

Chapter 1: Introduction
Chapter 2: Dense and Banded BLAS
Chapter 3: Sparse BLAS
Chapter 4: Extended and Mixed Precision BLAS
Chapter 5: Interval BLAS

* Discussion of model implementations and test suites.

Please send us items you would like to see on the agenda
or that you would like to discuss.

http://www.netlib.org/blas/blast-forum/blast-forum.html

Jack Dongarra and Sven Hammarling

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

From: Karsten Decker <decker@cscs.ch>
Date: Sun, 14 Feb 1999 08:07:40 +0100
Subject: SPEEDUP Workshop

SECOND ANNOUNCEMENT AND PRELIMINARY PROGRAM
25th SPEEDUP Workshop
Trends in High-Performance Computing:
Systems, Services, and User Requirements

Switzerland
March 25 - 26, 1999

Summary

High-performance computing (HPC) has become an important tool in science
and engineering. Over the last ten years new methods, new hardware, and
new software have stretched the bounds of what can be done. However,
the development did not always follow straight lines. What can we
expect in the near future and how should we direct our efforts?

This workshop will address the current trends in HPC systems, but also
raise the question whether these trends are going to persist in the
future. Will there be new types of computer systems emerging? What
about HPC center services? Which products and services should be
offered to respond to end-user requirements?

The 25th SPEEDUP Workshop is intended to bring together members of
academic institutions, business enterprises, and government who are
interested to learn about trends in hardware architecture, software
infrastructure, libraries and tools, HPC center products and services,
user requirements, and performance evaluation.

Topics

Topics will include, but will not be limited to the discussion of trends in

Hardware architecture
Software infrastructure
Libraries and tools
Products and services
End-user requirements

Preliminary program

M.H. Gutknecht, ETH Zurich
W.J. Camp, Sandia National Laboratory, Albuquerque, USA
R.C. Whaley, University of Tennessee, Knoxville, USA
J. Towns, National Computational Science Alliance, USA
E. Benz, ABB Corporate Research, Baden, Switzerland
W. Schmidt, DaimlerChrysler AG, Stuttgart, Germany
A. Geiger, RUS, Stuttgart, Germany
U. Harms, Harms Supercomputing Consulting, Munich, Germany
K.M. Decker, President SPEEDUP Society, CSCS
R. Gruber, EPF Lausanne, Lausanne, Switzerland
A. Gunzinger, ETH Zurich, Zurich, Switzerland
P. Kuonen, EPF Lausanne, Lausanne, Switzerland
H.-P. Wessels, CSCS, Manno, Switzerland

General information and registration

The latest information on the program can be found at URL

http://www.speedup.ch/Workshops/Workshop25Ann.html

On-line registration is available at URL

http://davos.cscs.ch:8080/Workshop25Reg.html.

Since the number of participants is limited, we urge you to register
promptly to ensure that a space is reserved in your name. In any case,
please register until March 15, 1998 at the latest.

Further information on the SPEEDUP Society can be obtained from URL

http://www.speedup.ch

Martin H. Gutknecht
General Workshop Chair
ETH Zentrum, RZ
CH-8092 Zurich
Tel: +41 (1) 632 3464
Fax: +41 (1) 632 1104
Email: mhg@scsc.ethz.ch

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

From: Mike Boucher <Mike.Boucher@Central.Sun.COM>
Date: Fri, 12 Feb 1999 08:51:35 -0700 (MST)
Subject: Software Quality Position at Sun Microsystems

The High Performance Libraries and Tools Group (HPG) at Sun is looking
for a numerical analyst or applied mathematician to fill an opening for
a Software Quality Engineer (SQE). The SQE's responsibilities are to:

o Develop tests of parallel numerical libraries for the SPARC and
x86 platforms
o Interact with developers to ensure that software is developed
in a way that facilitates testability and verifiability
o Prove (or disprove) the validity of proposed new algorithms and
optimizations to existing algorithms
o Help developers analyze test failures

Experience with LAPACK is a plus, and candidates with software testing
experience are preferred.

The SQE will be an integral part of every phase of the product life cycle
by participating in initial design reviews, proposing changes to aid
testability and verifiability, validating the libraries on new hardware
and software configurations, and participating in acceptance and shipping
decisions on the final product.

Please send a resume for prompt consideration to Diane.Boucher@Sun.com
or mail to:

Diane Boucher
Manager, HPG
303B South Technology Court
UBRM11-202

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

From: Richard Brualdi <brualdi@math.wisc.edu>
Date: Fri, 12 Feb 1999 13:07:57 -0600 (CST)
Subject: Pre- and Postdoctoral Positions at University of Wisconsin

VIGRE
Vertical Integration of Research and Education at
The University of Wisconsin - Madison

The Department of Mathematics at the University of
Wisconsin-Madison anticipates receiving a VIGRE award from
the National Science Foundation. Such an award would
provide support for

(1) up to thirteen new predoctoral fellowships, and

(2) up to three new postdoctoral fellowships.

These new fellowships are restricted to U.S. citizens and
permanent residents. It is expected that postdoctoral
fellowship recipients will have received their PhD since
1997. Additional awards are anticipated in future years.

The predoctoral fellowships are intended for graduate students
in the graduate program in Mathematics and in Applied
Mathematical Sciences, and are expected to provide two years of
fellowship support (including two years of summer support) and
three years of teaching assistantship support, contingent on
satisfactory progress. Full tuition remission and health
insurance is included.

The postdoctoral fellowships are three year fellowships,
including both academic year and summer support. The
teaching load is one course per semester, with no summer
teaching. Postdoctoral fellows will be mentored by a
faculty member and will participate in a variety of
instructional and research activities involving
Supplementary money is available each year for travel,
equipment, and supplies. Qualified applicants who have
already applied this year for a Van Vleck (Visiting)
Assistant Professorship or other position will
automatically be considered for a "VIGRE Van Vleck."

Applicants who have not already applied for a "Van Vleck"
should send an AMS Application Cover Sheet and a CV, and
have at least three letters of recommendation sent to:

Hiring Committee (VIGRE)
Department of Mathematics
480 Lincoln Drive
University of Wisconsin

Applications will continue to be considered until all
positions are filled.

Areas of focus for "VIGRE Van Vlecks" are:

Discrete Mathematics: cryptography, codes, combinatorial
analysis and probability, splines, wavelets, and related
areas;

Physical Applied Mathematics: continuum and statistical
mechanics, rheological flows, turbulence, and related
areas;

Dynamical Systems and Partial Differential Equations:
variational methods for ODE and PDE, hyperbolic systems of
conservation laws, viscosity methods for degenerate
parabolic PDEs, and related areas;

Stochastic Analysis: stochastic equations, mathematics of
finance, probability theory, and related areas.

The University of Wisconsin is an Affirmative Action, Equal
Opportunity Employer, and encourages applications from
women and minorities. Unless confidentiality is requested
in writing, information regarding the applicant must be
released upon request. Finalists cannot be guaranteed
confidentiality.

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

From: Kerry Anne Landman <k.landman@ms.unimelb.edu.au>
Date: Sat, 13 Feb 1999 11:57:31 +1000
Subject: Research Fellow at University of Melbourne

University of Melbourne Department of Mathematics and Statistics
Research Fellow
Industrial Applied Mathematics

The position: A research position is available to work on the ARC
industry partner grant on mathematical modelling of the cooking process of
cereal grains. This is a focussed collaborative project which requires
research outcomes to be achieved according to a research plan.

The person: You should have a PhD in applied mathematics, physics,
chemical engineering, mechanical engineering or their equivalent. You
will have a high level of mathematical modelling skills, with an ability
to understand physical processes and translate these into a mathematical
model. You will have a high level of analytic and numerical computing
skills. Good written and oral skills are essential, as well as the
ability to work in a team.

Contact: Associate Professor Kerry Landman, Telephone: 9344 6762, Fax:
9344 4599, Email: k.landman@ms.unimelb.edu.au, for further information and
a position description.

The benefits: The position has a salary of \$43042 - \$46203.

Employment type: The position is a fixed-term (research) position for 2 years.

Applications close: 31 March 1999

Position number: RY 0007541

Applications to: Director, Human Resources, The University of Melbourne,
Parkville, Victoria, 3052; fax +61 3 9344 4694. Quote the position number
and include the names, phone and facsimile numbers and E-mail addresses of
three referees in your application.

The University of Melbourne is an equal opportunity employer.

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

From: Jan Korvink <korvink@informatik.uni-freiburg.de>
Date: Tue, 09 Feb 1999 16:11:38 +0000
Subject: Graduate Student Positions at Albert Ludwig University

Graduate study opportunities in

Microsystem Modelling and Simulation

at the Institute for Microsystem Technology, Faculty of Applied Science,
Albert Ludwig University Freiburg, Germany.

Aim of the project, funded by the German DFG, is the development of new
modelling tools for microsystems (MEMS). We are seeking candidates with
programming skills in C++; a background in numerical analysis would be
beneficial. A master or diploma in computational science, computer science,
or related fields is required. The positions are open for three years.

The salary is regulated by the German BAT contract: BAT IIa or less
depending on qualifications.

Applications will be accepted until the positions are filled. Interested
applicants should send a resume and a cover letter stating research
interests as soon as possible to

Prof. Jan G. Korvink
Chair for Microsystem Simulation
IMTEK - Institute for Microsystem Technology
University of Freiburg
Am Flughafen 17
D-79110 Freiburg Tel. (+49 761) 203 8040
Germany Fax. (+49 761) 203 8042

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

From: Margot Gerritsen <m.gerritsen@auckland.ac.nz>
Date: Wed, 10 Feb 1999 15:19:36 +1300
Subject: Research Position at University of Auckland

PhD fellowship for TIDAL FLOW MODELLING in the Department of Engineering
Science, School of Engineering at the University of Auckland, New Zealand

Recently our research group received a Marsden Fund grant (similar to the
NSF grants in the US) to develop improved computational methods for efficient
and accurate computer modelling of tidal flows, and the associated transport
of chemicals and particles, in coastal regions. The computer model will be
used to study a wide variety of problems such

as the effects of sewage disposal, the effects of the marine dumping of
sediments and the movement of fish larvae. The model will be based on
composite grid techniques in combination with grid refinement strategies.
The project is conducted in collaboration with the National Institute
for Water and Atmospheric Research in New Zealand.

Besides Professor Mike O'Sullivan and myself, the project team consists
of two PhD students and one postgraduate student. We are looking for a third
PhD student. We offer a stipend and tuition support.

Auckland is a beautiful city surrounded by water, with a climate similar
to that of the bay area in California (although it rains a bit more),
and offers many opportunities for outdoor activities. It also has a vibrant
cultural scene.

For more information contact me at m.gerritsen@auckland.ac.nz

Margot Gerritsen
Lecturer Engineering Science
University of Auckland

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

From: Abhijit Bose <abose@engin.umich.edu>
Date: Wed, 10 Feb 1999 14:58:11 -0500
Subject: Scientific Computing Position at University of Michigan

The Center for Parallel Computing at the University of Michigan
is seeking to hire a person with background in scientific computing
and parallel algorithms for a user support and consulting position.
http://www.engin.umich.edu/labs/cpc .

Desired Qualifications

Ph.D. or equivalent combination of education and experience in
high-performance parallel computing, knowledge of the following:
parallel algorithms, MPI, OpenMP, PVM, AFS, Unix, SQL,
numerical and statistical libraries

Minimum Qualifications

M.S. Engineering or Computer Science or equivalent combination of
education and experience in parallel code development, algorithm
decomposition for parallelizing serial codes using MPI/PVM,
programming skills in Fortran, HPF, C, HTML; strong oral/written
skills and desire to help users; ability to teach others and work
within a team environment.

Duties

Provide scientific programming and consulting services to
Center for Parallel Computing (CPC) and University of Michigan
users; provide consulting services for national NPACI users
via email/Remedy system and occasional phone calls, create
training materials and conduct training sessions on parallel
programming using supercomputers such as IBM SP2, SGI Origin,
Cray T3E/T90; install, maintain and support application package
used on CPC supercomputers; maintain, develop and update CPC
web pages.

Resumes and questions with regard to this position can be sent to:
abose@engin.umich.edu (preferred) or via US mail:

Dr. Abhijit Bose
The University of Michigan
2356 Media Union
2281 Bonisteel Blvd.
Ann Arbor, MI 48109-2094
Phone: (734) 615-1490
Fax: (734) 936-3107

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

From: Hans Schneider <hans@math.wisc.edu>
Date: Tue, 9 Feb 1999 09:07:29 -0600 (CST)
Subject: Contents, Linear Algebra and Its Applications

URL: http://www.elsevier.nl/locate/jnlnr/07738
Journal: Linear Algebra and Its Applications
ISSN : 0024-3795
Volume : 289
Issue : 1-3
Date : 01-Mar-1999

Seventh Special Issue on Linear Algebra and Statistics
Special Editors: S.PUNTANEN, G.P.H.STYAN, H.J.WERNER

pp 1-2
Preface

pp 3-24
Improved biased estimation in an ANOVA model
SE Ahmed

pp 25-39
Mimimax adjustment technique and fuzzy information
BF Arnold

pp 41-54
Monte Carlo estimates of the log determinant of large sparse matrices
RP Barry

pp 55-74
Linear restrictions, rank reduction, and biased estimation in linear
regression
JS Chipman

pp 75-94
Spherical functions on the Grassmann manifold and generalized Jacobi
polynominals-Part 1
AW Davis

pp 95-119
Spherical functions on the Grassmann manifold and generalized Jacobi
polynominals-Part 2
AW Davis

pp 121-126
A class of statistical estimators related to principal components
RW Farebrother

pp 127-130
Solution to a rank equation
J Gross

pp 131-134
Explicit solutions to the matrix inverse problem AX=B
J Gross

pp 135-139
Idempotency of the Hermitian part of a complex matrix
J Grob

pp 141-150
On the product of orthogonal projectors
J Grob

pp 151-160
A note on the rank-subtractivity ordering
J Gross

pp 161-168
Partial orderings, preordenings, and the polar decomposition of
matrices
J Gross

pp 169-182
On semi-orthogonality and a special class of matrices
J Gross

pp 183-201
Frechet distance as a tool for diagnosing multivariate data
ALIS Hadi, H Nyquist

pp 203-224
Use of the Gibbs sampler to invert large, possible sparse, positive
definite matrices
DA Harville

pp 225-242
Changes in the general linear model: a unified approach

pp 243-259
The Marcus-de Oliveira conjecture, bilinear forms, and cones
A Kovacec

pp 261-266
A note on the parameter set for factor analysis models
WP Krijnen

pp 267-277
Matrix results on the Khatri-Rao and Tracy-Singh products
S Liu

pp 279-284
Simultaneous polar decomposition of rectangular complex matrices
A Markiewicz

pp 285-296
Softly unbiased estimation part 1: the Gauss-Markov model
B Schaffrin

pp 297-310
On oblique projectors
Y Takane

pp 311-318
Some new results on correlation-preserving factor scores prediction
methods
JMF Ten Berge

pp 319-332
The perturbed nonhomogeneous Markov systems
PCG Vassiliou

pp 333-342
Two-way selection of covariables in multivariate growth curve models
SG Wangi, T Nummi

pp 343-344
Author index to volume 289

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

End of NA Digest

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