Send mail to:             for the digests or bakeoff
            for comments or help
 Current editor:  Craig Douglas       
Anonymous ftp repository: (

World Wide Web: or

Today's editor:  Craig Douglas (

Volume 8, Number 12 (approximately December 31, 1998)

Today's topics:

     Important Date
     New Mirror Site
     Virtual Proceedings Started
     Preprint (John and Tobiska)
     2 Papers Related to Bonn Workshop (Tai et al)
     Bonn Paper by Brandt
     Bonn Paper by Reitzinger
     EPSICODE '99
     Copper Mountain Conference Special Themes
     Call for Papers (High Performance Scientific Computing)
     Conference in honor of Richard Varga
     Job opportunity in Linz
     Postdoctoral Opportunity (Southern Mississippi)
     Tenure Track Position in Applied Mathematics (Wyoming)


Date: Wed, 31 Dec 1998 12:19:99 -0500 (EST)
From: Craig Douglas 
Subject: Important Date

February 1, 1999 is the deadline for

    Author abstracts
    Early registration

for the Ninth Copper Mountain Comference on Multigrid Methods.  See

for more details.


Date: Wed, 16 Dec 1998 01:22:56 -0500 (EST)
From: Craig Douglas 
Subject: New Mirror Site

The National Center for High-Performance Computing, Hsinchu, Taiwan (ROC),
is now officially mirroring MGNet.  You will find links to their pages
in the MGNet web pages.  The web address is

This is courtesy of Dr. Daniel Lee (


Date: Wed, 31 Dec 1998 13:14:15 -0500 (EST)
From: Craig Douglas 
Subject: Virtual Proceedings Started

There are two new virtual proceedings that have been started.  The first one
is for the Tenth GAMM Workshop on Parallel Multigrid Methods (ParMGM98).  It
can be found at

(or mgnet/Conferences/ParMGM98 using anonymous ftp).

The second one is for the Ninth Copper Mountain Conference on Multigrid
Methods (CMCMM99).  It can be found at

(or mgnet/Conferences/CopperMtn99 using anonymous ftp).

Both proceedings currently have four papers in them, but will grow during


Date: Thu, 5 Nov 1998 12:54:34 +0100 (MET)
From: Volker John 
Subject: Preprint (John and Tobiska)

Smoothers in Coupled Multigrid Methods for the Stokes and Navier-Stokes

Volker John and Lutz Tobiska
Otto-von-Guericke-Universitaet Magdeburg, Germany


Coupled multigrid methods have been proven as efficient solvers for the
incompressible Navier-Stokes equations in recent benchmark computations.  This
paper presents a numerical study of two classes of smoothers in these methods.
The class of Vanka-type smoothers is characterized by the solution of small
local linear systems of equations in a Gauss-Seidel manner in each smoothing
step whereas the Brass-Sarazin-type smoothers solve a large global saddle
point problem.  The behaviour of these smoothers with respect to computing
times and parallel overhead is studied on 2d DFG benchmark problems of flows
around a cylinder.

Keywords.  Incompressible Navier-Stokes equations, parallel coupled multigrid
methods, Vanka-type smoothers, Braess-Sarazin-type smoothers.

    Editor's Note: This can be found through
    -------------  or mgnet/papers/John-Tobiska/


Date: Mon, 7 Dec 98 9:37:34 MET
From: Tai Xue-Cheng 
Subject: 2 Papers Related to Bonn Workshop (Tai et al)

Subspace Correction Methods for Convex Optimization Problems

Xue-Cheng Tai
Mathematics Institue, University of Bergen, Norway

Jinchao Xu
Department of Mathematics, Pennsylvania State University, USA


Domain decomposition and multigrid methodshave been intensively studied for
linear partial differential equations.  Recent research reveals that domain
decomposition and multigrid methods can be analysed using a same framework.
The present work uses this framework to analyse the convergence of two
algorithms for convex optimization problems.  Our emphasis is on nonlinear
problems instead of linear problems.  The algorithms reduce to the standard
additive and multiplicative Schwarz methods when used for linear partial
differential equations.

Researches for domain decomposition and multigrid methods have been mostly
concentrating on linear elliptic and parabolic partial differential equations.
Extension to more difficult problems have been considered by some recent
works.In this work,a general nonlinear convex minimization problems is
considered.  The proposed algorithms can be used for nonlinear partial
differential equations, optimal control problems related to partial
differential equations and eigenvalue problems.  The space decomposition can
be a domain decomposition method, a multigrid method or some other
decomposition techniques.

Domain decomposition methods and multigrid methods have been studied for
nonlinear partial differential equations by some earlier works.  In comparison
with the existing works, our approach has several features.  For example, the
proposed algorithms canbe used for certain degenerated or singular nonlinear
diffusion problems, i.e., the nonlinear diffusion coefflcient can be zero or
infinity and our approach do not need extra assumption on the smoothness of
the solutions.  The methods work for natural domain decomposition and
multigrid meshes.  Moreover, only small size nonlinear problems need to be
solved on the decomposed subspaces.  We also emphasis that our approach is
valid for general space decomposition techniques.  So the applications is not
restricted to domain decomposition and multigrid methods.  Other space
decomposition techniques can also be considered.  The two algorithms given in
this work were first proposed elsewhere, where the qualitative convergence of
the algorithms was proved, but the uniform rate of convergence was not given

    Editor's Note: This can be found through
    -------------  or mgnet/Conferences/ParMGM98/Papers/

                                  * * * * *

An Asynchronous Space Decomposition Method

Xue-Cheng Tai
Mathematics Institue, University of Bergen, Norway

Paul Tseng
Department of Mathematics, University of Washington, Seattle, WA, USA

No abstract.

    Editor's Note: This can be found through
    -------------  or mgnet/Conferences/ParMGM98/Papers/

Dr Xue-Cheng Tai,                          Professor of Mathematics,
University of Bergen,                      Tel. 47 55 58 4868 (office) 
Department of Mathematics,                 Fax. 47 55 58 9672
Johannes Brunsgate 12,                     www: 
5007, Bergen, Norway. 


Date: Wed, 23 Dec 1998 11:54:28 +0200
From: Fliegelmann Sarah 
Subject: Bonn Paper by Brandt

Achieving Textbook Multigrid Efficiency (TME) in CFD

Achi Brandt
Weizmann Institute of Science, Israel


``Textbook multigrid efficiency'' means solving a discrete PDE problem in a
computational work which is only a small (less than 10) multiple of the
operation count in the discretized system of equations itself.  As a road map
for attaining this optimal performance for general CFD problems, we list in a
table every foreseen kind of computational difficulty for achieving that goal,
together with the possible ways for resolving that difficulty, their current
state of development, and references.

Included in the table are staggered and nonstaggered, conservative and
nonconservative discretizations of viscous and inviscid, incompressible and
compressible flows at various Mach numbers, as well as a simple (algebraic)
turbulence model and comments on chemically reacting flows.  The listing of
associated computational barriers involves:  non-alignment of streamlines or
sonic characteristics with the grids; recirculating flows; stagnation points;
discretization and relaxation on and near shocks and boundaries; far-field
artificial boundary conditions; small-scale singularities; large grid aspect
ratios; boundary layer resolution; and grid adaption.

    Editor's Note: This can be found through
    -------------  or mgnet/Conferences/ParMGM98/Papers/ and


Date: Wed, 23 Dec 1998 18:34:26 -0500 (EST)
From: Craig Douglas 
Subject: Bonn Paper by Reitzinger

Algebraic Multigrid and Element Preconditioning

Stefan Reitzinger
University of Linz, Austria


This paper presents a solution strategy for poor conditioned, large linear
systems with a sparse matrix arising from an FE-discretization.  Especially
robustness has to be gained, if anisotropic elements are used.  To achieve
this, the AMG method of Ruge/Stuben is used.  This algorithm is robust for
M-matrices, but unfortunately the `region of robustness' between s.p.d.
M-matrices and general s.p.d.  matrices is very fuzzy .  For this reason the
so called element preconditioning technique is introduced to obtain a spectral
equivalent M-matrix with respect to the original stiffness matrix.  AMG, with
the spectral equivalent M-matrix instead of the original stiffness matrix, is
then applied as preconditioner to the conjugate gradient method.

Numerical studies are done for a magnetic shielding, a sandwich and a boundary
layer problem in 2D(and 3D) which show the numerical robustness of the new
preconditioning method.

    Editor's Note: This can be found through
    -------------  or mgnet/Conferences/ParMGM98/Papers/


Date: Tue, 15 Dec 1998 19:33:53 +0100 (MET)
From: Volker John 
Subject: EPSICODE99 - Conference on Numerical Methods for Transport-Dominated 
         Problems, 1. Announcement

                EPSICODE '99

International Conference on Numerical Methods
for Transport-Dominated and Related Problems

This International Workshop on analytical and numerical methods for
convection-dominated differential equations will be organized in cooperation
with the GAMM Committees ``Efficient numerical methods for pde'' and
``Scientific computing''.  It will provide a forum for communication and
interaction between applied mathematicians, numerical analysts and the
scientific computing community.
Date:  20 -- 23 September 1999

Schloss (Castle) Wendgr\"aben, near Magdeburg, Germany

Conference themes:

- Discretization techniques : 
         FDM/FEM/FVM/spectral methods/collocation methods
- Solution of the discrete algebraic problems :
         multigrid/domain decomposition/parallelization
- Analytical approaches :
         asymptotic expansions/Shishkin-type decompositions
- Error estimates and adaptive approaches
- Modelling and applications

Invited speakers:
   V.F. Butuzov (Moscow), 
   C. Canuto (Torino), 
   P.W. Hemker (Amsterdam),
   R.B. Kellogg (Maryland), 
   G. Lube (G\"ottingen), 
   R. Rannacher (Heidelberg), 
   H.-G. Roos (Dresden), 
   G.I. Shishkin (Ekaterinburg),	
   E. S\"uli (Oxford), 
   R. Verf\"urth (Bochum), 
   P. Wesseling (Delft)

Organizing committee:
   Lutz Tobiska (Magdeburg), 
   Martin Stynes (Cork, Ireland),
   Lutz Angermann (Magdeburg), 
   Volker John (Magdeburg)

Conference proceedings:
The conference proceedings will be published as a special issue of
``Computing''.  Articles will be collected after the conference, not later
than 26 November 1999.	All submitted papers will be reviewed in the standard

Contributed papers:
Abstracts (in Tex or Latex) should be submitted by 28 May 1999 to our email
address below.  Presentations can be in lecture or poster format.  No parallel
sessions will be timetabled at the conference, so the number of contributed
lectures is limited.

Dates to remember:
- Deadline for abstracts and registration: 28 May 1999
- Deadline for submission of manuscripts: 26 November 1999
- Notification of acceptance or rejection:  28 January 2000
- Final deadline for submission of revised manuscript: 31 March 2000

The registration fee will be approximately 80 DM.  This includes the volume of
abstracts, coffee breaks and reception.  Please send the registration form to
the address below or register on-line at our Conference website; in both
cases, the deadline for registration is 28 May 1999.

The conference will take place in the historical castle of Wendgr\"aben, which
provides full-board lodging.  The price of a double/single room is about
140/165 DM per night.

- email:
- Conference website:


Date: Wed, 16 Dec 1998 08:57:34 -0800 (PST)
From: Van Henson 
Subject: Copper Mountain Conference Special Themes

Dear Colleagues,

The Ninth Copper Mountain Conference on Multigrid Methods will be held April
11-16, 1999 in Copper Mountain, Colorado.  The program committee wishes to
place special emphasis on the theme "General Scalable Multigrid Methods".  In
particular, we are hoping to gather researchers interested in "Algebraic
Algorithms" and "Parallel Techniques".  Both topics have become prominent in
large-scale applications, and it is our intent to bring together researchers
and practitioners to share their ideas, challenges, and knowledge, with the
hope of furthering the field.

To emphasize the theme, we will be grouping talks into special theme sessions
and organizing informal workshops during the week.  Also, Algebraic Algorithms
and Parallel Techniques will be discussed in a day-long tutorial on April 11
(the Sunday before the conference talks begin).

We encourage everyone interested in these areas to participate, by attending
and hopefully by submitting an abstract to one of the theme sessions.

The web page
contains detailed information about the conference, including information on
how to submit abstracts (Deadline Feb.  1 1999).

We hope to see you in April.


         Van Emden Henson, Theme Chair: "Algebraic Algorithms"
         Jim E. Jones, Theme Chair: "Parallel Techniques"

         Center for Applied Scientific Computing (CASC)
         Lawrence Livermore National Laboratory



Date: Mon, 28 Dec 1998 13:27:30 -0500 (EST)
From: Jun Zhang 
Subject: Call for Papers (High Performance Scientific Computing)

                     C A L L   F O R   P A P E R S

                          Special Session on

  *** High Performance Scientific Computation with Applications ***
The above special session is organized in "The 1999 International Conference
on Parallel and Distributed Processing Techniques and Applications
(PDPTA'99)", June 28 - July 1, 1999 Monte Carlo Resort, Las Vegas, Nevada,

Today's scientific and engineering problems demand computational power that is
far beyond that can be provided by conventional computers.  Large scale
numerical simulations require high performance computation.  Here the high
performance computation is meant in its broadest sense, i.e., any computation
that utilizes special architectures and power of the high performance
computers is considered as high performance computation or nonconventional

Scientific computations has been one of the most important components of high
performance computation.  It is the initial and continuous driving force
behind the development of high performance computers.  The goal of this
session is to bring together researchers in various areas of high performance
scientific computation and applications to exchange ideas and communicate new

The HPSCA session focuses on the high performance numerical computation
techniques and applications.  Anything that is related to this topic is
relevant.  Please e-mail Jun Zhang at if you are interested
in presenting a paper.  A web page for this special session has been created

where various deadlines and a tentative speaker list can also be found.

This special session is organized by:

Jun Zhang   
Department of Computer Science
University of Kentucky
Lexington, KY 40506-0046


Date: Fri, 18 Dec 1998 23:44:42 -0500 (EST)
From: Lothar Reichel 
Subject: Conference in honor of Richard Varga


               a conference on the occasion of 
              Richard S. Varga's 70th birthday

The meeting will take place at Kent State University on March 25-27, 1999, and
will focus on the many research areas in which Richard Varga has made
important contributions.  The conference will provide an opportunity for
researchers in these different yet related areas to exchange ideas.  More than
30 speakers have already agreed to give presentations.  A banquet dinner will
be held on Friday, March 26.  The journal Numerical Algorithms will publish a
special issue dedicated to Richard Varga.

Further information about the conference is available at the web site 


If you are interested in participating in the conference or coming to the
banquet, please notify Daniela Calvetti ( or Lothar Reichel
( as soon as possible.  Your e-mail message should
indicate whether you would like to

___   participate in the meeting,

___   present a talk at the meeting,

___   submit a paper to the special issue.

Please contact Daniela Calvetti or Lothar Reichel if you have any


Date: Mon, 21 Dec 1998 16:45:14 +0100
From: Gundolf Haase 
Subject: Job opportunity in Linz

         Ph.D. Student Position for Visualization

The ``Spezialforschungsbereich'' SFB F013 ``Numerical and Symbolic Scientific
Computing'' offers a Ph.D.  student position for research in ``Graphical
Scientific Computing'' funded by the ``Austrian Research Fund'' FWF for at
least 2 years.  The successful candidate will be responsible for the graphical
pre- and postprocessing and the according services.  Besides the
qualifications in the fields of software design and visualization we expect
the ability to work in an interdisciplinary research environment.  The
research activities will be part of the work of the SFB which is concerned
with the development of new numerical, symbolic and coupled methods for the
solution of field problems arising in mechanics, electromagnetics and image
processing.  The methods are implemented in innovative software packages.  In
particular, the large amount of data arising from solving large scale problems
(also in parallel) demands for efficient methods for data partitioning and

Interested candidates are invited to send a CV together with a list of
publications as soon as possible to

    O.Univ.-Prof. Dr. Ulrich Langer
    Speaker of the  SFB
    Johannes Kepler University Linz
    Institute for Analysis and Computational Mathematics
    Altenbergerstr. 69
    A--4040 Linz
    Tel. ++43-732-2468-9168
    Fax ++43-732-2468--10

For additional information send an email to :


Date: Thu, 3 Dec 1998 13:52:35 -0600 (CST)
From: "Dr. Michael Mascagni" 
Subject: Postdoctoral Opportunity (Southern Mississippi)

                   Post-Doctoral Employment Announcement

Post-doctoral research associate sought to contribute to a research project in
random number generation and Monte Carlo methods for parallel and distributed
computing.  The project is supported by an Accelerated Strategic Computing
Initiative (ASCI) Level 3 contract between the University of Southern
Mississippi and the Department of Energy's (DOE) Lawrence Livermore National
Laboratory (LLNL) and Los Alamos National Laboratory (LANL).  The DOE's ASCI
program is a multi-year multi-billion dollar high-performance and scientific
computing effort that supports the DOE's nuclear stockpile stewardship effort.
The ASCI program is larger than all other US Government high-performance
computing and communication programs, combined.

Duties will include:

   * Support and continued development of the Scalable PseudoRandom Number
     Generation (SPRNG) library (URL: within
     a DOE/ASCI context
   * Carry out research on pseudo- and quasi-random number generation in
     collaboration with a well-established researcher in the field and his
     graduate students
   * Collaborate with DOE scientists at LLNL and LANL on ASCI-class Monte
     Carlo applications
   * Publish research results and present findings at national and
     international meetings
   * Lead the technical aspects of this highly visible ASCI project

The work location is the Stennis Space Center, located on the Mississippi Gulf
Coast in the University of Southern Mississippi's Center of Higher Learning
(CHL).  More specifically, this project is housed in the Trent Lott
Supercomputing and Visualization Institute.  This will provide access to a
wide variety of visualization and high-performance computing equipment and
know-how located both at the Institute and at the Naval Oceanographic Office's
(NAVO) Major Shared Resource Center (MSRC), located across the street.
Currently the NAVO MSRC is the 10th most powerful computing site in the world.

The project is expected to begin in 1999, and continue for up to four years.
The successful applicant must have a Ph.D.  in Mathematics, Computer Science,
Nuclear Engineering, Physics or a related field, and must have experience with
Monte Carlo methods and/or random number generation, as well as familiarity
with programming optimized and portable code in "C/C++" and "Fortran." While
US citizenship is not required, the possibility of employment at DOE following
this project is enhanced with it.

Interested parties should send a CV, names of three references, and a letter
of application to:

                    Dr. Michael Mascagni
                    Box 10057
                    University of Southern Mississippi
                    Hattiesburg, MS 39406-0057
                    Tel.: +1.601.266.6516 (USM)
                    +1.228.688.7123 (Stennis)
                    FAX: +1.601.266.4741 (USM)
                    +1.228.688.7454 (Stennis)



Date: Thu, 17 Dec 1998 15:08:39 -0700
From: " Benito Chen" 
Subject: Tenure Track Position in Applied Mathematics (Wyoming)

University of Wyoming
Department of Mathematics
Tenure Track Position in Applied Mathematics

The University of Wyoming Mathematics Department (Web site invites applications for a tenure-track Assistant or
Associate Professorship in Applied Mathematics to begin August, 1999.  We seek
candidates with an earned doctorate, proven teaching ability and strong
research in areas of interest in the department.  A strong commitment to
undergraduate and graduate advising and service is also necessary.  Areas of
particular interest include computational mathematics, mathematical modeling,
and numerical analysis.  For Associate Professor we require an outstanding
record in research and funding.

Applicants should arrange to send a vitae, research plan, teaching philosophy,
and three letters of recommendation to the Applied Search Committee,
Department of Mathematics, University of Wyoming, P. O. Box 3036, Laramie,
Wyoming 82071.  Review of applications begins February 1, 1999.

The University of Wyoming is an affirmative action/equal opportunity employer,
and encourages women and underrepresented minorities to apply.

Benito Chen-Charpentier   | (307)766-2280
Department of Mathematics | (307)766-6838 (FAX)
University of Wyoming     |
P.O. Box 3036
Laramie, WY 82071-3036


End of MGNet Digest