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 9, Number 6 (approximately June 30, 1999)

Today's topics:

     A Call for Help
     Two Papers (Brandt)
     Copper Mountain Contribution (Paraschivoiu and Cai)
     Copper Tutorial: Parallel Multigrid
     Copper Tutorial: Algebraic Multigrid
     Copper Transparencies (Johannsen)
     Summer School on CFD
     EMG99: Final Announcement


Date: Fri, 18 Jun 1999 17:00:26 -0400 (EDT)
From: Dejan Vinkovic 
Subject: A Call for Help

Dear Sirs/Madams,

I am a PhD student at Dept. of Phys. and Astro., Univ. of Kentucky.  I'm
currently working on the code for 2D radiative transfer in dust around stars.
I would be very thankful if somebody can help me with problems related to the
grid generation.

Any suggestion, comment, advice or reference about the grid generation with
described properties (see below) would be very helpful.  Also, suggestions and
comments about methods for solving the radiative transfer equation (or
integro-differential equations in general) are appreciated.

More extensive description of the problem (4 pages, with equations) is
available here:

Thank you very much.

Dejan Vinkovic
University of Kentucky
Department of Physics and Astronomy
177 Chem-Phys Bldg
Lexington KY 40506
          2D Radiative Transfer in Dusty Stellar Environments

The dust configuration around stars usually has rotational symmetry.  It can
absorb, scatter and re-emit the radiation from the star, that is, change the
stellar spectral continuum.  To solve the integro-differential radiation
transfer equation in the dusty circumstellar cloud, we need the temperature
distribution at all points in the cloud.  But the temperature is determined
from the local energy density, obtained from the transfer equation.  Thus the
emerging spectrum is calculated by iterations between calculated temperature
in one step, and energy density in another.

The dust consists of several different chemical components, each one with
different sublimation temperature, above which specific type of dust can not
exist.  Thus the dust density has discontinuities at the points with
sublimation temperatures.  Moreover, we don't know exact position of these
points because we don't know a priori the correct temperature distribution.

Obviously, the most important part of the calculation process is generation of
the computational grid.  A grid generator has to be applied after each
temperature iteration because the sublimation surface will be changed.  Thus
we can impose some demands on the grid generator:
- it has to follow the density gradient, with discontinuities
- it has to be fast and applied several times during the calculation (or somehow 
move the grid points at sublimation surface toward new position after the
temperature is updated)

There is a concern that the discrete ordinate method, usually applied in the
radiative heat transfer, would not be able to recognize possible abrupt
changes in intensities during the angular integration.  Therefore, some other
method could be more appropriate after the grid generation.


Date: Thu, 20 May 99 09:00:13 +0300
From: "Prof. Achi Brandt" 
Subject: Two Papers

My secretary Sarah will be sending you for MGNET the paper I have for the
Copper Proceedings.

             General Highly Accurate Algebraic Coarsening Schemes

    Editor's Note: See or access it at

I also have a new summary paper, extending by much my 1997 Copper
paper. It appears in my homepage at

        Multiscale Scientific Computation:  Six Year Research Summary

                                 Achi Brandt
                       The Weizmann Institute of Science
                            Rehovot, 76100, Israel


The Gauss Center research on multiscale computational methods is reported,
emphasizing main ideas and inter-relations between various fields, and listing
the relevant bibliography.  The reported areas include:  top-efflciency
multigrid methods in fluid dynamics; atmospheric flows and data assimilation;
feedback optimal control; PDE solvers on unbounded domains; wave/ray methods
for highly indefinite equations; rigorous quantitative analysis of multigrid;
many-eigenfunction problems and ab-initio quantum chemistry; fast evaluation
of integral transforms on adaptive grids; multigrid Dirac solvers; fast
inverse-matrix and determinant updates;multiscale Monte-Carlo methods in
statistical physics; molecular mechanics (including fast force summation, fast
macromolecular energy minimization, Monte-Carlo methods at equilibrium, both
for macromolecules and for large ensembles of small molecules, and the
combination of small-scale equilibrium with large-scale dynamics); image
processing (edge detection and segmentation); and tomography (medical imaging
and radar reconstruction).

Key words.  Scientific computation, multiscale, multi-resolution, multigrid,
fluid dynamics, atmospheric flows, data assimilation, optimal control, wave
problems, Dirac equations, inverse matrix, Schrodinger operator, Monte-Carlo
algorithms, critical slowing down, molecular mechanics, fast force summation,
energy minimization, integro-differential equations, tomography , medical
imaging, radar, image processing, edge detection, segmentation, algebraic

AMS subject classification.  34A50, 35A40, 44-04, 45{04, 65C05, 65F10, 65F15,
65F40, 65K10, 65M30, 65M50, 65M55, 65N22, 65N25, 65N38, 65N55, 65R10, 65R20,
65Y05, 68U10, 70-08, 76-04, 76M20, 81-08, 81T80, 82-08, 82B80, 82C80, 92E99

    Editor's Note: See or access it at


Date: Tue, 22 Jun 1999 14:33:22 -0600
From: Marius Paraschivoiu 
Subject: Copper Mountain Contribution (Paraschivoiu and Cai)

A Unigrid Multi-Model Full Potential and Euler Formulation for Transonic Flows

                             Marius Paraschivoiu
             Department of Mechanical and Industrial Engineering
                              University Toronto
                           Toronto, Canada M5S 3G8

                                Xiao-Chuan Cai
                        Department of Computer Science
                            University of Colorado
                            Boulder, CO 80309, USA


In this paper we present a unigrid multi-model formulation for transonic flow
calculations based on solving, in sequence, the full potential equation and
the the Euler equations.  The goal is to minimize the overall computation time
to simulate steady flows by using a more computational efficient physical
model in the early iteration steps.  The proposed method is based on two
steps.  In the first step, the full potential equation is solved to obtain a
intermediate solution which is ``improved'', in the second step, by solving
the Euler equations.  The full potential equation and the Euler equations are
discretized on the same unstructured mesh to avoid new coarse grid generation
and to simplify the transfer, between the two models.  The resulting nonlinear
systems are solved by a Newton-Krylov algorithm.  To demonstrate the
feasibility of this multi-model formulation, we present computational results
for a three-dimensional transonic flows around a nonlifting wing created from
a NACA0012 airfoil.

    Editor's Note: See or access it at


Date: Thu, 24 Jun 1999 17:42:26 -0700 (PDT)
From: "Jim E. Jones" 
Subject: Copper Tutorial: Parallel Multigrid

I've put the slides for my Copper tutorial on MGNet.

This presentation focuses on the issues involved in parallelizing a
multigrid algorithm. Assuming no experience with parallel computing, but an
understanding of the principles of multigrid, the tutorial introduces some
of the standard and efficient techniques for developing a parallel multigrid


        * Algorithmic and implementation scalability.
        * Parallelization of multigrid by domain partitioning.
        * Performance models and metrics for parallel multigrid
        * Novel parallel multigrid algorithms: multiple coarse grids
          and concurrent multigrid.

    Editor's Note: See or
    ------------- or access it at


Date: Thu, 24 Jun 1999 09:10:32 -0700 (PDT)
From: Van Henson 
Subject Copper Tutorial: Algebraic Multigrid

Here are the cleared slides for my tutorial.  They are in Powerpoint and you
can figure out how to make them into an html document.

This introduction focuses primarily on the "classical" AMG of Brandt,
McCormick, and Ruge. An understanding of the principles of multigrid is
assumed, but the tutorial introduces algebraic multigrid in a simple,
practical manner.


        * "Classical" AMG
             o Hammers and Wrenches: the Required AMG Toolkit
             o The Ardent Quest: Seeking Algebraic Smoothness
             o Get to the Point: Coarse Grid Selection
             o Building a Better Mousetrap: Prolongation

        * Other Algebraic Approaches: an Overview
             o Smoothed Aggregation
             o Multigraph methods
             o AMGe
             o Energy-Minimizing Basis methods

    Editor's Note: See or
    ------------- or access it at
                   I am still trying to get the html version to work...

Van Emden Henson                        | "All intelligent conversation is
Center for Applied Scientific Computing |  playing on words. The rest is
Lawrence Livermore National Laboratory  |  definition and instruction."
PO Box 808 L-560                        |                    - Herman Wouk
Livermore, CA 94551                     |
                                        |                        |    Phone:   (925) 423-4283  |      Fax:   (925) 422-6287


Date: Sun, 30 May 1999 12:07:03 +0200
From: klaus 
Subject: Summer School on CFD

EMS-WiR Summer School on
Numerical Simulation of Flows
Heidelberg September 6-21, 1999
G. Wittum, Heidelberg

The numerical simulation of flows is one of the central problems in Scientific
Computing.  Complexity of flow simulations is so high that a realistic
description requires sophisticated mathematical methods and models.  In
particular modeling and simulation of turbulent flows, nearly incompressible
flows, and multi-phase flows are challenging problems for mathematical models
and numerical methods.  There is a strong interest in this topic by numerous
groups from mathematical modelling and numerical simulation.  Recently a
number of new mathematical models and methods have been introduced which are
highly relevant for flow simulations.  Amongst others these are multiscale
modelling and numerics, homogenization, finite-element and finite volume
methods, spectral and h-p discretizations, grid adaptivity and error
estimators, multigrid and conjugate-gradient type methods and wavelets.
Another field of increasing importance is the development of methods for the
visualization of flows.  The numerical simulation of flows requires
cooperation of several mathematical disciplines as Analysis, Numerics,
Mathematical Physics and Computational Science.  The European Mathematical
Society (EMS) together with the research network WiR will organize a summer
school on Numerical Simulation of Flows from Sept.  6 - 21, 1999 in
Heidelberg.  The summer school is a joint event with AMIF (Applied Mathematics
for Industrial Flow Problems) and SFB 359 of Heidelberg University.

The Summer school will consist of a theoretical and a practical part, each one
lasting a week.  The first week (Sept. 6-10, in Heidelberg) is devoted to
basic instruction.  During this week mathematical models and methods are
presented in lectures given by specialists.  In the second week (Sept. 13-17)
the participants will work on problems posed by the lecturers in different
places (Heidelberg, Freiburg, Stuttgart, Zurich).  Finally the results of this
work will be presented in a plenary meeting (Sept. 20-21, in Heidelberg).

Scientific Comittee: P. Bastian, Heidelberg, G. Dziuk, Freiburg,  W.
Hackbusch, Leipzig, R. Jeltsch, Zurich,
D. Krner, Freiburg, C.-D. Munz, Stuttgart, R. Rannacher, Heidelberg, W.
Rodi, Karlsruhe,
S. Sauter, Leipzig S. Wagner, Stuttgart, G. Wittum, Heidelberg, H.
Yserentant, Tubingen

Local Organizer: G. Wittum, Technische Simulation, IWR, Universitat
Heidelberg, Im Neuenheimer Feld 368, D-69120 Heidelberg, email:,

Registration: Online via

Registration fee:  DM 300,? payed until July 15th, DM 400,? later. The
fee covers participation, lunches, coffee breaks, and materials. All
payments, net of all charges, are to be made in Deutsche Mark by bank
transfer to account No. 50 302 787 600, Baden-Wuerttembergische-Bank
Heidelberg, BLZ 672 700 20, with the address: Universitaet Heidelberg,
EMS, Kap 1412/TG86/BA1103.

Registration deadline: July 15th 1999.

Lodging reservation: Please book rooms for the first week in Heidelberg
as early as possible using for
booking rooms for the first week in Heidelberg. A limited  number of
cheap rooms is availble for the first week in Heidelberg (see
registration form).

Financial support of participation: A limited number of scholarships by
EMS is available. To apply for such a scholarship, please add a short
curriculum vitae, a sketch of your research interests and a letter of

Prof. Dr. Gabriel Wittum

tel/fax:  +49 6221 54 8860
tel/work: +49 6221 54 8855
org:      Technische Simulation, IWR, Universitaet Heidelberg
adr:      Im Neuenheimer Feld 368, D-69120 Heidelberg, Germany


Date: Wed, 30 Jun 1999 17:48:14 +0200
From: Erik Dick 
Subject: EMG99: Final Announcement

Final announcement

Sixth European Multigrid Conference
Universiteit Gent, Belgium 
September 27 - 30, 1999

A preliminary program of the conference is now available.  54 papers have been
selected for presentation.  For information and registration please consult
our website at (press reload if


End of MGNet Digest