Send mail to:  mgnet@cs.yale.edu             for the digests or bakeoff
               mgnet-requests@cs.yale.edu    for comments or help
 
Anonymous ftp repository:  www.mgnet.org (128.163.209.19)
 
Current editor:  Craig Douglas douglas-craig@cs.yale.edu
 

WWW Sites:  http://www.mgnet.org or
            http://casper.cs.yale.edu/mgnet/www/mgnet.html or
            http://www.cerfacs.fr/~douglas/mgnet.html or
            http://phase.hpcc.jp/mirrors/mgnet or
            http://www.tat.physik.uni-tuebingen.de/~mgnet

Today's editor:  Craig Douglas (douglas-craig@cs.yale.edu)

Volume 12, Number 8 (approximately August 31, 2002)

Today's topics:

     Full MG Code for optimality systems - A. Borzi 
     parallel Algebraic Recursive Multilevel Solver (pARMS) available
     Pieter Wesseling's Book Images
     Reading Level
     MGNet's Conference Page
     Help in Finding Software
     Benchmarking elementary functions
     MAFELAP 2003, 21 - 24 June 2003 (First Announcement)

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

Date: Fri, 16 Aug 2002 09:29:24 +0200
From: "Dr. Alfio Borzi" 
Subject: Full MG Code for optimality systems - A. Borzi 

I would be very pleased if you would consider the following submission to
MGNet Software.

I would like to contribute to the MGNet Software with a full multigrid code
for optimality systems.  As we are experiencing, there is growing interest in
fast methods for optimization problems, especially multigrid methods.

With the present code, called 'controlla.for' (sorry, I keep choosing italian
names), I wish give a contribution to the development of multigrid methods for
optimal control problems.  I hope that people interested in optimal control
problems and wishing to use multigrid methods may benefit from having this
code at disposal.

Brief description:

CONTROLLA.FOR:  A full multigrid method for the solution of an optimality
system arising from optimal control of the solid fuel ignition model.

Optimal control (OPC) problems are receiving increasing attention in the
scientific community.  As an example of optimal control formulation consider
the minimization of the following cost functional

J = |U - Z|^2/2 + a*|F|^2/2

where U satisfies

-DELTA U  = F

in a given domain Omega.

A way to solve this OPC problem is to reformulate it as an unconstrained
minimization problem by introducing the Lagrangian

L = J + < -DELTA U - F, P >

where P is a Lagrangian multiplier.  Derive the necessary conditions for a
minimum to obtain:

 -DELTA U  = F
 -DELTA P  = - ( U - Z )
  a*F - P  = 0

This is called the optimality system.  For a partial list of references
concerning the use of multigrid for solving optimality systems see Ref.  (1)
below.

The code CONTROLLA.FOR solves the optimality system of the following OPC
problem

min ( |U - Z|^2/2 + B*|EXP(U)-EXP(Z)|^2/2 + a*|F|^2/2 )

-DELTA U - D*EXP(U) = F

This reaction-diffusion model describes explosive chemical reaction.  Control
is required to drive combustion in a desired way or to avoid explosion.

A detailed discussion of this problem and of its multigrid solution is given
in

(1)  Alfio Borzi' and Karl Kunisch,
     The Numerical Solution of the Steady State Solid
     Fuel Ignition Model and Its Optimal Control,
     SIAM J. Sci. Comp., 22(1) (2000), pp. 263-284.

In the linear case (B=0,D=0) CONTROLLA.FOR solves the optimality system given
above.  Convergence results in the framework of local mode analysis and in the
framework of the theory given in

     J.H. Bramble, Do Y. Kwak, and J.E. Pasciak,
     Uniform convergence of multigrid V-cycle iterations
     for indefinite and nonsymmetric problems,
     SIAM J. Numer. Anal. 31 (1994), pp. 1746-1763.

are presented in

(2)  A. Borzi, K. Kunisch, and Do Y. Kwak,
     Accuracy and Convergence Properties of the Finite
     Difference Multigrid Solution of an Optimal Control
     Optimality System,
     Univ. of Graz & Tech. Univ., SFB F003, Rep. No. 245,
     July 2002, Graz.
     To appear in SIAM J. Control Optim..

This code has been developed based on techniques and software given in

  Achi Brandt,
  Multi-level adaptive solutions to boundary-value problems,
  Math. Comp., 31 (1977), pp. 333-390.

  Achi Brandt,
  Multi-level adaptive techniques (MLAT) for partial
  differential equations: Ideas and software,
  Mathematical Software III, J.R. Rice Ed., Academic Press,
  New York, 1977, pp. 277-318.


CONTROLLA.FOR is a public domain code.  I provide the program ``AS IS''
without warranty of any kind.

ALFIO BORZI'
INSTITUT FÜR MATHEMATIK
KARL-FRANZENS-UNIVRSITÄT GRAZ
HEINRICHSTR. 36, A-8010 GRAZ
AUSTRIA
ALFIO.BORZI@UNI-GRAZ.AT

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

Date: Fri, 30 Aug 2002 18:51:09 -0500 (CDT)
From: Yousef Saad 
Subject: parallel Algebraic Recursive Multilevel Solver (pARMS) available

Dear colleagues:

We are pleased to announce the availability of version 0.2 of the parallel
Algebraic Recursive Multilevel Solver (pARMS).  pARMS is a package for solving
large sparse linear systems on parallel computers.  It includes a fairly wide
variety of preconditioning options:

* Additive Schwarz type using ilu0, ilut, iluk, or arms factorization

* Schur complement - based techniques using ilu0, ilut, iluk, or arms
  factorization (left, right versions)

* Schur complement-based techniques using global ilu(0) or symmetric GS for
  the Schur complement

Configuration options are available for the IBM SP, the SGI Origin, and a
cluster of Linux machines.

Visit the pARMS web-site at

    http://www.cs.umn.edu/~saad/software/pARMS

for the software, related articles, and viewgraphs.

Main contributors to the project:

YS, Zhongze Li, Masha Sosonkina, Brian Suchomel, Devdatta Kulkarni, Yu Liang.

Acknowledgments: work supported by NSF/ACI

PS:  For those interested - a sequential versionof ARMS in matlab is available
in the matlab demos of a recent tutorial:

  http://www-users.cs.umn.edu/~saad/Brisbane/home.html

Yousef Saad                               saad@cs.umn.edu
Dept of computer Science & Eng.           http://www.cs.umn.edu/~saad
University of Minnesota                   ph:  (612) 624 7804
200 Union st. SE.                         fax: (612) 625 0572
Minneapolis, MN 55155

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

Date: Fri, 16 Aug 2002 20:28:51 +0200 (CEST)
From: Christian Cornelssen 
Subject: Pieter Wesseling's Book Images

Dear Mr Douglas,

I have downloaded the Wesseling book image files placed at

    http://www.mgnet.org/mgnet-books-wesseling.html

When preparing the PDFs for printing, I found the following:

* The Webpage lists the page range for chapter 5 as 60-94.
  This should be 60-78.

* preface.pdf: Some left text flows off the second (print) page.
  (Anyway, I don't need it.)

* ch8.pdf: As you mentioned, pages 194,195 pose problems.
  They should be merged onto one print page.  Currently,
  they are separate and each on the wrong side (left<>right).
  I have attached a file `p194+195.pdf' that has these pages merged
  as desired (by using some Ghostscript tools and PS utilities).

* index.pdf: The last page is on the wrong side (should be left).
  I have attached a file `p284.pdf' containing the image justified
  as indicated above.

To approximately center the PDFs on A4 paper, I recommend displacing
the typearea about 6pt rightwards and 8pt downwards. For letter paper,
shifting 14pt rightwards and 33pt downwards should give fine results.
These measures apply to the non-rotated (i.e. portrait-oriented)
files.

I haven't centered the attached PDFs.  If you are interested in
details on how to center the PDFs for printing, I can provide further
information.

Best regards,
Christian Cornelssen

    Editor's Note: The book has been updated with the better scans.
    -------------

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

Date: Tue, 6 Aug 2002 22:59:38 -0600 (MDT)
From: Jan Mandel 
Subject: Reading Level

Look at

http://www.amazon.com/exec/obidos/ASIN/0198536062/qid=1028696183/sr=1-30/ref=sr_1_30/103-8782815-9300662

"reading level"

    Editor's Note: This is an amazingly well written book to get this rating.
    -------------

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

Date: Wed, 21 Aug 2002 11:20:58 +0200
From: Michael Gee 
Subject: MGNet's Conference Page

There's a little bug on the mgnet's conference page
The link to the 7th european multigrid conference should be to

    http://cox.iwr.uni-heidelberg.de/Conferences/emg02.html

Yours, Michael Gee

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

Date: Thu, 22 Aug 2002 13:14:08 -0700 (PDT)
From: laura nora 
Subject: Help in Finding Software

I am wondering if you can help me find some free software on "using muligrid
method in slid mechanics".  It seems that there is no such code in the free
software section.

I am interested in using multigrid method to do structural static and dynamic
analysis.  I hope some day I can contribute something to this site.

Thanks
Laura

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

Date: Wed, 7 Aug 2002 14:45:01 -0400 (EDT)
From: Ren-Cang Li 
Subject: Benchmarking elementary functions

I am seeking applications or code pieces that heavily involve the evaluations
of elementary functions (exp, log, trig., sqrt, x^y, etc.) to do a performance
benchmark of available elementary math libraries (libm) from today's computer
vendors.  If you know or have any, please email me at rcli@cs.uky.edu.  You
help is greatly appreciated and will be acknowledged.

Ren-Cang Li
http://www.ms.uky.edu/~rcli

    Editor's note: Rencang works primarily for an HP Lab, on very long term
    -------------  leave from his university position, so check if you are
                   uneasy about giving code to a computer vendor employee.

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

Date: Tue, 27 Aug 2002 14:22:09 +0100 (BST)
From: John R Whiteman 
Subject: MAFELAP 2003, 21 - 24 June 2003 (First Announcement)

        THE MATHEMATICS OF FINITE ELEMENTS AND APPLICATIONS 2003

                www.brunel.ac.uk/~icsrbicm/mafelap03

The ELEVENTH conference on the Mathematics of Finite Elements and 
   Applications will be held at Brunel University during

                        21-24 JUNE, 2003

The internationally recognised MAFELAP conferences have been run by BICOM 
at Brunel University at three year intervals since 1972.   In these 
conferences exposure is given to research on the theory and practical 
application of finite element methods.

Invited speakers:

        M AINSWORTH, University of Strathclyde
        H T BANKS, North Carolina State University
        C CARSTENSEN, Technische Universitat, Vienna
        S N CHANDLER-WILDE, Brunel University
        L DEMKOWICZ, University of Texas at Austin
        J T ODEN, University of Texas at Austin
        C SCHWAB, Eidgenoessische Technische Hochschule, Zurich
        E SULI, University of Oxford
        M F WHEELER, University of Texas at Austin
        J R WHITEMAN, Brunel University

The 2003 ZIENKIEWICZ LECTURE Lecture will be given by

        K MORGAN, University of Wales, Swansea

The 2003 Isaac Newton Lecture will be given by

        F BREZZI, University of Pavia

                THE PROGRAMME AND CALL FOR PAPERS

In addition to the invited lectures, the programme will contain LECTURES IN 
PARALLEL SESSIONS, SESSIONS OF POSTER PAPERS and MINI-SYMPOSIA.

        ANYONE WISHING TO ORGANISE A MINI-SYMPOSIUM IS REQUESTED
        TO CONTACT THE MAFELAP SECRETARY AS SOON AS POSSIBLE AT 
        mafelap2003@brunel.ac.uk OR MAFELAP, BICOM, BRUNEL UNIVERSITY,
        UXBRIDGE, UB8 3PH, ENGLAND, OR TELEPHONE: 01895 203270 (UK)
        OR: -44-1895-203270 (INTERNATIONAL), OR FAX: 01895-203303.

        ANYONE WHO WISHES TO PRESENT EITHER A LECTURE IN A PARALLEL
        SESSION OF A POSTER PAPER IS ASKED TO SUBMIT AN ABSTRACT
        AS SOON AS POSSIBLE AND, IN ANY CASE, NOT LATER THAN
        26 FEBRUARY 2003. THE CONFERENCE ORGANISERS WILL MAKE EVERY
        EFFORT TO ASSOCIATE TALKS WITH SUITABLE MINI-SYMPOSIA.

URL:  www.brunel.ac.uk/~icsrbicm/mafelap03

Contact:  mafelap2003@brunel.ac.uk

Another feature of MAFELAP 2003 will be the Isaac Newton Institute 
Mini-Symposium which will run throughout the conference and is a satellite 
event of the Newton Institute Programme Computational Challenges in Partial 
Differential Equations; 20 January - 4 July, 2003; for further details of 
the INI programme see http://www.newton.ac.uk/programs/CPD.

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

End of MGNet Digest
**************************