Send mail to:    mgnet@cs.yale.edu             for the digests or bakeoff
Current editor:  Craig Douglas                 douglas-craig@cs.yale.edu
Anonymous ftp repository:    casper.cs.yale.edu (128.36.12.1)

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

Volume 3, Number 4 (April 30, 1993)

Today's topics:

Important Dates
LaTeX starter kit for NASA Proceedings of the 6th Copper Mountain Conf.
Papers by Pasciak et el
September'92 GAMM Seminar on multigrid methods
Latex smoothing-factor manuscript

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

Date: Thu, 30 Apr 1993 10:52:43 -0400
From: douglas-craig ("Craig C. Douglas")
Subject: Editorial (the first in 21 issues)

This year there will be many multigrid papers written, thanks to the Copper
Mountain and European multigrid conferences and the yearly domain
decomposition meeting.  In a period where funding is being scrutinized more
and jobs are scarser than in most years, I would like to point out that having
a more generous list of citations could help our field as a whole.

It has been quite common in the past for a good theorem to be proven in
slightly different forms by 3 or 4 groups at about the same time.  Citing all
of them helps our field.  Also, if a new theorem is proven that extends an
older theorem incrementally, citing the new one is not always correct if its
new features are not used in your paper.  However, citing both helps our
field.

One of the black marks in our field is that a new school of multigrid starts,
writes a few papers, and then only cites their own works thereafter.
Multigrid is by no means unique to this.  However, it means that no other
group has any motivation to cite the new school's papers (ever, by the way).

All of us are aware that the Citations Index is both our friend and foe at
times.  A simple use of a computer is to search the CD version of the
index using a structured query language.  Find all of the citations to an
author, delete all that came from people who ever wrote a paper or studied
under the author, and look at the total.  This is sometimes referred to as the
number of "unbiased citations" (there are also less flattering terms for
this).  Doing this type of search is utterly trivial and is being done I am
sorry to report.  It also means that citations coming from other groups are
much more important to your career than citations from people you regularly
(or even occasionally) write papers with.

During one 2 year period, I looked up or received reprints of over 300 journal
papers with multigrid or multilevel as a key word.  Roughly 85% of the
citations were for the books of either Hackbusch [1] or Briggs [2] (this is in
order of quantity of citations).  Journal articles were approximately 8%.  I
would like to see the 8% number increase in the next two years.

Finally, I would like to never again hear a professor inform one of his
students that it is not a good idea politically to cite someone's nearly
identical multigrid or domain decomposition results.  This advice to a student
is destructive, not constructive.

some of you to change yours, however.  In the long run, members of our field
will benefit by improved citations.

[1] W. Hackbusch, Multigrid Methods and Applications, Springer-Verlag, Berlin,
1985.
[2] W. L. Briggs, A Multigrid Tutorial, SIAM books, Philadelphia, 1987.

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

Date: Thu, 29 Apr 1993 16:10:34 -0400
From: douglas-craig ("Craig C. Douglas")
Subject: Important Dates

7th Domain Decomposition Conference:  May 15th for 1 page abstracts (please
send to ddm7@math.psu.edu).

6th Copper Mountain Conference:  May 7th for revisions to the accepted papers

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

Date: Thu, 29 Apr 1993 16:05:46 -0400
From: douglas-craig ("Craig C. Douglas")
Subject: LaTeX starter kit for NASA Proceedings of the 6th Copper Mountain Conf.

This was donated by one of the conference attendees.  I claim no credit for it
(nor does the submitter either).  I am using it myself, so I can attest to the
fact that it works.  Please fill in the title, address and grant information,
and the body of the paper.  This is not a style file, but would be your LaTeX
file (I called mine douglas_nasa.tex).  A copy of this can be ftp'ed from MGNet
(mgnet/copper93/NASA.tex).

send me e-mail that it is there.  If you download a LaTeX paper (.tex file),
NASA will not have to type the page numbers in on a typewriter (are these
things actually made anymore?).  If you submit a PostScript or .dvi file,
please do not put page numbers in (except maybe in pencil very lightly).

Good luck,
Craig Douglas

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%                                               %
%  The following is a LaTeX version of a paper  %
%  for the Sixth Copper Mountain Conference on  %
%  Multigrid Methods                            %
%                                               %
%  Blank lines count here.                      %
%                                               %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
\documentstyle[12pt]{article}
\topmargin=-.59375in
\textheight=9.1875in
\oddsidemargin=-.3in
\textwidth=7.1in
\pagestyle{empty}
\renewcommand{\thefootnote}{\fnsymbol{footnote}}

\newtheorem{thm}{Theorem}
\newtheorem{define}{Definition}
\newtheorem{cor}[thm]{Corollary}
\newtheorem{lem}[thm]{Lemma}
\newenvironment{Algorithm}{\begin{algorithm}\begin{description}
}{\end{description}\end{algorithm}}
\newtheorem{algorithm}{Algorithm}

\newcommand{\thmref}[1]{Theorem~\ref{#1}}
\newcommand{\assref}[1]{Assumption~\ref{#1}}
%\newcommand{\secref}[1]{\S\ref{#1}}
\newcommand{\secref}[1]{Section~\ref{#1}}
\newcommand{\lemref}[1]{Lemma~\ref{#1}}

\def\proof{\par{\it Proof}. \ignorespaces}
\def\endproof{{\ \vbox{\hrule\hbox{%
\vrule height1.3ex\hskip0.8ex\vrule}\hrule
}}\par}

\renewcommand{\section}[1]{\vglue.3in
\noindent ~ \hfill #1 \hfill ~
\vglue.3in}

\renewcommand{\subsection}[1]{\vglue.3in
\noindent ~ \hfill #1 \hfill ~
\vglue.3in}

\begin{document}

\vglue.4in
\noindent ~\hfill THE FIRST LINE OF THE TITLE\hfill ~

\noindent
~\hfill AND THE SECOND HERE\footnote{This
work was supported in part by the International Multigrid Foundation under
Grant No. IMG-0000001.} \hfill ~

\vglue.3in
\noindent ~ \hfill Sue Donim\hfill ~

\noindent ~ \hfill Department of Multiple Grids\hfill ~

\noindent ~ \hfill Multilevel University\hfill ~

\noindent ~ \hfill Copper Mountain, Colorado \hfill ~

\section{SUMMARY}
\raggedright
\parindent=.2in

\section{INTRODUCTION}
We consider the convergence properties of ...

\def\thebibliography#1{\section{{REFERENCES}}\list
{[\arabic{enumi}]}{\settowidth\labelwidth{[#1]}\leftmargin\labelwidth
\usecounter{enumi}}
\def\newblock{\hskip .11em plus .33em minus .07em}
\sloppy\clubpenalty4000\widowpenalty4000
\sfcode\.=1000\relax}
\let\endthebibliography=\endlist

\bibliography{mg}
\bibliographystyle{unsrt}

\end{document}

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

Date: Fri, 2 Apr 93 09:37:17 EST
From: pasciak@jep.das.bnl.gov (Joe Pasciak)
Subject: Papers by Pasciak et al

Alright!  You finally shamed me into doing it.  I put abstracts and ps files
into mgnet.  These include my paper for my copper mountain presentation as
well as five other reprints.

Editor's Note: Thank you, Joe...
-------------  in mgnet/papers/Pasciak_et_al (initially)
mgnet/copper93/bramble_kwak_pasciak.ps

FILE :: bgp1.ps bgp1.abs

@article{JHBramble_CIGoldstein_JEPasciak_93,
author =        "J. H. Bramble and C. I. Goldstein and J. E. Pasciak",
title =         "Analysis of V--cycle multigrid algorithms for forms
journal =       "SIAM J. Sci. Stat. Comput. (to appear)  also Brookhaven
Nat. Lab. Tech. #BNL-47533"
}

FILE :: bpxdd.ps bpxdd.abs

@incollection{JHBramble_JEPasciak_JXu_92,
author =        "J. H. Bramble and J. E. Pasciak and J. Xu",
title =         "A multilevel preconditioner for domain decomposition
boundary systems",
booktitle =     "Proceedings of the 10'th Inter. Conf. on Comput. Meth.
in Appl. Sci. and Engr.",
publisher =     "Nova Sciences",
year =          "1992"
}

FILE :: bpl.ps bpl.abs

@article{JHBramble_ZLeyk_JEPasciak_93,
author =        "J. H. Bramble and Z. Leyk and J. E. Pasciak",
title =         "The analysis of multigrid algorithms for
pseudo-differential operators of order minus one",
journal =       "Math. Comp..",
pages =         "Brookhaven Nat. Lab. #BNL-47335 "
}

FILE :: bpreg.ps bpreg.abs

@article{JHBramble_JEPasciak_92,
author =        "J. H. Bramble and J. E. Pasciak",
title =         "Uniform convergence estimates for multigrid V-cycle
algorithms with less than full elliptic regularity",
publisher =     "Brookhaven Nat. Lab. #BNL-47892",
year =          "1992"
}

FILE :: bpv.ps bpv.abs

@article{JHBramble_JEPasciak_93,
author =        "J. H. Bramble and J. E. Pasciak",
title =         "New estimates for multigrid algorithms including the
V-cycle",
journal =       " Math. Comp.(to appear) also Brookhaven Nat. Lab. #BNL-46730",
}

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

Date:  Wed, 7 Apr 1993 16:40:46 +0200
From: FUHRMANN@IAAS-BERLIN.DBP.de
Subject:  September'92 GAMM Seminar on multigrid methods

Organized by the Institute of Applied Analysis and Stochastics (IAAS) Berlin
and supported the German GAMM, a Seminar on multigrid methods took place  in
Gosen near Berlin in September, 1992. Among the participants were   U.Langer
(Chemnitz), V.V.Shaidurov (Krasnoyarsk), Yu.A.Kuznetsov (Moscow),  L.Tobiska
(Magdeburg) U.Ruede and M.Griebel (Munich).
This seminar continued the tradition of the GDR  multigrid  seminars  mainly
inspired  by U.Langer.
Now, a report on this seminar is ready, it  exists  in  paper  form  in  the
preprint series of the IAAS institute and can be ordered from

preprint@iaas-berlin.dbp.de

or

Institut fuer Angewandte Analysis und Stochastik (IAAS)
Hausvogteiplatz 5-7
D-O-1086 Berlin  (D-10117 Berlin after July 1st, 1993)
Germany
fax:   +49 (030) 2004975

Here, I only want to  announce the contents of this report, because
it has  been produced in a rather heterogeneous way:

GAMM - Seminar  on  Multigrid Methods
Gosen, Germany, September 21 - 25, 1992
S.Hengst (ed.)

IAAS Berlin,
Report No. 5, 1993

J.Fuhrmann (Berlin):
Calculation of saturated - unsaturated flow in porous
media with a Newton - multigrid method

J.Molenaar (Delft):
Multigrid for a groundwater flow problem

O.Dorok and L.Tobiska (Magdeburg):
A multigrid method for solving the non-isothermal
incompressible Navier-Stokes equation

W.Queck (Chemnitz/Freiberg):
The Finite-Element-Multigrid-Package FEMGP - A software tool
for solving boundary value problems on personal computers

T.Steidten (Chemnitz):
FEMGPM - a multilevel program for 2D problems

R.Blaheta and R.Kohut (Ostrava):
Inexact Newton-like methods with application
in elasto-plastic incremental analysis

M.Jung (Chemnitz):
On adaptive grids in multilevel methods

V.V.Shaidurov (Krasnoyarsk):
New convergence estimates for the symmetric V-cycle

U.Ruede (Muenchen):
Data abstraction techniques for multilevel algorithms

H.Regler (Muenchen):
Algebraic multilevel methods in chip design

E.Katzer (Kiel):
A subspace decomposition twogrid method for hyperbolic equations

Juergen Fuhrmann
fuhrmann@iaas-berlin.dbp.de

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

Date: Wed, 14 Apr 93 13:48:45 MDT
Subject: Latex smoothing-factor manuscript

Smoothing Factors of Two-Color Gauss-Seidel Relaxation for a
Class of Elliptic Operators

NCAR Oceanography & GTP

Abstract

Analytic formulae are obtained for the smoothing factors yielded by
Gauss-Seidel relaxation in two-color ordering for a class of scalar elliptic
operators.  Block and point relaxation, in conjunction with full or partial
coarsening are encompassed, for operators with general (constant, positive)
coefficients in general dimensions and arbitrary number of relaxation sweeps.
It is found that there is no direct dependence of the dimension on the
smoothing factors, and that the effect of the number of relaxation sweeps
performed on the smoothing factor is usually independent of the coefficients
in the operator and of the relaxation scheme.  The results are compared to
computed results of two-level analyses.  Smoothing strategies implied by the
formulae are discussed.

Editor's Note: mgnet/papers/Yavneh/smoothing.tex
-------------  mgnet/papers/Yavneh/smoothing.abs

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

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