These conversations bring up an interesting thought; we have many linear
elliptic solvers (but the three ones we really think about are BAM PETSc
and SOR). Wouldn't it be nice if we could do

maximal_solver = "BAM_precond petsc"

or

maximal_solver = "BAM sor_postcond"	(which is what Mark wants)

to use BAM as a pre-conditioner to PETSc or SOR as a post condition to
BAM.

Anyone in cactus land want to think about this? Seems it would be a change
to linear elliptic solve (basically a parser) and a change to each of the
elliptic solvers (they have to realize they are called as a pre- or post-
conditioner and therefore behave differently, and we have to define how).

Just randomly inventing features for other people to implement :-)

	- Paul

________________________________________________________________________

  Paul Walker                              <pwalker@aei-potsdam.mpg.de>
  Max-Planck-Institut fuer GravitationsPhysik  Albert-Einstein-Institut
  Schlaatzweg 1                                  +49-331-27537-55 (tel) 
  14473 Potsdam, GERMANY                         +49 331 27537-98 (fax)
  http://www.aei-potsdam.mpg.de/~pwalker


On Tue, 31 Mar 1998, Project ELLIPTIC/John Towns NCSA Cog wrote:

>  Certainly it would be better to get a few more sweeps of SOR within the 
>  BAM solver, but a better thing to try first would be to trun the SOR 
>  solver prior to the BAM solver.  You will lose some efficiency in 
>  running SOR after BAM, but you may have more high frequency noise in 
>  there this way too. 
>   
>  -John 
>   
>   
>  Project ELLIPTIC/Mark Miller wrote: 
>  _>  _ 
>  _>  Elliptic dudes (esp. Bernd),_ 
>  _>  _ 
>  _>  I am using BAM to implement maximal slicing for 2 NS runs._ 
>  _>  However, BAM seems to give bad high frequency noise, as seen_ 
>  _>  by this snapshot of trK along the x line, which_ 
>  _>  is the axis joining the center of the 2 neutron stars (trK_ns.gif)._ 
>  _>  _ 
>  _>  Wai-Mo remembered that I did a high-resolution run with misner_ 
>  _>  BH data, and we see the same thing (high frequency noise getting_ 
>  _>  worse in time).  Again, this is using maximal slicing with BAM._ 
>  _>  (trK_misner.gif)._ 
>  _>  _ 
>  _>  I notice in the 2 NS case, the trK is much smoother when using_ 
>  _>  SOR.  My question:  Bernd, is there any way to make BAM_ 
>  _>  spend more time at the top of the V cycle?  That is, do_ 
>  _>  a few more sweeps on the fine grid?  Or shall we code up_ 
>  _>  _ 
>  _>  maximal = "bam-sor"_ 
>  _>  _ 
>  _>  that would call bam as a preconditioner to SOR?_ 
>  _>  _ 
>  _>  Mark_ 
>  _>  _ 
>  _>  ------------------------------------------------_ 
>  _>  Mark Miller_ 
>  _>  Department of Physics_ 
>  _>  Washington University_ 
>  _>  St. Louis, MO  63130-4899_ 
>  _>  E-mail: mamiller@wugrav.wustl.edu_ 
>  _>  office phone: (314) 935-4617,  fax: (314) 935-6219_ 
>  _>  ------------------------------------------------_ 
>  _>  _ 
>  _>     [1][LINK]Filename: trK_ns.gif_ 
>  _>     Size :57329_ 
>  _>  _ 
>  _>     [2][LINK]Filename: trK_misner.gif_ 
>  _>     Size :83717_ 
>  _>  _ 
>  _>  References_ 
>  _>  _ 
>  _>     1. [1]file://localhost/home2/web/Util/projects/trK_ns.gif_ 
>  _>     2. [2]file://localhost/home2/web/Util/projects/trK_misner.gif_ 
>  _>  ________________________________________________________________________ 
>  _>  _ 
>  _>  CoCoBoard Information:_ 
>  _>    From: Mark Miller  <mamiller@void.wustl.edu>_ 
>  _>    URL : [3]http://wugrav.wustl.edu/Util/projects/ELLIPTIC/00022/_ 
>  _>  _ 
>  _>  Cross-Posted to: NS_ELLIPTIC,ELLIPTIC_ 
>   
>  References 
>   
>     1. file://localhost/home2/web/Util/projects/trK_ns.gif 
>     2. file://localhost/home2/web/Util/projects/trK_misner.gif 
>     3. http://wugrav.wustl.edu/Util/projects/ELLIPTIC/00022/ 
>  _______________________________________________________________________ 
>   
>  CoCoBoard Information: 
>    From: John Towns NCSA Cog  <jtowns@ncsa.uiuc.edu> 
>    URL : http://wugrav.wustl.edu/Util/projects/ELLIPTIC/00023/ 
>   
>   

Created for the WUGRAV CoCoBoard Page Projects Page.

Created by The CoCoBoard.