We need a set of options for black hole which:
1)reduces HPF to essentially zero. Continues
library work which I assume is low level of effort.
2)includes runtime work as necessary (clustering)
3)includes options for BBH work in WebFlow as
integration and Matt's language R... done as
webFlow scripted little language
4)Includes option of leaving collaboration
if activities 2) and 3) not accepted
In this case need orderly termination plan.
Please come with preliminary ideas/plans/levels
of effort.
Thanks

PS I consider it unwise to encourage David Rideout
that there is money -- even now BBH doesn't balance
with you two (partially) and Miller. I am sure
in best of circumstances, our budget will be cut



1. HPF activities. 
   We need 3 months to wrap things up. This includes working, limited
   implementation of AMR T2, posted on the Web, and documented as an
   SCCS-document. Limitations comes from the current limitations of the 
   HPF compilers. The most severe is lack of support for pointers to 
   distributed arrays that makes implementation of the tree stucture
   of meshes difficult. Result: limitation of number of grids than can 
   be handled. David Rideout can continue to work on this (funded/unfunded),
   as this can be part of his research.  Minimal support will be needed
   by us to help support David.  This means more work for David, which
   probably will be unfunded work.  His work can then go into WebAMR before
   he defends.

2. Library work.  
   By this I mean the runtime support for AMR/MLAT. Options:
    a) library of algorithms 
    b) library of codes available within the Alliance and outside
    c) reference implementations, such as interpolation and clustering
       libraries developed by Scott
    d) implementations for DAGH
    e) implementation for HPF
    f) Multigrid solvers for HPF and DAGH

   2a useful always
   2a,2b,2c  necessary for WebFlow version of AMR
   2a,2b,2c,2d usefull for DAGH (if funded)
   2e probably irrelevant in this context, necessary if any follow on
      activities start (No use to the GC community if AMR HPF is unfunded)
   2f Possible uses for WebAMR, DAGH, HPF version ...

-------------------------------------
3. BBH work in WebFlow
   1. AMR driver (scripting language)
   2. WebFlow Module generator
   3. PSE 
-------------------------------------
3. RNPL work
   a) Completely a waste of time!  I polled Texas, Cornell, NCSA, and 
      no one wants to use RNPL, and they will not use it.  For our 
      purposes to put into the Web, its a waste of effort! Has no use
      for us.  A better idea is to build drivers to include Maple input.
      In our methodology RNPL will not fit. 
   b) If we have to use this, then we will have to rewrite the whole thing,
      since it doesn't support AMR/DAGH/HPF/WebAMR.


4. Leaving collaboration.
   - HPF should be completed, as specified in 1
   - Library work contiuned as long as funded.
   - Internally: transition from BBH activities to WebVM (how funded?)
   - Charging scheme for CS involvement for GC