[aspect-devel] ASPECT with PETSc

Wolfgang Bangerth bangerth at math.tamu.edu
Fri Mar 9 05:11:37 PST 2012


Hi Matt,
so it gets archived, let me write this to the mailing list right away.

In order to compile and run ASPECT with PETSc, you'll need to do the 
following:

1/ Apply the attached patch that switches us from using Trilinos 
explicitly to using typedefs that are defined in one central place, 
namely include/aspect/global.h.

[Timo: Question -- do we want this anyway? If so, please apply to the 
repository since I will be offline for the next couple of days.]

2/ To switch to PETSc, simply redefine these typedefs to the 
corresponding PETScWrappers.

3/ You will likely get a number of compile errors that need to be 
adjusted one-by-one.

4/ There are also 4 places left in the code where we still explicitly 
use the TrilinosWrappers:: classes -- workaround needed, in particular 
in the place where we initialize the Trilinos AMG (ML) as this needs to 
be switched to PETSc hypre.

That's the first step.

The second part you talked about, namely not building the matrix as 
individual blocks but instead as one matrix from which you let PETSc 
pick and choose is likely going to be much more difficult. I'll try to 
explain how this currently works when we meet in 20 minutes down at the 
AGU building. For now I've gotta run :-)

Cheers
  W.

-- 
------------------------------------------------------------------------
Wolfgang Bangerth               email:            bangerth at math.tamu.edu
                                 www: http://www.math.tamu.edu/~bangerth/

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: diffs
Url: http://geodynamics.org/pipermail/aspect-devel/attachments/20120309/35146fa0/attachment-0001.txt 


More information about the Aspect-devel mailing list