[CIG-LONG] Gale general questions

Walter Landry walter at geodynamics.org
Mon Apr 23 08:51:54 PDT 2012


Edie Miglio <edie.miglio at polimi.it> wrote:
> Hi Walter,
>     we have some general questions:
> 1) how do you get images like those in Figure C31 in the
> manual of Gale version 1.6.1 ?

When you open an xdmf file In Paraview, scroll down before hitting
"Apply".  There will be a section "Blocks".  Uncheck the v-mesh and
pressure-mesh, leaving only the materialSwarm files.  Hit "Apply",
then select materialSwarm-MaterialIndex.

> 2) do you have a more detailed description (or even better
> an example file) about the usage of FaultingMoresiMulhaus2006
> rheology ?

No, I have never really used it.

> 3) we have followed your suggestion concerning the use of
> a parallel (MUMPS) direct solver for a problem involving strong
> viscosity gradient. In your experience which the highest resolution
> you have been able solve for using direct method ? On how many
> processors ? We are having (apparently) strange "Segmentation
> fault" errors.

It sounds like you are running out of memory.  The direct solver will
really only work in 2D.  In 3D the memory requirements quickly become
prohibitive.

For the current version, 2.0.0, the largest runs I have done in 2D
with Mumps is 1024x128, and I think I used 64 processors.

Cheers,
Walter Landry


More information about the CIG-LONG mailing list