[CIG-LONG] [Gale] EulerDeform: coordinate out of bounds when increasing model resolution

[Guillaume.Duclaux at csiro.au]

Hi Walter,

Thanks for your answer. I run a slightly different model (in terms of geometry), without fixing the timestep (dt=0, and dtFactor=1) and still having the same kind of error:


TimeStep = 66, Start time = 1.1986e+14 + 1.90721e+12 prev timeStep dt
Coordinate out of bounds 1867 594128.051477 108429.153138 240000.000000
Gale: build/Underworld/plugins/EulerDeform/EulerDeform.c:776: EulerDeform_InterpVar: Assertion `0' failed.
MPI: On host cognac, Program /short/cg01/duc025/gale_svn/bin/Gale, Rank 21, Process 19842 received signal SIGABRT/SIGIOT(6)


MPI: --------stack traceback-------
sh: idb: command not found

MPI: -----stack traceback ends-----
MPI: On host cognac, Program /short/cg01/duc025/gale_svn/bin/Gale, Rank 21, Process 19842: Dumping core on signal SIGABRT/SIGIOT(6) into directory /short/cg01/duc025/GSV_inputs/fromcognac
MPI: MPI_COMM_WORLD rank 21 has terminated without calling MPI_Finalize()
MPI: aborting job


Do these may help:

- inscreasing the number of particlesPerCell?
- changing the number of CPUs?

Cheers,

Guilly

Dr Guillaume Duclaux
Structural Geologist/Modeller
CSIRO Exploration and Mining
Visiting address: 6G13, ARRC, 26 Dick Perry Avenue, Kensington, Western Australia, 6151
Postal address: PO Box 1130, Bentley, Western Australia, 6102, Australia
Ph: + 61 8 6436 8728  Fax: + 61 8 6436 8555  Web: www.em.csiro.au


-----Original Message-----
From: cig-long-bounces at geodynamics.org [mailto:cig-long-bounces at geodynamics.org] On Behalf Of Walter Landry
Sent: Thursday, 13 November 2008 5:29 AM
To: CIG-LONG at geodynamics.org
Subject: Re: [CIG-LONG] [Gale] EulerDeform: coordinate out of bounds when increasing model resolution

<Guillaume.Duclaux at csiro.au> wrote:
> G'day all,
>
> I'm running compression models with similar geometries and
> mechanical properties at different resolution and a fixed timestep
> (dt=100 ka).
> An error with EulerDeform seems to occur when increasing the
> resolution. The higher the resolution, the earlier the error
> happens.

This is expected.  When a new mesh is created, Gale needs to
interpolate values onto the new mesh.  In your case, the mesh has
moved so far that it needs data on other processors which it does not
have.

So your time step is too big, and you need to reduce it.  With the
current timestep, your results are inaccurate anyway.  Note that the
timestep needs to get smaller for higher resolution grids.  The
simplest way is to let Gale choose the timestep for you.

Cheers,
Walter Landry
walter at geodynamics.org
_______________________________________________
CIG-LONG mailing list
CIG-LONG at geodynamics.org
http://geodynamics.org/cgi-bin/mailman/listinfo/cig-long