Hello Peter,
I've experienced a bug in new release ET_2012_05. After first regridding NaNs appear in gravitational fields. I have GRhydro at revision 374. Does anyone experienced the same problems?
Not this precise problem but others have also found odd values after regridding as well. GRHydro/trunk is currently very much in development, it will hopefully settle down a bit when we get closer to the next release.
Are your other thorns on release version or also from trunk (or the git/hg equivalent of the "newest" version)?
There is a current issue with HydroBase and Carpet where changes were made to HydroBase and Carpet that can cause memory corruption (both by me). This requires that you have both HydroBase r55 and Carpet 55759e108807 (see https://trac.einsteintoolkit.org/ticket/971). I don't know of a good fix for this yet, but a workaround would be to
* schedule Hydro_Base::InitiExcision_Mask with "OPTIONS: global loop-local" * add additional schedule items to schedule Hydro_Base::InitiExcision_Mask in local mode (no options) BEFORE HydroBase_ExcisionMaskSetup AT PostRegrid and PostRegridInitial
The strange scheduling requirements are due to interactions between SphericalSurface::SphericalSurface_Set, SetMask_SphericalSurface and HydroBase.
The cleanest solution might be (baring unforseen complications) to make SphericalSurface::SphericalSurface_Set global-early then everything else can be local mode and things are fine, though that mean that all routines that set spherical surfaces should be global-early as well (NB they are currently global-late in POSTSTEP since global==global-late there but are global-early in BASEGRID).
Yours, Roland