On Wed, Jul 20, 2011 at 6:32 PM, Roland Haas roland.haas@physics.gatech.edu wrote:
Hello Luca, Ian,
Is any of you using CarpetRegrid (as opposed to CarpetRegrid2)? Is CarpetRegrid still maintained (and tested) in the new versions of Carpet?)
I used to use CarpetRegrid at Penn State because it allowed me to set the grid structure that I wanted explicitly in a much easier manner than using Carpet's bbox classes etc. It's possible that the GT group still use it. At AEI we use CarpetRegrid2.
We do still use CarpetRegrid at GT but do not maintain it (since it is part of Carpet). Since CarpetRegrid2 can do (almost) everything that we need of CarpetRegrid we will likely transition to Regrid2 when we transition to Carpet/Hg (though Tanja will have to decide on this). At that point we will likely start working around things that CarpetRegrid did and Regrid2 does no longer do (eg. easy steering of parameters when recovering). On the other hand having full and easy to use AMR in Regrid2 is very likely worth any associated pain of transitioning to a new thorn :-)\
If there is any missing functionality in CR2, we should implement it.
You cannot steer CR2's parameters upon recovery, since these parameters are ignored. However, there are grid scalars you can change instead, e.g. in the schedule group post_recover_variables. The values of these grid scalars are used at the next regridding. You would inherit from CR2 to get access to these grid scalars.
-erik