Change in behavior of smoothed covering grids and ghost zones
Hi all, Dave and I were chatting a couple weeks ago, and it became clear to me that yt currently does something incorrectly with ghost zones and smoothed covering grids. As it stands, in the smoothed covering grids and the vertex centered data, the fields that are interpolated between points are logged, then interpolated, then 10^values is returned. This is not how Enzo does it, and I think yt should change too. This may change some user-facing behavior, so I wanted to clear it with everyone here first. This would change primarily the generation of ghost zones, which are used for some finite stencils and for the volume rendering. (However, I have been unable in my tests to see any changes in the volume rendering.) What do you all think? -Matt
Hi all,
and the vertex centered data, the fields that are interpolated between points are logged, then interpolated, then 10^values is returned. This is not how Enzo does it, and I think yt should change too. I am of the opinion that yt should mirror Enzo here. I would like derivatives taken in yt to be exactly the same as those in whatever hydro code the data came from. Any conserved quantity involving differentials of primitive variables (like vorticity in Eulerian hydrodynamics) should be the same in yt as in Enzo. I think that this is worthwhile, even if some user results change.
j
Hi everyone,
I agree. Interpolation should be done in linear space. If for some reason
a user wants the interpolation to be done in log space it can always be done
on a derived field.
Sam
On Wed, May 26, 2010 at 8:29 PM, j s oishi
Hi all,
and the vertex centered data, the fields that are interpolated between points are logged, then interpolated, then 10^values is returned. This is not how Enzo does it, and I think yt should change too. I am of the opinion that yt should mirror Enzo here. I would like derivatives taken in yt to be exactly the same as those in whatever hydro code the data came from. Any conserved quantity involving differentials of primitive variables (like vorticity in Eulerian hydrodynamics) should be the same in yt as in Enzo. I think that this is worthwhile, even if some user results change.
j _______________________________________________ Yt-dev mailing list Yt-dev@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-dev-spacepope.org
-- Samuel W. Skillman DOE Computational Science Graduate Fellow Center for Astrophysics and Space Astronomy University of Colorado at Boulder samuel.skillman[at]colorado.edu
participants (3)
-
j s oishi
-
Matthew Turk
-
Sam Skillman