[yt-users] Change in behavior of smoothed covering grids and ghost zones

Matthew Turk matthewturk at gmail.com
Wed May 26 22:51:40 PDT 2010


Hi everyone,

After a bit of discussion on and off the yt-dev list, the mechanism
for interpolation in smoothed covering grids and vertex centered data
has changed in r1734.  Previously, it would interpolate logged fields
in log-space -- so, for instance, Density would be interpolated in log
space between adjoining cells, rather than in linear space.  At the
end of the interpolation it was restored.

This may result in small changes in results for a small set of
analysis types, but I believe that this change -- to interpolating
exclusively in linear space in the unmodified fields -- is the correct
change to make.  To restore the old behavior, a derived field can be
created that operates by returning only the logged fields, and then
requesting smoothed or vertex centered data in that field.

Best,

Matt



More information about the yt-users mailing list