[XtalOpt-devel] New cartesian tolerances in XtalOpt, new tolerance recommendations

[David Lonie]

Hi all,

I just pushed a change that modifies xtalopt to use cartesian
tolerances instead of fractional tolerances. This should improve
spacegroup detection in cell that have "long" dimensions, e.g. 2x2x10.

This adds a new option in the "Optimization Settings" tab that allows
the user to specify the spacegroup tolerance. I've experimented a bit
with the tolerance values, and I've found the following to work quite
well with the new duplicate matching method:

enthalpy=0.005
volume=1.0
spacegroup=0.05

This is a much higher volume setting than previously recommended,
since inspection of several runs using our VASP optimization scheme
shows that volumes vary more than we previously thought. Please update
your code and tolerance settings to those recommended above.

Thanks,

Dave