Discontinuity of energy in reaxFF

Thank you.

If I run into any other issue, should I post it on the mailing list or directly on the github?

Michal

Thank you.

If I run into any other issue, should I post it on the mailing list or
directly on the github?

either is ok. we've just started the process to use github more
consistently for tracking and managing (reproducible) bugs and
suggestions/discussions of new features. we want to integrate this
more systematically into the whole development process, so that by the
time changes get included into the official version, they've already
been reviewed and tested.

at the moment, going directly to github will save me a little effort,
but until all core developers are systematically monitoring it, you
will get a little less visibility.
posting to the mailing list will cause (me) a little more effort, but
will get you more visibility (for now). the main disadvantage of
posting to the mailing list is that sometimes such bug reports get
overlooked and forgotten. that is why i now copy them to github, when
i see one.

so the short answer is: your choice. :wink:

thanks,
     axel.

I took a look at this. It is not something that is in the original ReaxFF Fortran code, nor in the PureMD stand-alone code from which it is derived. It seems to have been added in this revision:

r11904 | sjplimp | 2014-05-02 15:38:42 -0600 (Fri, 02 May 2014) | 1 line

Somebody thought it was necessary to add the numbonds>0 condition, but I can’t say why.

I will copy Ray on this, as I suspect he may have made the change.

Aidan