Possible bug in pair_style gayberne

Hi there,
I found that the pair_style gayberne does not conserve the energy if one or both the exponents mu and upsilon are negative. I am sending a minimal example reproducing the incorrect behaviour on the latest stable release, as well as in version 11Aug17.

Best,
Otello

gb_bug.in (1.28 KB)

Possible bugs are best reported in the issue tracker of the lammps project on GitHub: https://github.com/lammps/lammps/issues

Otherwise they may get lost when nobody has time to work on them right away.

Thanks, Axel.