Avoid NaN in rlist buffer calculation
authorBerk Hess <hess@kth.se>
Thu, 16 Oct 2014 12:03:49 +0000 (14:03 +0200)
committerMark Abraham <mark.j.abraham@gmail.com>
Mon, 24 Nov 2014 16:40:53 +0000 (17:40 +0100)
commit5f5e94ad571b8e04081182af6e23c6960370fdf0
treeeb9c56a289277c28cb9b92b552c32a7e83e0179d
parentac83c85a9943c8215382652655d32af0bca94b7a
Avoid NaN in rlist buffer calculation

With constraints on large masses and very low tolerance, the Verlet
buffer calculation could produce 1/0. This could lead to a somewhat
too smaller buffer.
Also added a missing scaling factor to the contribution of the third
derivative of the potential. This issue only caused a minor
overestimate for systems with constraints and without electrostatics.

Change-Id: I97e9d428a83f1b4954012ebd39bc49d397574f8c
src/gromacs/gmxpreprocess/calc_verletbuf.c