Open mattwthompson opened 1 year ago
Something's pretty significantly off here.
Can you reprint what the Amber mdout provides? Amber differentiates between 1:4 vdW / electrostatics and "the rest" of nonbonded vdW / electrostatics. From the energies you reported, there is probably an interaction that should be 1:2 excluded masquerading as a 1:4 scaled non-bonded pair.
I can't reproduce this anymore, but I'm pretty sure I've seen similar behavior in other systems. I'll try to remember to repost them here.
Something is wrong with the interplay between setting the third atom negative (its 1-4 interaction was already accounted for) and the fourth atom negative (this torsion is an improper). It makes no sense that they'd ever both be negative but I also can't reason through what it should be.
Based on b29ed479acfbfc0df7cd733304b3807f67ae0998 in #685:
The file (prior to popping the impropers):