I think that the results from mdgx actually come out all right even if I
don't try to catch the zeros and nullify the scaling they'd technically
create. But, I don't like nan's running around in my code, and mdgx is
NOT based on sander so the internal data structures and the way it handles
everything isn't the same. I've tried to write things so that mdgx and
sander will produce equivalent results, and if there's a case where they
are not I will try to make the behavior similar to the extent that sander
can accomplish something a certain way. A lot of the time the mdgx
solution is probably convergent with the sander solution, but that's not a
result of deliberate imitation.
Dave
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Aug 12 2011 - 14:00:02 PDT
