Re: [AMBER] Negative DIHEDRAL_PERIODICITY in prmtop for amber20-benchmark FactorIX benchmark

From: David A Case <>
Date: Wed, 26 Jan 2022 15:34:53 -0500

On Wed, Jan 26, 2022, David Case wrote:

>4. My current bottom line is that the *timings* one gets from the factor_ix
>benchmark (whcih only runs a few hundred steps) are probably fine, but that
>no one should rely on the actual energies.

OK, as is common, I spoke too soon. Both the timings and the energies
for factor_ix are probably fine. The way the code is written, one can't
just swap the phase for its absolute value, even though cosine is an even

We *should* check for negative values, however. What Amber does when it
finds a negative periodicity in the prmtop file is not well commented,
and never tested (per John's report about the prmtop files in the test
suite.) And conversion of Amber prmtop files to other codes (gromacs,
CHARMM, OpenMM, Desmond...) would need to be needlessly complex if we tried
to continue to support the (very) old option of allowing negative values.


AMBER mailing list
Received on Wed Jan 26 2022 - 13:00:02 PST
Custom Search