Re: [AMBER] Interesting findings regarding water density, ntf values, and total potential energy

From: David A Case <david.case.rutgers.edu>
Date: Tue, 23 Nov 2021 15:07:15 -0500

On Wed, Nov 24, 2021, Liao wrote:

>Here is the mdin file, and the program is pmemd.cuda. In the 3 scenarios,
>ntf=1,2 or 8 is being changed. When ntf=2 or 8, shake must be turned on. If
>ntf=1 also has SHAKE turned on, the results are identical to ntf=2.

OK: the results look a little less strange now. As I understand it:

     ntf=1, ntc=2 is the same as ntf=2, ntc=2 (as it should be)

     ntf=1, ntc=1 is a very different model, since the water is now
        flexible. This can't really be called "TIP3P" any more, and one
        would expect all kinds of properties, including density, to be
        different

     ntf=8, ntc=2: I've never used this, and the extrememly abbreviated
        explanation in the manual of what ntf=8 means doesn't help much.
        (Does this really turn off all nonbonded interactions?) This option
        was put in for use with the debug facility in sander. From a
        brief look at the pme_force.F90 code, it looks like all forces are
        set to zero, i.e. that this is nearly useless for MD. Others may
        want to chime in here. Have you visualized what the final state
        looks like?

As for the density you get for ntc=2, ntf=1,2: again the experience of
the list may help here. I'm sure Amber has been used to simulation pure
TIP3P water. Izadi et al, J Phys Chem Lett 5:3863, 2104 report a density
of 0.980 for TIP3P water in Amber. (I've lost your original email that had
the ntc=2,ntf=2 result.)

TIP3P was originally parameterized with a non-bonded cutoff, so it's not
a very good water model to use for PME simulations. An 8 Lennard-Jones
cutoff is also a bit short -- you could see if increasing that a bit makes
a difference. 50 ps is a pretty short simulation time, unless you have
extensively equilibrated your boxes before starting these run.

But as I said, there are "water experts" lurking here.

....dac


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Nov 23 2021 - 12:30:02 PST
Custom Search