[AMBER] Potential on QM atoms - details about cutoff

From: Elisa Pieri <elisa.pieri90.gmail.com>
Date: Fri, 10 Mar 2017 16:13:19 +0100

Hello everyone,

I need to get the electrostatic potential acting on each QM atom in my
system, so I built a simple Fortran code based on the Sander API and I've
applied it to a minimalistic model of two water molecules in a very large
cuboid box ( l =100 angstroms) for testing.

Hence I could also calculate the exact electrostatic potential created by
one molecule (MM) onto the other one (QM), including the very
small contribution of the images (for both molecules).

When qm_inp%qmcut is lower than 11 angstroms, the Sander values of the
electrostatic potential (mmpot and qmpot, the latter being negligible in
that example) matches quite well the exact values, at least in magnitude
and sign (I discovered that the minimal difference is due mainly to
conversion factors). However, when qmcut > 11 angstroms, it is no longer
the case, see the example below.

Exact pot on O : -0.0014143 au

mmpot on O (qmcut = 10 angstroms and lower) : -0.0014032 au

mmpot on O (qmcut = 11 angstroms) : 0.0180207 au

mmpot on O (qmcut = 12 angstroms and greater) : 0.000007 au

Given the large size of the box and the distance between the two water
molecules, I would have not expected such a dependence to the qmcut value !
I don't see any clear explanation to this behavior, do you? Could you tell
me a little bit more on what is going on here?

I'm attaching the files I'm using, so you will be able to reproduce the
results if you want.

Thank you for your time,

Elisa


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

Received on Fri Mar 10 2017 - 07:30:02 PST
Custom Search