AMOEBA water model: best practice for calculating the total dipole moment

[rowanwalker96]

Hello,

I've been using OpenMM to run simulations of the AMOEBA14 water model in a capacitor-type set up and would like to calculate a dipole moment profile across the simulation box.

There are three contributions: monopole charge weighted relative positions from a molecular reference (q*r), lab frame permanent dipoles and induced dipoles, however I am a little confused as to which molecular reference point I should be using to bin the location of the resulting dipole vector.

I came across a published study (https://www.biorxiv.org/content/10.1101/2020.03.01.971853v1.full) which uses the center of geometry of the water molecules to calculate the charge weighted positions from. They then bin the vector sum of all three contributions on the center of geometry (their code is here: https://github.com/Inniag/openmm-scripts-amoeba/blob/master/scripts/postproc/dipole_profile_amoeba.py).

Ideally, should the monopole charge weighted positions term not be calculated using the oxygen atom and binned at the location of the oxygen atom, and the induced/lab frame permanent dipoles also be binned at their respective atomic centers?

This is a problem for me since I am calculating the dipole moment density across my simulation box and integrating from the bulk to one of the walls to get the potential there. By changing the molecular reference point on which the dipole moment sits I get quite different results for the final potential at the capacitor walls.

Thanks for reading!

[peastman]

No idea! This seems like something @jayponder might have insight on.

[jayponder]

If I understand your question (which I may not...), and from looking at the Sansom paper you cited, I think you want to use the molecular center of mass as the reference point (presumably this is what Sansom, et al. mean by "center of geometry" in their paper and code).

[rowanwalker96]

Thanks so much for getting back to me! So do you think that if I calculate each water molecule's individual dipole moment, made up of the three contributions, and then bin the z component of this dipole moment on the centre of mass of each water molecule, this would be the correct approach in creating a profile of the dipole moment density P(z)?

[rowanwalker96]

I find for example when studying AMOEBA water confined between two LJ walls at z=1nm and z=5nm, that I recover the dipole moment profile for SPC water (red line) only when I bin the monopole charged weighted positions term (Σqr) on the oxygen atom and all the induced and permanent dipole vectors on the center of geometry of the water molecule. Is this just a coincidence or the most accurate reflection of the dipole moment profile for the AMOEBA model? Small variations in the binning location of the dipole moment vectors (e.g, all vectors binned on the center of geometry) unfortunately leads to different results for the dipole moment profile and hence derived properties such as the surface potential...