phast-forcefields

PHAST (Potentials with High Accuracy, Speed, and Transferability) provides general purpose small molecule forcefields with physically motivated functional forms. These forms are distinct from traditional Lennard-Jones based forcefields. Transferability is achieved by the default inclusion of explicit polarization and fitting parameters soley to electronic structure calculations.

Functional Forms

Repulsion-Dispersion Potential

The repulsion-dipsersion potential is a combination of exponential repulsion and the three leading dispersion coefficients, which are damped at short ranges to avoid singularities,

$$U_{rd} = \sum_{i \neq j} \frac{F_0}{\beta_{ij}} e^{\beta_{ij}(r_{ij}-\rho_{ij} )} + \sum_{n=3}^5 f_{2n}(\beta r_{ij}) \frac{C_{2n}}{r_{ij}^{2n}}$$

$$f_{2n}( \beta r_{ij} ) = 1 - e^{- \beta r_{ij}} \sum_{k=0}^{2n} \frac{(\beta r_{ij})^k}{k!}$$

Mixing rules for the dispersion coefficients are the usual geometric mean mixing rule from London and the repulsion mixing are from Smith 1972.

$$C_{2n, ij} = \sqrt{C_{2n, ii} * C_{2n, jj}}; n = 3, 4, 5$$

$$\rho_{ij} = \frac{1}{2} ( \rho_{ii} + \rho_{jj} )$$

$$\beta_{ij} = 2 \frac{\beta_{ii}\beta_{jj}}{\beta_{ii}+\beta_{jj}}$$

Explicit Polarization

Many-body polarization is included via the induced dipole method of Silberstein, Thole and Applequist. The induced dipoles are obtained by scaling the atomic electric field by the atom's polarizability.

$$\vec{\mu}_{i} = \alpha_i \left( \vec{E}_i + \vec{E}^\prime_i \right)$$

Each induced dipole then induces an electric field on every other induced dipole, making the model many-body.

$$\mu_{i} = \alpha_i \left( E_i - \sum_{j \neq i} T_{ij} \mu_j \right)$$

Installation

git clone https://github.com/space-group-research/phast-forcefields.git
cd phast-forcefields/devtools/conda-envs
conda env create --file test_env.yaml
conda activate phastff-test
cd ../..
python3 -m pip install .

Usage

from openff.interchange import Interchange
from openff.toolkit import Molecule, ForceField
from openff.units import unit
from openff.units.openmm import to_openmm, from_openmm
import numpy as np
import openmm
import sys
ff = ForceField('PHAST-H2CNO-2.0.0.offxml', load_plugins=True)
mol = Molecule.from_smiles('CCC')
mol.generate_conformers()
cubic_box = unit.Quantity(30 * np.eye(3), unit.angstrom)
interchange = Interchange.from_smirnoff(topology=[mol], force_field=ff, box=cubic_box)
openmm_sys = interchange.to_openmm(combine_nonbonded_forces=False)
openmm_top = interchange.topology.to_openmm()
openmm_positions = interchange.positions.to_openmm()
openmm_integrator = openmm.LangevinMiddleIntegrator(
    to_openmm(300*unit.kelvin),
    to_openmm(1/unit.picosecond),
    to_openmm(0.002*unit.picoseconds)
)
simulation = openmm.app.Simulation(openmm_top, openmm_sys, openmm_integrator)
simulation.context.setPositions(openmm_positions)
simulation.minimizeEnergy()
simulation.reporters.append(openmm.app.PDBReporter('output.pdb', 1000))
simulation.reporters.append(openmm.app.StateDataReporter(sys.stdout, 1000, step=True,
        potentialEnergy=True, temperature=True))
simulation.step(10000)

Limitations

Currently, a water model is not provided, this is a work in progess. OpenFF Interchange does not support export to any simulation software other than OpenMM due to the custom repulsion-dispersion potential. AmoebaMultipoleForce in OpenMM is being used behind the scenes, however this choice is not ideal due to some particular choices made early in AMOEBA's development.

Publications

The PHAST 2.0 Force Field for General Small Molecule and Materials Simulations Adam Hogan, Logan Ritter, and Brian Space in press

PHAHST Potential: Modeling Sorption in a Dispersion-Dominated Environment Logan Ritter, Brant Tudor, Adam Hogan, Tony Pham, and Brian Space Journal of Chemical Theory and Computation 2024 20 (13), 5570-5582 DOI: 10.1021/acs.jctc.4c00226

Next-Generation Accurate, Transferable, and Polarizable Potentials for Material Simulations Adam Hogan and Brian Space Journal of Chemical Theory and Computation 2020 16 (12), 7632-7644 DOI: 10.1021/acs.jctc.0c00837

History

Version 2.1.0

WIP

Version 2.0.0

Published in JCTC soon. Note that a water model is not included in this version of the forcefield.

• PHAST-H2CNO-2.0.0.offxml - Recommended for general use, mostly element-typed forcefield with 2 atom types for hydrogen
• PHAST-H2CNO-direct-2.0.0.offxml - Direct polarization version of the above, not included in published benchmarking
• PHAST-H2CNO-nonpolar-2.0.0.offxml - Nonpolar version of the above, only recommended for systems lacking significant permanent polarity
• PHAST-HCNO-2.0.0.offxml - Element-typed forcefield
• PHAST-HC4NO-2.0.0.offxml - Forcefield with 4 different carbon atom types (sp3, sp2, sp and aromatic sp2)
• PHAST-HC2NO-2.0.0.offxml - Forcefield with 2 different carbon atom types (aromatic and nonaromatic)
• PHAST-H2C4NO-2.0.0.offxml - Forcefield with 4 different carbon atom types and 2 hydrogen atom types
• PHAST-H2CNO-ecut-1000-2.0.0.offxml - Hyperparameter testing forcefield (not recommended)
• PHAST-H2CNO-ecut-100-2.0.0.offxml - Hyperparameter testing forcefield (not recommended)

General versioning guidelines

Force fields moving forward will be called name-X.Y.Z

• X denotes some major change in functional form or fitting strategy.
• Y is the parameterization epoch / generation, or a minor change that can affect energy.
• Z is a bugfix version -- e.g. something we've caught and corrected.

Copyright

Copyright (c) 2023-2025 Adam Hogan