AP-Net

AP-Net is a python package for modeling intermolecular interactions with atomic-pairwise neural networks. The AP-Net architecture provides smooth and asymptotically correct interaction potentials.

Getting Started

Environment

It is strongly recommended that you install this package in some kind of virtual environment (like a conda environment). Create and activate a new conda environment with the following commands:

>>> conda create --name apnet python=3.8
>>> conda activate apnet

Installation

Next, clone this repository and cd into the top level of the repository (the same level as this README). Run the following command to install the apnet package (and dependencies) into your current environment:

>>> pip install -e .

This will take a few minutes.

Testing

You can now call import apnet from any python script. (There is no further need to work out of this repository). Verify that the installation was succesful by running the following snippet of code, which predicts the interaction energy of a water dimer:

import apnet
import qcelemental as qcel

dimer = qcel.models.Molecule.from_data("""
0 1
O 0.000000 0.000000  0.000000
H 0.758602 0.000000  0.504284
H 0.260455 0.000000 -0.872893
--
0 1
O 3.000000 0.500000  0.000000
H 3.758602 0.500000  0.504284
H 3.260455 0.500000 -0.872893
""")

prediction, uncertainty = apnet.predict_sapt(dimer)
print(prediction)

The output should look like this:

[array([-2.87016274, -3.77428398,  2.41356459, -0.50601828, -1.00342507])]

AP-Net predicts an interaction energy of -2.87 kcal / mol. The predicted SAPT depcomposition is -3.77 (electrostatics), +2.41 (exchange), -0.51 (induction) and -1.00 (dispersion).

For reference, the actual SAPT interaction energy of this dimer is pretty close: -2.66 kcal / mol. The actual SAPT decomposition is also in good agreement: -3.46 (electrostatics), +2.31 (exchange), -0.53 (induction), and -0.97 (dispersion).

Using the Code

The docs/ directory of this repository contains detailed documentation for using this package. In the near future, this documentation will be hosted online for easy reference. For now, you can navigate this local copy of the documentation by opening docs/build/html/index.html in a web browser. (Unforunately, GitHub doesn't render these documentation pages, you have to download them to your machine in order to view them).

Note that the installation instructions in docs/ are different from the instructions in this README. The docs contain install instructions that will take effect when the apnet package is made public. For now, stick to the installation instructions in this README.

Common Errors

A message such as "NotImplementedError: Cannot convert a symbolic Tensor to a numpy array. This error may indicate that you're trying to pass a Tensor to a NumPy call, which is not supported" can occur for incompatible versions of tensorflow and numpy. Make sure to avoid changing the version of numpy in your environment after installation.