Quick start

The OpenFF Recharge framework aims to offer a comprehensive solution for refitting different charge models to QC data, especially electrostatic potential (ESP) and electric field data.

Tip

For more information about the types of charge model available in the framework, as well as the underlying theory that backs this framework, see the theory chapter.

The recommended way to install openff-recharge is via the conda package manager:

conda install -c conda-forge openff-recharge

although several other methods are available.

The fastest way to learn the framework is to look over the examples directory on the frameworks website. Included are examples of how to

  • generate QC electrostatic potential and electric field data using the Psi4 package and store that data in a local SQLite database for future use

  • retrieving already available QC calculations from the QCArchive and reconstructing the ESP and electric field from the stored wave functions

as well as

  • training a new set of bond charge correction (BCC) parameters on to ESP data using numpy

  • simultaneously training both BCC and virtual site (v-site) parameters to ESP data using pytorch

  • performing Bayesian sampling of v-site parameters using pyro