Skip to main content

Software & Data / Machine learning

PACMAN.

PACMAN: crystal to partial atomic charges via graph convolution network

Accurate DFT-quality partial atomic charges for MOFs and COFs. PACMAN is a fast and easy application that predicts density-derived electrostatic and chemical (DDEC06), Charge Model 5 (CM5), Bader, and REPEAT partial atomic charges based on a crystal graph convolution neural network (CGCNN) model.

Get started

Input
Periodic crystal structures in CIF format.
Output
Predicted partial atomic charges; choose a supported charge model for your material.
Environment
Python package; command-line and notebook examples are provided.
First step
Install PACMAN-charge and start with the supplied CIF example.
How to cite

01

G. Zhao, Y.G. Chung, "PACMAN: A Robust Partial Atomic Charge Predicter for Nanoporous Materials Based on Crystal Graph Convolution Networks," Journal of Chemical Theory and Computation, 20, 12, 5352–5367 (2024)