Constructing Accurate Quantitative Structure-Property Relationships via Materials Graph Networks
Category
Published on
Abstract
This tutorial covers materials graph networks for modeling crystal and molecular properties. We will introduce the graph representation of crystals and molecules and how the convolutional operations are carried out on the materials graphs. Then we will introduce the megnet package for constructing the graph network models, setting up the model for training and the model evaluation. We will show examples of graph models using computational data from the Materials Project database.
The nanoHUB tool "megnet: Materials Graph Network" used in this hands-on tutorial.
Sponsored by
Cite this work
Researchers should cite this work as follows: