From 6461ea355f2c655c36259b617f8fe9a123c60834 Mon Sep 17 00:00:00 2001 From: John Gardner Date: Sat, 11 Jan 2025 08:21:41 +0000 Subject: [PATCH] summary --- paper/paper.md | 19 +++++++++++++++---- 1 file changed, 15 insertions(+), 4 deletions(-) diff --git a/paper/paper.md b/paper/paper.md index ca6c1107..fabf211f 100644 --- a/paper/paper.md +++ b/paper/paper.md @@ -13,19 +13,26 @@ authors: orcid: 0000-0001-6873-0278 affiliation: 1 affiliations: - - name: Inorganic Chemistry Deparment, University of Oxford + - name: Department of Chemistry, Inorganic Chemistry Laboratory, University of Oxford, Oxford OX1 3QR, United Kingdom index: 1 date: 10/01/2025 bibliography: paper.bib + --- # Summary -TODO +`graph-pes` is an open-source toolkit that accelerates the development, training and deployment of machine-learned potential energy surfaces (ML-PESs) that act on graph-based representations of atomic structures. The `graph-pes` toolkit comprises three components: + +1. **the `graph_pes` Python package**: a modular Python library containing all the functionality required to build and train graph-based ML-PES models. This includes a mature data pipeline for converting atomic structures into graph representations, a fully featured ML-PES base class with automatic force and stress inferences, and a suite of common data manipulation and model building blocks. + +2. **the `graph-pes-train` command-line interface**: a unified tool for training graph-based ML-PES models on datasets of labelled atomic structures. Several popular model architectures are provided out of the box, but the interface is designed so that custom, end-user defined architectures can be used (alongside custom loss functions, optimizers, datasets, etc.) + +3. **the `pair style graph_pes` LAMMPS integration**: a pair style for using LAMMPS to perform GPU-accelerated molecular dynamics simulations using any graph-based ML-PES model defined and trained using the `graph_pes` package. + # Statement of need -TODO # Related work @@ -37,6 +44,8 @@ The core functionality of `graph-pes` builds upon - `ase` [@HjorthLarsen-17-06] +- `LAMMPS` + `graph-pes` also builds upon the `e3nn` [@Geiger-22-07] package for implementing the `NequIP` [@Batzner-22-05] and `MACE` [@Batatia-23-01] architectures. Other note-worthy softwares that offer similar functionality to `graph-pes` include: @@ -45,14 +54,16 @@ Other note-worthy softwares that offer similar functionality to `graph-pes` incl - `mace-torch` -- `deepmd` +- `deepmd-kit` - `schnetpack` - `torchmd` + # Acknowledgements Krystian Gierczak, Daniel Thomas du Toit and Zoé Faure Beaulieu for early testing and feedback. + # References \ No newline at end of file