Amp: Atomistic Machine-learning Package

Amp is an open-source package designed to easily bring machine-learning to atomistic calculations. This project is being developed at Brown University in the School of Engineering, primarily by Andrew Peterson and Alireza Khorshidi, and is released under the GNU General Public License.

The latest stable release of Amp is version 1.0.1, released on January 25, 2023; see the Release notes page for a download link. Please see the project’s git repository for the latest development version or a place to report an issue.

You can read about Amp in the below paper; if you find this project useful, we would appreciate if you cite this work:

Khorshidi & Peterson, “Amp: A modular approach to machine learning in atomistic simulations”, Computer Physics Communications 207:310-324, 2016. DOI:10.1016/j.cpc.2016.05.010

An amp-users mailing list exists for general discussions about the use and development of Amp. You can subscribe via listserv at:

https://listserv.brown.edu/?SUBED1=AMP-USERS&A=1

Manual:

• Introduction
• Installation
• Using Amp
• Community
• Theory
• Credits
• Release notes
• Example scripts
• Analysis
• Building modules
• More on descriptors
• More on models
• Gaussian descriptor
• TensorFlow
• Bootstrap statistics
• Nearsighted force training
• Electronically grand-canonical learning
• Fingerprint databases
• Fast Force Calls
• Development

Module autodocumentation:

• Main
• Descriptor
• Model
• Regression
• Utilities
• Analysis
• Stats
• Convert
• Preprocess
• Nft

Indices and tables

• Index

• Module Index

• Search Page