Hi there! This is the github page of the COSMO lab. In this repository you can find the software developed by our group, collaborators and friends.
We publish our code because we believe in open science and want the community to both benefit from our work and contribute to it.
Clone, enjoy... and cite
Besides the repositories we pin below we develop the following projects.
metatrain: a command line tool to train atomistic machine learning model for any kind of target
metatomic: a standard interface to load and execute ML models in various simulation engines
metatensor: a data storage and exchange format created for atomistic machine learning
i-pi a universal force engine designed to be used together with an ab-initio, empirical or machine-learning force field (or a combination of these!) to run molecular dynamics simulations, including nuclear quantum effects.
scikit-matter: a collection of scikit-learn compatible utilities that implement methods born out of the materials science and chemistry communities.
featomic: a library to compute representations for atomistic machine learning
atomistic-cookbook an online, interactive collection of "recipes" to perform different kinds of atomistic simulations, from path integral dynamics to the training of a ML model
kernel-tutorials a set of utilities and pedagogic notebooks for the use of linear and kernel methods in atomistic modeling
sampling-tutorial a set of exercise files and handouts for a sampling tutorial, from thermostatting to non-linear dimensionality reduction
We list below a collection of legacy codes and tools developed in our lab. They are functional and working, but no longer maintained.
gle4md colored-noise thermostats for molecular dynamics: http://gle4md.org/
pamm tools for the Probabilistic Analysis of Molecular Motifs method
sketchmap a suite of programs to perform non-linear dimensionality reduction: http://sketchmap.org
