Skip to content
Chris Iacovella edited this page Nov 7, 2023 · 11 revisions

Welcome to the modelforge wiki!

Mission statement & overview

We aim to to create an interoperable ecosystem compatible with various applications, primarily focused on differentiable physical modeling. To achieve this we will focus on neural network potentials (NNPs) and, intially, free energy calculations. Our architecture is designed with clean APIs that provide developers with simplicity and minimal overhead to implement custom routines and workflows. Packages will be designed with interoperability in mind, ensuring that they can be used in conjunction with one another to maximize versatility and functionality.

The modelforge package is aimed at implementing, training, distributing, storing, and using NNPs for molecular simulations, as well as providing curated training datasets. We will further provide a automated NNP testing and benchmark package auditorium and our NNP-first Markov Chain Monte Carlo (MCMC) state sampler engine chiron.

Clone this wiki locally