AMS2024 released: active learning ML potentials, reaction discovery

We are thrilled to unveil the latest advancements in the Amsterdam Modeling Suite for 2024. Our new release brings a host of powerful features designed to step up your computational chemistry and material science research:

On-the-Fly Machine Learning Potentials

Train and fine-tune M3GNet ML potentials with ParAMS, improving prediction accuracy through simple active learning workflows.

Quantum ESPRESSO integration

Utilize Quantum ESPRESSO as an AMS engine for exploring potential energy surfaces and training machine learning models. Including new reliable pseudopotentials.

Advanced reaction discovery

Leverage enhanced reactive MD for quick generation of plausible reactions and products, along with tools for transition state searches and reaction enforcement during MD simulations.

GUI improvements

Experience enhanced usability in AMSinput and AMSview, such as viewing Powder X-ray Diffraction (PXRD), visualizing uncertainties for ML committee models, and exporting input to a PLAMS python script.

Check out the release notes for more information. Let us know what you would like to see in future releases!