AI & ML interests
None defined yet.
Recent Activity
Atomistic Simulation of Thermodynamic Properties
Research group at ICAMS, Ruhr-Universität Bochum, within the Department of Atomistic Modelling and Simulation.
We develop and apply atomistic simulation methods to calculate thermodynamic properties of materials and construct phase diagrams, predicting formation energies, phase stability, and free energy landscapes directly from atomic interactions.
What we work on
- Free energy calculations. Efficient algorithms for Helmholtz and Gibbs free energies as functions of temperature and pressure, together with the tooling to run them without hand-holding.
- Phase diagrams from simulation. Molecular dynamics and Monte Carlo workflows covering the full chain: ab initio data generation → potential parametrisation → validation → binary phase diagrams.
- Machine learning interatomic potentials. The atomic cluster expansion and its extensions (ACE, GRACE), delivering ab initio accuracy at greatly reduced cost.
- Reproducible software and workflows. We treat the code and the workflow as part of the scientific result, not as scaffolding to be discarded.
What we publish here
Reference datasets and interatomic potentials that come out of the group's work. Datasets carry enough metadata to be reused directly for fitting: energies, forces, cells, and a documented structure generation protocol.
Datasets
CuAu |
DFT energies and forces for roughly 96,000 small periodic Cu, Au, and Cu-Au cells (1 to 8 atoms), labelled with three functionals (PBE, LDA, r2SCAN) as separate subsets. Built for fitting and benchmarking machine learning interatomic potentials. CC-BY-4.0 |
Code
Our software lives on GitHub at thermoatoms, including calphy for automated free energy calculations.
Get in touch
Discussions on individual datasets and models are welcome. For questions about the group, collaborations, or open positions: sarath.menon@rub.de · group page · GitHub · ICAMS open positions