ase_uhal Workshop
Date: Thursday 26th March 2025
Time: 11:00–15:30
Location: Room F2.10, School of Engineering, University of Warwick
Schedule
11:00–12:00 — Introduction to ase_uhal
12:00–13:00 — Lunch
13:00–14:00 —Invited and Contributed Talks
14:00–15:30 — Hands-on workshop
About
ase_uhalLink opens in a new window is a Python package funded by a PSDI Pilot Project that streamlines the development of DFT training datasets for machine learning interatomic potential (MLIP) models. It builds on the HyperActive Learning (HAL) approach (van der Oord et al., 2023) and provides ASE calculators that bias the potential energy surface to avoid structures already well-covered in an existing dataset.
The morning session will give an introduction to the package and its core ideas. After lunch, invited talks will cover a range of topics relevant to MLIP users, including automated workflow infrastructure, novel potentials, uncertainty quantification, and advanced sampling methods. The final session is a hands-on workshop where attendees can begin applying ase_uhal to their own research problems with guidance from the package developers.
Invited and Contributed Talks
Talks will cover topics relevant to MLIP users, which may include automated workflow infrastructure, novel potential architectures, uncertainty quantification, and advanced sampling methods. Speakers to be confirmed. If you are interested in giving a contributed talk please indicate this on the registration form.
Who should attend?
The workshop is aimed at anyone developing or using MLIPs in their research. No prior experience with ase_uhal is required, but familiarity with ASE and Python will be helpful. If you'd like to work on your own system during the hands-on session, please bring a laptop with a working Python environment.
Registration
Please register using the form below. Places are limited.