| Time: | October 10, 2025, 9:45 a.m. – 10:30 a.m. |
|---|---|
| Lecturer: | Dr. Julija Zavadlav, Technical University of Munich, Germany |
| Venue: | ICP Seminarraum 1.079 Allmandring 3 70569 Vaihingen |
| Download as iCal: |
|
ML potentials: from efficient coarse-grained models to large-scale deployment
Molecular dynamics simulations are essential for understanding complex phenomena in soft matter physics. A prominent research area is the development of machine learning potentials (MLPs), particularly those based on Graph Neural Networks (GNNs), which have emerged as a powerful tool for bridging the gap between quantum-mechanical accuracy and classical atomistic or even coarse-grained force field efficiency. In this presentation, I will showcase the significant achievements of both atomistic and coarse-grained MLPs in effectively capturing many-body interactions. I will address the current challenges of MLP development, including the broad and accurate training dataset generation, capturing long-range interactions, and numerical stability. To address these challenges, we propose a range of innovative strategies that encompass synergistic integration of diverse data sources [1,2], novel training objectives [3], physics-based GNN architectures, and advanced Bayesian methods for uncertainty quantification [4]. Through insightful case studies of various molecular systems, I will demonstrate the practical effectiveness and versatility of our approaches. Lastly, I will introduce our software platform, chemtrain [5], designed to streamline the training of machine learning potentials with customizable routines and advanced training algorithms, as well as the extension chemtrain-deploy [6], enabling scalable parallelization across multiple GPUs and million-atom simulations.
[1] S. Thaler, J. Zavadlav, Nat. Commun., 12, 6884 (2021)
[2] S. Röcken, A. Burnet, J. Zavadlav, The Journal of Chemical Physics, 161, 234101 (2024)
[3] S. Thaler, M. Stupp, J. Zavadlav, The Journal of Chemical Physics, 157, 244103 (2022)
[4] S. Thaler, G. Doehner, J. Zavadlav, J. Chem. Theory Comput., 19, 4520-4532 (2023)
[5] P. Fuchs, S. Thaler, S. Röcken, J. Zavadlav, Computer Physics Communications, 310, 109512 (2025)
[6] P. Fuchs, W. Chen, S. Thaler, J. Zavadlav, J. Chem. Theory Comput., 21, 7550-7560 (2025)