Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
BP: Fachverband Biologische Physik
BP 32: Computational Biophysics II
BP 32.10: Vortrag
Freitag, 21. März 2025, 12:00–12:15, H46
Leveraging quantum data to advance machine-learning in (bio)molecular simulations — •Leonardo Medrano Sandonas1, Mirela Puleva2, Gianaurelio Cuniberti1, and Alexandre Tkatchenko2 — 1TUD Dresden University of Technology, Germany. — 2University of Luxembourg, Luxembourg.
The rapid advancement of machine learning (ML) applications in chemistry and physics has been driven by the increasing availability of comprehensive quantum-mechanical (QM) datasets. Recently, we introduced high-fidelity property data at the PBE0+MBD level of theory for both small [Sci. Data 8, 43, (2021)] and large [Sci. Data 11, 742, (2024)] drug-like molecules in equilibrium and non-equilibrium states. These datasets have been instrumental in advancing QM-based ML interatomic potentials [10.26434/chemrxiv-2024-bdfr0, (2024)] and enhancing semi-empirical (SE) methods [J. Phys. Chem. Lett., 11, 6835 (2020)], enabling accurate (bio)molecular simulations. In this presentation, we will discuss our recent efforts to improve the transferability and generalizability of the ML-corrected density functional tight-binding method. We demonstrate that equivariant neural networks significantly enhance the accuracy and scalability of ML-based many-body repulsive potentials trained on energies and forces of small organic systems. This approach facilitates the investigation of the energetic and structural properties of large drug-like molecules and molecular dimers. Hence, our findings indicate that combining ML with SE methods achieves both high accuracy and computational efficiency, paving the way for diverse applications in (bio)molecular simulations.
Keywords: bimolecular simulations; Quantum mechanics; machine learning; data-driven methods; semi-empirical methods