Solvation free energies from neural thermodynamic integration

Article Article scientifique Article Solvation free energies from neural thermodynamic integration eng 2025-03-25 2025-03-28 author Mate Balint 0000-0001-9144-9206 Department of Computer Science, University of Geneva, Carouge, Switzerland.,Department of Physics, University of Geneva, Geneva, Switzerland. 951340 author Fleuret François Jean 0000-0001-9457-7393 Department of Computer Science, University of Geneva, Carouge, Switzerland.,Fundamental AI Research, Meta AI, Paris, France. 766513 Département d'informatique (DINFO) author Bereau Tristan 0000-0001-9945-1271 Institute for Theoretical Physics, Heidelberg University, 69120 Heidelberg, Germany.,Interdisciplinary Center for Scientific Computing (IWR), Heidelberg University, 69120 Heidelberg, Germany. 124107 10.1063/5.0251736 40130791 0021-9606 https://pubs.aip.org/jcp/article/162/12/124107/3340485/Solvation-free-energies-from-neural-thermodynamic We present a method for computing free-energy differences using thermodynamic integration with a neural network potential that interpolates between two target Hamiltonians. The interpolation is defined at the sample distribution level, and the neural network potential is optimized to match the corresponding equilibrium potential at every intermediate time step. Once the interpolating potentials and samples are well-aligned, the free-energy difference can be estimated using (neural) thermodynamic integration. To target molecular systems, we simultaneously couple Lennard-Jones and electrostatic interactions and model the rigid-body rotation of molecules. We report accurate results for several benchmark systems: a Lennard-Jones particle in a Lennard-Jones fluid, as well as the insertion of both water and methane solutes in a water solvent at atomistic resolution using a simple three-body neural-network potential. Dewey 500.2 Dewey 025.063 Swiss National Science Foundation Robust Deep Density Models for High-Energy Particle Physics and Solar Flare Analysis (RODEM) 193716 CH Deutsche Forschungsgemeinschaft STRUCTURES: A unifying approach to emergent phenomena in the physical world, mathematics, and complex data 390900948 DE 32 Département d'informatique 18 27 Section de physique 13 false The Journal of chemical physics 162 12 27b9cd45-a27b-43b5-9829-b4387f199f7c Article (Published version) Public 124107_1_5.0251736.pdf 6130466 application/pdf CC BY-4.0 58e58a471df902031f19bd873cf7510d256af6350c03351a6f9c3dbae6d5db53 openAccess Faculté des sciences / Département d'informatique Faculté des sciences / Section de physique

Title	Published in	Year	Views	Downloads

Solvation free energies from neural thermodynamic integration	The Journal of chemical physics	2025	5	6
Flowification: Everything is a normalizing flow	UNIGE Data Science Day 2022	2022	135	2
Fast detector simulation and detector design	UNIGE Data Science Day 2022	2022	128	2

Archive ouverte UNIGE

Mate, Balint