Neural-network based simulation of rare event processes at the water/oxide interface

Subproject P12

Atomistic computer simulations of processes occurring at the water/oxide interface are challenging in several ways. The calculation of atomic forces based on ab initio methods is computationally very demanding, and barrier crossing events may lead to long computation times. Both these aspects severely limit accessible system sizes and simulation times.

Building on the neural network potentials and the rare events simulation methods developed in the first funding period, project P12 will simulate complex dynamical processes occurring at the oxide/water interface. In particular, one central objective will be to investigate heterogeneous ice nucleation on various mineral surfaces that are of atmospheric significance. The studies will involve tight interactions with projects P03 Kresse and P02 Diebold. In addition, project P12 will continue to explore the use of machine learning approaches for trajectory-based rare events sampling. Here, the main objectives are to develop efficient latent space methods to sample path distributions and the on-the-fly optimization of transition path sampling simulations based on information encoded in learned committor functions.

Christoph Dellago
PI

Expertise

Our research efforts focus on the development of simulation algorithms and their application to investigate dynamical processes in condensed matter systems based on the principles of equilibrium and non-equilibrium statistical mechanics. In particular, we have helped to create the transition path sampling methodology for the simulation of rare but important events, such as nucleation aprocesses, chemical reactions and biomolecular reorganizations. More recently, we have worked on applying machine learning methods to molecular structure recognition and the representation of potential and free energy surfaces.

Recent research topics include: