진학프로석·박사 채용 접수 플랫폼

Keywords: molecular dynamics, machine learning, rare event sampling

The aim of this project is to efficiently and accurately compute reaction rates of chemical transformations, typically in the field of catalytic chemistry (for instance alcohol isomerization). 

A first task will be to quantitatively study the limitations of approximate formulas obtained from harmonic transition state theory (Eyring-Kramers or Eyring-Polanyi formulas), by comparing the predictions of these approximate theories to reference results obtained either by direct numerical simulations or using dedicated numerical methods such as the adaptive multilevel splitting (AMS) algorithm[1]. 

A second task will be to improve the performance of AMS by combining it with importance sampling techniques and building good reaction coordinates, relying on neural networks such as bottleneck autoencoders, which have proved successful for free energy computations [2,3]. 

Improved reaction coordinate can be obtained by minimizing the total reaction rate of the associated effective dynamics [4].

This postdoctoral fellowship is funded by the PEPR B-best, through the project MAMABIO, whose overall goal is to develop new numerical methods to study reactions of transformation of biosourced molecules, in collaboration with IFP Energies nouvelles (Pascal Raybaud, Thomas Pigeon, Manuel Corral-Valero), and the group of Christoph Chipot (CNRS, University of Lorraine).

[1] T. Pigeon, G. Stoltz, M. Corral-Valero, A. Anciaux-Sedrakian, M. Moreaud, T. Lelièvre, P. Raybaud, Computing surface reaction rates by Adaptive Multilevel Splitting combined with machine learning and ab initio molecular dynamics, J. Chem. Theory Comput. 19(12), 3538–3550 (2023) https://doi.org/10.1021/acs.jctc.3c00280

[2] Z. Belkacemi, M. Bianciotto, H. Minoux, T. Lelièvre, G. Stoltz and P. Gkeka, Autoencoders for dimensionality reduction in molecular dynamics: collective variable dimension, biasing and transition states, J. Chem. Phys 159, 024122 (2023) https://doi.org/10.1063/5.0151053

[3] Z. Belkacemi, P. Gkeka, T. Lelièvre, G. Stoltz, Chasing collective variables using autoencoders and biased trajectories, J. Chem. Theory Comput. 18(1), 59-78 (2022) https://doi.org/10.1021/acs.jctc.1c00415

[4] W. Zhang, C. Hartmann and C. Schutte, Effective dynamics along given reaction coordinates, and reaction rate theory, Faraday Discussions, 195, 365-394 (2016) https://doi.org/10.1039/C6FD00147E