link to homepage

Institute for Advanced Simulation (IAS)

Navigation and service

Seminar by Dr GiovanniMaria Piccini

Swiss Federal Institute of Technology (ETH) and Università della Svizzera italiana (USI) (Switzerland)

27 May 2019 15:00
27 May 2019 16:00
Lecture room 2009, Jülich GRS building (16.15)

Molecular dynamics is a powerful tool for studying the evolution of complex chemical reactions. However, reactants and products states are often separated by very large activation energy barriers. Therefore, the typical time scales of a chemical reaction cannot be reached in a standard simulation. Metadynamics[1] allows overcoming these barriers by applying a bias potential to the underlying potential energy surface enhancing the fluctuations between reactants and product states. At the core of a Metadynamics simulation lies the choice of the collective variables describing the essential thermodynamics of the process.

Recently we have introduced a new method named Harmonic Linear Discriminant Analysis (HLDA)[2,3] allowing to extract efficient yet physically meaningful collective variables from the local equilibrium fluctuations within the free energy basins. This allows extracting essential information without having any specific knowledge on the reaction mechanism and at the same time operating a drastic dimensionality reduction of the collective variables. The method has been applied successfully to fundamental organic reactions up to complex processes such as enzymatic catalysis.

Although enhanced sampling methods allow reaching long time scales the convergence of the free energy surfaces associated to the process of interest often requires sampling of millions of different configurations. This limits the affordable accuracy of the method used to describe the potential energy and forces and restricts the applicability of high-level quantum chemical methods to very few examples. Due to this fact the conclusions derived from these kinds of simulations can only be qualitative. To overcome this problem we introduce a new method combining Metadynamics and Free Energy Perturbation (FEP)[4]. This new approach to enhanced sampling allows the perturbative estimation of accurate quantum chemical free energy surfaces starting from a trajectory obtained using a lower level of theory with a computational cost of several orders of magnitude less than traditional approaches.



[1] A. Laio, M. Parrinello, PNAS, 99 (20), 12562-12566  

[2] D. Mendels, GM. Piccini, M. Parrinello, J Phys. Chem. Lett. 9 (11), 2776-2781

[3] GM. Piccini, D. Mendels, M. Parrinello, J. Chem. Theory and Comput. 14 (10), 5040-5044

[4] GM. Piccini, M. Parrinello, arXiv:1904.02004