Dr. Omar Valsson, ETH Zurich
2:00pm - 3:00pm
Many enhanced sampling methods are based on constructing a bias potential as a function of a small number of collective variables (CVs) that is designed to facilitate movements between metastable states. However, constructing such a bias is often not easy.
Here we will present a new CV-based enhanced sampling method named variationally-enhanced sampling . In this approach the bias potential is constructed by variationally minimizing a convex functional of the bias potential. A great advantage of the method is the considerable flexibility in choosing the functional from of the bias potential that is minimized. The free energy surface as a function of the CVs can be directly obtained from the bias that results from the minimization. Furthermore, the minimization will results in the CVs being sampled according to a target distribution that can be either predefined or iteratively updated during the minimization . A considerable advantage of the method is the freedom in selecting this target CV distribution.
We will present numerous examples that demonstrate the usefulness and flexibility of the approach. We will furthermore discuss how the variationallyenhanced sampling approach can be used in innovative and novel ways. For example, for constructing effective bias potentials for obtaining kinetics properties , or to bias high-dimensional CV spaces .
 O. Valsson and M. Parrinello, Phys. Rev. Lett. 113 (2014) 090601
 O. Valsson and M. Parrinello, J. Chem. Theory Comput. 11 (2015) 1996-2002
 J. McCarty, O. Valsson, P. Tiwary. and M. Parrinello, Phys. Rev. Lett. 115 (2015) 070601
 P. Shaffer, O. Valsson, and M. Parrinello, Under preparation