Exploring high-dimensional free energy landscapes of chemical reactions

Molecular dynamics (MD) techniques are widely used in computing free energy changes for conformational transitions and chemical reactions, mainly in condensed matter systems. Most of the MD‐based approaches employ biased sampling of a priori selected coarse‐grained coordinates or collective variables (CVs) and thereby accelerate otherwise infrequent transitions from one free energy basin to the other. A quick convergence in free energy estimations can be achieved by enhanced sampling of large number of CVs. Conventional enhanced sampling approaches become exponentially slower with increasing dimensionality of the CV space, and thus they turn out to be highly inefficient in sampling high‐dimensional free energy landscapes. Here, we focus on some of the novel methods that are designed to overcome this limitation. In particular, we discuss four methods: bias‐exchange metadynamics, parallel‐bias metadynamics, adiabatic free energy dynamics/temperature‐accelerated MD, and temperature‐accelerated sliced sampling. The basic idea behind these techniques is presented and applications using these techniques are illustrated. Advantages and disadvantages of these techniques are also delineated