Local Sampling in Steered Monte Carlo Simulations Decreases Dissipation and Enhances Free Energy Estimates via Nonequilibrium Work Theorems

Configurational freezing (<i>J. Chem. Theory Comput.</i> <b>2011</b>, <i>7</i>, 582) is a method devised for steered Monte Carlo simulations aimed at improving free energy estimates via nonequilibrium work theorems (see Jarzynski in <i>Phys. Rev. Lett.</i> <b>1997</b>, <i>78</i>, 2690 and Crooks in <i>J. Stat. Phys.</i> <b>1998</b>, <i>90</i>, 1481). The basic idea is to limit the sampling to particles located in the region of space where dissipation occurs, while leaving the other particles fixed. Therefore, the method is based on the reasonable assumption that dissipation is a local phenomenon in single-molecule nonequilibrium processes, a statement which holds for many processes including, for example, folding of biopolymers and protein–ligand binding/unbinding. In this article, the configurational freezing approach, based on the sampling of particles located around hot-spot sites encompassing the high dissipation domain, is supplemented by the possibility of selecting such particles (for trial Monte Carlo moves) dependent on their distance from the hot spots. This is accomplished by exploiting an extension of the Owicki’s preferential sampling (<i>J. Am. Chem. Soc.</i> <b>1977</b>, <i>99</i>, 7413) in the original configurational freezing machinery. The combined strategy is shown to improve the accuracy of free energy estimates in physically sound cases: the calculation of the water to methane relative hydration free energy and the calculation of the potentials of mean force of two solvated methane molecules and two solvated benzene molecules along the direction connecting the centers of mass