Recent simulation advances inspired by transition path sampling have led
to powerful and reliable ways to obtain accurate one dimensional
reaction coordinates for complex activated processes. After a brief
review of Likelihood Maximization and Aimless Shooting (LM/AS), already
featured by Bernhardt Trout, we discuss applications and further
methodological advances. First, we discuss quantitative measures of
reaction coordinate error and the effects of error on computed rate
constants. The remainder of the talk focuses on applications of LM/AS
to understand nucleation processes. The first example is nucleation of
crystals in a Lennard-Jones melt. In collaboration with Gregg Beckham,
3000 accepted transition paths were obtained for this highly diffusive
system. An accurate reaction coordinate was constructed from measures
of nucleus size, shape, and structure that have been used in the
literature. Amazingly, from 67 candidate variables, the optimal
reaction coordinate involves the two components that were identified by
Moroni, ten Wolde, and Bolhuis in 2005. A second example is a simple
lattice model that captures the phase behavior of crystallization from
solution. The lattice model suggests that the characteristics of
critical nuclei strongly depend on the difference between the
temperature during nucleation and the solute melting temperature.
Copyright. All Rights Reserved.