Simulating Crystallization\nin a Colloidal System Using\nState Predictive Information Bottleneck Based Enhanced Sampling

We investigate crystal nucleation in supersaturated colloid\nsuspensions\nusing enhanced molecular dynamics simulations augmented with machine\nlearning techniques. The simulations reveal that crystallization in\nthe model colloidal system studied here, with particles interacting\nthrough a repulsive screened Coulomb Yukawa potential, proceeds from\nvapor to dense liquid droplet to crystalline phases across multiple\nhigh barriers. Employing a one-dimensional reaction coordinate derived\nfrom the State Predictive Information Bottleneck framework, our simulations\ncapture back-and-forth phase transitions across multiple barriers\neffectively in biased metadynamics simulations. We obtain relative\nfree energy differences between different phases and also quantify\nthe roles of different molecular level features in driving the phase\nchanges.