One trick that is commonly used to resolve this problem is to add a simulation bias in order to force the interesting, rarer events to occur. This simulation bias forces the value of some particularly interesting collective variable (CV) to change more rapidly than it otherwise would. This is a very good strategy when you have a reasonable understanding of what occurs during your rare event and when you can encode what happens mathematically. For instance, to study a process such as nucleation you simply have to find a differentiable quantity that distinguishes between configurations where the ions are distributed uniformly across the simulation cell from those in which the ions are bunched together in a crystalline nucleus. To develop coordinates for complicated problems such as these we are involved with the development of a code called plumed 2.0. Already we are using this code to find new CVs that can be used to study nucleation and to study how water molecules enter the ionic channels in large protein molecules.
Researcher: Gareth Tribello