From June 30 – July 3, I attended a conference on multiscale molecular modelling (M3) in Edinburgh, and had the opportunity to talk about a rare event acceleration algorithm we’ve recently developed (more below). The conference was extremely interesting, and I had the opportunity to listen to and meet a number of experts in the field of molecular dynamics, from those who focus on very small molecular systems, like Bill Miller, to those who focus on much larger ones, like David Chandler, David E. Shaw, and Vijay Pande. Links to some photos can be found here.
The method I presented is called Boxed MD (or BXD), and it’s arisen from recent work I’ve done with Dmitry Shalashilin and Emanuele Paci. It’s sort of like a formally exact, perfect form of umbrella sampling with no messy numerical renormalization procedures involved. The biggest advantage of BXD over umbrella sampling is that it simultaneously preserves both kinetic and thermodynamic information – allowing you to obtain both rate coefficients and free energies directly from the dynamics. Also unlike umbrella sampling, it involves no modifications to the potential energy, and it works equally for both canonical and microcanonical ensembles. I’ve implemented it in CHARMM, and it will be available in the next release version. The method is described in detail on Dmitry’s website.