Speaker
Description
Free energy has always served as a necessary measure for understanding
chemical reactions, particularly knowing whether a reaction will occur
spontaneously, but it is also a useful quantity when considering equilibrium
systems, revealing qualities such as mean distances and preferred orientations.
However adequate sampling of full multidimensional free energy surfaces from
molecular dynamic simulations presents several problems the most important of
which is that regions of high energy are rarely, if ever, sampled on normal
timescales.
The Scientific Computing Research Unit has recently developed a new method
[1] to generate complete multidimensional free energy surfaces from MD
simulations, based upon umbrella sampling and flat histogram methods. It also
integrates the weighted histogram analysis method to allow for multiple
simultaneous parallel simulations to run and combine free energy information.
Intended to easily integrate into most common MD software packages, this
method is capable of utilizing HPC hardware, such as multicluster CPUs running
OpenMP, to pot
