The vision of this workshop was to explore mechanisms to build common infrastructure that can help the community broadly, allowing for the rapid development of new methods as well as rapid dissemination into a broad range of community codes.
There are already many successful codes and packages for driving simulation for Molecular Dynamics, Monte Carlo, etc (eg CHARMM, AMBER, NAMD, GROMACS, LAMMPS, TINKER, OPENMM, etc). However, there is a paradigm shift emerging as massively parallel compute resources are becoming available. Running on hundreds to thousands of cores is now common, with tens of thousands to millions on the horizon. Traditional schemes for parallelization will clearly fail to scale to those resources for many simulations of interest.
However, a natural scheme to use these resources and to propel more efficient and advanced science is to employ novel sampling schemes on top of these codes. Our vision is to provide such infrastructure, allowing for a rapid development and dissemination of new methods, as well as cross fertilization between different approaches.
Organizer: Vijay Pande
Dates: 6-7 April 2017
Location: Stanford University