Project Goals

This project has three major goals:

$1)$ Design, implement, and test a standalone MC code that is robust, extensible, and capable of performing PT as well as traditional Metropolis MC.

$2)$ Implement an efficient parallelization scheme for the PT algorithm. Naive implementation of PT on parallel processors is dramatically inefficient, resulting in $50\%$ or more idle processor time [9]. This is because, in general, replicas run at lower temperature require more computational effort. Implementation of a job scheduler that manages and distributes replica runs over the available processors makes the method ideally efficient.

$3)$ Study the efficiency of PT in simulations of liquid water. Water is a glassy, complex liquid [10,11], and typical MC simulations require millions of MC steps to evolve the system into an uncorrelated configuration. We will use the TIP5P empirical potential, a rigid five-site geometry parametrized for classical MC simulations [12].