Poster Session I: COMPUTATIONAL PHYSICS (DCOMP)
Path Integral Monte Carlo Simulations of Flexible Water Clusters with Normal-Mode Sampling
Poster 477In this work, we present a novel importance sampling scheme for path integral Monte Carlo simulations of water clusters that emphasize the inclusion of vibrational degrees of freedom. This approach takes advantage of the vibrational normal modes that naturally occur in clusters of water molecules. We can perform efficient Monte Carlo sampling by proposing new configurations according to the harmonic oscillator density matrices associated with these normal modes. We show that using normal-mode sampling generates decorrelated samples much faster than traditional Monte Carlo sampling methods such as the Metropolis-Hastings algorithm. Therefore, this allows us to accurately compute expectation values using fewer Monte Carlo iterations.
This normal-mode sampling technique is applied to path integral Monte Carlo simulations of the water monomer, water dimer, and several equilibrium geometries of the water hexamer. For each of these water clusters, we calculate energetic and structural properties, such as ground state energies, binding energies, positional distribution functions, and pair correlation functions. We then validate our methodology by showing that our results are in excellent agreement with experimental data and existing results from path integral molecular dynamics and diffusion Monte Carlo simulations. Future work will extend this methodology to simulate larger water clusters, as well as lattices of confined molecules with dipole-dipole interactions.