Stochastic boundary conditions for molecular dynamics simulations.

• Main Author: Suen, Whei Yeap.
• Other Authors: Cheong Siew Ann
• Format: Final Year Project
• Language: English
• Published: 2011
• Subjects: DRNTU::Science::Physics
• Online Access: http://hdl.handle.net/10356/44755
• Institution: Nanyang Technological University

At present, the cutting edge molecular dynamics simulation can be performed fr a system of approximately 10$^{10}$ to 10$^{11}$ particles over about 10$^{3}$ nodes. Nevertheless, such systems are still profoundly undersized compared to a real physical systems that contains particle number at the order of 10$^{23}$. As a result, finite size effect can undermine the validity of studies of physical system. In minimizing the finite size effects, periodic boundary conditions have been widely used in molecular dynamics simulations. However, due to the artifical correlation caused by the time reversal invariance of the periodic boundary conditions, the periodic boundary conditions has very limited applications. Therefore, we strive to develop stochastic boundary conditions that will not only rid such artificial correlation, but also has a wide area of applications. By using the statistics gathered from the periodic boundary condition simulation, we perform a numerical cumulative distribution transform and implement the first order stochastic boundary conditions into our system. Henceforth, the thermodynamical properties of the system is calculated and compared to the existing canonical and grand canonical ensemble properties. It is shown in our project that our system does not belong to the canonical ensemble but more data is required to compare our system to the grand canonical ensemble more accurately.