Master of Science in Computer Science (MSCS)
In this report, we have reported some preliminary results in the development of a parallel computing-visualization framework for large-scale molecular dynamics simulations of polycrystals of minerals, which are geophysically relevant for Earth’s mantle. First, we have generated the input configurations of atoms belonging to various grains distributed in the space in a way, which resembles the polycrystalline structure of the minerals. The Input configuration is developed using Voronoi geometry. Thus generated polycrystalline system is simulated using the PolyCrystal Molecular Dynamics algorithm. Performance tests conducted using up to 256 processors and a couple of millions of atoms have shown that the computation time per MD step remains under 20 seconds. The other important part is the development of an efficient visualization system to interactively explore the massive three dimensional and time-dependent datasets produced by MD simulations. Some results are presented for the simulation of two-grain structure. The proposed framework is expected to be useful in simulations of more realistic and complex rheological (mechanical) properties of important Earth forming mineral phases under different conditions of stresses and temperatures.
Document Availability at the Time of Submission
Release the entire work immediately for access worldwide.
Yerraguntla, Venkatasrirama PavanKumar, "A parallel computing-visualization framework for polycrystalline minerals" (2005). LSU Master's Theses. 602.
Bijaya B Karki