This project combines a real application to NIST problems namely, molecular beam epitaxy, with work on particularly ITL problems, namely developing methods for parallel computing using arrays of workstations. It is not known how to deal with truly parallel problems efficiently, that is, problems where processors communicate with other processors randomly and the entire machine looks to the user like one large flat memory space. We address these questions in our development and implementation of a parallel algorithm to simulate molecular beam epitaxy (MBE) using a domain decomposition approach.
ACMD is involved in this because these are fundamental applied and computational mathematical problems and mathematics is the only hope for a real solution. NIST is involved because one can't do testing before the basic physics is understood. In addition, this is where the next generation of integrated circuits will come from but the basic physics is not understood.
In the past year the Monte Carlo tree algorithm was refined, an extension to real face-centered cubic geometry was made, and an enlargement of the phenomena studied was made to include edge diffusion, cascading and barriers. This year the program will be moved to a network of workstations, and realistic calculations will be made for cases that no single processor could do, as there now exists real data for comparison.
We are in the process of moving the simulation to the SP-2 using MPI and plan to move the code to the ATM network when it is ready.