Ongoing Project, collaborate with George Stadler and Michael Shelley
We are developing a parallel algorithm for the large-scale fish schooling simulation using the vortex sheet method. In our preliminary algorithm, the boundary integral is evaluated directly using shared memory parallelism of OpenMP. The test performance of a 10 fish school is improved by a factor of 12 on 64 processors using OpenMP. Parallel algorithm using distributed memory and the Kernel-Independent Fast Multiple Method (FMM) is going to be developed and implemented.