An Improved Adaptive Decomposition Method in Parallel DSMC

A general parallel direct simulation Monte Carlo method may result in a strongly unbalanced distribution of work, which leading to very low speedups. In this paper, load balancing between processors is achieved based on an improved adaptive decomposition technique. The method has been implemented on a cluster, using Master-Slave architecture to minimize the communications. Applications were made for a two-dimensional supersonic leading-edge problem. Results show that that the algorithm can greatly reduce the computational cost while maintaining essentially the same accuracy and high speedups.