GPU Acceleration of Swendson-Wang Dynamics

When simulating a lattice system near its critical temperature, local algorithms for modeling the system’s evolution can introduce very large autocorrelation times into sampled data. This slowing down places restrictions on analysis that be completed in timely manner of behavior systems around point. Because it is often desirable to study such this point, new algorithm must introduced. Therefore, we turn cluster algorithms, as Swendsen–Wang and Wolff clustering algorithm. They incorporate global updates which generate configurations with little correlation previous states, even We look accelerate rate at these are capable running by implementing benchmarking parallel implementation each designed run GPUs under NVIDIA’s CUDA framework. A 17 90 fold increase computational was, respectively, experienced when measured against equivalent implemented serial code.