Parallel Computation

The Monte Carlo simulation of the Ising model lends itself easily to parallel computation. The parallel algorithm we adopt in this class simply splits the $x$ coordinate equally among all the processors (requiring that $n_x$ be a multiple of the number of processors). Since there are only nearest neighbor interactions, the processors need only exchange information about their spins on the edges of their $x$ domains. For the interior sites the processors can work independently, resulting in good speedup and scaling compared with a single processor.