Statistical Cooling is an optimization technique based on Monte-Carlo techniques. Here we propose two parallel formulations of the statistical cooling algorithm, i.e. a systolic algorithm and a clustered algorithm. Both algorithms are based on the requirement that quasi-equilibrium is preserved throughout the optimization process. It is shown that the parallel algorithms can be executed with a polynomial-time complexity. Performance of the algorithms is discussed by means of implementations on an experimental multi-processor architecture. It is concluded that substantial reduction of computation time can be achieved by both parallel algorithms compared to the sequential algorithm.
Aarts, E. H. L., Bont de, F. M. J., Habers, E. H. A., & Laarhoven van, P. J. M. (1986). Parallel implementations of the statistical cooling algorithm. Integration : the VLSI Journal, 4(3), 209-238. https://doi.org/10.1016/0167-9260(86)90002-7