Benchmarking of Probabilistic-bit based Algorithm for Max-cut Problem

In the age of big-data and internet of things combinatorial optimization problems (COP) have found widespread applications in both research and industry. However, solving these problems efficiently using deterministic approaches on conventional computers becomes challenging as the problem size and complexity increases. Recently, the concept of probabilistic computing utilizing probabilistic bits has gained attraction as an energy-efficient hardware accelerator for randomized algorithms. In this paper, we explore simulated annealing using probabilistic bits in solving the most popular COP - max-cut problem. The performance of the algorithm is benchmarked against 50 G-set connection graphs, where the best cut was found for 11 graphs and the deviations for the others were minimal.