Clustering to improve bi-partition quality and run time

Circuit partitioning is a very extensively studied problem. Traditional heuristic methods used to solve these problems degrade in performance when the problem size becomes large. Clustering has been proposed as a method to reduce the problem size, allowing the problem to be solved faster and with more accuracy. We proposed a novel bipartitioning method based on a mathematical formulation of the problem which can be solved using commercial combinatorial optimization (CO) tools. In this paper we continue to formulate the problem as a nonlinear program (NLP) but also apply clustering to determine if run-time and quality can be improved.