Dense Sub-Circuit Reduction in RC Circuits

Circuit simulation is a crucial but time-consuming step. Dense circuits and sub-circuits are significantly more expensive to simulate than sparse ones. In this paper, a novel dense sub-circuit reduction system is introduced. The proposed methodology uses dense sub-graph mining techniques on top of the model order reduction techniques to reduce the dense sub-circuits in the designs. This paper focuses on comparing four different dense sub-graph mining algorithms in the context of circuit reduction. The conducted experiments showed that the greedy approximation algorithm for edge density maximization is the most suitable for our circuit reduction problem since it provides the best results in terms of scalability on large graphs as well as the density and the size of the detected dense sub-graphs.