Average-Approximation SIP Algorithms Used to Improve Accuracy of RC Model Order Reduction

This paper presents the Average-Approximation Sparse Implicit Projection (AASIP) algorithm, which enhances the accuracy of the lossless correction factor in Sparse Implicit Projection (SIP) by employing the mean fitting correction. AASIP improves the approximate accuracy of RC-netlist reduction compared to conventional SIP, while also addressing the challenges of high-order moment matching in SIP. The correctness and feasibility of the proposed algorithm are demonstrated through netlist examples.