Statistical Sensitivity Analysis in Distributed Circuits using Compact Polymorphic Polynomial Chaos Surrogates

This paper presents a novel polynomial chaos (PC) based approach to calculate the sensitivity of the statistical moments of distributed circuit responses with respect to design variables. The proposed approach is based on the concept of polymorphic variables which can compactly represent both the aleatory (probabilistic) uncertainty and epistemic (design) variability present in the input parameters of the circuit. Orthogonal basis functions of these polymorphic variables are used to efficiently construct the PC metamodel of the circuit response of interest. Thereafter, the sensitivity of the statistical moments of the circuit response of interest can be analytically derived from the PC metamodel.