Global sensitivity based dimension reduction for fast variability analysis of nonlinear circuits

In this paper, a dimension reduction methodology for expedited polynomial chaos (PC) based variability analysis of nonlinear circuits is presented. The key feature of this work is the development of an efficient global sensitivity approach to quantify the relative impact of each random dimension on the variance of the circuit outputs. This global sensitivity measure is then used to guide the truncation of the original high-dimensional random space into a reduced dimensional random subspace. Performing the PC expansion of the circuit model on this reduced dimensional subspace leads to far faster variability analysis with only marginal loss of accuracy.