A Multivariate Adaptive Sampling Scheme for Passivity Characterization of Parameterized Macromodels

We introduce a multivariate adaptive sampling algorithm for the passivity characterization of parameterized macromodels. The proposed approach builds on existing sampling methods based on adaptive frequency warping for tracking pole-induced variability of passivity metrics, which however are available only for univariate (non-parameterized) models. Here, we extend this approach to the more challenging parameterized setting, where model poles hence passivity violations depend on possibly several external parameters embedded in the macro-model. Numerical examples show excellent performance and speedup with respect to competing approaches.