Convex Density Design for Regional Phase-Volume Contraction of Nonlinear Systems

Abstract

Regional weighted phase-volume contraction for nonlinear systems is studied using positive densities selected from a finite-dimensional log-affine family. The affine dependence of the density divergence on the design parameters turns density selection into a semi-infinite convex optimization problem that maximizes a certified uniform contraction margin on a compact region. This provides a constructive alternative to testing a preselected density and ensures a well-posed design problem under compact parameter constraints. Finite active-point and minimax characterizations identify the worst-case states governing the optimum, while rigorous sampling bounds allow continuum contraction guarantees to be recovered from finitely many inequalities. For scalar power densities, feasibility is characterized by an exact interval condition. The same structure is preserved in discrete time and admits distance-weighted and conformal-geometric interpretations. Overall, the framework provides computable certificates for detecting regional contraction that may remain hidden from ordinary divergence. A two-parameter limit-cycle example demonstrates this effect when both ordinary phase volume and a natural single-basis density fail.