Robustness analysis: nonlinear and growing

A numerical algorithm for analysis of disturbance rejection for nonlinear systems is proposed. The algorithm seeks solutions to Euler-Lagrange equations and is similar to power algorithms for mu-analysis lower bounds, beating conventional optimization on speed and global convergence properties. This is a special case of earlier work and focuses on a problem for which a simple proof of global convergence is available for the case of linear dynamics. Thus we can focus our attention on the convergence problems and lack of convexity created by nonlinearities, and benchmark examples are used to explore these issues.