Stability Analysis of Data Driven Complex Systems

In this work we develop a mathematic and algo­ rithmic framework for data-driven stability analysis of black box switching systems by using adaptive sampling and on­ line optimization. Our algorithm learns the optimal sam­ pling strategy to build more accurate stability guarantees with fewer sampling points. This work is an extension of [ l], where an algorithm is presented to bound the Joint Spec­ tral Radius (JSR) [2] of an unknown switched linear system within some probability from a finite number of trajectories