Scalable Control Design for K-positive Linear Systems

Abstract: Systems whose variable are constrained to be positive allow computationally efficient control design. We generalize these results to linear systems which leave a cone invariant. This is a wider class of systems than positive systems. We revisit classical results on stability and dissipativity of positive linear systems and show how scalable conditions on linear programming can be extended to cone invariant linear systems. Our results are illustrated by scalable stabilizing controller design for mass-spring systems.