A comparison of homotopies for alternative formulations of the \(L^ 2\) optimal model order reduction problem (Q1919479)

The problem considered in this paper is to find, for a given state space model \[ \dot x(t)=Ax(t)+Bu(t), \qquad y(t)=Cx(t), \] a reduced order approximation minimizing the \(L^2\) norm of the error between the responses of the two systems. The paper compares two previous homotopy algorithms based on the use of input normal forms and optimal projections and introduces a new overparametrization formulation less sensitive to ill-conditioning. Many numerical examples are also reported.