1: \begin{abstract}

2: In a recent article we presented a framework to control nonlinear partial differential equations (PDEs) by means of Koopman operator based reduced models and concepts from switched systems. The main idea was to interpret a control system as a set of autonomous systems for which the optimal switching sequence has to be computed. These individual systems can be approximated by reduced order models obtained from data, and one can guarantee convergence under certain assumptions. In this article we extend these results to continuous control inputs using linear interpolation between two Koopman operators corresponding to constant controls, which results in a bilinear control system. Although convergence can be proved when the PDE depends linearly on the control, we show that this approach is also valid in other scenarios using a flow control example.

3: %\keywords{Optimal Control \and Reduced Order Modeling \and Koopman Operator \and Dynamic Mode Decomposition}

4: %\subclass{49J20 \and 65P99 \and 47B33}

5: \end{abstract}

6: