Algebraic Riccati Tensor Equations with Applications in Multilinear Control Systems

In a recent interesting paper [8], Chen et al. initialized the control-theoretic study of a class of discrete-time multilinear time-invariant (MLTI) control systems, where system states, inputs and outputs are all tensors endowed with the Einstein product. Criteria for fundamental system-theoretic notions such as stability, reachability and observability are established by means of tensor decomposition. The purpose of this paper is to continue this novel research direction. Specifically, we focus on continuous-time MLTI control systems. We define Hamiltonian tensors and symplectic tensors and establish the Schur-Hamiltonian tensor decomposition and symplectic tensor singular value decomposition (SVD). Based on these we propose the algebraic Riccati tensor equation (ARTE) and show that it has a unique positive semidefinite solution if the system is stablizable and detectable. A tensor-based Newton method is proposed to find numerical solutions of the ARTE. The tensor version of the bounded real lemma is also established. A first-order robustness analysis of the ARTE is conducted. Finally, a numerical example is used to demonstrate the proposed theory and algorithms.