Relaxation of the greatest common divisor condition for the controllability of bilinear systems (Q6110302)

This manuscript deals with the greatest common divisor condition for the controllability of bilinear systems. In this paper the authors consider homogeneous single-input one bilinear systems of the form \[ x(t+1)=Ax(t)+u(t)ch^T x(t), \] where \(x(t) \in \mathbb{R}^n\) and \(u(t) \in \mathbb{R}\) are the state vectors and scalar input, respectively, at time step \(t \in \mathbb{N}\), \(A \in \mathbb{R}^{n \times n}\) is the state matrix and \(c, h \in \mathbb{R}^n\). The authors study the controllability conditions for this type of bilinear systems. They show that when the greatest common divisor condition is not met \((k>1)\), the states/nodes of the system are partitioned into \(k\) groups of \(\alpha = n/k\) states/nodes and starting from each of these partitions only a determined sub-sequence of inputs can affect the system. The concept of axis-aligned near controllability is presented which discusses the fact that if in each partition there exists at least one nonzero state the system can be driven into any desired point in the state space and, hence, is controllable. The relaxation of the greatest common divisor condition reduces the controllability problem of the bilinear system to the controllability and observability of its associated linear system.