Application of the Drazin inverse to the analysis of pointwise completeness and pointwise degeneracy of descriptor fractional linear continuous-time systems (Q2023622): Difference between revisions

The subject is the continuous time fractional linear system \[ E \frac{d^\alpha}{dt^\alpha} x(t) = Ax(t) \tag{1} \] where \(0 < \alpha < 1\), \(x(t)\) is a \(n\)-vector function and \(E, A\) are \(n \times n\) matrices. The derivative is the \textit{Caputo fractional derivative} \[ \frac{d^\alpha}{dt^\alpha} x(t) = \frac{1}{\Gamma(1 - \alpha)} \int_0^t \frac{1}{(t - \tau)^\alpha} \frac{d}{d\tau} x(t) d\tau \tag{2} \] The choice of this derivative for equations like (1) over the Riemann-Liouville version \[ \frac{d^\alpha}{dt^\alpha} x(t) = \frac{1}{\Gamma(1 - \alpha)} \frac{d}{d t}\int_0^t \frac{1}{(t - \tau)^\alpha} x(t) d\tau \tag{3} \] is due (among other things) to the fact that the Laplace transform of (2) involves \(x(0)\) rather than fractional derivatives of \(x(\cdot)\) at \(t = 0\) as in (3), thus allowing pointwise initial conditions. The system (1) is \textit{pointwise complete} for \(t = t_f\) if for every \(x_f\) there exists an initial condition \(x(0)\) such that the solution \(x(\cdot)\) satisfies \(x(t_f) = x_f\). The case \(E\) invertible is elementary via an explicit solution of (1). The authors only assume that \(\det (Ec - A) \ne 0\) for some \(c\) and prove a version of pointwise completeness where \(x_f, x(0)\) belong to a subspace. They also examine the degenerate case, where pointwise completeness does not hold.