Classical and generalized solutions of time-dependent linear differential-algebraic equations (Q1923198): Difference between revisions

The subject of the paper is the time-dependent linear differential-algebraic equation of the form (e) \(A(t)\dot x+B(t)x=b(t)\), \(t\in I\), where \(I\subseteq \mathbb{R}\) is an open interval of reals and for each \(t\), \(A(t)\) and \(B(t)\) are \(n\times n\) matrices. By using the hypothesis that \(A\) and \(B\) are analytic functions a reduction method is developed to prove that both the classical and distribution solutions of (e) are of the form \(x(t)= \Gamma(t) x_v(t)+ u(t)\), where \(x_v\) solves a ``smaller'' system \(A_v(t)\dot x_v+B_v(t)x_v= g_v(t)\), \(t\in I\) where \(A_v(t)\) is invertible for all but isolated values of \(t\). Such a reduction method is also applied for the case when \(A\), \(B\) are \(C^\infty\) but not analytic. In this case some additional hypotheses (like \(\text{rank } A(t)\oplus B(t)=n\) and \(\text{rank } A(t)=r=\text{const.}\)) are needed to compensate for the lack of analyticity.