Minimal representations of inverted Sylvester and Lyapunov operators (Q1870039): Difference between revisions

For the \(m\times m\) and \(n\times n\) complex matrices \(A\) and \(B\) consider the Sylvester operator \({\mathcal S}_{A,B}(X):=AX-XB\) where \(X\) is \(m\times n\). If \(n=m\) and \(B=-A^*\) this is by definition the Lyapunov operator \({\mathcal L}_A\). It is known that \({\mathcal L}^{-1}(Y)=\sum _{i=1}^NV_iYW_i\) for some \(m\times m\) and \(n\times n\) matrices \(V_i\) and \(W_i\). The author shows that the minimal possible value of \(N\) is the degree \(\nu _A\) of the minimal polynomial of \(A\) and that for \(N=\nu _A\) one can choose \(W_i=\pm V_i^*\). He also shows that for a Sylvester operator one has \({\mathcal S}^{-1}(Y)=\sum _{i=1}^{\nu}V_iYW_i\) where \(\nu =\min (\nu _A,\nu _B)\).