Finite iterative method for solving coupled Sylvester-transpose matrix equations (Q741404)

The paper gives an iterative algorithm to solve real coupled Sylvester-transpose matrix equations \[ \sum_{\eta=1}^p(A_{i\eta}X_\eta B_{i\eta}+C_{i\eta}X_\eta^T D_{i\eta})=F_i,~~i=1,\dots,p, \] for the unknown matrices \(X_\eta\), \(\eta=1,\dots,p,\) where all matrices may have different but compatible dimensions. It is shown that if the equations have a solution, then in exact arithmetic, the algorithm converges to the solution in a finite number of steps. For appropriate starting values, the computed solution minimizes \(\sum_\eta\|X_\eta\|^2\) (Frobenius norms), and because of the linearity, it is also possible to find the solution closest to a given set of matrices in the same Frobenius sense.