An iterative method based on ADMM for solving generalized Sylvester matrix equations

From MaRDI portal
Publication:2676138





The generalized Sylvester matrix equation is \(AXB+CXD=E,\) with \(A\) and \(C,\) \( B\) and \(D\), and \(E\) being matrices of sizes \(m\times n\), \(r\times p\) and \( m\times p\), respectively;\ accordingly, the unknown matrix \(X\) is of size \( n\times r\). The authors assume that \(A\) and \(C\) (\(B\) and \(D\)) are full column (row, respectively) rank. In the solution method they propose, submatrix equations of the type \(AXB=E_{1}\) and \(CXD=E_{2}\) are considered. An equation of the type \(AXB=F\) is reformulated as the optimization problem consisting in minimizing \(\frac{1}{2}\left\Vert ZB-X\right\Vert ^{2}\) subject to \(AX-Z=0\). Then, an alternating direction augmented Lagrangian method, using the decomposition \(f(X)+g(Z)\) of the objective function given by \(f(X):=0\) and \(g(Z):=\frac{1}{2}\left\Vert ZB-X\right\Vert ^{2}\), is proposed. A convergence theorem is proved and numerical examples are presented.



Cites work


Cited in
(18)


Describes a project that uses

Uses Software






This page was built for publication: An iterative method based on ADMM for solving generalized Sylvester matrix equations

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2676138)