Numerical Methods for Nearly Singular Constrained Matrix Sylvester Equations
A. Ghavimi, Alan J. Laub · SIAM Journal on Matrix Analysis and Applications · 1996
A recently published result describes a numerical procedure for solving a matrix Sylvester equation that is subject to certain constraints. It is quite possible that this Sylvester equation, or another intermediate one in the solution process, is nearly singular. As a result, certain computed parameters can have unexpectedly large norms and be very inaccurate. This paper incorporates an implicit deflation method for nearly singular matrix Sylvester equations to implement a reliable version of the published algorithm.