Dykstra's algorithm for constrained least-squares rectangular matrix problems (Q1130433)

An application of \textit{R. L. Dykstra's} projection algorithm [J. Am. Stat. Assoc. 78, 837-842 (1983; Zbl 0535.62063)] to solve a constrained least-squares \(n\times n\) matrix problem is considered. A new improved implementation of the projection algorithm onto an \(\varepsilon\)-positive definite set of matrices is proposed. This implementation does not require the computation of all eigenvalues and eigenvectors of a matrix per iteration, as it is adopted in the original one proposed by \textit{R. Escalante} and \textit{M. Raydan} [Dykstra's algorithm for a constrained least-squares matrix problem, Numer. Linear Algebra Appl. 3, No. 6, 459-471 (1996; Zbl 0908.90207)] and still guarantees the convergence.