rdbeoap

%function [UH, p] = rdbeoap(k, r, U, P, mtol1) % % Objective: The revised direct batch generation of desired eigenvectors % lying in the ei

%function [UH, p] = rdbeoap(k, r, U, P, mtol1)

%

% Method: Revised Direct Batch Evaluation of submatrix UH by a constrained

% Optimization Algorithm using the notion of Pseudoinverse (RDBEOAP).

%

% Input arguments:

% k: the serial number of the eigenspace Ek (numbered starting from 1).

% r: the dimension of the eigenspace Ek.

% U: a matrix whose columns are approximate eigenvectors close to Ek.

% P: the orthogonal projection matrix on the eigenspace Ek.

% mtol1: Multiplying factor of the tolerance. (tol = mtol * eps)

%

% Output arguments:

% UH: the desired orthonormal eigenvectors lying in the eigenspace Ek.

% (The matrices U and UH have the same size).

% p: rank of matrix PU = P*U.