Eigenvalue analysis orthogonal matrix

The matrix K is a square and orthogonal matrix of the eigenvectors. The different ways of computing the dispersion matrix by Q or R analysis techniques lead to different sets of eigenvalues, as we will see next in the comparison with SVD. 

All the various options influence the result of a population analysis. However, before a final assessment can be made we must address the more basic problem of a category for the population analysis. The most popular analysis is the Mulliken analysis for nonorthogonal AOs and equal partitioning of overlap between the two atoms involved. If the ZDO assumption is maintained for the Fock matrix eigenvalue problem, an orthogonalized AO basis must be considered. In such semiempirical methods the symmetric orthogonalization is most frequently implied. 

---