Comment on "Gradient-based direct normal-mode analysis" [J. Chem. Phys. 122, 184106 (2005)]

A comment on the efficient protocol for the direct gradient-based calculations of normal modes, described by Kaledin, is presented. It is emphasized that the same algorithm had been presented earlier in the name of 'mode-tracking method' which was extensively tested in all its facets to demonstrate its efficiency and range of applicability. The calculation of certain parts of the eigenvalue spectrum of a Hermitian operator through a block-Lanczos algorithm is also possible. It is found that no reference to the form of the converged eigenvectors is made in the mode-trackig protocol by Kaledin.