The reduced mass does not appear to be included in the printout of a frequency analysis in DMOL3. The normal mode output appears to be normalized to 1, in the same manner that Gaussian 2009 does so.  In Gaussian, the reduced mass is used to normalize the normal mode vectors.  I assume this is the same for DMOL, except the rotational and translational degrees of freedom are not removed from the Hessian matrix.  Is there a way to get the reduced masses?  I have attempted to do so using the hessian matrix output and the reported normal mode frequencies, yet operating on the hessian matrix in the *.hessian file with a matrix with the normal mode vectors as columns does not diagonalize the hessian.  Can someone provide guidance on how to get the reduced mass?

I wish to estimate the overtone intensities by calculating the dipole moment per unit cell, approximate the nuclear wavefunction along a normal coordinate using the Morse oscillator wavefunctions and the reduced mass.