dipole derivatives in normal modes

[olga]

Hi everybody,
I need to calculate transition dipoles for all 0 -> 1 vibrational transitions of some molecules. I have written a simple script doing transformation of dDIP/dX (from file "dipgrad") to dDIP/dQ using pseudoinverse of truncated (translations and rotations are excluded) transformation matrix L from file "vib_normal_modes". And then, using analytic expressions for harmonic oscillator wavefunctions, obtain the needed transition dipoles assuming that dipole depends only linearly on normal modes.
I did a simple test with water and found, that the values of |dDIP/dQ| (a.u.) from TURBOMOLE output file did not agree with my calculated values. The ratio is in agreement, but absolute values are approximately 45 times larger. I assume that dDIP/dX in "dipgrad" are given in a.u. and L-matrix is dimensionless. Am I right? What could be a possible reason of disagreement? At the first stage, I've used only dDIP/dX and L-matrices to obtain dDIP/dQ matrix and no conversion prefactors.
Thanks in advance,
Olga

[mike]

Hi Olga, and everyone,

Does the file dipgrad contain gradients of transition dipoles (i.e. between excited states and ground states for Herzberg-Teller vibronic coupling) or only gradients of the state dipole moment?

If you now have a script to calculate dDip/dQ from a turbomole excited state frequency calculation and would be willing to share it, I'd love to make use of it!

A reply to a related post mentioned that obtaining the dP/dQ gradients would require modification of the NumForce script:

http://forum.marek-sierka.com/index.php?topic=408.msg1112#msg1112

Regards,

Mike