Page **1** of **1**

### Excited to excited states dipole moments

Posted: **26 Mar 2014, 14:36**

by **jd140**

Dear all,

I have been a number of quadratic response calculations with double residue, using DFT (B3LYP and CAM-B3LYP), in order to obtain the excitation energies and transition dipole moment for the calculation of excited-state absorption spectra.

I have compared the results provided by Dalton2013 with two other chemistry packages. It turns out that the excitation energies and ground-state to excited-state transition dipole moments are essentially the same between the three quantum chemistry packages; however, all the transition dipole moments between excited-state differ greatly between Dalton and the two other quantum chemistry packages (which provide basically the same results). Do you have any idea/suggestion why this is the case?

Thanks very much in advance,

JCD

### Re: Excited to excited states dipole moments

Posted: **26 Mar 2014, 15:19**

by **lyzhao**

You should tell us "what the other two quantum chemistry packages are".

Lan

### Re: Excited to excited states dipole moments

Posted: **26 Mar 2014, 16:00**

by **jd140**

Indeed, I wanted to avoid giving names, but I realise it's difficult to help without knowing which packages I'm talking about. So that's:

- Firefly (ex- PC-GAMESS):

http://classic.chem.msu.su/gran/gamess/index.html
- Gaussian09:

http://www.gaussian.com/g_prod/g09.htm
Thanks,

JCD

### Re: Excited to excited states dipole moments

Posted: **27 Mar 2014, 01:44**

by **taylor**

So before trying to analyze this issue of comparing the results from the codes, can I ask if you are \emph{sure} that all three codes are setting about the calculation the same way? I ask because as far as I know Dalton is considerably more sophisticated than most codes in doing these high-level elaborate response-type calculations. It would be possible, for instance, for a code to use a response calculation to generate, in essence, a set of excited-state wave functions, and then to calculate transition dipole moments between those states by taking transition matrix elements of the dipole operator between those excited-state wave functions.

However, if I understand things correctly, it will not always be the case that the full response approach taken in Dalton will agree with such a "matrix element-based" approach. It will I think depend on the particular type of wave function and other issues, and I suspect this is only one way among many so-called "response" calculations may differ from the Dalton approach. So before anyone starts to chase this, I think it's essential to establish whether in principle Dalton and your other two codes are doing exactly the same calculation, or at least claim they are. I don't mean that they simply say "we can do transition moments between excited states". I mean that they are doing this using quadratic response the same way it is described in the various papers cited in the Dalton documentation. I think you may have to dig into the literature cited by the other programs a bit for this.

Best regards

Pete

### Re: Excited to excited states dipole moments

Posted: **27 Mar 2014, 09:41**

by **taylor**

P.S. You are aware that some excited-state properties are \emph{differences} from ground-state values, so they are relative, not absolute numbers? (It is in the documentation...!)

Best regards

Pete

### Re: Excited to excited states dipole moments

Posted: **27 Mar 2014, 11:22**

by **jd140**

Hi Peter,

Thanks for your reply it gives me a direction of where to start looking to explain this discrepancy. I will keep you updated.

I'm not sure I understand what you mean in your P.S. though.

Best regards,

JCD

### Re: Excited to excited states dipole moments

Posted: **27 Mar 2014, 13:44**

by **olav**

He means that if you form a matrix of dipole transition moments the diagonal values are

<k|z|k> - <0|z|0>

i.e. the change in dipole moment of the excited state relative to the ground state.

Regards,

Olav

### Re: Excited to excited states dipole moments

Posted: **27 Mar 2014, 14:11**

by **jd140**

Hi Olav,

Thanks for the clarification. Indeed, I was aware of this.

JCD