NTOs

Hello,
I was not able to find in the user manual that MRCC can calculate NTOs, but I am asking anyway, is there a way to extract information to plot NTOs?

If yes, is there anyway to associate those NTOs to the ADC(2) states, the way for example Turbomole does, in which the double excitation contributions and correlation contributions to the ground state are ignored for ADC(2), or other correlated methods? Thanks!

Edit: I see now that it is possible to obtain the transition density matrix using 'dens'. NumPy (scientific computing in Python) provides Singular Value Decomposition, so one may save the TDM and use Numpy to compute the NTOs. So, is it possible to save the TDM in a suitable way to be read by Numpy?

At this stage, only the natural orbitals can be visualized and analysed. For more details, I would recommend to read this topic:
mrcc.hu/index.php/forum/general-questions/150-adc-2-ntos#667

The calculation of NTOs from only the singles contribution is easy to implement. I will do it at your request in days. I would note that, an interface program to the TheoDORE package is under development, but it will take more time.

Hi David,
thanks for the clarification. I would really benefit from the calculation of NTOs, so it would be a great plus if you can include them in the code. Thanks a lot!

I think that if the NTOs can be dumped in cube files, then they may be plotted easily with standard packages, for example VMD. Otherwise, Molden files are of general use too. I am looking forward for the TheoDORE interface, but a local implementation would be fine for me at this moment.

If it makes things easier, another possibility would be to dump the transition density matrix and the basis info in a format suitable for NumPy, but it all depends if the interface for NumPy is already done for other purposes. MRCC local implementation is still easier for me, but if this route is easier for you, that's another thing to consider.

Thanks very much, David

Dear Edums,
the new files (cis.f source code and cis binary file) have been uploaded to the following folder: mrcc.hu/index.php/download-mrcc/category/5-mrcc-patches
Please recompile the code or overwrite the binary with the corresponding new file.

For the NTOs, please, set the molden=on and verbosity=3 keywords in your MINP file. Thereafter the corresponding files will be generated automatically. The files are compatible with the common packages (molden, jmol, etc.) and the orbital energies are replaced by the singular values.

Some additional notes:
1. In the case of CIS/TDHF/TDDFT/TDA-TDDFT calculation, the MOLDEN_CIS_NTO.# files are generated for each state and labelled via # in the order of the excitation energies.

2. In the case of ADC(2)/CC2/CIS(D_Inf) calculation, the MOLDEN_CORR_NTO.# files are generated. In the case of canonical calculation, the files are labelled in the order of the excitation energies of the second order method. In contrast, if a state-specific, for example, reduced-cost calculation, is performed the files are labelled in the order of the CIS excitation energies! The ADC(2) and CIS roots can be swapped. You can check if such rearrangement occured at the end of the output in the summary.

3. In the case of double hybrid calculation, the MOLDEN_DH_NTO.# files are generated and labelled in the order of the excitation energies.

If you have any further questions or something is not clear, please do not hesitate to ask!

Best regards,
Dávid

Dear David,
thanks a lot for your super quick help.
However, this solution is not working for me. I must be doing something wrong, or is my lack of training with this software, but I am not seeing the NTOs listed after running MCRR with the indicated parameters. I tried several combinations, but none of them gave me the NTOs. I am attaching a very short example of a test that did not work, to see if I am doing something fundamentally wrong,


Thanks for your help!

Your example works perfectly for me. The generated file for the first excited state is attached. You need to find these files on your disk where the calculation is performed. If you use any cluster environment you should copy these files back.

Maybe went something wrong with the update. If you use the binary code of the release version you should just overwrite the original executable (cis) with the patch, i.e., copy the patch to the directory where the executable can be found.
If you use the source code format you had to install the program. In this case, you should overwrite the cis.f file with the patch and recompile the code.