Conversion of PBEPBE and pbex

[wanghong]

Hello：
I recently used the method of PBEx to calculate phosphorescence with Dalton. I used the method of pbepbe when I used Gaussin optimization, but I couldn't use pbepbe when Dalton calculated phosphorescence, so I chose pbex. I wonder what's the difference between pbepbe and pbex? and What keywords do I need to match them?
Looking forward to your reply.

[hjaaj]

Gaussian PBEPBE is just PBE in Dalton input (and in many other programs).

PBEx is just the exchange part, without the correlation part of the PBE functional.

[wanghong]

@ MPI MASTER, node no.: 0
@ Reason: functional not fully implemented/tested for QR

Total CPU time used in DALTON: 10 minutes 33 seconds
Total wall time used in DALTON: 10 minutes 34 seconds


QTRACE dump of internal trace stack

========================
level module
========================
3 RSPDRV
2 DALTON
1 DALTON main
========================



When I use a function like PBE, the output document will have errors, as shown above. How can I modify the key words in the input file to make the output result normal?

[kennethruud]

As the error message states, it is not possible to use this functional in quadratic response calculations. Please consider using a different functional.

[wanghong]

Is there no way to use this functional PBE by mediating the parameters in the input file?

[kennethruud]

Not in a quadratic response calculation. The problem is that there you will need derivatives of the exchange-correlation kernels which have not been implemented for this functional. LSDalton will be able to do this in principle, since it uses the XCFun library to generate the necessary XC kernel derivatives, but then instead LSDalton cannot calculate phosphorescence (only singlet quadratic response). The development version of ReSpect should now be able to calculate the require transition moments using the PBE functional, but then instead you would do the calculations at the relativistic two- or four-component level of theory. But at the same time, it is bleeding-edge code, and we have not really thoroughly tested it for spin-forbidden transitions.

Kenneth

[wanghong]

OK, I got it.
Thank you for your reply,