Molcas Forum

dismas

Member

Registered: 2022-02-18

Posts: 2

[SOLVED] Orbitals obtained from EXPBAS, read via FILEORB

Dear OpenMolcas-Users,
I have performed an SCF-calculation at the ANO-RCC-MB level of theory, followed by a CASSCF-calculation in the minimal basis, from which I obtained a RasOrb-file. Inspection of this file shows qualitatively correct orbitals and the correct type-indices for each orbital in their respective representation (within the C2v-group). At the MB-level of theory, RAS2 contains 4 orbitals in the a1-representation.

I have now tried to expand this orbital file into the ANO-RCC-VTZP basis and to use the obtained ExpOrb-file as input for further calculations.
EXPBAS terminates without error, but the active-space I obtain from it is incorrect. Inspection via pegamoid shows that at the VTZP-level  the previously highest a1-orbital was exchanged by an orbital of the same representation. It is now incorrectly the first of the Secondary orbitals.

How can I avoid this behavior and expand my MB-calculations into a higher basis while maintaining the qualitatively correct composition of the active space I chose at ANO-RCC-MB? Or is it possibly an error I am making while reading in the orbital-file via FILEORB and TYPEINDEX?

&GATEWAY
	Title=Basis Expansion + CASSCF
	coord=coord.xyz
	basis=ANO-RCC-VTZP
	group= XY Y

>> COPY $WorkDir/$Project.RunFile $CurrDir/$Project.RunFil2

>> COPY $CurrDir/MB_orbitals INPORB
>> COPY $CurrDir/RunFil1 RUNFIL1
>> COPY $CurrDir/$Project.RunFil2 RUNFIL2

&EXPBAS
>>COPY $WorkDir/$Project.ExpOrb $CurrDir/$Project.ExpOrb

&SEWARD
&RASSCF
	FILEORB=$CurrDir/$Project.ExpOrb
	TYPEINDEX
	spin=2
	nactel=9

The attempt to manually set the typeindex of the erroneous orbital to "s" and the relevant orbital to "2" and then calling FILEORB and TYPEINDEX most confusingly did not change the result. As did invoking the ALTER keyword and switching the order of these two orbitals. The program terminates normally, but these options do not change the result.

I am deeply grateful for your help in correcting my mistakes and misunderstandings.

Last edited by dismas (2022-02-18 12:59:06)

Ignacio

Administrator

From: Uppsala

Registered: 2015-11-03

Posts: 1,079

Re: [SOLVED] Orbitals obtained from EXPBAS, read via FILEORB

EXPBAS terminates without error, but the active-space I obtain from it is incorrect.

Do you mean the active space in the ExpOrb file, or the active space after you run RASSCF on it?

If it's the former, it may be a bug, and I'd like to see a full example.
If it's the latter, it's probably just that your RASSCF calculation doesn't "like" that orbital in the active space and prefers another one. Maybe the occupation is too close to 0 and a lower energy is obtained by including the other orbital instead. This can usually be avoided by using a state-average method, and making sure you include some electronic state with significant occupation for the problematic orbital.

dismas

Member

Registered: 2022-02-18

Posts: 2

Re: [SOLVED] Orbitals obtained from EXPBAS, read via FILEORB

Dear Ignacio,
thank you for your quick help and expertise! It was exactly how you outlined it.
The ExpOrb-file with a correct active-space was obtained but then lost in the subsequent RASSCF-calculation for the reasons you mentioned. SA-RASSCF helped in avoiding the intrusion of the undesired orbital.
Thanks again.