[molpro-user] Instability problems

Kirk Peterson kipeters at wsu.edu
Thu Nov 10 01:57:06 GMT 2005


this is not a bug.  MCSCF natural orbitals are always ordered by  
occupation, not energy, since in general there is not a suitable Fock  
operator as in a HF calculation to define orbital energies.  What you  
can do is generate pseudo-canonical orbitals which uses an effective  
Fock operator.  This might "fix" your problem.  Just use the keyword  
"canonical" instead of "natural" when you save/generate your orbitals.



On Nov 9, 2005, at 5:20 PM, Yann P. Tambouret wrote:

> I have had a similar issue.
> I can rotate the orbitals so that they are in the correct energy- 
> ordered order. I use the merge program.
> When I run a subsequent CI calc. with the NOEXC option, I no longer  
> see the instabilities.
> My concern is that the MCSCF program is  optimizing the orbitals  
> for the wrong CI wavefunction(s), because as you pointed out, if  
> the orbitals are not in the correct order, then the active space is  
> not defined correctly.
> Is this actually a bug, as it seems?
> Attached is a pdf file showing what I've seen. This is a 7 B2g SA  
> MCSCF calc. for molecular oxygen. Point (1) shows from the start  
> that the B1u orbitals are miss labeled, but this does not have a  
> great affect on the surfaces, as you can see in the before (bottom)  
> and after (top) reordering. Point (2) shows a case where the  
> surfaces are greatly affected by a rapid adjustment in the labeling.
> Any help in this area would be greatly appreciated.
> Yann Tambouret
> Boston University
> Constantine Koukounas wrote:
>> Dear Molpro list,
>> When i run a CASSCF calculation the orbitals (of a particular  
>> symmetry
>> block) are
>> not sorted in ascending energy order that causes problems since  
>> some core
>> orbitals penetrate in my active space and valence orbitals in the  
>> core
>> space.The
>> problem of course is more severe at the CI level since wrong  
>> electrons are
>> excited.
>> Rotation does not work.What is the generic problem for that sorting?
>> Instability problems result as a natural consequence of it.
>> Sincerely,
>> C. Koukounas
>> <compare.pdf>

More information about the Molpro-user mailing list