Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly  
We are NATO-free zone

Re^4: Is it possible to print out MCSCF orbitals at the end of geometry optimization

Alex Granovsky

Dear Evgeniy,
Dear Thom,

Indeed, the SA-MCSCF optimization or conical intersection location
runs do not print converged orbitals at the end of geometry search.
This is by code design and was done intentionally. The reason is
exactly as guessed by Thom - final orbitals usually correspond to
a converged SA-MCSCF with slightly perturbed weights, i.e. these
orbitals are not exactly the orbitals that user wants to be printed.

However, these orbitals are punched so one can find them at the end
of a punch file. One can use these orbitals as a starting guess
(very good guess actually) for a single point energy calculation,
which will print converged orbitals.    

All the best,

On Wed Aug 31 '16 10:38am, Thomas Pijper wrote
>Dear Evgeniy,

>I've try it myself and observed the same behavior. A RHF geometry optimization with $STATPT NPRT=0 causes orbitals to be printed at the final geometry. Final orbitals are also printed for the same job at the SS-MCSCF level. However, at the SA-MCSCF level final orbitals are not printed.

>I do not know why this is the case. If I were to hazard a guess, I'd say that the final orbitals of the current geometry are no longer present in the system memory at the end of the gradient calculation (as calculation of the gradient requires two additional SCF procedures) and thus aren't available for printing -- but this is speculation on my part. Perhaps Alex can comment on this.
>Kind regards,
>On Tue Aug 30 '16 6:19pm, GrEv wrote
>>Dear Thom,

>>Thanks for your reply. In my case there are only population analysis printed out at the end of geometry optimization. Actually, I though that the problem/inconvenience can relate to that I do a SA-MCSCF geometry optimization (state-averaging over 2 states) rather than a state-specific one. Can this be the reason?

>>Best regards,

>>On Mon Aug 29 '16 9:36pm, Thomas Pijper wrote
>>>Dear Evgeniy,

>>>If I recall correctly MCSCF orbitals are always printed at the end of a successful geometry search, provided you set NPRT=0 in $STATPT. Note that '0' is the default value of this option, so not specifying any value for NPRT should already cause MCSCF orbitals to be printed.
>>>Kind regards,
>>>On Mon Aug 29 '16 5:06pm, GrEv wrote

>>>>I would like to print out MCSCF orbitals at the end of MCSCF geometry optimization run. Any point how to do that will be greatly appreciated.

>>>>Best regards,

[ Previous ] [ Next ] [ Index ]           Wed Aug 31 '16 8:05pm
[ Reply ] [ Edit ] [ Delete ]           This message read 141 times