Firefly and PC GAMESS-related discussion club


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


Re: MCSCF for a large system.

Pavlo Solntsev
pavlo.solntsev@gmail.com

On Wed Sep 10 '14 3:19am, alex wrote
------------------------------------
>Dear QChemists,
>Would you please share some ideas of treating of protein-substrate complex with mcscf method? I have some problems with setting up of active space, maybe you can advise something:
>1. Is it possible to print transposed (and even better then sorted) matrix of
eigenvectors as well? It’s not a problem to find interesting molecular orbitals if the system consists of the only few but the selection is just overwhelming if there are thousands orbitals. My life would be much easier if firefly highlight the several orbital to consider for the active space instead of all that mess.
>2. How to suggest amount of states (NSTATE) and their weights (WSTATE) from knowledge about the chemical reaction?

From my experience, run first a job using say 5 states, but putting only first in wstate array (wstate(1)=1,-0). If you see states close to your ground state, they may affect your ground state and you should consider them by putting to the wstate array. However, geometry optimization for SA-CASSCF was discussed here. Do a search.

How it is related to the active space?

I afraid, there is no relationship.

>3. How to carry out geometry optimization for mcscf? As I understand the orbitals will change at each step of the search. But I can reorder/select them only at the initial configuration… How the correct active space is determined/maintained during geometry search?

Every next step during a geometry optimization uses orbitals from previous. With good starting geometry and the correct active space it should not be a problem.


>4. My system is oxidase, i.e. triplet. But if I am trying to set MULT=3 and NAOS=2 NBOS=0 it complains “NAOS.NE.NBOS IN DRT INPUT DOES NOT MAKE SENSE.” How to model the triplet then and why there is no sense?

relevant part of the output file would be very informative for this case.

>Thanks,
>Alex

>


Pavlo.

[ Previous ] [ Next ] [ Index ]           Thu Sep 11 '14 5:30am
[ Reply ] [ Edit ] [ Delete ]           This message read 568 times