Firefly and PC GAMESS-related discussion club
Learn how to ask questions correctly
Re^2: Orbitals sorting with firefly
Thanks for your comments, patience and sorry for my fragmented data.
From the manual about JORDER
-------------- CITATION --------------------
Functions the same as IORDER, but applies to the beta MOs of a UHF wavefunction.
This keyword also has an alternative meaning. By default, the IORDER and JORDER reordering instructions are applied after any processing of orbitals like orthogonalization or symmetry adaptation. However, for inputs with no separate beta orbital set (e.g., RHF or MCSCF), the JORDER array, if given, allows one to reorder orbitals prior to the orbital processing stages. This feature is intended to work together with IORDER which additionally reorders orbitals after processing them. An example of a typical usage scenario is the use of MOs generated for one molecular geometry on a different molecular geometry. By putting the most important orbitals (say, orbitals forming the active space in the MCSCF) to the beginning of the orbital list before executing the processing steps, one can minimize the impact of the processing on these particular orbitals. The correct order can then be recovered using the IORDER array.
-------------- END OF CITATION --------------------
I am trying to:
1. Upload NBO orbitals.
2. Construct the active space firstly with JORDER in position 1-7 (source orbitals are 30, 32, 104,122,119,477,475; i know them from nbo log, see previous output file )
3. Recover the active space secondary with IORDER in position 116-123
4. Try to move electron from carbon double bond (30) into oxygen-Fe interface (104,122,119) or into antibonding orbitals to simultaneously break O-O bond. In addition, because O-O is triplet, 104 and 122 are two alpha occupied orbitals.
5. I believe that there should be a (weak) joint between reaction coordinate and route through states in mcscf so I freeze my reaction coordinate with DLC with 0.1A sampling rate and consider two states in each point (planning to screen ration between states to find a Ďminimalí route in energy/wstates/reaction coordinate space).
6. Because itís a part of a protein, several molecular fragment as frozen in space as it is observed in xray images. The geometry is however from optimized RHF of significantly bigger protein fragment, then optimized GVB for Fe, O2, and several atoms of cofactor only (rest are frozen) and then planning to use optimized mcscf for these atoms for energy evaluations with XMCQDPT2.
Thatís what my attached input should do. Does this make any sense?
Thanks for the help,
On Sat Sep 27 '14 6:20pm, Alex Granovsky wrote
>frankly, I do not understand what exactly you are trying to achieve.
>I'm afraid you are trying to use iorder/jorder directives for purpose
>they are not designed for. Specifically I do not understand why you
>believe that the ordering of the initial guess orbitals should affect
>NBOs and their order. In addition it is not clear to me why exactly
>are you using jorder array. It is seldom used for non-UHF calculations
>and it serves very specific purposes when used.
>I'm guessing your previous posts on problems with NBOs and and with
>iorder/jorder directives are not correct as you used these orbitals
>and the reordering instructions by a wrong way. In particular I mean
>the following threads:
>Please help us to help you by explaining us in details what is the
>exact goal you was trying to achieve with your sample input fragments.
| ||This message contains the 3260 kb attachment|
[ nbo_problem.tar.bz2 ] unconverged out and inp for the orbital ordering problem
Sun Sep 28 '14 2:08am
This message read 570 times