OK, so I tried this myself. Doing the following I can get to a set of orbitals in which the e is not split (the concern you expressed):
charge +1
dft
odft
iterations 1000
vectors output test.movecs
mult 1
xc hfexch
end
task dft
charge -1
dft
vectors input test.movecs
mult 3
xc hfexch
end
task dft
dft
vectors input test.movecs
mult 3
xc pbe0
max_ovl
end
task dft
What I effectively did is a Hartree-Fock calculation, using a different functional effectively reorder the orbitals enough to get the ordering I was trying to get. Then I do the PBE0 with max_ovl to force the same character on the occupied orbitals.
As you can see the two e-orbitals are lower in energy than the a1 orbital in the alpha space. I wonder if you really are having a e-state that is more of a multireference character.
Bert
Quote:Bock788 Jul 6th 9:32 pmSwapping the orbitals has not been working very well. Maybe I am doing something wrong? Here is my process:
-Establish permanent directory, run calculation with no specifications
-Look at output on ECCE, see which vectors need to be swapped
-Run calculation again, inputting .movecs and swapping the appropriate orbitals
The swap occurs, but sometimes, for example, I want to swap vector 199 with 202. These vectors do swap, but the vectors between them become disordered. This can be a problem, especially for the doubly degenerate LUMO of my molecule, which should not be split up. Is there a way to avoid this? |