I’m trying to generate cube files for ground state and delta-SCF state densities ($rem and $plot sections of the jobs below). However, when I later subtract them using gabedit, the difference density is zero. What could be the reason for such behaviour?
I don’t use gabedit, usually just subtract with a command-line awk script. Have you looked at the cube files? They are human-readable ASCI. Are they identical? Maybe run a test with a smaller grid to make it easier to see what’s going on.
Looking at your input, however, I don’t understand why the 2nd job should produce anything different from the first job. What are you trying to do here?
Thank you for replying. As I mentioned, I am performing deltaSCF calculation - I change the occupation manually so that it corresponds to HOMO-LUMO exctitation. My bad that I didn’t include the $occupied section of the second job above, I’ll append it below.
The cube files are indeed identical, but I don’t know why that is - given that the orbitals for two distinct states that I’m working with are actually different (I checked that through fchk file).
Are the SCF energies the same? My guess is that the 2nd calculation undergoes variational collapse to the ground state. To avoid that, you need a SCF convergence algorithm that is designed to converge non-Aufbau solutions. Choices in Q-Chem are MOM, IMOM, STEP, or SGM. In Q-Chem manual, see here: https://manual.q-chem.com/latest/sec_DeltaSCF.html
and in the literature, see here: https://doi.org/10.1021/acs.jctc.0c00502
I recommend starting with IMOM (not regular MOM). If that fails, try STEP or SGM, which are more robust but a bit more expensive.
It’s a bit hard to believe that this just works without some alternative convergence algorithm. What about the orbital energies? For successful delta-SCF, you should be able to see a virtual level that is below the HOMO, corresponding to the hole.
Well, this is a system with many excited states in close proximity to each other. Followng your advice I have run additional calculations with 1) IMOM, 2) STEP and 3) MOM. Using IMOM I get the same excitation energy as without any additional procedures, while with STEP and MOM I obtain a slightly lower lying state, with E(exc) = 3,23 eV.
As to the orbital energies, in all the cases the unoccupied 147 (hole) beta orbital is in fact higher than 148 alpha, so this part is fine.
However, I figured out my problem: the cube files weren’t generated for dSCF state at all, it worked only after I had removed “state_analysis true” from the second job.
That makes some sense in hindsight, sorry I didn’t catch that. The original cube file code (my code) is sort of orthogonal to the new libwfa code (STATE_ANALYSIS, Felix Plasser’s code).