Questions regarding libwfa job control

Questions
#1

I was trying to learn how to plot NTOs and difference densities for TDDFT and EOM-CCSD excited states using libwfa but got quite puzzled by the job control. While I have already mentioned two things in trac ticket #2890, I would like to ask two additional questions to which there might be obvious answers:

(1) I ran Example 10.5 on the manual (10.2.9 General Excited-State Analysis‣ 10.2 Wave Function Analysis ‣ Chapter 10 Molecular Properties and Analysis ‣ Q-Chem 5.4 User’s Manual), which says “export of NOs, NTOs and NDOs in MolDen format, densities in cube format, and computation of the ESPs of these densities”. However, what I noticed is that in the “.plots” folder the job generated, there are only cube files but no .mo files. Similar behavior was observed for sample job “sp_adc_wfa.in” (1/2). These seem to indicate that the orbital plots in MolDen format is somehow suppressed by “make_cube_files = true”.

(2) I’m not quite sure how to generate the GS-EX difference density plots for EOM-CCSD. I ran the following calculation:
$molecule
0 1
C 1.331746 0.495076 -0.018037
O 1.023200 -0.680562 0.005977
H 0.567546 1.281583 -0.032606
H 2.385108 0.805133 -0.025252
$end

$rem
jobtype = sp
method = eom-ccsd
basis = 6-31+g(d)
symmetry = false
sym_ignore = true
scf_convergence = 8
thresh = 14
ee_states = 1
cc_state_to_opt = [1,1]
cc_eom_prop = true
cc_trans_prop = true
make_cube_files = true
state_analysis = true
$end

$plots
comment
50 -5.0 5.0
50 -5.0 5.0
50 -5.0 5.0
0 0 1 1
1
1
$end

The only cube file I got is called “_eomee-ccsd_transition_1_a_u_dens.cube”. When I change the method to “adc(2)”, then the behavior is much more like that of CIS/TDDFT (a whole lot of cube files for transition/difference/attach/detach densities). I was thus wondering if these density plots are available for EOM-CC methods.

Comments and suggestions will be appreciated. Thank you in advance!

#2

P. S. I suppose that the difference/attachment/detachment density plots for EOM-CCSD excited states should not be a problem given Figure 1 in this recently published paper https://wires.onlinelibrary.wiley.com/doi/full/10.1002/wcms.1595

#3

If you activate cc_eom_prop, then ccman2 should actually compute the excited state densities. And if those are there, then you can also do the attachment/detachment analysis. But I don’t know if anyone ever wrote the interface.

#4

Just to update the status here since some of my conversations with Felix are on the developer site:
For (1), it looks like the current setting of libwfa in Q-Chem supports one format of output at a time (cube files/molden format/fchk), so indeed the description of that sample job doesn’t exactly match what it actually gives.

For (2), it is definitely true that with cc_eom_prop = true, the state 1PDMs are computed and so are the properties related (e.g. the state dipoles). However, the 1PDM-related plots are not generated.

#5

One interesting observation: when I get rid of the line “cc_state_to_opt = [1,1]” in the input, I was able to get the trans/elec/hole densities. This is actually quite surprising (not entirely sure why).

Still no diff/attach/detach densities at the moment for the reasons as Felix commented in trac ticket #2890.

#6

Yes, one would really have to think about the whole workflow.

This has been written by different people at different times and no one really has had the big picture in mind.