EOM-CCSD Natural Transition Orbitals

Questions
#1

I’m currently using EOM-EE-CCSD with a ROHF reference to describe the excited states of NO+C2H2 with Q-Chem 5.3.2 I can successfully export the NTOs in molden format and view them in MacMolPlt. However, when I instead save the NTOs to the .fchk file and open it in IQmol, the NTOs for some states are missing from the list in IQmol (only 11 out of 16 are present). When I search the .fchk file for “NTO” I only find 44 instances (alpha NTO coefficients, beta NTO coefficients, alpha NTO amplitudes, and beta NTO amplitudes for 11 states). This suggests to me that NTO data is not being written out into the .fchk file for some of the states.

The input file is pasted below. Is there something I should do to get all of the states written out into the .fchk file or is this a bug?

Thanks!

$molecule
0 2
C 5.2500134027 -0.5996502380 -0.0000000000
C 5.1404579984 0.6190952452 0.0000000000
H 5.3452594616 -1.6723344318 -0.0000000000
H 5.0426819765 1.6916193813 0.0000000000
N -4.7876413442 -0.5779194909 -0.0000000000
O -4.9021600544 0.4886851805 0.0000000000
$end

$rem
jobtype sp
method eom-ccsd
basis aug-cc-pvdz
ee_states [16]
cc_symmetry false
max_scf_cycles 150
scf_algorithm diis_gdm
thresh 14
max_sub_file_num 512
mem_static 2000
cc_memory 58000
mem_total 60000
scf_print 0
scf_guess gwh
scf_final_print 0
GUI 2
symmetry false
sym_ignore true
cc_eom_prop true
cc_eom_prop_te true
cc_trans_prop=1
cc_ref_prop=1
state_analysis=true
nto_pairs=2
$end

#2

I wrote the NTO code for TDDFT but I’m unfamiliar with how it’s implemented for EOM-CC. That said, a couple of things come to mind: (1) NTOs are usually state-specific, so are you sure you should be getting them for all states at once? (2) Have you checked Q-Chem manual for ‘libwfa’ analysis options?

#3

I double checked and NTOs for all 16 states are produced when I used molden format. I glanced through them and they are what I would expect (the two molecules are 10 Angstroms apart in this geometry so I know what the states are).

I also double checked and the input file used to produce the .fchk which is missing random states is identical to that used to generate the molden format files (outside of removing the molden_format true line and adding GUI 2).

#4

We usually visualize NTOs using molden files that libwfa generates and not *fck file. These files (written in the directory from which Q-Chem is run) can be visualized with Gabedit or jmol. There could be an issue with what is written to fck… I suggest to check these files.

#5

I am definitely late to this thread, but the above input file runs UHF EOM-EE-CCSD calculations, correct?

My purpose in asking that is I have been looking at diatomic clusters (eg N2-CN) using ROHF EOM-EE-CCSD and the .mo files have “UHF” in the file name. I just wanted to make sure I am getting the correct results.
Thanks,
Benj

$molecule
0 2
C
N 1 1.147224
N 2 3.334215 1 97.989916
N 3 1.096642 2 162.318545 1 178.025654 0
$end

! SP of EOM-EE-CCSD/aug-cc-pvtz
$rem
mem_static 1000
mem_total 2000
cc_memory 1000
JOBTYPE SP
THRESH 14
SYM_IGNORE True
SYMMETRY FALSE
CC_SYMMETRY False
METHOD EOM-CCSD
UNRESTRICTED false
BASIS aug-cc-pvdz
PURECART 1111
SCF_convergence 8
SCF_algorithm gdm
MaxSCF 1000
SCF_GUESS sad
EOM_DAVIDSON_CONVERGENCE 9 EOM amplitude convergence
CC_T_CONV 9 CCSD amplitudes convergence
EE_STATES [5]
CC_STATE_TO_OPT [1,1]
EOM_NGUESS_SINGLES 3
CC_T_CONV 8 CCSD amplitudes convergence
CC_MAX_ITER 500
CC_DOV_THRESH 75001
CC_SCALE_AMP 90
CC_TRANS_PROP true
STATE_ANALYSIS true
NTO_PAIRS 5
MOLDEN_FORMAT true
$end

#6

What files are you referring to specifically? plot.mo is almost certainly Hartree-Fock. Suggest using libwfa capabilities (as described in the manual) to make sure you are getting the EOM-CCSD states.

#7

Hi Dr. Herbert,
My apologies for the ill-posed question. I am trying to get the NTO files to view in Molden.

eomee_ccsd_uhfref_1_1_eomee_ccsd_uhfref_1_2_nto.mo

and the other natural transition orbital files (and NO files) are generated at the end of the run, though they don’t end up in the plots folder and are in the work directory instead. The above has “uhfref” in the file name even if I use “unrestricted false” to switch over to ROHF.
Thanks,
Benj