SOC calculation in EOM-CCSD

Questions
#1

I want to calculate SOC elements between singlet and triplet states from EOM-CCSD calculation, but I get the following lines:

State OPDM analysis using libwa module will be excecuted

contents of the context
/canmo/ao/dx_bb : ??? (libtensor::expr::btensor_i<2ul, double>)
/canmo/ao/dxx_bb : ??? (libtensor::expr::btensor_i<2ul, double>)
.
. (There is a long list of various parameters here)
.
/rawmo/symmetry/pg/irreps : ??? (std::map<unsigned long, std::string, std::less, std::allocator<std::pair<unsigned long const, std::string> > >)
/rawmo/symmetry/pg/name : C1 (std::string)

Q-Chem fatal error occurred in module /scratch/qcdevops/jenkins/workspace/build_qchem_linux_distrib/tags/qc601rc1/qchem/ccman2/qchem/ccman2_main.C, line 26:

Key ‘g1p_bb_aa’ is not known.

What is making the error here?

#2

If this is still causing you problems, please provide your Q-Chem version number along with an input file that exhibits the problem. Use the preformatted text box for the latter, </>

#3

I am using Q-Chem 6.0 and the input file is

$molecule
0 1
O       -0.340078    0.171394   -0.000000
H        0.627988    0.226708    0.000000
H       -0.611448    1.102297   -0.000000
$end

$rem
  INPUT_BOHR FALSE
  MAX_SCF_CYCLES 30
  METHOD     EOM-CCSD
  BASIS      cc-pvdz
  SYM_IGNORE  TRUE
  EE_SINGLETS 1
  EE_TRIPLETS 1
  JOBTYPE SP
  SCF_ALGORITHM DIIS
  STATE_ANALYSIS TRUE
  CC_EOM_PROP TRUE
  CC_CONVERGENCE 5
  CALC_SOC 1
  EOM_NGUESS_SINGLES 5
  EOM_NGUESS_DOUBLES 1
  EOM_DAVIDSON_CONVERGENCE 4
  EOM_DAVIDSON_THRESHOLD 00102
  EOM_DAVIDSON_MAX_ITER 32
  EOM_DAVIDSON_MAXVECTORS 240
$end
#4

The issue is the STATE_ANALYSIS keyword, which triggers wave function analysis via the libwfa module but is somehow crashing this job. With that removed, the job runs to completion.

It’s usually a good idea, in an input like this with many different input options, to test some combinations and see what is causing the problem. In this case the last line of useful text was a clue:

State OPDM analysis using libwa module will be executed

Further testing suggests that CALC_SOC and STATE_ANALYSIS appear to be incompatible. I will submit a bug ticket.

#5

Thank you for the answer!

#6

No, I coded a special type of analysis in libwfa for SOC, which should be activated passing both CALC_SOC and STATE_ANALYSIS. It used to be working. Here is the paper describing it: P. Pokhilko, A. I. Krylov. J. Phys. Chem. Lett. 10, 4857 – 4862 (2019)
May I ask which transition is of interest for analysis? EOM-to-EOM or REF-to-EOM?

#7

I can reproduce the crash with v. 6.1, so it seems to be broken. I have created a ticket.

#8

I am actually interested in EOM-to-EOM transition but both REF-to-EOM and EOM-to-EOM cases seem to crash when STATE_ANALYSIS is set to TRUE.

#9

I looked into the issue. Please modify the rem section of the input as shown below. It works for Q-Chem 6.1. If you want EOM-to-EOM transitions, you can specify CC_TRANS_PROP 2.

$rem
  INPUT_BOHR FALSE
  MAX_SCF_CYCLES 30
  METHOD     EOM-CCSD
  BASIS      cc-pvdz
  SYM_IGNORE  TRUE
  EE_SINGLETS 1
  EE_TRIPLETS 1
  JOBTYPE SP
  SCF_ALGORITHM DIIS
  STATE_ANALYSIS TRUE
  MOLDEN_FORMAT TRUE
  CC_EOM_PROP TRUE
  CC_REF_PROP TRUE
  CC_TRANS_PROP 1                                                                                                          
  CC_CONVERGENCE 5
  CALC_SOC 1
  EOM_NGUESS_SINGLES 5
  EOM_NGUESS_DOUBLES 1
  EOM_DAVIDSON_CONVERGENCE 4
  EOM_DAVIDSON_THRESHOLD 00102
  EOM_DAVIDSON_MAX_ITER 32
  EOM_DAVIDSON_MAXVECTORS 240
$end