I am trying to run an excited-state calculation in Q-Chem 6.3 using EOM-CCSD with the def2-TZVP basis set. My input is:
$rem
METHOD eom-ccsd
BASIS def2-TZVP
JOBTYPE SP
GUI 2
CC_TRANS_PROP 1
EE_TRIPLETS 3
EE_SINGLETS 3
AUX_BASIS RIJK-def2-TZVP
MEM_TOTAL 300000
CC_MEMORY 290000
CC_BACKEND XM
FROZEN_CORE TRUE
USE_LIBPT TRUE
CC_SINGLE_PREC 2
EOM_SINGLE_PREC 1
CC_SP_T_CONV 4
CC_SP_E_CONV 6
CC_ERASE_DP_INTEGRALS 1
LIBPT_MIXED_PRECISION 1
STATE_ANALYSIS TRUE
CC_EOM_PROP TRUE
CC_EOM_PROP_TE TRUE
CC_REF_PROP 1
NTO_PAIRS 2
SYM_IGNORE TRUE
CC_SYMMETRY FALSE
$end
The job never passes the SCF step — it just hangs without converging. I have tried tightening and loosening convergence thresholds, but no progress.
- Are there known issues with SCF convergence for EOM-CCSD with
wb97x-d3
orbitals or with large memory allocation? - Should I switch to a different reference (e.g. RHF → MP2 → CCSD)?
- Are there recommended keywords for improving SCF stability before the EOM-CCSD part starts (e.g.
SCF_ALGORITHM
,SCF_CONVERGENCE
,GUESS_CORE
,DIIS_GUESS
)?
Any advice even for pbs file ?
Any advice on how to get the SCF to converge in this setup would be greatly appreciated.