B3LYP+XDM and RI problem

Dear all,

I’m trying to run a B3LYP+XDM/aug-cc-pVTZ calculation for a dimer of 63 atoms using Q-Chem 6.1 on a single node with 128 cores and 200 GB. These are my input keywords

$rem
JOBTYPE SP
EXCHANGE B3LYP
DFTVDW_JOBNUMBER 1
DFTVDW_METHOD 1
DFTVDW_PRINT 1
DFTVDW_KAI 800
DFTVDW_USE_ELE_DRV 0
MEM_STATIC 200
MEM_TOTAL 180000
AO2MO_DISK 200000
SCF_PRINT 1
THRESH 14
RI_J TRUE
RI_K TRUE
BASIS aug-cc-pvtz
AUX_BASIS gen
$end

$aux_basis

$end

I’m using the aug-cc-pvtz-rifit aux_basis from the basissetexchange website to speed up the calculations, which I didn’t add here just for simplicity. However, the calculation crashes after 50 minutes with the following message:

 -------------------------------------------------------
 OpenMP Integral computing Module
 -------------------------------------------------------
CFUNC_VDW_BR89 not done here.
 One-Electron    Energy = -12245.2982673104
 Total Coulomb   Energy =  5633.5862809485
 Alpha Exchange  Energy =   -26.8966862335
 Beta  Exchange  Energy =   -26.8966862335
 DFT   Exchange  Energy =  -206.5282101176
 DFT Correlation Energy =   -12.9188521726
 Kinetic         Energy =  2242.1387431056
 Nuclear Repu.   Energy =  4612.4686577497
 Nuclear Attr.   Energy = -14487.4370104160
 Ewald QM/MM     Energy =     0.0000000000
 ---------------------------------------
  Cycle       Energy         DIIS Error
 ---------------------------------------
    1   -2272.4837633695      9.42E-03  Gap = 3.14 eV

 Q-Chem fatal error occurred in module progman/main.C, line 178:

  Problem running Q-Chem: vector::_M_fill_insert

 Please submit a crash report at q-chem.com/reporter

Error: in the serial run

For comparison, a B3LYP/aug-cc-pVTZ for the same systems finished correctly in just 10 min. I noticed that for B3LYP+XDM, it seems that Q-Chem is not using RI.

I would appreciate any help in solving this issue. Please let me know what is the correct way to run with XDM corrections for large systems.

Best regards,

XDM is presently implemented only in the old version of Q-Chem’s SCF code, which does not have RI available. Apologies, we should backstop this but there is no RI for DFT+XDM.

Dear Professor Herbert,

Thank you for the clarification.

Best,