PBE0/ADMM single point error - Cholesky decompose failed

[Quentin Pessemesse]

Hello again,

I encounter an issue when running a hybrid single point of a Ni nanoparticle with a carbene ligand. I start from a PBE wavefunction with the same parameters, and get the error "Cholesky decompose failed" before the first SCF step.

I tried :

- Switching to a Davidson diagonalization algorithm

- Turning off CHOLSEKY in the &SCF section

- Reducing the size of the admm aux_fit basis set (from dzp to dz)

None of which solved the issue. Does anyone have advice on how to proceed?

Best,

Q.

NB: Here is the force_eval section of my input.

&FORCE_EVAL
 METHOD QS
 
  &DFT
   LSD
   BASIS_SET_FILE_NAME   ${BASIS_FILE}
   BASIS_SET_FILE_NAME ./BASIS_ADMM_PBE0
   POTENTIAL_FILE_NAME ${POTENTIAL_FILE}
   RESTART_FILE_NAME  ./${PROJECT}_PBE-RESTART.wfn
   CHARGE ${CHARGE}
   MULTIPLICITY ${MULTIPLICITY}

    &MGRID
     NGRIDS ${PW_NG}
     CUTOFF ${PW_CUTOFF}
     REL_CUTOFF ${PW_REL_CUTOFF}
    &END MGRID

    &QS
     EPS_DEFAULT ${EPS_DEF}
     EPS_PGF_ORB 1.0E-32
     EXTRAPOLATION ${EXTRAPOLATION}
     EXTRAPOLATION_ORDER ${EXTRAPOLATION_ORDER}
    &END QS

    &SCF
     SCF_GUESS ${GUESS}
     EPS_SCF   ${EPSSCF}
     MAX_SCF ${MAX_SCF_IN}
     ADDED_MOS ${ADDED_MOS}
     CHOLESKY INVERSE

     @if ${SMEAR_MIX_YN} == 1

      &SMEAR
       METHOD FERMI_DIRAC
       ELECTRONIC_TEMPERATURE [K] 1200
      &END SMEAR
      &DIAGONALIZATION
       ALGORITHM STANDARD
      &END DIAGONALIZATION
      &MIXING
       METHOD BROYDEN_MIXING
       ALPHA   0.04
       BETA    1.5
       NBROYDEN  8
      &END

     @endif

      &PRINT
        &RESTART
          &EACH
           QS_SCF 50
          &END
         ADD_LAST NUMERIC
        &END
      &END

    &END SCF

    &AUXILIARY_DENSITY_MATRIX_METHOD
     METHOD BLOCKED_PROJECTION
     ADMM_PURIFICATION_METHOD NONE
    &END AUXILIARY_DENSITY_MATRIX_METHOD

    &XC

      &XC_FUNCTIONAL PBE0
        &PBE
         SCALE_X 0.75 #75 % GGA Exchange
         SCALE_C 1.0 #100 % GGA Correlation
         PARAMETRIZATION REVPBE
        &END PBE
      &END XC_FUNCTIONAL

      &HF
       FRACTION 0.25 #25 % HF Exchange
       &SCREENING
        EPS_SCHWARZ 1.0E-6 #Ignore contributions smaller than 1.0E-6
        SCREEN_ON_INITIAL_P TRUE
       &END SCREENING

       &MEMORY
        MAX_MEMORY 3500
        EPS_STORAGE_SCALING 0.1
       &END MEMORY

       &INTERACTION_POTENTIAL
        POTENTIAL_TYPE TRUNCATED
        CUTOFF_RADIUS 9.0
        T_C_G_DATA ./t_c_g.dat
       &END INTERACTION_POTENTIAL

     &END HF

      &XC_GRID
        XC_SMOOTH_RHO NN50
        XC_DERIV NN50_SMOOTH
      &END
   
      @if ${D3_YN} == 1

      &vdW_POTENTIAL
       DISPERSION_FUNCTIONAL PAIR_POTENTIAL
        &PAIR_POTENTIAL
         TYPE DFTD3
         CALCULATE_C9_TERM .FALSE.
         REFERENCE_C9_TERM .TRUE.
         PARAMETER_FILE_NAME ./dftd3.dat
         REFERENCE_FUNCTIONAL PBE0
         R_CUTOFF [angstrom] 24.
         EPS_CN 1.0E-6
        &END PAIR_POTENTIAL
      &END vdW_POTENTIAL

      @endif
     
    &END XC
  &END DFT

  &SUBSYS
    &CELL
     ABC ${CELL_SIZE_XYZ}
     MULTIPLE_UNIT_CELL 1 1 1
    &END CELL
   
    &TOPOLOGY
     COORD_FILE_NAME ${PROJECT}.xyz
     COORDINATE XYZ
     CONNECTIVITY OFF    
    &END TOPOLOGY
   
    &KIND Ni
     BASIS_SET ORB DZVP-MOLOPT-SR-GTH-q18
     BASIS_SET AUX_FIT admm-dz-q18
     POTENTIAL GTH-PBE0-q18
    &END KIND

    &KIND C
     BASIS_SET ORB DZVP-MOLOPT-SR-GTH-q4
     BASIS_SET AUX_FIT admm-dz-q4
     POTENTIAL GTH-PBE0-q4
    &END KIND

    &KIND N
     BASIS_SET ORB DZVP-MOLOPT-SR-GTH-q5
     BASIS_SET AUX_FIT admm-dz-q5
     POTENTIAL GTH-PBE0-q5
    &END KIND

    &KIND H
     BASIS_SET ORB DZVP-MOLOPT-SR-GTH-q1
     BASIS_SET AUX_FIT admm-dz-q1
     POTENTIAL GTH-PBE0-q1
    &END KIND

  &END SUBSYS  

&END FORCE_EVAL

[Quentin Pessemesse]

Update: I solved this specific issue by setting EPS_PGF_ORB 1.0E-20 instead of 1.0E-32