HSE06 with periodic systems

[Nóra Kovács]

Dear cp2k community,

I have been using the cp2k.9.1 version and I want to use for a periodic solid HSE06, but no matter what cutoff parameter I put in the input file it gives the same error:
*** WARNING in hfx_types.F:1968 :: Periodic Hartree Fock calculation      ***
 *** requested with use of a truncated or shortrange potential. The cutoff ***
 *** radius is larger than half the minimal cell dimension. This may lead  ***
 *** to unphysical total energies. Reduce the cutoff radius in order to    ***
 *** avoid possible problems.

Can someone tell me if is it a problem with the cp2k version or I have some mistakes in my input file?

Thank you for your help!

This is my input file:
&GLOBAL
  PROJECT 32_hse06
  RUN_TYPE ENERGY
  PREFERRED_DIAG_LIBRARY SL
  PRINT_LEVEL medium
&END GLOBAL


&FORCE_EVAL
  METHOD Quickstep

&DFT
    BASIS_SET_FILE_NAME BASIS_MOLOPT
    BASIS_SET_FILE_NAME BASIS_ADMM
    BASIS_SET_FILE_NAME BASIS_ADMM_MOLOPT
    LSD
    POTENTIAL_FILE_NAME GTH_POTENTIALS
    WFN_RESTART_FILE_NAME pbe_standard-RESTART.wfn
    SURFACE_DIPOLE_CORRECTION .TRUE.
    &MGRID
      CUTOFF 250
    &END MGRID

    &QS
      METHOD GPW
      EPS_FILTER_MATRIX 0.0e0
      EPS_DEFAULT 1.0E-17
      EPS_PGF_ORB 1.0E-30
    &END QS

    &POISSON
       PERIODIC XYZ #Direction(s) of PBC for calculating electrostatics
       PSOLVER PERIODIC #The way to solve Poisson equation
    &END POISSON

    &AUXILIARY_DENSITY_MATRIX_METHOD
            METHOD BASIS_PROJECTION
            ADMM_PURIFICATION_METHOD MO_DIAG
            !EXCH_CORRECTION_FUNC PBEX
    &END AUXILIARY_DENSITY_MATRIX_METHOD

    &SCF
      SCF_GUESS RESTART #Use wavefunction from WFN_RESTART_FILE_NAME file as initial guess
      MAX_SCF   500
      EPS_SCF   5.0E-5
      CHOLESKY INVERSE

      &OT .TRUE.
      &END OT

    &END SCF

    &PRINT
       &MO_CUBES
           WRITE_CUBE F
           NHOMO 1
           NLUMO 1
       &END MO_CUBES
    &END PRINT

    &XC
      &HF
        FRACTION 0.25
        &INTERACTION_POTENTIAL
          OMEGA 0.11
          POTENTIAL_TYPE SHORTRANGE
          CUTOFF_RADIUS 7.0
        &END INTERACTION_POTENTIAL
        &MEMORY
          EPS_STORAGE_SCALING 0.1
          MAX_MEMORY 900
        &END MEMORY
        &SCREENING
          EPS_SCHWARZ 1.0E-6
          SCREEN_ON_INITIAL_P FALSE
        &END SCREENING
      &END HF
      &XC_FUNCTIONAL
        &PBE
          SCALE_C 1.0
          SCALE_X 0.0
        &END PBE
        &XWPBE
          OMEGA 0.11
          SCALE_X -0.25
          SCALE_X0 1.0
        &END XWPBE
      &END XC_FUNCTIONAL
    &END XC
  &END DFT

  &SUBSYS
    &KIND W
            POTENTIAL GTH-PBE
           ! POTENTIAL_TYPE GTH
            BASIS_SET DZVP-MOLOPT-SR-GTH
            BASIS_SET AUX_FIT cFIT10
    &END KIND
    &KIND Nb
            POTENTIAL GTH-PBE
           ! POTENTIAL_TYPE GTH
            BASIS_SET  DZVP-MOLOPT-SR-GTH
            BASIS_SET AUX_FIT cFIT10
    &END KIND
    &KIND O
            BASIS_SET TZV2P-MOLOPT-GTH
            BASIS_SET AUX_FIT SZV-MOLOPT-GTH
            POTENTIAL GTH-PBE
            !POTENTIAL_TYPE GTH
    &END KIND
    &KIND Li
            BASIS_SET GTH-def2-QZVP
            BASIS_SET AUX_FIT cFIT4
            POTENTIAL GTH-PBE
    &END KIND
    &CELL
      A     18.79885064 0.0000000000000000 0.0000000000000000
      B     0.0000000000000000  18.79885064 0.0000000000000000
      C     0.0000000000000000 0.0000000000000000 42.63355178
      PERIODIC XYZ
    &END CELL
    &TOPOLOGY
        COORD_FILE_NAME     prst_32.xyz
        COORD_FILE_FORMAT   XYZ
    &END TOPOLOGY
  &END SUBSYS
&END FORCE_EVAL

[Frederick Stein]

Dear Nóra,

This warning is not of concern in case of HSE06 (and similar functionals). CP2K raises this warning to inform the user of a potential violation of the minimum image convention meaning that the range of the interaction potential is supposed to smaller than half of the distance to the nearest neighbor. Its violation leads to unphysical double-counting of interacting reducing the numerical stability, but may be ignored in case of HSE. It is mostly of concern if you employ a truncated Coulomb potential with a cutoff radius larger than half the distance to the nearest neighbor.

Considering your input file, I wonder about the tight EPS_DEFAULT and especially EPS_PGF_ORB. Did you check that you actually need these tight thresholds? Unfortunately, some regtests employ these tight values although they are usually not necessary. I have never used EPS_PGF_ORB lower than 1.0E-7. EPS_DEFAULT also determines the range of the short-ranged Coulomb potential. In your case, you will probably not be able to reduce it to prevent the warning. You may get a first guess by testing with PBE.

Best,

Frederick

[Frederick Stein]

PS: The CUTOFF_RADIUS is only relevant for the truncated Coulomb potential but ignored in your case. The internal cutoff radius r_c of the short-ranged Coulomb potential is chosen such that V(r_c)=EPS_DEFAULT.

[Nóra Kovács]

Thank you!