[CP2K-user] [CP2K:21583] Suspect CAM-B3LYP NMR Shieldings with GAPW + Linear Response

[Jürg Hutter]

Hi

Hybrid functionals are not supported for NMR chemical shift calculations.

regards
JH

________________________________________
From: cp2k at googlegroups.com <cp2k at googlegroups.com> on behalf of Sergio Tavares <sergiortf at gmail.com>
Sent: Wednesday, June 25, 2025 4:19 PM
To: cp2k
Subject: [CP2K:21577] Suspect CAM-B3LYP NMR Shieldings with GAPW + Linear Response

Dear CP2K developers and users,

I am performing linear response NMR calculations using the GAPW method and the CAM-B3LYP functional.

I constructed linear regression curves for a set of small organic molecules (benzene, toluene, cyclohexane, hexane, acetone, ethyl acetate, dimethoxyethane, diethyl ether, dioxane, and TMS), comparing PBE and CAM-B3LYP. These regressions were done using both cc-pVQZ and aug-cc-pVQZ basis sets, with and without ADMM. I’ve attached the plots for reference.

Surprisingly, CAM-B3LYP slopes are around 1.6 and RMS/MAD errors are significantly higher than PBE. This is unexpected, especially since CAM-B3LYP should in principle provide improved results.

Additionally, when testing larger molecules (e.g., see test.xyz, also attached), the computed isotropic shieldings seem unphysical. In some cases, values are much larger than those for TMS carbon atoms or even highly negative, which contradicts expectations.

Below is the input file used:

&FORCE_EVAL
 METHOD QS

 &DFT
  CHARGE 0
  MULTIPLICITY 1
  BASIS_SET_FILE_NAME /home/sergio/Desktop/pseudopotenciais/BASIS_SET_CP2K
  POTENTIAL_FILE_NAME /home/sergio/Desktop/pseudopotenciais/PSEUDOTENTIAL_CP2K
  WFN_RESTART_FILE_NAME  NMR-Frame-01_aug-cc-pVQZ-RESTART.wfn

  &QS
   METHOD GAPW
   EPS_DEFAULT 1.0E-18
   MIN_PAIR_LIST_RADIUS -1
  &END QS

  &XC
      &XC_FUNCTIONAL
        &HYB_GGA_XC_CAM_B3LYP
          _AC 0.81
          _ALPHA 0.65
          _BETA -0.46
          _OMEGA 0.33
        &END HYB_GGA_XC_CAM_B3LYP
      &END XC_FUNCTIONAL
      &HF
        &INTERACTION_POTENTIAL
            POTENTIAL_TYPE MIX_CL_TRUNC
            CUTOFF_RADIUS 12
            OMEGA 0.33
            SCALE_COULOMB 0.19
            SCALE_LONGRANGE 0.46
            T_C_G_DATA /home/sergio/Desktop/pseudopotenciais/t_c_g.dat
        &END
        &SCREENING
          EPS_SCHWARZ 1.0E-12
        &END
        &MEMORY
          MAX_MEMORY  1000
          EPS_STORAGE_SCALING 0.1
        &END MEMORY
        FRACTION 1.00
      &END
  &END XC

  &AUXILIARY_DENSITY_MATRIX_METHOD
   ADMM_TYPE ADMMS
  &END AUXILIARY_DENSITY_MATRIX_METHOD

 &SCRF
  EPS_OUT 4.81
  &SPHERE
   &CENTER
    ATOM_LIST 1..82
    WEIGHT_TYPE UNIT
   &END CENTER
   RADIUS 10.8
  &END SPHERE
 &END SCRF

  &MGRID
   CUTOFF 600
   REL_CUTOFF 60
  &END MGRID

  &SCF
   EPS_SCF 1.0E-6
   MAX_SCF 20
   SCF_GUESS RESTART
   &OT T
    MINIMIZER DIIS
    PRECONDITIONER FULL_ALL
    LINESEARCH 3PNT
   &END OT
   &OUTER_SCF  T
    EPS_SCF  1E-06
    MAX_SCF  20
   &END OUTER_SCF
  &END SCF

 &END DFT

 &SUBSYS
  &CELL ! box size
   ABC [angstrom] 25.00000 25.00000 25.00000
  &END CELL

  &TOPOLOGY
   COORD_FILE_NAME  test.xyz
   COORD_FILE_FORMAT XYZ
  &END TOPOLOGY

  &KIND O
   LEBEDEV_GRID 120
   RADIAL_GRID 250
   BASIS_SET ORB aug-cc-pVQZ
   BASIS_SET AUX_FIT admm-3
   POTENTIAL ALL
  &END KIND

  &KIND C
   LEBEDEV_GRID 120
   RADIAL_GRID 250
   BASIS_SET ORB aug-cc-pVQZ
   BASIS_SET AUX_FIT admm-3
   POTENTIAL ALL
  &END KIND

  &KIND H
   LEBEDEV_GRID 120
   RADIAL_GRID 250
   BASIS_SET ORB aug-cc-pVQZ
   BASIS_SET AUX_FIT admm-3
   POTENTIAL ALL
  &END KIND

 &END SUBSYS

&PROPERTIES
 &LINRES
  &LOCALIZE
  &END LOCALIZE
  PRECONDITIONER FULL_ALL
  MAX_ITER 2000
  &CURRENT
   GAUGE ATOM
   ORBITAL_CENTER ATOM
  &END CURRENT
  &NMR
   INTERPOLATE_SHIFT T
   &PRINT
    &CHI_TENSOR
    &END CHI_TENSOR
   &END PRINT
  &END NMR
 &END LINRES
&END PROPERTIES

&END FORCE_EVAL

&GLOBAL
 PRINT_LEVEL MEDIUM !(or SILENT/LOW/HIGH/…)
 PROJECT_NAME NMR-Frame-01_aug-cc-pVQZ
 RUN_TYPE LINEAR_RESPONSE
 WALLTIME 10000000000
&END GLOBAL

I'd appreciate any suggestions or insights into these points. Looking forward to any advice.

Best regards,
Sergio Tavares

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com<mailto:cp2k+unsubscribe at googlegroups.com>.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/b19bf001-7dd6-422d-b699-a8d4e94bb842n%40googlegroups.com<https://groups.google.com/d/msgid/cp2k/b19bf001-7dd6-422d-b699-a8d4e94bb842n%40googlegroups.com?utm_medium=email&utm_source=footer>.

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/ZR0P278MB0759FADB9806DD914B5D3FE39F7AA%40ZR0P278MB0759.CHEP278.PROD.OUTLOOK.COM.