[CP2K-user] [CP2K:21605] CP2K calculation on a large system facing FFT error
Sayed Ahmad Almohri
aalmohri at umich.edu
Tue Jul 1 23:01:22 UTC 2025
I have a bit of a long question, I am running a system which is a little
bit large (19.175bx56x12.953, 624 atoms) and I faced an issue. When I ran
it with my regular inputs I was getting this error:
*******************************************************************************
* ___
*
* / \
*
* [ABORT]
*
* \___/ Index to radix array not found.
*
* |
*
* O/|
*
* /| |
*
* / \
pw/fft_tools.F:270 *
*******************************************************************************
I first assumed its an NGRID problem and used a higher value but that
wasn't the fix, I looked up the error and on the forum I saw that we should
use the EXTENDED_FFT_LENGTHS flag to fix this. I was able to run this with
that flag. Then I did some testing on my previous runs to see how different
the results could be with and without the flag. I took a system with box
dimensions (19.175x24.899x19.954, 312 atoms) ran it once with
EXTENDED_FFT_LENGTHS and once without and I see a difference of 0.15
Hartree in the energy (without is lower); is there a way to fine tune the
grid size without using EXTENDED_FFT_LENGTHS, since I am getting a lower
value without EXTENDED_FFT_LENGTHS on ran on the same system my assumption
if that its lowering accuracy.
Input file:
! CP2K Library file locations
@SET data_dir (not putting in message)
! Additional input block file locations
@SET cell_file incfg.box
@SET crds_file incfg.xyz
@SET intg_file cp2k.md_int-block.inp
@SET mdio_file cp2k.md_io-block.inp
@SET qmps_file cp2k.qm_psuedo-block.inp
@SET qmbs_file cp2k.qm_basis-block.inp
! Specify the simualtion temperature
@SET temper 1000
!Note:
!
! cell_file: Text file with cell vectors (cp2k format)
! crds_file: XYZ file with system coordinates (xyz format)
! intg_file: MD integrator specification, and energy I/O print frequency
! mdio_file: I/O file names/frequencies for MD (except energy!)
! qmps_file: QM Psuedo potential/dispersion/etc settings
! qmbs_file: QM basis sets for each atom type
&GLOBAL
PROJECT ALD
RUN_TYPE MD
PRINT_LEVEL HIGH
SEED 1111
PREFERRED_DIAG_LIBRARY ELPA
EXTENDED_FFT_LENGTHS T
&END GLOBAL
&MOTION
&MD
@INCLUDE ${intg_file}
STEPS 0
TIMESTEP 0.5
TEMPERATURE ${temper}
&END MD
&CONSTRAINT
&LAGRANGE_MULTIPLIERS OFF
&END LAGRANGE_MULTIPLIERS
&END CONSTRAINT
&PRINT
@INCLUDE ${mdio_file}
&END
&END MOTION
&FORCE_EVAL
METHOD QS
@INCLUDE ${qmps_file}
&SUBSYS
@INCLUDE ${qmbs_file}
&CELL
@INCLUDE ${cell_file}
CELL_FILE_FORMAT cp2k
PERIODIC XYZ
&END
&TOPOLOGY
COORD_FILE_NAME ${crds_file}
COORD_FILE_FORMAT xyz
&END TOPOLOGY
&END SUBSYS
STRESS_TENSOR ANALYTICAL
&END FORCE_EVAL
&EXT_RESTART
#RESTART_FILE_NAME cp2k.restart
#RESTART_DEFAULT F
#RESTART_POS T
#RESTART_VEL T
&END EXT_RESTART
psuedo potential block:
&DFT
BASIS_SET_FILE_NAME ${data_dir}/GTH_BASIS_SETS
POTENTIAL_FILE_NAME ${data_dir}/GTH_POTENTIALS
! Tell CP2K to stop printing so many damn files
&PRINT
&E_DENSITY_CUBE OFF
&END E_DENSITY_CUBE
&EFIELD_CUBE OFF
&END EFIELD_CUBE
&V_HARTREE_CUBE OFF
&END V_HARTREE_CUBE
&TOT_DENSITY_CUBE OFF
&END TOT_DENSITY_CUBE
&ELF_CUBE OFF
&END ELF_CUBE
&MO_CUBES OFF
&END MO_CUBES
&NEIGHBOR_LISTS OFF
&END NEIGHBOR_LISTS
&END PRINT
&MGRID
CUTOFF 1200
COMMENSURATE
NGRIDS 4
&END
&QS
EPS_DEFAULT 1.0E-12
EXTRAPOLATION ASPC
EXTRAPOLATION_ORDER 6
&END QS
&SCF
# Smearing is not compatible with OT
# If smearing is requested need diagonalization
# OT allows for faster SCF than diagonalization, so only use smearing
if required!
# Doesn't work well for high T systems ... couldn't converge molten
carbon at 6000 K even with 500 MOs within 200 SCF...
#&SMEAR
# ELECTRONIC_TEMPERATURE ${temper}
# METHOD FERMI_DIRAC
#&END SMEAR
#ADDED_MOS 100 # Required for smearing ... set to at least 20% of
occupied orbitals, may need more (check for error messages from single
point test jobs)
EPS_SCF 1.0E-7
SCF_GUESS RESTART
MAX_SCF 200
&OUTER_SCF
EPS_SCF 1.0E-7
MAX_SCF 20
&END OUTER_SCF
# OT is not compatible with Smearing
&OT
MINIMIZER DIIS
PRECONDITIONER FULL_SINGLE_INVERSE
N_DIIS 7
&END OT
# Use diagonalization if smearing requested
#
#&DIAGONALIZATION
# ALGORITHM DAVIDSON
#&END DIAGONALIZATION
&PRINT
&RESTART ON
&END RESTART
&RESTART_HISTORY OFF
&END RESTART_HISTORY
&END PRINT
&END SCF
&XC
&XC_FUNCTIONAL PBE
&END XC_FUNCTIONAL
&XC_GRID
#XC_DERIV NN10_SMOOTH
XC_DERIV SPLINE2_SMOOTH
XC_SMOOTH_RHO NN10
&END XC_GRID
&VDW_POTENTIAL
DISPERSION_FUNCTIONAL PAIR_POTENTIAL
&PAIR_POTENTIAL
&PRINT_DFTD OFF
&END PRINT_DFTD
TYPE DFTD3
LONG_RANGE_CORRECTION .TRUE.
PARAMETER_FILE_NAME ${data_dir}/dftd3.dat
REFERENCE_FUNCTIONAL PBE
R_CUTOFF [angstrom] 10.0
EPS_CN 1.0E-6
&END PAIR_POTENTIAL
&END VDW_POTENTIAL
&END XC
&END
--
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/b0c588ff-0f5a-49f4-aa6b-b7fcba751f68n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20250701/c6f44f19/attachment.htm>
More information about the CP2K-user
mailing list