Cut-off problrm
Matt W
mattwa... at gmail.com
Mon Apr 30 16:47:11 UTC 2018
Oh, come on now. Typical papers with PW codes report convergence to 1 meV
per formula unit or similar. I think you are quite safe!
It is a definite improvement, no?
On Monday, April 30, 2018 at 2:10:45 PM UTC+1, simin pahlavi wrote:
>
> Dear Matheu,
> As you guide, I changed the XC_SMOOTH_RHO NONE XC_DERIV PW
> parts , and also I am asking for ngrids 1. but still I dont see a clear
> convergence in cuttoff. Is it possible that there is a random number
> generator somewhere in my input?
> I really appreciate any helps in advance.
>
> cutoff energy
> 50.0000000000 -230.8714667076 0
> 100.0000000000 231.7994590012 0
> 150.0000000000 -231.8316639575 0
> 200.0000000000 -231.8369041137 0
> 250.0000000000 -231.8375305477 0
> 300.0000000000 -231.8377198370 0
> 350.0000000000 -231.8375224535 0
> 400.0000000000 -231.8375088884 0
> 450.0000000000 -231.8375120028 0
> 500.0000000000 -231.8375120028 0
> 550.0000000000 -231.8375100426 0
> 600.0000000000 -231.8375088384 0
>
>
> &GLOBAL
> PRINT_LEVEL MEDIUM
> PROJECT HH
> RUN_TYPE ENERGY
> # SEED 300
> &END GLOBAL
>
>
> &FORCE_EVAL
> METHOD Quickstep
> &DFT
> # WFN_RESTART_FILE_NAME HH
> BASIS_SET_FILE_NAME ../BASIS_MOLOPT
> POTENTIAL_FILE_NAME ../GTH_POTENTIALS
> CHARGE 0
> MULTIPLICITY 6
> &MGRID
> NGRIDS 1
> CUTOFF LT_cutoff
> REL_CUTOFF LT_rel_cutoff
> # CUTOFF LT_cutoff
> &END
> &QS
> # ! use the GPW method (i.e. pseudopotential based calculations with
> the Gaussian and Plane Waves scheme).
> # METHOD GPW
> # ! default threshold for numerics ~ roughly numerical accuracy of
> the total energy per electron,
> # ! sets reasonable values for all other thresholds.
> # EPS_DEFAULT 1.0E-10
> # ! used for MD, the method used to generate the initial guess.
> EXTRAPOLATION ASPC
> EPS_DEFAULT 1.0E-10
> # WF_INTERPOLATION PS
> # EXTRAPOLATION_ORDER 3
> METHOD GPW
> &END
> &POISSON
> PERIODIC XYZ !
> &END
> &SCF
> SCF_GUESS ATOMIC
> EPS_SCF 5.e-7
> MAX_SCF 1
> &OT
> !an accurate preconditioner suitable also for larger systems
> PRECONDITIONER FULL_SINGLE_INVERSE
> !the most robust choice (DIIS might sometimes be faster, but not
> as stable).
> MINIMIZER DIIS
> &END OT
> &OUTER_SCF
> TYPE NONE
> OPTIMIZER NONE
> EPS_SCF 5.0E-7
> MAX_SCF 1
> &END OUTER_SCF
> &END SCF
>
> &XC
> &XC_FUNCTIONAL PBE
> &END XC_FUNCTIONAL
> &XC_GRID
> # XC_SMOOTH_RHO NN50
> # XC_DERIV NN50_SMOOTH
> XC_SMOOTH_RHO NONE
> XC_DERIV PW
>
> &END XC_GRID
> &VDW_POTENTIAL
> DISPERSION_FUNCTIONAL PAIR_POTENTIAL
> &PAIR_POTENTIAL
> TYPE DFTD3
> PARAMETER_FILE_NAME ../../dftd3.dat
> REFERENCE_FUNCTIONAL PBE
> CALCULATE_C9_TERM TRUE
> REFERENCE_C9_TERM TRUE
> LONG_RANGE_CORRECTION TRUE
> VERBOSE_OUTPUT TRUE
> R_CUTOFF 15.0
> &END PAIR_POTENTIAL
> &END VDW_POTENTIAL
> &END XC
> &END DFT
>
> &SUBSYS
> &CELL
> ABC 16.0 16.0 16.0
> PERIODIC XYZ
> &END CELL
> &TOPOLOGY
> COORDINATE XYZ
> COORD_FILE_NAME ../structure
> CONNECTIVITY OFF
> # &TOPOLOGY
> # COORDINATE XYZ
> #COORD_FILE_NAME Fin-struct.xyz
> # CONNECTIVITY OFF
> &END TOPOLOGY
> &END SUBSYS
> &END FORCE_EVAL
>
>
>
>
> On Monday, April 23, 2018 at 10:41:00 PM UTC+2, Matt W wrote:
>>
>> Sorry, I posted too quickly without really looking at your setup.
>>
>> The big problem for you is the XC_GRID settings.
>>
>> &XC_GRID
>> XC_SMOOTH_RHO NN50
>> XC_DERIV NN50_SMOOTH
>> &END XC_GRID
>>
>> You really do not want to be using heavy smoothing, especially for this
>> purpose. These settings (see original paper) help get reasonable forces
>> with a rather low cutoff, but as you increase the cutoff lead to you
>> continuously redefining the XC functional you use (you are smoothing over
>> different areas of real space), so they do not converge.
>>
>> Try changing to the defaults
>>
>> &XC_GRID
>> XC_SMOOTH_RHO NONE
>> XC_DERIV PW
>> &END XC_GRID
>>
>> you should see at least a significant improvement.
>>
>> Matt
>>
>> On Monday, April 23, 2018 at 6:38:35 PM UTC+1, simin pahlavi wrote:
>>>
>>> Also for ngrid 1 the output is:
>>> shouldn't we not see any changes less than 10e-7 in energy with changing
>>> cutoff?
>>> I relay appreciate any help here
>>>
>>> 50.0000000000 -230.3665655634 0
>>> 100.0000000000 -231.5201721491 0
>>> 150.0000000000 -231.6333696011 0
>>> 200.0000000000 -231.6954662742 0
>>> 250.0000000000 -231.7211472374 0
>>> 300.0000000000 -231.7441677975 0
>>> 350.0000000000 -231.7546966069 0
>>> 400.0000000000 -231.7608341845 0
>>> 450.0000000000 -231.7712553463 0
>>> 500.0000000000 -231.7712553463 0
>>> 550.0000000000 -231.7762288308 0
>>> 600.0000000000 -231.7833659880 0
>>>
>>> On Monday, April 23, 2018 at 6:55:33 PM UTC+2, simin pahlavi wrote:
>>>>
>>>>
>>>> Dear Matteu,
>>>> Thank you very much for your quick answer,
>>>> I am not sure what did you mean by " Your set up is converged at the
>>>> lowest setting you report",
>>>> because after first few step with convergence I had a different energy
>>>> in next cutoffs.
>>>> but the result for lower cutoff is as the following.
>>>> Can I ask where is the suitable cutoff is now? and why you say so?
>>>>
>>>> Grid cutoff vs total energy
>>>> # Date: Mon Apr 23 18:44:42 CEST 2018
>>>> # REL_CUTOFF = 60
>>>> # Cutoff (Ry) | Total Energy (Ha) | NG on grid 1 | NG on grid 2 | NG on
>>>> grid 3 | NG on grid 4
>>>> 50.0 -230.3665656028 88679 13322 48
>>>> 0
>>>> 100.0 -231.5201721496 69162 25725 7162
>>>> 0
>>>> 150.0 231.6333696411 66211 22468 13322
>>>> 48
>>>> 200.0 -231.6954662441 48271 28520 25210
>>>> 48
>>>> 250.0 -231.7211472373 45554 27891 28476
>>>> 128
>>>> 300.0 -231.7441677949 43072 26090 25725
>>>> 7162
>>>> 350.0 -231.7546966111 43072 23982 22834
>>>> 12161
>>>> 400.0 -231.7608341848 40587 25752 23501
>>>> 12209
>>>> 450.0 -231.7712553838 36594 29617 22468
>>>> 13370
>>>> 500.0 -231.7712553419 31682 21399 35598
>>>> 13370
>>>> 550.0 -231.7762288293 24651 26716 28756
>>>> 21926
>>>> 600.0 -231.7833659579 24651 23620 28520
>>>> 25258
>>>>
>>>> with best regards
>>>> Simin
>>>>
>>>> On Monday, April 23, 2018 at 2:57:15 PM UTC+2, Matt W wrote:
>>>>>
>>>>> To see really clean convergence you need to turn off multigrids (set
>>>>> ngrids 1).
>>>>>
>>>>> Your set up is converged at the lowest setting you report.
>>>>>
>>>>> We need a new tutorial on how to get happy pw settings.
>>>>>
>>>>> On Monday, April 23, 2018 at 11:38:03 AM UTC+1, simin pahlavi wrote:
>>>>>>
>>>>>> Dear aal,
>>>>>> Hello.
>>>>>> I am trying to find a suitable cutoff for my DFT calculation on
>>>>>> hydrocarbons according to the cp2k manual
>>>>>> https://www.cp2k.org/howto:converging_cutoff. I appriciate any help
>>>>>> in advance.
>>>>>> The results never converge and the calculated output for REL_CUTOFF
>>>>>> = 60, is as following:
>>>>>>
>>>>>> # Grid cutoff vs total energy
>>>>>> # Date: Fri Apr 20 20:39:04 CEST 2018
>>>>>> # REL_CUTOFF = 60
>>>>>> # Cutoff (Ry) | Total Energy (Ha) | NG on grid 1 | NG on grid 2 | NG
>>>>>> on grid 3 | NG on grid 4
>>>>>> 800.0 -231.7994673423 73445 28476 128
>>>>>> 0
>>>>>> 850.0 -231.7994673423 73445 28476 128
>>>>>> 0
>>>>>> 900.0 -231.8023753152 69162 25725 7162
>>>>>> 0
>>>>>> 950.0 -231.8023753152 69162 25725 7162
>>>>>> 0
>>>>>> 1000.0 -231.8065368280 69162 25725 7162
>>>>>> 0
>>>>>> 1050.0 -231.8065368280 69162 25725 7162
>>>>>> 0
>>>>>> 1100.0 -231.8065368280 67054 27833 7162
>>>>>> 0
>>>>>> 1150.0 -231.8129239556 67054 22834 12161
>>>>>> 0
>>>>>> 1200.0 -231.8129239556 67054 22834 12161
>>>>>> 0
>>>>>> 1250.0 -231.8129239556 67054 22834 12161
>>>>>> 0
>>>>>> 1300.0 -231.8129239556 66339 23501 12209
>>>>>> 0
>>>>>> 1350.0 -231.8129239556 66339 23501 12161
>>>>>> 48
>>>>>> 1400.0 -231.8148206637 66259 22420 13322
>>>>>> 48
>>>>>> 1450.0 -231.8148206637 66211 22468 13322
>>>>>> 48
>>>>>> 1500.0 -231.8148206637 66211 22468 13322
>>>>>> 48
>>>>>> 1550.0 -231.8158559068 63323 25356 13322
>>>>>> 48
>>>>>> 1600.0 -231.8175292623 57217 31462 13322
>>>>>> 48
>>>>>>
>>>>>> ********************************************************************************
>>>>>> Also my input file is :
>>>>>>
>>>>>> ******************************************************************************
>>>>>> &GLOBAL
>>>>>> PRINT_LEVEL MEDIUM
>>>>>> # WALLTIME 36:00:00 # changed
>>>>>> PROJECT cuttoff-test
>>>>>> RUN_TYPE ENERGY
>>>>>> SEED 300
>>>>>> &END GLOBAL
>>>>>>
>>>>>>
>>>>>> &FORCE_EVAL
>>>>>> METHOD Quickstep
>>>>>> &DFT
>>>>>> BASIS_SET_FILE_NAME ../data/BASIS_MOLOPT
>>>>>> POTENTIAL_FILE_NAME ../data/GTH_POTENTIALS
>>>>>> CHARGE 0
>>>>>> MULTIPLICITY 1
>>>>>> &MGRID
>>>>>> NGRIDS 4
>>>>>> CUTOFF LT_cutoff
>>>>>> REL_CUTOFF LT_rel_cutoff
>>>>>> &END
>>>>>> &QS
>>>>>> EPS_DEFAULT 1.0E-10
>>>>>> WF_INTERPOLATION PS
>>>>>> EXTRAPOLATION_ORDER 3
>>>>>> METHOD GPW
>>>>>> &END
>>>>>>
>>>>>> &SCF
>>>>>> SCF_GUESS ATOMIC
>>>>>> EPS_SCF 5.e-7
>>>>>> MAX_SCF 1
>>>>>> &OT
>>>>>> MINIMIZER DIIS
>>>>>> &END OT
>>>>>> &OUTER_SCF
>>>>>> TYPE NONE
>>>>>> OPTIMIZER NONE
>>>>>> EPS_SCF 5.0E-7
>>>>>> MAX_SCF 1
>>>>>> &END OUTER_SCF
>>>>>> &END SCF
>>>>>>
>>>>>> &XC
>>>>>> &XC_FUNCTIONAL PBE
>>>>>> &END XC_FUNCTIONAL
>>>>>> &XC_GRID
>>>>>> XC_SMOOTH_RHO NN50
>>>>>> XC_DERIV NN50_SMOOTH
>>>>>> &END XC_GRID
>>>>>> &VDW_POTENTIAL
>>>>>> DISPERSION_FUNCTIONAL PAIR_POTENTIAL
>>>>>> &PAIR_POTENTIAL
>>>>>> TYPE DFTD3
>>>>>> PARAMETER_FILE_NAME ../data/dftd3.dat
>>>>>> REFERENCE_FUNCTIONAL PBE
>>>>>> CALCULATE_C9_TERM TRUE
>>>>>> REFERENCE_C9_TERM TRUE
>>>>>> LONG_RANGE_CORRECTION TRUE
>>>>>> VERBOSE_OUTPUT TRUE
>>>>>> R_CUTOFF 15.0
>>>>>> &END PAIR_POTENTIAL
>>>>>> &END VDW_POTENTIAL
>>>>>> &END XC
>>>>>> &END DFT
>>>>>>
>>>>>> &SUBSYS
>>>>>> &CELL
>>>>>> ABC 16.0 16.0 16.0
>>>>>> PERIODIC XYZ
>>>>>> &END CELL
>>>>>>
>>>>>> &KIND H
>>>>>> BASIS_SET DZVP-MOLOPT-SR-GTH-q1
>>>>>> POTENTIAL GTH-PBE-q1
>>>>>> &END
>>>>>>
>>>>>> &KIND C
>>>>>> BASIS_SET DZVP-MOLOPT-SR-GTH-q4
>>>>>> POTENTIAL GTH-PBE-q4
>>>>>> &END KIND
>>>>>>
>>>>>>
>>>>>> &TOPOLOGY
>>>>>> COORDINATE XYZ
>>>>>> COORD_FILE_NAME ../data/structure
>>>>>> CONNECTIVITY OFF
>>>>>> &END TOPOLOGY
>>>>>> &END SUBSYS
>>>>>> &END FORCE_EVAL
>>>>>>
>>>>>>
>>>>>>
>>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180430/08ab08e3/attachment.htm>
More information about the CP2K-user
mailing list