[CP2K:8806] very long calculation time

Amin amin.jaf... at gmail.com
Mon Mar 13 12:07:09 CET 2017


To add up some thing: I think that is because of the slow convergence of 
the SCF.
The interesting point is that with 48 atoms in my supercell, SCF converged 
in less than one hour! but now with 192 atoms, it takes more than a week 
(even more!, because it hasn't converged yet!)

On Monday, March 13, 2017 at 11:57:14 AM UTC+1, Amin wrote:
>
> Thank you very much for your swift reply.
> I am replicating a calculation that has been done previously by VASP. at 
> first I tried to calculate adsorption energy with 48 atoms in my supercell, 
> but there was a major difference between my results and the previous one. 
> Then I modified my structure to include more atoms. I'll try to apply your 
> suggestions but I also attach my complete input file in case that it might 
> help to give a better perspective. 
> -Because of this point that calculation stopped after reaching the 
> wall-time, I set the new input file to use the Restart wave-function. But 
> I'm not sure that it's been done correctly! Sorry if my questions are so 
> rudimentary! 
> I sincerely appreciate your time and considerations.
>
> &GLOBAL
>   PROJECT  Chemisorption-newconf
>   RUN_TYPE GEO_OPT
>   PRINT_LEVEL LOW
>   EXTENDED_FFT_LENGTHS
> &END GLOBAL
>
> &FORCE_EVAL
>   METHOD Quickstep             
>   &DFT
>     WFN_RESTART_FILE_NAME Chemisorption-newconf-RESTART.wfn 
>     BASIS_SET_FILE_NAME  BASIS_MOLOPT
>     POTENTIAL_FILE_NAME  GTH_POTENTIALS
>     LSD
>     CHARGE 0
>     MULTIPLICITY 1
>     &MGRID
> CUTOFF 800
> REL_CUTOFF 60
>     &END MGRID
>     &QS
> METHOD GPW
> EPS_DEFAULT 1.0E-12
>     &END QS
>     &POISSON
> PERIODIC XYZ
>     &END POISSON
>     &SCF                        
>       SCF_GUESS RESTART
>       EPS_SCF 1.0E-6
>       MAX_SCF 100
>       &OT
> PRECONDITIONER FULL_ALL
> MINIMIZER CG
>       &END OT
>       &OUTER_SCF
> MAX_SCF 20
> EPS_SCF 1.0E-6
>       &END OUTER_SCF
>     &END SCF
>     &XC                       
>       &XC_FUNCTIONAL PBE
>       &END XC_FUNCTIONAL
>       &VDW_POTENTIAL
> DISPERSION_FUNCTIONAL PAIR_POTENTIAL
> &PAIR_POTENTIAL
> TYPE DFTD3(BJ)
> REFERENCE_FUNCTIONAL PBE
> CALCULATE_C9_TERM .FALSE.
> PARAMETER_FILE_NAME dftd3.dat
> R_CUTOFF 15.0
> &END PAIR_POTENTIAL
>       &END VDW_POTENTIAL
>     &END XC
>   &END DFT
>
>   &SUBSYS
>     &CELL
>       ABC 15.138 15.138 30.0000
>       PERIODIC XYZ 
>     &END CELL
>     &TOPOLOGY
>       COORD_FILE_FORMAT XYZ
>       COORD_FILE_NAME  ./Anatase001-192.xyz
>     &END
>     &KIND C
>       ELEMENT C
>       BASIS_SET DZVP-MOLOPT-SR-GTH
>       POTENTIAL GTH-PBE-q4
>     &END KIND 
>     &KIND Ti
>       ELEMENT Ti
>       BASIS_SET DZVP-MOLOPT-SR-GTH
>       POTENTIAL GTH-PBE-q12
>     &END KIND
>     &KIND O
>       ELEMENT O
>       BASIS_SET DZVP-MOLOPT-SR-GTH
>       POTENTIAL GTH-PBE-q6
>     &END KIND
>   &END SUBSYS
>   &PRINT
> &TOTAL_NUMBERS ON
> &END TOTAL_NUMBERS
>   &END PRINT
> &END FORCE_EVAL
> &MOTION
> &GEO_OPT
>  OPTIMIZER BFGS
>  MAX_FORCE 4.50000000E-004
> &END
> &CONSTRAINT
>  &FIXED_ATOMS
> LIST 0..95
>  &END FIXED_ATOMS
> &END CONSTRAINT
> &END MOTION
>
> On Monday, March 13, 2017 at 11:35:58 AM UTC+1, jgh wrote:
>>
>> Hi 
>>
>> is your 'long' simulation time caused by: 
>>
>> - slow convergence of geometry optimization 
>> - slow convergence of SCF 
>> - high CPU time per SCF cycle? 
>>
>> To improve CPU time per SCF I would suggest (guessing without 
>> knowing your full input) 
>> - use only MOLOPT-SR basis sets 
>> - reduce EPS_DEFAULT to 1.e-10 
>> - Preconditioner -> FULL_SINGLE_INVERSE 
>>
>> Do you really need LSD? 
>>
>> regards 
>>
>> Juerg Hutter 
>> -------------------------------------------------------------- 
>> Juerg Hutter                         Phone : ++41 44 635 4491 
>> Institut für Chemie C                FAX   : ++41 44 635 6838 
>> Universität Zürich                   E-mail: hut... at chem.uzh.ch 
>> Winterthurerstrasse 190 
>> CH-8057 Zürich, Switzerland 
>> --------------------------------------------------------------- 
>>
>> -----cp... at googlegroups.com wrote: -----To: cp2k <cp... at googlegroups.com> 
>>
>> From: Amin 
>> Sent by: cp... at googlegroups.com 
>> Date: 03/13/2017 10:28AM 
>> Subject: [CP2K:8806] very long calculation time 
>>
>> Hi all,I am trying to calculate the adsorption energy of co2 on Tio2 
>> surface (with 192 atoms in my supercell). but the problem is that it takes 
>> too long time to be calculated. I'm using 80 processors (2.8 GHz - each) 
>> but as I set the wall time for 7 days, it reached that wall time and 
>> calculations stopped. I have done similar calculations for fewer atoms 
>> (48,98) in super cell but in those cases (even with fewer CPUs) it takes at 
>> most 1 day to calculate everything. Because I'm a beginner in CP2K, I don't 
>> have any idea weather this amount of time for calculation is reasonable. I 
>> have attached critical parts of my input file. I will be grateful if 
>> someone gives a comment regarding my question.- Any suggestion for 
>> decreasing the calculation time with the same number of atoms is warmly 
>> appreciated.. :) 
>> &GLOBAL  PROJECT  Ch  RUN_TYPE GEO_OPT  PRINT_LEVEL LOW  
>> EXTENDED_FFT_LENGTHS&END GLOBAL 
>> &FORCE_EVAL  METHOD Quickstep       &DFT    BASIS_SET_FILE_NAME 
>>  BASIS_MOLOPT    POTENTIAL_FILE_NAME  GTH_POTENTIALS    LSD    CHARGE 0    
>> MULTIPLICITY 1    &MGRID        CUTOFF 800        REL_CUTOFF 60    &END 
>> MGRID    &QS        METHOD GPW        EPS_DEFAULT 1.0E-12    &END QS    
>> &POISSON        PERIODIC XYZ    &END POISSON    &SCF                        
>>     SCF_GUESS ATOMIC      EPS_SCF 1.0E-6      MAX_SCF 100      &OT       
>>  PRECONDITIONER FULL_ALL        MINIMIZER CG      &END OT      &OUTER_SCF   
>>      MAX_SCF 20        EPS_SCF 1.0E-6      &END OUTER_SCF    &END SCF    
>> &XC                  &XC_FUNCTIONAL PBE      &END XC_FUNCTIONAL      
>> &VDW_POTENTIAL        DISPERSION_FUNCTIONAL PAIR_POTENTIAL       
>>  &PAIR_POTENTIAL                TYPE DFTD3(BJ)               
>>  REFERENCE_FUNCTIONAL PBE                CALCULATE_C9_TERM .FALSE.         
>>        PARAMETER_FILE_NAME dftd3.dat                R_CUTOFF 15.0       
>>  &END PAIR_POTENTIAL      &END VDW_POTENTIAL    &END XC  &END DFT 
>>   &SUBSYS    &CELL      ABC 15.138 15.138 30.0000      PERIODIC XYZ     
>> &END CELL    &TOPOLOGY      COORD_FILE_FORMAT XYZ      COORD_FILE_NAME 
>>  ./Anatase001-192.xyz    &END 
>> &END FORCE_EVAL&MOTION        &GEO_OPT          OPTIMIZER BFGS         
>>  MAX_FORCE 4.50000000E-004        &END        &CONSTRAINT         
>>  &FIXED_ATOMS                LIST 0..95          &END FIXED_ATOMS       
>>  &END CONSTRAINT&END MOTION 
>>
>>
>>
>>
>> -- 
>>
>> 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+... at googlegroups.com. 
>>
>> To post to this group, send email to cp... at googlegroups.com. 
>>
>> Visit this group at https://groups.google.com/group/cp2k. 
>>
>> For more options, visit https://groups.google.com/d/optout. 
>>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20170313/ae0f4362/attachment.html>


More information about the CP2K-user mailing list