Problematic CUTOFF convergence for bulk wurtzite ZnO

[Soumya Ghosh]

Dear all CP2K users,


I am a new CP2K user. I am trying to converge CUTOFF and REL_CUTOFF values 
for bulk wurtzite ZnO structure with CP2K 4.0. I followed the detailed 
protocol provided in the website 
(https://www.cp2k.org/howto:converging_cutoff). I find that the convergence 
with respect to CUTOFF is very slow. I am using PBE functional with GTH 
pseudopotentials and double zeta quality DZVP-MOLOPT-SR-GTH basis set. For 
some reason I had to turn CHOLESKY OFF. I also tried increasing the 
EPS_DEFAULT to 1.0E-14 but that did not help. In accordance with a previous 
post I have also included USE_FINER_GRID in &XC_GRID section. The set of 
single point calculations with 80 Ry for REL_CUTOFF is given below:

# REL_CUTOFF = 80
# Cutoff (Ry) | Total Energy (Ha) | NG on grid 1 | NG on grid 2 | NG on 
grid 3 | NG on grid 4
    200.00  -657.4213933987   36684   21860   10104      56
    250.00  -657.4335430328   36240   16020   16388      56
    300.00  -657.4341365218   34908   15184   18556      56
    350.00  -657.4384146252   32228   16680   14764    5032
    400.00  -657.4324674408   25868   22944   14820    5072
    450.00  -657.4347000952   25332   20208   18092    5072
    500.00  -657.4356129316   23548   17320   22708    5128
    550.00  -657.4347921404   23548   14568   20428   10160
    600.00  -657.4332115807   23548   13136   21860   10160
    650.00  -657.4335385743   23264   13420   21860   10160
    700.00  -657.4316842547   22684   14000   16772   15248
    750.00  -657.4316842544   22684   13556   16020   16444
    800.00  -657.4330871943   19676   15232   17352   16444
    850.00  -657.4339224245   18868   16040   15184   18612
    900.00  -657.4339224241   17812   17096   15184   18612
    950.00  -657.4332111532   15764   19144   15184   18612
   1000.00  -657.4333137052   14388   20424   14096   19796

 I have also attached the corresponding template inputfile that I am using.


I have also tried changing the basis set on "O" from DZVP-MOLOPT-SR-GTH to 
DZVP-MOLOPT-GTH. The results are more or less similar:


# REL_CUTOFF = 80
# Cutoff (Ry) | Total Energy (Ha) | NG on grid 1 | NG on grid 2 | NG on 
grid 3 | NG on grid 4 | NG on grid 5
    200.00  -657.0308787529   61432   30637   30882     120       0
    250.00  -657.0429305605   54024   31358   34084    3605       0
    300.00  -657.0434513084   51699   28121   31728   11523       0
    350.00  -657.0479451098   46119   28694   31660   16548      50
    400.00  -657.0417893365   43438   29367   33668   16548      50
    450.00  -657.0440489204   42768   25833   33054   21366      50
    500.00  -657.0449114959   41900   21637   32434   27050      50
    550.00  -657.0441812146   35446   27566   32959   26980     120
    600.00  -657.0426019141   34868   26564   30637   30882     120
    650.00  -657.0429345818   33961   24151   32456   28898    3605
    700.00  -657.0410773531   33535   21779   30068   34084    3605
    750.00  -657.0410773529   33535   20489   31358   34084    3605
    800.00  -657.0425113625   28635   23730   30457   36644    3605
    850.00  -657.0432906095   26725   25640   30457   28726   11523
    900.00  -657.0432906090   26027   25672   28121   31728   11523
    950.00  -657.0425912354   26027   25672   24347   35452   11573
   1000.00  -657.0426760657   25416   23385   26012   36685   11573


Note that in the above set NGRIDS is set to 5 according to the comment 
section in BASIS_MOLOPT file.


Please advice/ comment whether this seems natural or if you have any 
further suggestion on the keywords I am using in the inputfile. Thanks in 
advance.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20160505/2bc3c808/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: template.inp
Type: chemical/x-gamess-input
Size: 2672 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20160505/2bc3c808/attachment.inp>