[CP2K:8422] problem with Wannier block in MD

tfran... at gmail.com tfran... at gmail.com
Wed Nov 23 10:21:20 UTC 2016


Dear prof. Hutter,
first of all thanks a lot for the quick reply. 

My system is an organic system. The single molecule is a radical, but I am 
simulating a bulk system, with 32 molecules. Each molecules is "coupled" 
with the first neighbor, so that the final result is that I have a neutral 
system. My system is an insulator. The states are mainly localized in the 
around of the nitrogen which formally host the unpaired electron. I will 
provide also an image of the system, so that You can have an idea on how it 
is.

The parallelization setup I am using is the following:

128 processors with 16 processes per node.

Ok, i will try to use the Jacobi option to see if it able to run run 
properly, as a first trial. 

Thanks a lot again,
Best,
Tommaso



Il giorno mercoledì 23 novembre 2016 09:45:36 UTC+1, jgh ha scritto:
>
> Hi 
>
> it is impossible to give a final answer to your question without 
> looking deeper into your system and setup. 
> Why is your system is so difficult to converge? Is it metallic, or has a 
> small gap? 
> Has it very delocalized states? 
> Another reason for the bad timing might be an unfortunate setup for 
> parallelization. What are you using? 
>
> In any case, I would first try to switch to Jacobi from the start or 
> reduce the maximum number of iterations to a much smaller number. 
>
> regards 
>
> Juerg   
> -------------------------------------------------------------- 
> 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 
> <javascript:> 
> Winterthurerstrasse 190 
> CH-8057 Zürich, Switzerland 
> --------------------------------------------------------------- 
>
> -----cp... at googlegroups.com <javascript:> wrote: -----To: cp2k <
> cp... at googlegroups.com <javascript:>> 
> From: tfra... at gmail.com <javascript:> 
> Sent by: cp... at googlegroups.com <javascript:> 
> Date: 11/22/2016 11:41AM 
> Subject: [CP2K:8422] problem with Wannier block in MD 
>
> Dear all,I am newbie in CP2K code. I am running a MD, and after a 3 ps of 
> equilibration, I added the Wannier block in order to compute the Wannier 
> functions associated with the atomic positions for post processing the 
> corresponding wannier.xyz output with TRAVIS. I added the Wannier block 
> and at the same time I changed the parameter of the thermostat in order to 
> start the full production. The point is the the time per cycle increase 
> abruptly and honestly I don't know if there are some parameters to tune 
> properly. In the output, also, the CRAZY method does not CONVERGE, and CP2K 
> automatically switches to JACOBI.  
> Can someone kindly help me fixing this problem? 
> Here I provide the input file, the part of the energy file in which the 
> time per cycle increased a lot and the line of the output file. 
>
>
>       3970         1985.000000         0.450372996       270.116570530     
> -2373.094227635     -2370.372586619        67.556102037      3980         
> 1990.000000         0.465900641       279.429460220     -2373.099082026     
> -2370.372684263        76.962266922      3990         1995.000000         
> 0.450246868       270.040923331     -2373.087063109     -2370.371947353     
>    68.042631149      4000         2000.000000         0.458024384       
> 274.705581585     -2373.082490350     -2370.371942815        67.633645058  
>     4010         2005.000000         0.455150729       272.982072598     
> -2373.099321281     -2370.372651569        67.668117046      4020         
> 2010.000000         0.492625498       295.458011450     -2373.139838152     
> -2370.373074594        67.646747828      4030         2015.000000         
> 0.473892952       284.222944031     -2373.121942090     -2370.373181966     
>    67.785763025      4040         2020.000000         0.502305148       
> 301.263497118     -2373.148017864     -2370.373927941        67.868869066  
>     4050         2025.000000         0.505149987       302.969723468     
> -2373.153554300     -2370.373384706      1286.867120981      4060         
> 2030.000000         0.485675769       291.289829048     -2373.133253656     
> -2370.372950823      1222.631000996      4070         2035.000000         
> 0.462755255       277.542977703     -2373.129295826     -2370.372541675     
>   687.871176958      4080         2040.000000         0.470894469       
> 282.424568277     -2373.126909142     -2370.371939148      2357.914371967  
>     4090         2045.000000         0.457195995       274.208745478     
> -2373.116683742     -2370.371460260      1175.034868956      4100         
> 2050.000000         0.439054488       263.328160394     -2373.098414834     
> -2370.370188867       638.787050962      4110         2055.000000         
> 0.463669118       278.091077567     -2373.115282563     -2370.370098537     
>  1164.508292913      4120         2060.000000         0.455513606       
> 273.199712436     -2373.110187483     -2370.370783224      1174.876551867  
>     4130         2065.000000         0.471881288       283.016424388     
> -2373.113322837     -2370.370647886      1185.692931890      4140         
> 2070.000000         0.466999351       280.088424858     -2373.107402878     
> -2370.370877377      1175.991256952      4150         2075.000000         
> 0.491675077       294.887985488     -2373.140123274     -2370.371007238     
>  1198.926747084      4160         2080.000000         0.468795115       
> 281.165455732     -2373.121103471     -2370.370412862      1233.599603891  
>     4170         2085.000000         0.502428311       301.337365690     
> -2373.155216771     -2370.371468193      1345.183628798 
>
>
> Crazy Wannier localization not converged after   3000 iterations, 
> switching to jacobi rotations         Localization by iterative distributed 
> Jacobi rotation                      Iteration            Functional       
>   Tolerance   Time               Localization  for spin   2 converged in   
>   79 iterations          Spread Functional     sum_in -w_i ln(|z_in|^2)   
>  sum_in w_i(1-|z_in|^2)    Total Spread (Berry) :               
>  2612.7381541523     2561.3576511947 
>
>
>
>
>
> Thanks in advance for the help!Best,Tommy 
> Tommaso FrancesePhD Student (ITN-EJD-TCCM)Universitat de 
> BarcelonaDepartamento de Química-FísicaMartí i Franquès 1Planta 4a08028 
> BarcelonaSpain 
>
>
>
>
>
> -- 
>
> 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 <javascript:>. 
>
> To post to this group, send email to cp... at googlegroups.com <javascript:>. 
>
>
> 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/20161123/29a60e8d/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: molecule.png
Type: image/png
Size: 6597 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20161123/29a60e8d/attachment.png>


More information about the CP2K-user mailing list