[CP2K-user] Unusual behavior of CI-NEB calculation

Jacob Jensen Neutr... at gmail.com
Tue Feb 4 12:50:48 UTC 2020


I have experienced the exact same problem. if anyone has had time to review 
this post, that interests me as well to know the source of the problem.

Cheers,

Jacob

On Friday, October 9, 2015 at 12:41:06 PM UTC+2, Tobias Kraemer wrote:
>
> Hi all,
>
>
>
> I am observing some strange behavior during a NEB calculation. Basically, 
> the job has been running for a long time, but didn't seem to get close to 
> converging. Checking the output it turns out that following step 20 (so 
> right at the beginning)
> there weren't any convergence checks any more. Hence, this is the last 
> convergence check I get:
>
>
>
>  *******************************************************************************
>  BAND TYPE                     =                                         
>  CI-NEB
>  BAND TYPE OPTIMIZATION        =                                           
>    SD
>  STEP NUMBER                   =                                           
>    20
>  NUMBER OF NEB REPLICA         =                                           
>    16
>  DISTANCES REP =        0.904376        0.970665        1.003149       
>  0.989286
>                         0.919178        0.879384        0.898267       
>  0.929957
>                         0.917761        0.958179        0.913835       
>  0.867431
>                         0.835008        0.833033        0.824020
>  ENERGIES [au] =    -4477.534580    -4477.534142    -4477.532777   
>  -4477.530610
>                     -4477.527827    -4477.524573    -4477.520887   
>  -4477.517698
>                     -4477.523689    -4477.538783    -4477.543540   
>  -4477.546473
>                     -4477.548883    -4477.550356    -4477.551290   
>  -4477.540294
>  BAND TOTAL ENERGY [au]        =                           
> -71640.56568602792686
>
>  *******************************************************************************
>
>  *******************************************************************************
>  RMS DISPLACEMENT =   0.00015     [ 0.00010]                               
> ( NO)
>  MAX DISPLACEMENT =   0.00872     [ 0.00020]                               
> ( NO)
>  RMS FORCE        =   0.00014     [ 0.00030]                               
> (YES)
>  MAX FORCE        =   0.00652     [ 0.00080]                               
> ( NO)
>
>  *******************************************************************************
>
>  NEB| Computing Energies and Forces
>  NEB| REPLICA Nr.    1- Energy and Forces
>  NEB|                                     Total Energy:         
>  -4477.534579955
>  NEB|    ATOM                            X                Y               
>  Z
>  NEB|      Rh                      0.000001025     -0.000015044     
>  0.000003010
>  NEB|      P                      -0.000001915      0.000000815     
> -0.000002622
>  NEB|      C                      -0.000000579      0.000005799     
>  0.000000635
>  NEB|      C                      -0.000003518     -0.000003207     
> -0.000002095
>
>
> What follows are Energies and Forces for the rest of the file. I can't 
> find any problems with SCF convergence in the BAND files, the energy 
> profile looks perfectly sensible, the trajectory of the last 16 images also 
> makes total sense to me. I find it hard to believe that this is a program 
> bug, since the very same input has successfully worked for a good number of 
> transition states in the meantime, on the very same system. However, it 
> strikes me that there appear no convergence reports after 20 steps. I also 
> noted that there isn't necessarily such a 
> convergence check after forces have been calculated for the 16 images I 
> have defined, so how does this work exactly? And why did I still get 
> sensible results for the other Transition States, while in the present 
> example the calculation seems to run into nowhere.
>
> Thanks for your comments, your help is much appreciated. 
>
>
> Tobias
>
>
>
>
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200204/e57bcf5f/attachment.html>


More information about the CP2K-user mailing list