[CP2K-user] Very strange behavior of CI-NEB calculations

[Travis]

Hi,

The atoms have to have the same ordering in i.xyz and f.xyz. You have NCH 
in one and NHC in the other.

-T


On Tuesday, February 4, 2020 at 7:37:46 AM UTC-5, Jacob Jensen wrote:
>
> Dear all,
>
> I'm doing CI-NEB calculations for a set of reactions on different metal 
> surfaces. During my calculations I encountered a very bizarre behavior of 
> the code which I haven't been able to find a reason for it.
> The problem is: I give starting (i.xyz) and end points (f.xyz) to the 
> CP2K to do a simple CI-NEB calculation, but in a completely random way 
> (only happens to some reactions) the code exchanges the last two lines of 
> the final configuration (f.xyz) and starts the calculation with wrong 
> coordinates. It has happened also for more than two lines in some 
> structures.
>
> For instance in the coordinates provided below, "tail f.xyz" shows this:
>
>  Cu         2.5607522840        3.6179848259       10.2114493134
>  Cu         5.1090632775        3.6199416787       10.2213122857
>  Cu         7.6670721044        3.6155408617       10.2040766974
>  Cu         0.0000430330        7.2313188455       10.2023231891
>  Cu         2.5586830958        7.2299962241       10.2055589609
>  Cu         5.1117660386        7.2291476249       10.2075861138
>  Cu         7.6675778217        7.2313677337       10.2036234566
>   N         6.0791607429        5.3490729602       14.3227064450
>   C         5.1523694228        5.0243296300       14.0410514879
>   H         4.1418019657        4.6566468208       13.6463880308
>
> But once the job starts to run, CP2K exchanges the last two atoms (C and 
> H) and starts with: 
>
>
>  NEB|***************************************************************************
>  NEB| Geometry for Replica Nr. 2 in Angstrom
>  NEB|         Cu         1.278231000    1.807692000    1.278231000
> .
> .
> .
>  NEB|         Cu         5.112380330   -0.000210681   10.202471653
>  NEB|         Cu         7.667615495   -0.000585929   10.201760866
>  NEB|         Cu         0.000556842    3.615646368   10.203009475
>  NEB|         Cu         2.560752284    3.617984826   10.211449313
>  NEB|         Cu         5.109063277    3.619941679   10.221312286
>  NEB|         Cu         7.667072104    3.615540862   10.204076697
>  NEB|         Cu         0.000043033    7.231318845   10.202323189
>  NEB|         Cu         2.558683096    7.229996224   10.205558961
>  NEB|         Cu         5.111766039    7.229147625   10.207586114
>  NEB|         Cu         7.667577822    7.231367734   10.203623457
>  NEB|          N         6.079160743    5.349072960   14.322706445
>  NEB|          H         5.152369423    5.024329630   14.041051488
>  NEB|          C         4.141801966    4.656646821   13.646388031
> .
> .
> .
> This happens without any error or warning and as you can imagine, the 
> result of this calculation would be nothing but garbage. I have provided my 
> input file, initial and final configurations together with the required 
> basis sets and pseudopotentials to provide a possibility of simple test run 
> (even with one core, cause the interpolation is at the very beginning of 
> the output file).
>
> I would be grateful if you can take a look and let me know about the 
> possible reason. Or if you have had the same experience with other 
> structures that would be beneficial for me to know.
>
> I have tried with different versions of CP2K (e.g. 5.1 and 7.0) and the 
> result is the same. Also, whenever I change i.xyz to f.xyz in :
>
>  &TOPOLOGY
>                         COORD_FILE_FORMAT XYZ
>                         COORD_FILE_NAME i.xyz
>  &END,
>
> the same exchange of lines happens to i.xyz and final replica (f.xyz) 
> stays intact! This really baffles me a lot. 
>
> Any remark or sharing of experience is highly appreciated.
>
>
> Kind regards,
>
> Jacob Jensen
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200204/4105df9b/attachment.htm>