[CP2K-user] NEB for diffusion energy barriers

[Marcella Iannuzzi]

Dear Stefano, 

1) You can fix atoms, but  this will not affect the computational time for 
the NEB simulation.
2) The default for K_SPRING is good in most of the cases. I would keep it 
unless you have a very good reason not to.
3) The NUMBER_OF_REPLICAS should be set such that the RMSD distance among 
neighbouring beads does not exceed 1-2 Å 
4) Be sure that you run on a total number of processors that is a multiple 
of NUMBER_OF_REPLICAS*NPROC_REP
5) This should be easy to check by comparing the absolute values of the 
coordinates. 

Kind regards,
Marcella

On Monday, December 16, 2019 at 1:46:28 PM UTC+1, Stefano Ferrero wrote:
>
> Dear CP2K experts,
>
> I have just started using CP2K and I would be really thankful if anybody 
> can help me.
> I am trying to calculate diffusion barriers of single atoms adsorbed on a 
> surface using the NEB method.
> I optimized at DFT level the initial and final positions of the diffusion 
> hop and I am providing those as first and last replicas.
> I have some questions about the set up of the calculation and the 
> computational strategy to follow to obtain good results.
>
> 1) My system contains more than 200 atoms. Can I fix the atoms underneath 
> the first layer of the surface to save computational time?
> 2) Should the value of K_SPRING be varied in order to see if there are 
> changes in the results?
> 3) Should the value of NUMBER_OF_REPLICAS be varied in order to see if 
> there are changes in the results?
> 4) Are there any other important parameters to take care of?
> 5) When plotting the energy profile of the NEB calculations I always find 
> that the last frame, which I gave as an input, changes in energy with 
> respect to the result found in the geometry optimization previously 
> performed.Is it due to the ALIGN_FRAMES and ROTATE_FRAMES keywords, which I 
> did not switched off? 
>
> Is this a good strategy to obtain the diffusion energy barriers or would 
> you suggest something else?
>
> I've attached the input file used and the geometries of the initial and 
> final replicas.
> Thank you for your help!
>
> Stefano Ferrero
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191216/73858fe8/attachment.htm>