[CP2K-user] [CP2K:20908] Problem with the convergency for gold nano plate

[Ebrahim Tayyebi]

Hi all,

I have built a hexagonal gold nanoplate and am performing cluster 
calculations on it. For the cluster calculation and the POISSON_SOLVER, I 
am using MT. I am currently calculating the binding energy of OH and H. 
While running these calculations, I’ve observed inconsistent behavior. Some 
calculations converge properly, but others encounter issues. Specifically, 
the total energy in some cases stabilizes at a certain level, remains 
monotonous for an extended period, and then suddenly diverges, causing the 
system to deform.

Below is the output from the optimization for your reference.
  --------  Informations at step =    77 ------------
  Optimization Method        =                 BFGS
  Total Energy               =    -13652.9754629492
  Real energy change         =        -0.0000036973
  Predicted change in energy =        -0.0000025237
  Scaling factor             =         0.3263774079
  Step size                  =         0.0043054959
  Trust radius               =         0.1889726133
  Decrease in energy         =                  YES
  Used time                  =              572.179

  Convergence check :
  Max. step size             =         0.0043054959
  Conv. limit for step size  =         0.0030000000
  Convergence in step size   =                   NO
  RMS step size              =         0.0014084418
  Conv. limit for RMS step   =         0.0015000000
  Convergence in RMS step    =                  YES
  Max. gradient              =         0.0000964461
  Conv. limit for gradients  =         0.0004500000
  Conv. in gradients         =                  YES
  RMS gradient               =         0.0000135281
  Conv. limit for RMS grad.  =         0.0003000000
  Conv. in RMS gradients     =                  YES

In most cases, the maximum step size does not converge. I am encountering 
this issue specifically with the cluster calculations.

Additionally, I would like to ask: how much can I reduce the threshold to 
consider the results reasonably converged?

Any help would be greatly appreciated.
Best,
Ebrahim

-- 
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+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/603f6338-9d02-4d7b-b11b-fe3b632ccd25n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20241124/6e6d80e2/attachment.htm>