[CP2K-user] [CP2K:20595] Re: Follow-up on AIMD Simulation of Incident EMIBF4 to Gold Surface

[Marcella Iannuzzi]

Hello Nick

The reduce.out output shows a decent convergence behaviour.
EPS_SCF at 1E-3 is definitely too low and the forces are going to be wrong. 
The cutoff at 300 Ry is too low and it is also going to affect the accuracy 
of the forces.
Not accurate forces might explain the high temperature and for sure lead to 
wrong dynamics.
If the temperature is very high, or anyway with large velocities, it is a 
good idea to use a variable time step, to avoid too large displacements.
Check the *DISPLACEMENT_TOL keyword in the MD section.*
Have you tested AIMD on your system without the modification of the 
velocities to induce the collision?

Regards
Marcella



On Tuesday, August 20, 2024 at 6:49:29 PM UTC+2 nichol... at gmail.com wrote:

> Hi all,
>
> I wanted to follow-up on this, because it seems I am still having a bit of 
> trouble isolating the ideal SCF convergence settings. I had first noticed 
> that the temperature using the surface.inp script was quite large, and 
> acknowledged that there was a conversion issue for the velocities when 
> going from Angstrom/ps to Bohr/a.u. time, and had changed the &VELOCITY 
> section to reflect this error. It should be noted that the incident EMIBF4 
> in this setup is moving with an impact KE of 50 eV. Now, the temperature 
> lies around 6700 K, which still seems pretty high to me. Furthermore, it 
> seems that unless I loosen the EPS_SCF convergence criteria to 1E-3, each 
> MD step can take up to 10 minutes to converge on 28 threads. I have also 
> since reduced the wall size to 38 atoms (and the EMIBF4 still at 24 atoms). 
> I was wondering if this is due to the &MGRID setup, the setup for 
> smearing/mixing/added MO's, and/or the basis sets selected for this 
> simulation. I've attached a newer .inp file and a partially executed .out 
> file for reference. Please let me know if you have any questions and I look 
> forward to hearing your advice.
>
> All my best,
> Nick
>
> On Sunday, August 18, 2024 at 9:45:09 PM UTC-4 Nicholas Laws wrote:
>
>> Hi all,
>>
>> I hope you are all doing well. I am following up on a prior message I 
>> sent regarding a CP2K simulation setup used to model the collisional 
>> surface interactions between an incident EMIBF4 neutral traveling with a 
>> known velocity at a gold surface. I have since reduced the size of the gold 
>> surface from 1024 atoms to 128 atoms. Although, it seems that the SCF 
>> wavefunction optimization is struggling to converge. I was wondering if 
>> there were any guidelines or suggested methodologies for operating with 
>> metallic systems and how to better ensure convergence for the system I've 
>> described. I've attached the input script (surface.inp). Please let me know 
>> if you have any additional questions and I look forward to hearing from you!
>>
>> All my best,
>> Nick
>>
>

-- 
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 on the web visit https://groups.google.com/d/msgid/cp2k/bf90b811-6f5c-4c3c-93fe-a877cd514aaan%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240821/c91fe83a/attachment.htm>