[CP2K-user] [CP2K:17864] Re: Conserved quantity changed and used time increase when during a time in AIMD when do restart with different EPS_SCF

Marcella Iannuzzi marci.akira at gmail.com
Fri Oct 14 07:26:16 UTC 2022

Dear X.J.Xu

Such total energy drifts indicate that the results of the simulation are 
not reliable, the errors in the integration of the equation of motion are 
to big. 
The conserved quantity must be conserved.
The increase in time per MD step is most probably related to an increase in 
the number of SCF iterations, if you are running BOMD. 
If this is the case, i.e., if it becomes more and more difficult to 
optimise the wave function, it might indicate that the computed forces are 
rather wrong and lead to the sampling of weird geometries.
By the way, have you considered that graphene is a semi-metal? This makes 
the convergence of the wave function even more problematic. 
Before worrying on how to speed up the sampling, it is recommendable to 
find settings that produce reliable results, for instance further reducing 
EPS_SCF, adding virtual MOS, increasing the cutoff, etc.

Kind regards

On Friday, October 14, 2022 at 4:22:59 AM UTC+2 pxp11... at gmail.com wrote:

> Hi cp2k groups,
> I'm a new cp2k user and recently i simulated the contact Angle of water on 
> graphene using AIMD. I found that different values of EPS_SCF lead to 
> different changes in cons qty, thus leading to different increases in used 
> time. As shown in the following three figures, when cons qty is almost 
> unchanged, the time taken by each MD step is the same, but when cons qty 
> rises, usedtime rises sharply, may I ask what is the problem? How can I 
> improve it? The following is my input file.
> EPS_SCF = 1.0*10^-5, cutoff = 320 Ry
> [image: 1E-5.png]
> 5.0*10^-6, cutoff =  380Ry
> [image: 5E-6.png]
> 2*10-6,  cutoff = 380 Ry
> [image: 2E-6.png]
> I used NVT 330 K and nose-hoover chain, and here is my input file.
> My questions are: 1. Does cons Qty remain constant over time once NVT 
> reaches equilibrium? 2. Why does the cons Qty in 2*10-6 fluctuate wildly 
> after it has remained unchanged for a long time? 3. How to accelerate AIMD. 
> Best regards, 
> X.J Xu

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/c06dbbea-0986-4f5e-8e62-c878ae42d8f8n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20221014/e02670a5/attachment-0001.htm>

More information about the CP2K-user mailing list