[QE-users] Best way to choose dt and nstep in AIMD/CPMD simulation using cp.x code in quantum espresso

[KRISHNENDU MUKHERJEE]

Dear Jayraj Anadani, 

Thank you for your reply. As far as I have understood there are several ways to keep the electron wave functions close to the Born-Oppenheimer surface. One of them is as you are doing by, electron dynamics = "cg". Another one is the ensemble-DFT approach. I have put my questions for the ensemble-DFT approach under the topic, Ensemble_DFT simulation. Ensemble-DFT also does cg-minimization of electron wavefunctions (in an outer loop) in combination with updating the electron occupancy (in an inner loop). As you I am also trying to optimize the atomic structure of a cluster placed inside vacuum in a box. My computational facility is limited. So, till now I have only been able to simulate till 0.1 ps. So far the change in atomic structure from the input perfect crystal-fragment is reasonable. 

However I am getting a note, 

NOTE : eigenvalues are not computed without ortho 
And the reported eigenvalues are all zeros 

I am concerned with this NOTE as, I think the eigen values are required to calculate the smearing of the electrons. 

So, today my question to you is the following. Did you get this NOTE too? 

Thank you, 
Best regards, 
Krishnendu 


-----------------------------------------------------------------------------------you wrote: 

It is a cluster inside a 20 A box to remove periodicity. But that cluster 
is a part of large MD simulated position coordinates file of 10000 atoms. 

Thank you 
Regards 
Jayraj Anadani 

On Sun, 2 Oct, 2022, 10:15 am KRISHNENDU MUKHERJEE, <krishnendu at nmlindia.org> 
wrote: 

> 
> Dear Dr. Jayraj Anadani, 
> 
> Is your input atomic structure is a cluster inside vacuum in a box or it 
> is a supercell? 
> 
> Thank you, 
> Regards, 
> Krishnendu 
> 
> 
> ------------------------------------------------------------------------------------------------- 
> You wrote: 
> In my case, I have a system of 54 zrcu atoms. i want to melt it at higher 
> temperatures above 2000K and then i will do quenching to get amorphorse 
> metallic alloys state. 
> But shorter time step in cg method give more time consuming. So i choose dt 
> = between 30 to 60 and in order to get equilibrium at high temperatures i 
> took nstep = above 500. 
> For equilibrium i have to run continue till 3 to 5 ps to get randomisation 
> so it took large time to complete. 
> Any suggestions ? 
> 
> Thank you 
> Jayraj Anadani 


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20221002/0f6e1080/attachment.html>