[QE-users] Structural stability problem

[Pietro Davide Delugas]

Hello
What value are you using for conv_thr in the electrons namelist ? You 
should use a value which is rather lower than the default at least 
something like 1.d-9. It this is the case the should be some warning 
just after the forces summary at the end of each force computation.
I hope this helps
Pietro


On 9/12/19 3:33 AM, Gui Wei wrote:
> Dear QE users,
>
> When I optimized the structure of a heterostructure (containing about 
> 130 atoms),it does converges as expected,and there is basically no 
> problem with the atomic structure.The convergence threshold I used was 
> *etot conv thr: 2.0D-4* *forc_conv_thr:2.0D-3*.By comparing with the 
> parameter design in the literature (similar model，usually using 
> software VASP),I think this convergence accuracy is relatively 
> rough,especially force.So I raised the threshold to *etot conv thr: 
> 1.0D-4 * *forc_conv_thr:2.0D-3.*At this point I found that the 
> structure optimization could not converge.I extracted the data from 
> the output document and plotted Energy difference-Scf cycles curve and 
> Max force-Scf cycles curve.I found that scf cycles =5 satisfies the 
> convergence of the former rough convergence(both energy and 
> force).After that,two curves rise synchronously,this means that the 
> latter's fine convergence accuracy will never be achieved.
> What I want to know is whether the atomic structure under the rough 
> precision can be counted as a stable structure.If so, how do I 
> understand the rise of the above two curves (this seems to be that the 
> structure is not stable, in my understanding, if it is a stable 
> structure, the two curves should continue to slowly decline, 
> eventually reaching an almost horizontal state).If not, what is the 
> solution?
>
> Wei gui
> School of Mechanical Engineering，Chongqing University, China
>
>
>
>
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20190912/dbd39472/attachment.html>