[QE-users] Query about forc_conv_thr in relax or vc-relax calculation

[Lorenzo Paulatto]

On 1/12/21 9:12 PM, Mainak Ghosh wrote:
> Dear all
>     In 'vc-relax' or 'relax' calculation for getting better results i 
> have to reduce the forc_conv_thr, but it will take longer time to be 
> completed. So,if  i reduce this threshold step by step say, i wanted 
> the forc_conv_thr tobe 1.0D-5. I can get this result in two way.
>    Firstly, i can run a direct calculation where this threshold is 1.0D-5.
>    Secondly, in frist run i can use 1.0D-2 as the threshold value and 
> use the output file for next step input where i can use 1.0D-5 as 
> forc_conv_thr.
>    Is these two things are identical ?

The final result yes, but it will be faster, in total, to start 
immediately with the smaller threshold as the algorithm uses the 
minimization history.

>   Another thing can i use LDA pseudopotential with forc_conv_thr 
> 1.0D-5 as trial run and use the output for GGA-PBE xc functional input ?

Yes, but I don't think you will save any time, GGA is only marginally 
more expensive than LDA. You may do a pre-relax with a lower cutoff or 
less k-points, again I seldom find this strategy to be very practical. 
It also depend on the value you give to CPU time with respect to human time

cheers


> Thanks in advance.
>                                   Mainak Ghosh
>                             University of Calcutta,India
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
> 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/20210113/6cadced5/attachment.html>