Error in GW: Self-consistent quasi-particle solution not found

[Jan Wilhelm]

Hi Jan-Niclas, 

thanks for your interest in GW in cp2k. The input as shown in the tutorial 
is only working properly for truly 3d systems without vacuum. For 1d 
periodic systems, I recommend to do finite-size calculations with finite 
strips of your polymer, see e.g. 
https://pubs.acs.org/doi/abs/10.1021/acs.jpclett.7b02740. In Fig. 3 (c) you 
see, that the gap is converged already for ~ 5 nm length, which is in your 
case roughly 80 atoms which should be doable on ~ 10 nodes. If you really 
want to do the periodic calculation, I rather would recommend to use 
BerkeleyGW, VASP or Jambo (but avoid plasmon-pole models!). 

I hope that helps. If you have further questions, you are welcome to write.

Cheers, 

Jan

------------------------------
-------------------------------- 
Jan Wilhelm, E-mail: wilhelmjan (AT) web.de


Am Mittwoch, 9. Mai 2018 14:48:57 UTC+2 schrieb Jan-Niclas Luy:
>
> Hello,
>
> doing a GW calculation I encounter the error *"Self-consistent 
> quasi-particle solution not found"*
>
> The innitial SCF calculation takes very long to converge (more than 300 
> cycles).
>
> I don't know if this is part of the problem or if it is memory related.
>
> The system I am trying to calculate is a 1D polymer using an input similar 
> to the example 
> https://www.cp2k.org/exercises:2017_uzh_cp2k-tutorial:gw#periodic_gw_calculations
>
> I am attaching the in- and out-files.
>
>
> With hope that a more experienced person can point me in the right 
> direction.
> Jan-Niclas Luy
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20180509/dd06d0ba/attachment.htm>