[CP2K-user] [CP2K:10318] Hints for a better convergence with transition metals

Nam Tran trann... at gmail.com
Tue Oct 22 06:02:55 UTC 2019


Dear Anton,

As I know the OT solver is not reliable for a metal system. How do I known 
when we can use the method instead of DIAGONALIZATION. I actually prefer OT 
since it is much faster in my test. My system consists of a graphene layer 
on top of a Ni (111) surface. 

Regards
Nam

On Tuesday, 5 June 2018 19:24:16 UTC+10, Anton S. Lytvynenko wrote:
>
> Dear Dr Junior,
>
> If you are looking for an ultimate guide to convergence of pathological 
> SCF cases, please consider the manual of ORCA software. Though they use a 
> bit other SCF solvers (DIAGONALIZATION-based in the CP2K glossary), the 
> general principles are thoroughly described and explained and especially 
> for transition metal complexes.
>
> Briefly, if you prefer OT solver, you should benefit from precondition 
> selection (btw, FULL_SINGLE_INVERSE was the most reasonable in my hands) 
> and from improved LINESEARCH variants.
>
> If your use diagonalization routines (you can try both OT and 
> DIAGONALIZATION and choose the best), you should use MIXING and decrease 
> the ALPHA value (0.2--0.05 would do). As a second step, try LEVEL_SHIFT 
> (0.1--2 a.u.). After that, I would try MIXING%METHOD KERKER_MIXING with 
> BETA ca. 1.5 and probably smearing. If that doesn't help, check the sanity 
> of the initial geometry of your system (and please show the example of the 
> system).
>
> Yours,
> Anton Lytvynenko.
>
> 2018-05-14 15:09 GMT+03:00 Henrique Junior <he... at gmail.com 
> <javascript:>>:
>
>> Hello, dear colleagues, I’m getting started with CP2K because I’m facing 
>> my first tasks dealing with solid state chemistry. Mostly, my work involves 
>> coordination compounds with transition metals and lanthanides (usually 
>> paramagnetic).
>>
>> Could you, please share your general hints to achieve a better 
>> convergence when performing GEO_OPT and CELL_OPT and vibrational spectra? 
>> Do we have any recommended literature available?
>>
>> Thank you very much
>>
>> -- 
>> 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 c... at googlegroups.com <javascript:>.
>> To post to this group, send email to c... at googlegroups.com <javascript:>
>> .
>> Visit this group at https://groups.google.com/group/cp2k.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191021/3ace05c1/attachment.htm>


More information about the CP2K-user mailing list