[CP2K-user] [CP2K:20704] Incorrect Band gaps from simple calculations

[Lucas Lodeiro]

Hi

there are many weird things from my point of view.
the main ones are:
*CUTOFF of 600 eV (44 Ry), which is less than the used RELCUTOFF of 60
Ry... you must use N times the RELCUTOFF for the CUTOFF.
*small cell without kpoints... you must use supercell (4x4x4 for example)
or kpoint mesh to properly get the electronic structure.
*your functional definition does not make sense to me... which functional
form are you pointing out?
*why add all virtual orbitals? it is not useful at all, and very expensive.

Regards - Lucas


El jue, 12 sept 2024 a las 12:48, Thomas Kasel (<tomkasel at gmail.com>)
escribió:

> The original files are in the first posting. But I suppose those are no
> longer relevant. I have attached my input,  basis set, and pseudopotential
> for silicon. Thank you for all your help.
>
> Cheers,
> Tom
>
> On Thursday, September 12, 2024 at 2:56:57 AM UTC-5 Jürg Hutter wrote:
>
>> Hi
>>
>> Obviously, you have a serious problem in your (secrete) input.
>> Without further information it is impossible to help.
>>
>> regards
>> JH
>>
>> ________________________________________
>> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of
>> Thomas Kasel <tomk... at gmail.com>
>> Sent: Wednesday, September 11, 2024 10:34 PM
>> To: cp2k
>> Subject: Re: [CP2K:20694] Incorrect Band gaps from simple calculations
>>
>> I managed to find a way to print the HOMO - LUMO gap from the pdos. The
>> only problem now is that for some reason I get a band gap of 85 eV for
>> silicon when using a hybrid functional.
>>
>> On Tuesday, September 10, 2024 at 1:54:25 PM UTC-5 Thomas Kasel wrote:
>> Thanks to your suggestions I managed to get the SCF to converge for PBE
>> based upon playing with the mixing. The only problem is now that CP2K is
>> not printing the HOMO-LUMO gap. Is there a setting I can specify to force
>> print the gap?
>>
>> Regards,
>> Tom
>>
>> On Tuesday, September 10, 2024 at 4:45:21 AM UTC-5 Jürg Hutter wrote:
>> Hi
>>
>> For silicon: if you are using the unit cell (with 2 atoms) you need a
>> large number of k-points.
>> In this case SCF needs diagonalization and an appropriate mixing scheme.
>> I would also add smearing
>> of occupation numbers. If you have this working, you can start
>> investigating basis sets for convergence.
>>
>> For graphene: The problem is the Dirac cone at the K special point.
>> Selecting a matching cell and k-point
>> sampling is mandatory to get correct results. There is ample literature
>> addressing this problem.
>>
>> regards
>> JH
>>
>> ________________________________________
>> From: cp... at googlegroups.com <cp... at googlegroups.com> on behalf of
>> Thomas Kasel <tomk... at gmail.com>
>> Sent: Monday, September 9, 2024 10:50 PM
>> To: cp2k
>> Subject: [CP2K:20677] Incorrect Band gaps from simple calculations
>>
>> Hello All,
>>
>> As a test I am trying to accurately (relatively) reproduce the band gaps
>> of silicon and graphene structures using PBE. I have however, run into an
>> issue where graphene gives a band gap of 1.2 eV while silicon gives a band
>> gap of 0.18 eV, the exact opposite of what one would expect. I would like
>> to ask what it is I am doing wrong in my calculations. Attached is a zip
>> file containing all my input files including basis set and
>> pseudopotentials.
>>
>> Additionally, I cannot seem to get the SCF to converge unless I specify:
>>
>> &OT
>> PRECONDITIONER FULL_ALL
>> MINIMIZER CG
>> &END OT
>>
>> Why is this setting so critical to SCF convergence in these cases?
>>
>> Cheers and thanks in advance,
>> Tom
>>
>> --
>> 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 cp2k+uns... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com>.
>>
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/cp2k/cf2a1d8d-c40c-4da4-9857-38cd2b7e7e91n%40googlegroups.com
>> <
>> https://groups.google.com/d/msgid/cp2k/cf2a1d8d-c40c-4da4-9857-38cd2b7e7e91n%40googlegroups.com?utm_medium=email&utm_source=footer>.
>>
>>
>> --
>> 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 cp2k+uns... at googlegroups.com<mailto:cp2k+uns... at googlegroups.com>.
>>
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/cp2k/519f408c-fb0c-4c48-9a70-a36fbdc02f23n%40googlegroups.com
>> <
>> https://groups.google.com/d/msgid/cp2k/519f408c-fb0c-4c48-9a70-a36fbdc02f23n%40googlegroups.com?utm_medium=email&utm_source=footer>.
>>
>>
> --
> 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 cp2k+unsubscribe at googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/cp2k/d83110d8-d806-4b3c-b35e-c3050ad192b1n%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/d83110d8-d806-4b3c-b35e-c3050ad192b1n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>

-- 
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 cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/CAOFT4PJW3r3gdm6Y3ek6AwCBafwsLADy6V0U-BR5LKGFqoq6SQ%40mail.gmail.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240912/cada3e13/attachment-0001.htm>