[CP2K-user] [CP2K:20165] Re: Hybrid functional issues

[xuan wang]

Oh, yes, I understand why use shortrange type
upon noticing the αerfc（wr），this is the shortrange because of α=-β！
heartful thanks to Frederick～


Frederick Stein <f.stein at hzdr.de>于2024年5月3日 周五21:48写道：

> I am not entirely sure what exactly you do not understand. If α = -β ,
> then the potential operator for HF simplifies to [α+βerf(wr)]/r = [α- α
> erf(wr)]/r = α [1-erf(wr)]/r = α erfc(wr)/r , i.e. the short-range
> operator. The shortrange operator has the advantage that it is shortranged,
> much simpler and does not need any artificial truncation whereas in the
> representation with the Coulomb and the longrange operator, you have to
> artificially truncate the Coulomb operator to obey the minimum image
> convention and ensure numerical stability (convergence issues and/or too
> large exchange energies).
> xuan wang schrieb am Freitag, 3. Mai 2024 um 14:32:33 UTC+2:
>
>> Thank you, Frederick,
>> In detail, for range-separated functional, I am allowed to use
>> POTENTIAL_TYPE MIX_CL_TRUNC for wB7XD, CAM-B3LYP. The _TRUNC is truncated
>> for HF exchange, and can be truncated by CUTOFF_RADIUS from our
>> self-definition. But for HSE06, I can only use  POTENTIAL_TYPE SHORTRANGE
>> rather than type with truncated condition, because I am acquired that the
>> CUTOFF_RADIUS is defined by CP2K, not users, where this  CUTOFF_RADIUS is
>> relevant with  EPS_DEFAULT,  omega=0.11,as you said.
>>
>> For range-separated functional, the coulomb operator can be seen as 1/r =
>> [1-α-βerf(wr)]/r + [α+βerf(wr)]/r. In terms of wB7XD, α = 0.222036，β =
>> 0.777964，of CAM-B3LYP, α = 0.19，β = 0.46. While in principle, HSE06 can
>> also be seen α = 0.25，β = -0.25, that we can use  POTENTIAL_TYPE
>> MIX_CL_TRUNC. But not the condition, only recommeded to use POTENTIAL_TYPE
>> SHORTRANGE.
>>
>> I don't know why. I am dying for understanding why.
>>
>> 在2024年5月3日星期五 UTC+8 20:13:30<Frederick Stein> 写道：
>>
>>> Dear Garrett,
>>> If I understand your question correctly, this affects all
>>> range-separated hybrid functionals, i.e. all functionals with an explicit
>>> short/long-range Coulomb or Gaussian potential contribution. This warning
>>> is relevant if you as a user have to choose some parameters of the
>>> functional on your own. If all parameters of the functionals
>>> (CUTOFF_RADIUS, OMEGA) are pre-defined by the functional like in HSE06, you
>>> may ignore this warning. In the case of HSE06 with omega=0.11, you will
>>> probably not be able to get rid of this warning in real-case scenarios
>>> because with usual values of EPS_DEFAULT of at most 1.0E-10, this cutoff
>>> radius is at least at the order of 20 A.
>>> HTH,
>>> Frederick
>>>
>>> xuan wang schrieb am Freitag, 3. Mai 2024 um 13:50:03 UTC+2:
>>>
>>>> Dear all,
>>>>
>>>> I learned that in CP2K, the use of HSE06 for periodic calculations is
>>>> not allowed to use POTENTIAL_TYPE TRUNCATED, but POTENTIAL_TYPE SHORTANGE.
>>>> However, even SHORTANGE will also assume a cutoff_radius which is not
>>>> defined by users.
>>>>
>>>> I am wondering about how many functionals behave like such HSE06?
>>>>
>>>>
>>>> Thank you!
>>>>
>>>> Best regards,
>>>> Garrett
>>>>
>>> --
> 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/b37aade8-2141-4f7c-bf69-b41637debce9n%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/b37aade8-2141-4f7c-bf69-b41637debce9n%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/CAEkxPfiU9f%2BqJ111tiEWA9zA0fzmG2X9rzPNV%2B%2BGwHXqtvQL0Q%40mail.gmail.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240503/2c5595f2/attachment-0001.htm>