[QE-users] Restarting a hybrid calculation

[Paolo Giannozzi]

Thank you for reporting this

Paolo

On Sat, Jun 2, 2018 at 7:38 PM, Ary Junior <aryjunior at gmail.com> wrote:

> Ok,
>
> Just for the record. Fixed in a easy and pragmatic way. Not sure if is the
> most appropriate, but I haven't noticed any loss in performance and the
> computed energies are exactly the same of a non-restarted run. I just
> called sum_band before exxinit, line 117 of PW/src/electrons.f90.
>
> Thank you very much!
>
> Ary Ferreira
>
> FAPESP postdoctoral fellow
> UFSCar - Brazil
>
> On Fri, Jun 1, 2018 at 4:04 PM, Ary Junior <aryjunior at gmail.com> wrote:
>
>> Hi,
>>
>> Please, for those working in the ACE implementation. I tried to restart a
>> non-gamma-only hybrid calculation with QE-6.2 and using USPP PAW datasets
>> (okvan and okpaw == .true.). However, the code crashes because when the
>> subroutine rotate_becxx (PW/src/us_exx.f90) tries to access becxx0 and it
>> has not been set yet. As commented in the code, the subroutine sum_bec
>> (sum_band.f90) should be called before.
>>
>> The point is that in the subroutine electrons (PW/src/electrons.f90),
>> there is a first conditional that gears up everything for a restarting
>> calculation before the self-consistent loop starts. And this is the part in
>> which the code crashes, when the sequence of subroutines implemented in the
>> file PW/src/exx.f90 is called: exxinit -> rotate_becxx. Whereas in a
>> 'from_scratch' run, before that same call, in the self-consistent loop, the
>> sequence of subroutines is called before: electrons_scf -> sum_band ->
>> sum_bec -> store_becxx0, and everything is fine.
>>
>> Please, I'm trying to figure out whether it is possible to set becxx0
>> before start the self-consistent loop when restarting a hybrid calculation.
>> If someone could help with a tip or with a sign that there is a possible
>> solution in a next release, I will be very thankful. Maybe something could
>> be done in the subroutines exxinit or aceinit, both implemented in
>> PW/src/exx.f90. From the source code, it seems that things still the same
>> in version 6.2.1.
>>
>> Thank you very much!
>>
>> Ary Ferreira
>>
>> FAPESP postdoctoral fellow
>> UFSCar - Brazil
>>
>> --
>> http://lattes.cnpq.br/8221674673413336
>>
>
>
>
> --
> http://lattes.cnpq.br/8221674673413336
>
> _______________________________________________
> users mailing list
> users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
>



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20180602/159a0fc3/attachment.html>