[Q-e-developers] Reporting bug in QE ph.x

[Paolo Giannozzi]

If nobody has any better idea, I'll commit reset_grid modified as in
the message below

Paolo

On Fri, Apr 7, 2017 at 10:07 PM, Paolo Giannozzi <p.giannozzi at gmail.com> wrote:
> Thank you Giuliana. While waiting for better solutions, the best way
> to communicate problems and proposed solutions is via the
> q-e-developers mailing list (CC:).
>
> It is likely sufficient to modify the function reset_grid as follows:
>
>        reset_grid = (nk1_*nk2_*nk3_ > 0)
>        IF ( .NOT. reset_grid ) RETURN
>        nk1 = nk1_
>        nk2 = nk2_
>        nk3 = nk3_
>        k1 = k1_
>        k2 = k2_
>        k3 = k3_
>
> but given the complexity of linear-response codes I cannot rule out
> unexpected side effects
>
> Paolo
>
>
> On Fri, Apr 7, 2017 at 7:38 PM, Giuliana Barbarino
> <giuliana.barbarino at polytechnique.edu> wrote:
>> Dear Paolo Giannozzi,
>> I am Giuliana Barbarino and I am a Postdoc at the Laboratoire des Solides
>> Irradies in Ecole Polytechnique. My supervisor is Nathalie Vast and I work
>> with Jelena Sjakste on a project about electron-phonon coupling.  Nathalie
>> suggested me to write to you because I found a problem on PH code. It
>> regards the possibility to reset a new grid of k-points inside Phonon.
>>
>> If the self consistent calculation is performed with an automatically
>> generated Monkhorst-Pack grid (nk1,nk2,nk3) and with the grid offset
>> k1=1,k2=1,k3=1 and in the phonon input it is specified a new k-points grid
>> to run the pw non-self consistent calculation without offset
>> k1=0,k2=0,k3=0, the phonon code doesn't change the grid and it performs
>> anyway the calculation with (nk1,nk2,nk3,1,1,1).
>> This also if any k1,k2,k3 is specified by input, even if the default
>> values of k1,k2,k3 variables should be 0.
>>
>> The reset_grid function, called in phq_readin and defined in the
>> subroutine  start_k of PW code, resets the automatic grid to new values
>> only if these are > 0. The (nk1,nk2,nk3,k1,k2,k2) defined in start_k are
>> then used to setup the non-self consistent calculation.
>>
>> This bug is not important in the case of a simple phonon calculations but
>> it affects my results in the case of electron-phonon calculation by using
>> Wannier functions.
>>
>> I would like to ask you a suggestion about how I should proceed to
>> communicate the corrections I have done to fix it.
>>
>> thank you for your help,
>> Best regards
>>
>>
>>
>>
>> Giuliana Barbarino
>> Laboratoire des Solides Irradies
>> Ecole Polytechnique
>> Route de Saclay
>> 91128 Palaiseau
>> Phone: +33 1 69 33 45 15
>>
>
>
>
> --
> 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



-- 
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