[Pw_forum] Inconsistent results of LDA+U calculation for v6.1 and v5.4

[Dmitry Korotin]

Dear all,
It was simple to get the patch, so the hotfix is attached.

2017-05-29 9:39 GMT+05:00 Dmitry Korotin <dmitry at korotin.name>:
> Dear Paolo,
> thank you for the reply and fixing first of all.
>
> 2017-05-26 22:20 GMT+05:00 Paolo Giannozzi <p.giannozzi at gmail.com>:
>> Your results for v.5.4 look very funny: you run in parallel on 4
>> processors a serial code.
>
> You are right. But there was a batch script that configures and builds
> several versions, then runs the code... I even didn't realize that
> some versions have not automatically configured in a parallel.
>
>> Anyway, results for 5.4 are correct, those
>> for 6.1 (and 6.0 as well) are unfortunately not. More exactly:
>> non-default values of U_projection_type have no effect. This affects
>> XSpectra as well. Explanations in this thread (note the date):
>> http://qe-forge.org/pipermail/q-e-developers/2017-May/001665.html
>> and fixes here:
>> https://github.com/QEF/q-e/commit/78860ea5de8da244c7ecb9363f82fcab27b2043e
>> Sorry about that, thanks for reporting this (you've lost the
>> possibility to report an unfixed bug for just 1 day)
>
> Well, since QE foundation do not pay for every new unreported bug yet,
> I'am not upset about the not being the first ))
>
> But it would be great to have a hotfix for ver 6.1 that fixes this
> particular problem and do not affect any other piece of the code
> (otherwise I'm able to use git diff, but not sure that other parts of
> the code would work smooth). I realy think that bug in the LDA+U
> implementation is a rather critical issue and deserve a release of
> v6.1.1. Of course, you are the boss, anyway )
>
>>
>> Paolo
>>
>> On Fri, May 26, 2017 at 6:30 PM, Dmitry Korotin <dmitry at korotin.name> wrote:
>>> Dear QE developers,
>>> I'm surprised that versions 6.1 and 5.4 (and, actually, 5.0.3) give
>>> different result in the same lda+u calculations. Please find attached
>>> one input file for NiO and two different out files. And a
>>> bandstructure picture.
>>> Was the LDA+U implementation changed some time ago?
>>>
>>> (LDA results are the same for both version)
>>>
>>> Thank you in advance,
>>> Dmitry Korotin.
>>>
>>> --
>>>
>>>
>>> Best regards,
>>> Dr. Dmitry Korotin
>>>
>>> Institute of Metal Physics
>>> S. Kovalevskaya, 18
>>> 620990 Yekaterinburg
>>> Russia
>>>
>>> _______________________________________________
>>> Pw_forum mailing list
>>> Pw_forum at pwscf.org
>>> http://pwscf.org/mailman/listinfo/pw_forum
>>
>>
>>
>> --
>> 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
>> _______________________________________________
>> Pw_forum mailing list
>> Pw_forum at pwscf.org
>> http://pwscf.org/mailman/listinfo/pw_forum
>
>
>
> --
> С уважением,
> Дмитрий Коротин
>
> Best regards,
> Dr. Dmitry Korotin
>
> Institute of Metal Physics
> S. Kovalevskaya, 18
> 620990 Yekaterinburg
> Russia



-- 
С уважением,
Дмитрий Коротин

Best regards,
Dr. Dmitry Korotin

Institute of Metal Physics
S. Kovalevskaya, 18
620990 Yekaterinburg
Russia
-------------- next part --------------
A non-text attachment was scrubbed...
Name: patch.6.1_ldau_fix.patch
Type: application/octet-stream
Size: 3192 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20170529/38e70c1e/attachment.obj>