aiida-quantumespresso icon indicating copy to clipboard operation
aiida-quantumespresso copied to clipboard

Published 20 hours ago • verified publisher icon aiidateam

fix lda_plus_u_calculation in output_parameters

Open superstar54 opened this issue 1 year ago • 8 comments

fix #935

The dftU is in the dft section of the output.

superstar54 avatar May 21 '23 04:05 superstar54

Thanks @superstar54! What version of QE are you running here? Could you maybe also provide the corresponding output files (aiida.out, data-file-schema.xml) so I can set up a test for this?

mbercx avatar May 21 '23 05:05 mbercx

PWSCF v.7.0. Here are the files dftU.zip

Thanks for testing!

superstar54 avatar May 21 '23 06:05 superstar54

Are we only supporting v7.0 and up now @mbercx ? And is the location of this keyword consistent for all other versions > v7.0?

sphuber avatar May 21 '23 15:05 sphuber

Are we only supporting v7.0 and up now @mbercx ? And is the location of this keyword consistent for all other versions > v7.0?

Both good questions. Indeed, according to our compatibility policy

Starting from aiida-quantumespresso==4.0, the last three minor versions of Quantum ESPRESSO are supported. Older versions are supported up to a maximum of two years.

We should support v7.0-v7.2. As for the hubbard key word, I'm assuming that things might have changed with the new hubbard inputs in v7.1. However, perhaps we should let someone answer that has actually been running pw.x calculations with Hubbard corrections, @bastonero?

mbercx avatar May 21 '23 15:05 mbercx

Indeed, from v7.1 all the Hubbard related keywords in SYSTEM are dropped. Only the HUBBARD card should be specified, which is done via the new HubbardStructureData, and not directly from parameters of the builder.

Nevertheless, I know that QE did not really change internally the logic, and many parameters are still the same.

Conclusion: it might be that the xml still has these keywords (😅). I will check the xml of a recent calculation of mine and update you

bastonero avatar May 22 '23 04:05 bastonero

Here you have an example of dataschema using the latest (v7.2) QE version. Indeed, the output_parameters has incorrectly lda_plus_u_correction: False in my PwCalculation node.

Shall we also think of dropping the VERY old convention of lda_plus_u naming? A boolean hubbard_correction would do the job.

data.tar.gz

bastonero avatar May 22 '23 05:05 bastonero

Here you have an example of dataschema using the latest (v7.2) QE version. Indeed, the output_parameters has incorrectly lda_plus_u_correction: False in my PwCalculation node.

Thanks!

Shall we also think of dropping the VERY old convention of lda_plus_u naming? A boolean hubbard_correction would do the job.

Yeah, the output_parameters could do with a good scrubbing. This will be part of the work that @sphuber has been doing on moving all possible parsing to the XML file. @sphuber perhaps you can open a PR with this work so we can start collaborating on it there? I remember I was still supposed to give you some files for additional testing, but forgot which ones. 😅

mbercx avatar May 22 '23 05:05 mbercx

https://github.com/aiidateam/aiida-quantumespresso/pull/940

Let's pull that over the line first indeed, before optimizing other changes in the output_parameters output. @mbercx is there anything we still want to release in the current major version? Or can we start merging stuff that can be considered breaking and that we will release with v5.0?

What about the ACWF stuff? Is that complete?

sphuber avatar May 23 '23 08:05 sphuber