Make a release compatible with verification-PBE-v1 study

[mbercx]

List of protocols which should be supported to reproduce the results in the verification study paper:

• ABINIT@PW|PseudoDojo-v0.5 - Abinit ACWF updates #287 Abinit validation protocol #309
• BigDFT@DW|HGH-K(Valence) - Update bigdft code #310
• CASTEP@PW|C19MK2 - db0f1ae
• CP2K/Quickstep@TZV2P|GTH - Cp2k: protocols revision #263
• GPAW@PW|PAW-v0.9.20000 - GPAW ACWF #301
• FLEUR@LAPW+LO - Adjustments to the protocols/generator for Fleur code #265
• Quantum ESPRESSO@PW|SSSP-prec-v1.3 - QE: Add verification-pbe-v1 protocol #283
• SIESTA@AtOrOptDiamond|PseudoDojo-v0.4 - Finalize the siesta protocol for verification study #308
• SIRIUS/CP2K@PW|SSSP-prec-v1.2 - SIRIUS-CP2K protocol and workchain #312
• VASP@PW|GW-PAW54* - Update VASP part to comply with ACWF study #303
• WIEN2k@(L)APW+lo+LO - Wien2k implementation #314

Other TODOs

• The SSSP v1.3 pseudopotential family has not been officially released yet, nor has support been provided in aiida-pseudo. - Adding SSSP 1.3.0 support aiida-pseudo#153
• Version 0.5 of the PseudoDojo NC SR is not yet supported in aiida-pseudo. - Add support for Pseudo-dojo v0.5 aiida-pseudo#155
• The comparison of CASTEP and QE with ABINIT for NC pseudos discussed in section S14 uses "trimmed" UPF files for CASTEP and QE generated by @zhubonan (see Add PseudoDojo results for QE and CASTEP acwf-verification-scripts#36). However, these have not been put in a publicly available repository yet, afaik. Protocols are added in Include PseudoDojo protocols for CASTEP #302 for trimmed pseudos as well, but it's not clear how to obtain/install the corresponding pseudo potential family.

EDIT: releasing the trimmed pseudopotentials, see this comment.

[broeder-j]

@mbercx I can confirm, that the Protocols 'verification-pbe-v1' and 'oxides_validation' on the master branch are the exact ones used for the result generation in the acwf-verification paper.

[bosonie]

@mbercx, Just made a pull request for the Siesta part. It was in a separate branch and I forgot to make the PR.
Regarding WIEN2K, I have to get back to Oleg to understand what was the situation. I will write an email

[yakutovicha]

• SIRIUS/CP2K@PW|SSSP-prec-v1.2 - @dev-zero @yakutovicha I don't think this one is included in Cp2k: protocols revision #263?

@hmhoseini is working on that, I believe. He should make a PR very soon.

[sphuber]

I have checked the to-do for SSSP v1.3 support in aiida-pseudo which has been released with aiida-pseudo==1.1.0: https://github.com/aiidateam/aiida-pseudo/blob/v1.1.0/CHANGELOG.md

[sphuber]

Version 0.5 of the PseudoDojo NC SR is not yet supported in aiida-pseudo.

Who is responsible for this? If they open a PR, I can review it and create a release. @sponce24 @zooks97 ?

[mbercx]

Who is responsible for this?

I had a crack at it earlier this week, but it seems some of the .djrepo files are missing some of the metadata, see:

abinit/pseudo_dojo#120

I built a very hack solution (https://github.com/mbercx/aiida-pseudo/tree/hacky-af/pseudo-dojo-v5), but it would be much better if we could simply update the .djrepo tarballs on the PseudoDojo website.

[bosonie]

@sphuber @sphuber when my last PR is merged, we can make a release. I do not have access to PyPI, so it is better than one of you will do it. Thanks.
The only point that has not been addressed yet is the PR #302, that should include the support for calculations with the CASTEP code using trimmed UPF potentials. These calculations are used for a comparison between CASTEP, QE and Abinit in the supplementary section S14 of the paper. Since this is a very little part of the paper and we are in an hurry to finish up, after discussions with Giovanni and Marnik, we decided that is not critical to support this calculations. We can proceed with the release and I will write in the paper that calculations of section S14 are not reproducible through the aiida-common-workflow package. So let's close this issue with the PR that makes the release. Thanks everybody for the great job!!!

[mbercx]

@sphuber @sphuber when my last PR is merged, we can make a release. I do not have access to PyPI, so it is better than one of you will do it.

It has finally happened, I have merged into @sphuber and we have become one Github entity. 🧘 😉

Re the trimmed pseudopotentials: if I understand correctly, we had to "trim" the UPF files to have a close match with ABINIT, which is using psp8. The issue was in the psp8 files however, their grids weren't big enough and so @zhubonan trimmed the UPF ones. So it wouldn't make sense to publish the trimmed UPF pseudo potentials, rather the psp8 ones should be "fixed".

[sphuber]

v1.0.0 has been released: https://pypi.org/project/aiida-common-workflows/1.0.0/

[bosonie]

Thanks @sphuber !