search

aiidateam / aiida-common-workflows

A repository for the implementation of common workflow interfaces across materials-science codes and plugins
https://aiida-common-workflows.readthedocs.io
MIT License
52 stars 31 forks source link

CASTEP protocol revision #269

Closed zhubonan closed 2 years ago

zhubonan commented 2 years ago

Contains the verification-PBE-v1 protocol.

Changes:

  1. Use a fixed cut off energy of 800 eV. This is to make sure the consistency between the oxides and the unaries set. Previously, the cut off energy is determined from the pseudopotentials on the code side automatically. However, to construct the convex hull, the same cut off energy should be used for all calculations. Previsou oxides are calculated at about 750 eV cut off.

  2. Fixed a bug with the soft_elements.yml file. This file provides a list of "soft" elements where the default cut off energy is very low - which can give some large pulay stress. If all of the elements are "soft" a defautl 326 eV cut of energy is set regardlessly. The yml files expands No and Y into False and True, which need to be properly quoted. In addition, this override should not act if cut_off_energy is set explicitly.

  3. Updated with new potential family ("C19V2"), containing pseudopotential for lanthanides and actinides.

bosonie commented 2 years ago

@zhubonan Hi Bonan, all good here? Thanks for the work, the only thing missing is a function to extract the TS contribution from a CastepCommonRelaxWorkChain, look at this template: https://github.com/aiidateam/aiida-common-workflows/blob/master/aiida_common_workflows/workflows/relax/siesta/extractors.py After that I can merge. Thanks!

zhubonan commented 2 years ago

@bosonie I have a question regarding the naming of the protocols. So the verification-PBE-v1 is the original one and verification-PBE-v1-mk2 is the one with improve pseudopotentials - is this the desired way to name those?

Alternatively I can have the one with the improved pseudopotentials as verficiation-PBE-v1, and the one without as verification-PBE-v1-a0 or somehting like it?

bosonie commented 2 years ago

We can ask also the opinion of @giovannipizzi, but I tend to say that the best set should be called verficiation-PBE-v1, and the other with another name. Actually this should be discussed in general since it is common to other codes.

zhubonan commented 2 years ago

hmm I see. I would go with verficiation-PBE-v1 (latest and better) and verification-PBE-v1-a0 (less accurate) then