Feat: Add run types for GW calculations

[yanghan234]

Added GW-related configurations and modified utilities

This PR is a dependency of a PR in the atomate2 repo [WIP] feat: GW workflow with VASP. I previously submitted a PR in #1099 . However, I realized that I did not properly modified the calculation types, enums, etc. In this PR, these have been corrected.

Contributor Checklist

• Added run types and calculations types for GW calculations
• Modified the utils to identify GW calculations
• I have run the tests locally and they passed.

Tagging @mkhorton for awareness.

[tsmathis]

Thanks for resubmitting @yanghan234, your last PR got submitted right as we were updating how we generate the calc_type/run_type enums, sorry for the extra work.

Re: the empty CONTCAR, when might that apply? Is that an edge case specific to the GW workflow?

[yanghan234]

Thanks @tsmathis, yes, this is the edge case that I only oberve in the G0W0 calculation. In the G0W0 calculation, ionic steps are never updated. In addition, there is only one electronic step in such calculations, i.e. NELM = 1 or NELMGW=1. This is probably the reason that the CONTCAR is not written.

[esoteric-ephemera]

Given your comment about when CONTCAR is not written, could you change this to:

if not os.path.isfile(contcar_file) and vasprun.parameters.get("NELM",60) == 1:
    contcar = Poscar(vasprun.final_structure)
else:
    contcar = Poscar.from_file(contcar_file)

This should help separate cases where something has gone wrong (CONTCAR missing for any reason) vs. expected behavior when NELM is 1

[esoteric-ephemera]

I'm not sure I would put GW Band Structure as a task type. I think you want to add GW to the run types based on the ALGO. Then the corresponding task type would be NSCF Line or NSCF Uniform, which is what we use to indicate band structure calculations along a line in the Brillouin zone (NSCF Line), or using a regular k-point grid (NSCF Uniform). 'Static' is also an acceptable fallback

Also there need to be features in emmet.core.vasp.calc_types.utils to parse these, they were in your original PR, can you migrate these over?

[yanghan234]

Thanks @esoteric-ephemera, please give me some time to fix them. Meanwhile, I realize that testing/check-enums check fails because it cannot find the branch in the forked repo under my personal account. Is this a problem with the CI setup, should I be concerned about this?

[tsmathis]

The issue with checking the enums for forks was an oversight on my part. You don’t need to worry about that.

…

On Wed, Sep 25, 2024 at 6:19 PM Han Yang ***@***.***> wrote: Thanks @esoteric-ephemera <https://github.com/esoteric-ephemera>, please give me some time to fix them. Meanwhile, I realize that testing/check-enums check fails because it cannot find the branch in the forked repo under my personal account. Is this a problem with the CI setup, should I be concerned about this? — Reply to this email directly, view it on GitHub <#1105 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AIPH7AB7KHDJ4VPMUM7U3VTZYNOKDAVCNFSM6AAAAABOQI2WTGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNZVGU2DQNJWHA> . You are receiving this because you were mentioned.Message ID: ***@***.***>