Abinit flows from abiflows

[davidwaroquiers]

Summary

Include a summary of major changes in bullet points:

• Scf and non-Scf jobs for abinit
• Band Structure flow with abinit
• Relaxation for abinit
• Supports restarts of scf, non-scf and relax jobs
• Supports restarts from a different previous job

Additional dependencies introduced (if any)

• abipy

TODO

• other flows
• add autoparal
• uniformize with vasp flows (e.g. prev_vasp_dir vs prev_dirs or prev_outputs or ..., band structures, static maker so that it can be used interchangeabily in the elastic flow, ...) => discuss this with @utf @computron Guido Petretto (@gpetretto), ... who else ?

[utf]

Hi @davidwaroquiers,

Thanks for all your hard work with this implementation! Its an impressive an amount of code and I'm excited to see where it goes.

I've added a few comments, the main ones are regarding the base abinit job maker. My primary concern is having roughly the same interaction model for VASP and abinit. I realise this won't be 100% possible, but the more aligned things are the easier it will be to create mixed workflows.

[codecov-commenter]

Codecov Report

Merging #81 (d11fec8) into main (277817a) will decrease coverage by 3.79%.
The diff coverage is 55.93%.

❗ Current head d11fec8 differs from pull request most recent head 6099b06. Consider uploading reports for the commit 6099b06 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #81      +/-   ##
==========================================
- Coverage   73.51%   69.71%   -3.80%     
==========================================
  Files          49       60      +11     
  Lines        4349     5544    +1195     
  Branches      686      889     +203     
==========================================
+ Hits         3197     3865     +668     
- Misses        980     1476     +496     
- Partials      172      203      +31     

Impacted Files	Coverage Δ
src/atomate2/abinit/flows/core.py	0.00% <0.00%> (ø)
src/atomate2/abinit/utils/settings.py	0.00% <0.00%> (ø)
src/atomate2/cli/dev.py	0.00% <0.00%> (ø)
src/atomate2/abinit/run.py	27.50% <27.50%> (ø)
src/atomate2/common/files.py	66.01% <33.33%> (-1.31%)	⬇️
src/atomate2/abinit/utils/common.py	48.33% <48.33%> (ø)
src/atomate2/utils/file_client.py	43.61% <50.00%> (+0.35%)	⬆️
src/atomate2/abinit/utils/history.py	57.89% <57.89%> (ø)
src/atomate2/abinit/sets/core.py	63.55% <63.55%> (ø)
src/atomate2/abinit/sets/base.py	79.01% <79.01%> (ø)
... and 6 more

[gpetretto]

Here are some comments on the code. I think it will be easier to discuss generic topics (like how to handle dependencies and restarts) in the upcoming meeting.

[utf]

This is coming along very nicely. Thanks for all your work @davidwaroquiers (and @gpetretto for reviewing).

I left a couple of minor comments I noticed.

[davidwaroquiers]

Hi @utf , we finished a first version for abinit. As it's a quite big block of changes, I'd like to merge it soon so that we can then work on smaller changes more easily after that. The base should be solid enough now. We may still change things so that we try to be even closer to the Vasp jobs/flows etc ... but at least it is a start. We still need to have the last version of abipy to be released (tests are passing on my computer but the released version of abipy will not work). We will tell you when this is done.
Thanks @gpetretto and @utf for the reviews!

[utf]

Great! I'll aim to do a full code review at the start of next week.

In the mean time, what is the plan for the tests? It would be good to have some kind of mocking set up like VASP does. I also made a script that made writing VASP tests easier (not that this is essential for this PR). I would like to at least get a rough version of the job/flow tests done before merging.

[davidwaroquiers]

Hi @utf , for the tests, we have two (maybe complementary) options. For ABINIT, we don't have the same issue as Vasp, we can actually run abinit. So we could indeed have a similar approach as the one for Vasp and then we can add the possibility to actually run integration tests instead of mocked tests. I will have a look at how things are done for Vasp and I will probably get back with some proposal (or questions).

[davidwaroquiers]

Hi @utf , I have a made a first draft for the testing infrastructure for abinit. Somewhat based on what is there for vasp but using pytest-mock and also allowing to actually run abinit instead of mocking it (this is an addoption to pytest in the main conftest.py).
With this, I think we have a first rough version of the job/flow tests infrastructure. There is currently just one job tested, but before going further I would like to get your opinion on this so that I do not spend time making the rest of the tests and then have to change everything :)
Thanks a lot!
David

[davidwaroquiers]

Hi @utf , I fixed the testing for abinit (was working on my computer but not on github because of environment, pseudos, versions etc ...). Now it should be fine.

[utf]

Please can you merge in the main branch? Its currently showing lots of changed files but I think they should go away.

[davidwaroquiers]

[davidwaroquiers]

Hi @utf, I'll reopen another (clean) PR. It seems I did something wrong somewhere and discussing with @gpetretto it is probably better to restart from scratch so that reviewing is easier. Sorry about this.