libAtoms / workflow

python workflow toolkit
GNU General Public License v2.0
24 stars 17 forks source link

Quantum espresso wrapper for new ASE version #321

Closed jungsdao closed 5 days ago

jungsdao commented 3 weeks ago

Some modifications on QE wfl wrapper and slight modification on minima hopping as it doesn't save energy and force etc in info/arrays.

bernstei commented 2 weeks ago

The tests are failing because profile is used in the tests but it is undefined. @gelzinyte can you take a look and do something analogous to what you did for aims? I don't exactly understand the whole new Profile business, and how you're calling Aims without passing AimsProfile object. Or do you want to close this one and do it from scratch for espresso separately?

jungsdao commented 2 weeks ago

In my local machine, profile is defined by ase_config file. I don't know how this config file should be defined for pytest. Also in my pytest in local machine, it actually fails because the workdir=tmp_path is defined as, for example, tmp_path = PosixPath('/tmp/pytest-of-hjung/pytest-2/test_qe_calculation0'), but actual workdir is /tmp/pytest-of-hjung/pytest-2/test_qe_calculation0/run_QE_9f6rsnhc. I'm not sure if this is expected behavior.

bernstei commented 2 weeks ago

I think I have a fix for the testing profile issue. Please apply the attached patch before you fix my comments above.

Also, there's no good reason for the pytest to fail for you. The workdir is supposed to be set to the directory in which the espresso calculator will be creating each run_QE_xxxxxx subdirectory, one for each config, and it will set self.directory to that, so the calculator should run in the right place. It certainly works for me, and when the github CI runs it.

[edited - new version of the patch]

new_espresso_test.patch

gelzinyte commented 6 days ago

And the tests seem to fail because of phono3py, also on the main branch. Let me see about fixing that.

jungsdao commented 6 days ago

And the tests seem to fail because of phono3py, also on the main branch. Let me see about fixing that.

Hi @gelzinyte, I'm a bit stuck with phono3py part of pytest because I had no idea how why this was happening. Feel free to finish this PR, sorry for not finishing it by myself.

bernstei commented 6 days ago

And the tests seem to fail because of phono3py, also on the main branch. Let me see about fixing that.

Hi @gelzinyte, I'm a bit stuck with phono3py part of pytest because I had no idea how why this was happening. Feel free to finish this PR, sorry for not finishing it by myself.

See https://github.com/libAtoms/workflow/pull/153 for why we're doing it this way now

bernstei commented 6 days ago

And the tests seem to fail because of phono3py, also on the main branch. Let me see about fixing that.

Hi @gelzinyte, I'm a bit stuck with phono3py part of pytest because I had no idea how why this was happening. Feel free to finish this PR, sorry for not finishing it by myself.

See #153 for why we're doing it this way now

Looks like we can probably do it (i.e. install from source rather than trying to fight with conda) with just python3 -m pip install -e . instead (based on https://phonopy.github.io/phono3py/install.html#installation-from-source-code). Assuming cmake is able to find the libraries it wants.

gelzinyte commented 6 days ago

Let me open a PR for phonopy, clone this PR and continue the discussion there?

bernstei commented 6 days ago

Let me open a PR for phonopy, clone this PR and continue the discussion there?

Sounds good.

bernstei commented 5 days ago

superseded by #325