RssMaker does not work as expected

[naik-aakash]

Steps to reproduce

Note: The config file used for this example is from our recent RSS preprint

from autoplex.auto.rss.flows import RssMaker
from jobflow import run_locally

rss_job = RssMaker(name="test rss").make(config_file="/home/anaik/Work/Dev_Codes/autoplex/tests/test_data/rss/si.yaml")

run_locally(rss_job)

Error Traceback

/home/anaik/miniconda3/envs/autoplex_dev/bin/python /home/anaik/Work/Dev_Codes/autoplex/rss_wf.py 
/home/anaik/miniconda3/envs/autoplex_dev/lib/python3.10/site-packages/nequip/__init__.py:20: UserWarning:

!! PyTorch version 2.2.1+cu121 found. Upstream issues in PyTorch versions 1.13.* and 2.* have been seen to cause unusual performance degredations on some CUDA systems that become worse over time; see https://github.com/mir-group/nequip/discussions/311. The best tested PyTorch version to use with CUDA devices is 1.11; while using other versions if you observe this problem, an unexpected lack of this problem, or other strange behavior, please post in the linked GitHub issue.

2025-01-10 20:24:57,918 - INFO - Started executing jobs locally
2025-01-10 20:24:57,925 - INFO - Starting job - test rss (c560d438-d381-429d-9bcf-d317e51c4797)
2025-01-10 20:24:57,918 INFO Started executing jobs locally
2025-01-10 20:24:57,925 INFO Starting job - test rss (c560d438-d381-429d-9bcf-d317e51c4797)
2025-01-10 20:24:58,141 - INFO - Finished job - test rss (c560d438-d381-429d-9bcf-d317e51c4797)
2025-01-10 20:24:58,143 - INFO - Starting job - initial_rss (d202bb3c-4bcd-47a5-b721-4706ee0c8149)
2025-01-10 20:24:58,148 - INFO - initial_rss failed with exception:
Traceback (most recent call last):
  File "/home/anaik/miniconda3/envs/autoplex_dev/lib/python3.10/site-packages/jobflow/managers/local.py", line 114, in _run_job
    response = job.run(store=store)
  File "/home/anaik/miniconda3/envs/autoplex_dev/lib/python3.10/site-packages/jobflow/core/job.py", line 600, in run
    response = function(*self.function_args, **self.function_kwargs)
  File "/home/anaik/Work/Dev_Codes/autoplex/src/autoplex/auto/rss/jobs.py", line 215, in initial_rss
    do_mlip_fit = MLIPFitMaker(
  File "/home/anaik/Work/Dev_Codes/autoplex/src/autoplex/fitting/common/flows.py", line 205, in make
    mlip_fit_job = machine_learning_fit(
TypeError: autoplex.fitting.common.jobs.machine_learning_fit() got multiple values for keyword argument 'database_dir'

2025-01-10 20:24:58,148 - INFO - Finished executing jobs locally
2025-01-10 20:24:58,141 INFO Finished job - test rss (c560d438-d381-429d-9bcf-d317e51c4797)
2025-01-10 20:24:58,143 INFO Starting job - initial_rss (d202bb3c-4bcd-47a5-b721-4706ee0c8149)
2025-01-10 20:24:58,148 INFO initial_rss failed with exception:
Traceback (most recent call last):
  File "/home/anaik/miniconda3/envs/autoplex_dev/lib/python3.10/site-packages/jobflow/managers/local.py", line 114, in _run_job
    response = job.run(store=store)
  File "/home/anaik/miniconda3/envs/autoplex_dev/lib/python3.10/site-packages/jobflow/core/job.py", line 600, in run
    response = function(*self.function_args, **self.function_kwargs)
  File "/home/anaik/Work/Dev_Codes/autoplex/src/autoplex/auto/rss/jobs.py", line 215, in initial_rss
    do_mlip_fit = MLIPFitMaker(
  File "/home/anaik/Work/Dev_Codes/autoplex/src/autoplex/fitting/common/flows.py", line 205, in make
    mlip_fit_job = machine_learning_fit(
TypeError: autoplex.fitting.common.jobs.machine_learning_fit() got multiple values for keyword argument 'database_dir'

2025-01-10 20:24:58,148 INFO Finished executing jobs locally

Process finished with exit code 0

[naik-aakash]

Seems this change on main branch is the issue. The argument is now a class attribute rather than arg to Make

autoplex/src/autoplex/fitting/common/flows.py

Line 87 in d2b96d9

database_dir: Path | str | None = None,

[JaGeo]

Independent of the bug, I think it makes sense to keep it as a part of the make to easily reuse this maker.

[QuantumChemist]

I guess this went under my radar when I made the changes because the RssMaker is not properly included in the unit tests and we are using mock_rss instead.

[JaGeo]

@QuantumChemist yes, this can surely happen. Thus, tests for this specific Maker are needed.

[YuanbinLiu]

Some new users have also encountered this issue, and I will check it as soon as possible. We may also need to find a way to fix the random seed in buildcell, otherwise it's impossible to provide a complete unit test for RssMaker.

[JaGeo]

As mentioned previously, we can ommit the test of the POSCAR. Then, we can provide a test