suggestions from pmd general assembly

[ligerzero-ai]

Suggestions from materialdigital general assembly.

Export an environment yaml inside the hdf5 file automatically?
Down to the granular (atomic dependencies... e.g. not just "pyiron_atomistics" but all the package versions that were used in a particular pyiron run?

[jan-janssen]

I like the idea, in particular when we move towards one HDF5 file per workflow rather than once HDF5 file per calculation.

[ligerzero-ai]

there was also a request to allow custom submission scripts - since there was a user that wanted to adjust parameters inside the job submission script across varying types of jobs.

[jan-janssen]

there was also a request to allow custom submission scripts - since there was a user that wanted to adjust parameters inside the job submission script across varying types of jobs.

In pyiron_base this functionality is a bit hidden, but you can specify additional arguments using:

job.server.additional_arguments = {"my_parameter_1": 1, "my_parameter_2": 2} 

https://github.com/pyiron/pyiron_base/blob/main/pyiron_base/jobs/job/extension/server/generic.py#L120

We might have to document it a bit better. The full discussion on this topic is available in the corresponding pull request #1082

For future reference, the documentation is a bit better for pysqa as we set the custom parameters using the **kwargs. You can set any key and as long as it appears in your submission script template it is parsed correctly. https://pysqa.readthedocs.io/en/latest/python.html#submit-job-to-queue

[ligerzero-ai]

so basically, replace the parameters that are changing anywhere in the jinja jobscript with the text
{{varying_parameter}}, {{varying_parameter2}} and specify it as

job.server.additional_arguments = {"varying_parameter": 1, "varying_parameter2": 2}}

right?

[jan-janssen]

right?

Correct.