Add optional argument degree_of_freedom for asymptotic_critical_value

[dachengx]

Resolve: #83

What does the code in this PR do / what does it improve?

Add optional argument degree_of_freedom for asymptotic_critical_value

Can you briefly describe how it works?

1. For models, if asymptotic_dof is provided as an __init__ argument, the StatisticalModel.asymptotic_dof will be used in _confidence_interval_checks.
2. For runners, if asymptotic_dof is in statistical_model_args, the Runner.confidence_interval_thresholds will be set according to asymptotic_dof, and further used in CL calculation.

Can you give a minimal working example (or illustrate with a figure)?

from alea.utils import asymptotic_critical_value
asymptotic_critical_value("central", 0.9)
>>> 2.705543454095414

What are the potential drawbacks of the codes?

When "confidence_interval_kind" is not "lower" or "upper", and degree_of_freedom is not None or 1, asymptotic_critical_value will raise an error.

Please include the following if applicable:

• Update the docstring(s)
• Update the documentation
• Tests to check the (new) code is working as desired.
• Does it solve one of the open issues on github?

[github-actions]

Pull Request Test Coverage Report for Build 6223968781

• 10 of 16 (62.5%) changed or added relevant lines in 3 files are covered.
• 1 unchanged line in 1 file lost coverage.
• Overall coverage decreased (-0.4%) to 91.81%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
alea/submitter.py	1	2	50.0%
alea/utils.py	4	5	80.0%
alea/model.py	5	9	55.56%

Files with Coverage Reduction	New Missed Lines	%
alea/model.py	1	86.1%

Totals
Change from base Build 6052911707:	-0.4%
Covered Lines:	1334
Relevant Lines:	1453

---

💛 - Coveralls

[kdund]

Can we change the None defaults to 1 for the DOF?

[kdund]

Why warning and not error?

[dachengx]

There will be no generate_values in real data, so we will set a set of data without generate_values. This was introduced along with https://github.com/XENONnT/alea/pull/79/files#diff-ccb14fbc2f6819c7774476fe82d0c97563c7006391df64c86b0fbea0706af8f5R421 store_real_data function.

[dachengx]

I need to submit a new PR to solve the deprecated readthedocs configuration. https://readthedocs.org/projects/alea/builds/21954754/

[kdund]

Thanks, @dachengx !

[dachengx]

Hey, @kdund . Sorry for the back-and-forth, I was validating the PR by non-beta study. I find that it is better to set asymptotic_dof of StatisticalModel.confidence_interval to None to avoid mis-assignment. But still the default asymptotic_dof of StatisticalModel.__init__ is 1. I think this will ease most of the pain, and one does not need to set it explicitly as 1 in most cases.

So though the default value of asymptotic_dof of StatisticalModel.confidence_interval is None, the asymptotic_dof can be readout from StatisticalModel.asymptotic_dof.