recover_marginals infers model from model context

[zaxtax]

This enables using recover_marginals in a model context. Just required making the model argument optional

Closes #610

[ricardoV94]

Just some minor suggestions, let me know if too much of a drag

[ricardoV94]

Suggested change

	raise TypeError("The first argument of `recover_marginals` must be an idata")
	raise TypeError("The order of arguments of `recover_marginals` changed. The first input must be an idata")

[zaxtax]

No it's all fairly minor and happy to do so. I'm mainly just stuck on why CI fails

…

On Thu, 18 Dec 2025, 00:31 Ricardo Vieira, ***@***.***> wrote: ***@***.**** commented on this pull request. Just some minor suggestions, let me know if too much of a drag ------------------------------ In pymc_extras/model/marginal/marginal_model.py <#619 (comment)> : > idata: InferenceData, + model: Model | None = None, Make all but idata kwarg only? ------------------------------ In pymc_extras/model/marginal/marginal_model.py <#619 (comment)> : > @@ -389,6 +389,11 @@ def recover_marginals( """ + if isinstance(idata, Model): Add a comment that this is temporary check due to API change and should be removed eventually ------------------------------ In pymc_extras/model/marginal/marginal_model.py <#619 (comment)> : > @@ -389,6 +389,11 @@ def recover_marginals( """ + if isinstance(idata, Model): + raise TypeError("The first argument of `recover_marginals` must be an idata") ⬇️ Suggested change - raise TypeError("The first argument of `recover_marginals` must be an idata") + raise TypeError("The order of arguments of `recover_marginals` changed. The first input must be an idata") ------------------------------ In tests/model/marginal/test_marginal_model.py <#619 (comment)> : > @@ -933,7 +936,7 @@ def test_nested(self): ) idata = InferenceData(posterior=dict_to_dataset(prior)) - idata = recover_marginals(marginal_m, idata, return_samples=True) + idata = recover_marginals(idata, return_samples=True) Are we testing the explicit model kwarg anywhere now? — Reply to this email directly, view it on GitHub <#619 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAACCUIITU4TOPFDO4CAKID4CHRVBAVCNFSM6AAAAACPEO2VISVHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZTKOJQGAZTMNBRGM> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[ricardoV94]

Need to ignore the arviz warning on project.toml and open an issue to address whatever it means.

[zaxtax]

How do we ignore the arviz warning? I don't know how the project toml file works?

…

On Thu, 18 Dec 2025, 09:22 Ricardo Vieira, ***@***.***> wrote: *ricardoV94* left a comment (pymc-devs/pymc-extras#619) <#619 (comment)> Need to ignore the arviz warning on project.toml and open an issue to address whatever it means. — Reply to this email directly, view it on GitHub <#619 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAACCULXW7MY3VSJSV7GZDL4CJP27AVCNFSM6AAAAACPEO2VISVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTMNRYHE4DSNZQGQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[ricardoV94]

It's done here:

pymc-extras/pyproject.toml

Lines 102 to 112 in 10d6765

	filterwarnings =[
	"error",
	# JAX issues an over-eager warning if os.fork() is called when the JAX module is loaded, even if JAX isn't being used
	'ignore:os\.fork\(\) was called\.:RuntimeWarning',
	# Preliz needs to update for pytensor > 2.35
	'ignore:.*`pytensor\.graph\.basic\.ancestors`.*`pytensor\.graph\.traversal\.ancestors`.*:FutureWarning:^preliz(\.|$)',
	# OpenMP library warning on windows CI
	'ignore::RuntimeWarning:threadpoolctl'
	]

[zaxtax]

Any idea why I'm getting this jax one? https://github.com/pymc-devs/pymc-extras/actions/runs/20270017341/job/58203008839#step:5:535

…

On Thu, 18 Dec 2025, 09:29 Ricardo Vieira, ***@***.***> wrote: *ricardoV94* left a comment (pymc-devs/pymc-extras#619) <#619 (comment)> It's done here: https://github.com/pymc-devs/pymc-extras/blob/10d6765ded4de2da9630e6f21540dcc8373da3e7/pyproject.toml#L102-L112 — Reply to this email directly, view it on GitHub <#619 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAACCUMVNBCN72EWMIA62VT4CJQVXAVCNFSM6AAAAACPEO2VISVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTMNRZGAZDKOJTGI> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[ricardoV94]

Not because of any changes you did, so mark the test as xfail and open an issue to track?

[ricardoV94]

We have two RTD jobs? Anyway failure unlikely to be related

[ricardoV94]

Thanks @zaxtax