Improve docstrings of aesaraf.join_non_shared_inputs

[ricardoV94]

This function could benefit from some more clear docstrings and example.

• Point is actually optional when make_shared is False
• Explain shared
• Make shared optional

pymc/pymc/aesaraf.py

Lines 565 to 586 in 761f77d

	def join_nonshared_inputs(
	point: Dict[str, np.ndarray],
	xs: List[TensorVariable],
	vars: List[TensorVariable],
	shared,
	make_shared: bool = False,
	):
	"""
	Takes a list of Aesara Variables and joins their non shared inputs into a single input.
	Parameters
	----------
	point: a sample point
	xs: list of Aesara tensors
	vars: list of variables to join
	Returns
	-------
	tensors, inarray
	tensors: list of same tensors but with inarray as input
	inarray: vector of inputs
	"""

[wd60622]

I am looking into this a bit but am currently understanding the function a bit more.

Is the point actually optional with make_shared = False?
I am seeing that point will be indexed in the for loop which seems like it will cause an error. Am I reading that wrong?

Seems like there would need some additional logic to make that happen.

[ricardoV94]

@wd60622 You're right. In that case the error message in the ifelse branch makes no sense. We can just make the point a required argument and remove the message.

[wd60622]

@wd60622 You're right. In that case the error message in the ifelse branch makes no sense. We can just make the point a required argument and remove the message.

Gotcha. Also noticing that when this function is called (smc and metropolis files in helper functions), the make_shared is never set from the default. i.e. raised error never touch with the currently implemented pymc code. Where you suggesting removing the error raising? Is there justification for keeping that ifelse branch in all together.

Obviously different scope. I can at least take a stab at adding to the type hints, better documenting what currently exists, and try to make some example based on the code I see for the docstring