pdf method of MarginalDistribution syntactically incorrect

[oscargus]

While cleaning up the code, I realized the following problem. However, as I am not that familiar with the stats module and therefore wasn't able to write a test that covers the potential issues, I open an issue.

Consider the method:

sympy/sympy/stats/joint_rv.py

Lines 312 to 328 in d4475ac

	def pdf(self, *x):
	expr, rvs = self.args[0], self.args[1]
	marginalise_out = [i for i in random_symbols(expr) if i not in self.args[1]]
	syms = [i.pspace.symbol for i in self.args[1]]
	for i in expr.atoms(Indexed):
	if isinstance(i, Indexed) and isinstance(i.base, RandomSymbol)\
	and i not in rvs:
	marginalise_out.append(i)
	if isinstance(expr, CompoundDistribution):
	syms = Dummy('x', real=True)
	expr = expr.args[0].pdf(syms)
	elif isinstance(expr, JointDistribution):
	count = len(expr.domain.args)
	x = Dummy('x', real=True, finite=True)
	syms = [Indexed(x, i) for i in count]
	expr = expression.pdf(syms)
	return Lambda(syms, self.compute_pdf(expr, marginalise_out))(*x)

The primary issue is

sympy/sympy/stats/joint_rv.py

Line 327 in d4475ac

expr = expression.pdf(syms)

where expression is not defined. Should probably be expr. However, looking at

sympy/sympy/stats/joint_rv.py

Line 322 in d4475ac

expr = expr.args[0].pdf(syms)

it may be that it should instead be expr.args[0].

On

sympy/sympy/stats/joint_rv.py

Lines 314 to 315 in d4475ac

	marginalise_out = [i for i in random_symbols(expr) if i not in self.args[1]]
	syms = [i.pspace.symbol for i in self.args[1]]

one may replace args[1] with rvs

Now,

sympy/sympy/stats/joint_rv.py

Line 315 in d4475ac

syms = [i.pspace.symbol for i in self.args[1]]

is redefined by

sympy/sympy/stats/joint_rv.py

Line 321 in d4475ac

syms = Dummy('x', real=True)

and

sympy/sympy/stats/joint_rv.py

Line 326 in d4475ac

syms = [Indexed(x, i) for i in count]

and, hence, it should probably be moved to an else: clause just before

sympy/sympy/stats/joint_rv.py

Line 328 in d4475ac

return Lambda(syms, self.compute_pdf(expr, marginalise_out))(*x)

First, though, it would be good to make sure that this code is covered by a few tests since there are many things that are not obvious when just browsing the code. For example, the arguments to pdf are redefined on line

sympy/sympy/stats/joint_rv.py

Line 325 in d4475ac

x = Dummy('x', real=True, finite=True)

which may be OK or just an unfortunate variable naming.

[supreet11agrawal]

I am studying the stats module so I would like to solve this one. I might take some time though

[oscargus]

Sure! Go ahead!

[Sc0rpi0n101]

@oscargus Is this still an issue?
I don't see it fixed in current master

[Sc0rpi0n101]

Also,
sympy/sympy/stats/joint_ry.py

Line 323

elif isinstance(expr, JointDistribution):
            count = len(expr.domain.args)
            x = Dummy('x', real=True, finite=True)
            syms = [Indexed(x, i) for i in count]
            expr = expression.pdf(syms)

We are checking if expr is an instance of JointDistribuion
and Line 327 is expr = expression.pdf(syms)
and in Line 158

    def pdf(self, *args):
        return self.density.args[1]

So, I think it will be expr which is an instance of JointDistribution and uses syms as args

[Sc0rpi0n101]

as for syms, it does not look like it is doing anything else in the code,
I tried putting it in else just before line 328
Did not have any test fails.

So, I'm trying to see if new tests are required

[oscargus]

I do not think that this is fixed, no.

Adding at least one test is required as the incorrect line for sure is not covered now.

[supreet11agrawal]

Currently, I am working on this. This is not as simple as it looks. Joint Random Variables in Sympy were implemented last year so they have a lot of vulnerabilities as of now. Tests are quite redundant in terms of checking the working of the module. That is why this will take some time.

[Sc0rpi0n101]

Yeah, no problem.
It was 3 months old and had no activity, so I was just asking.
If you're working on it, go ahead.

[Smit-create]

I am working on it.

[Smit-create]

sympy/sympy/stats/joint_rv.py

Lines 315 to 328 in 63ab534

	def pdf(self, *x):
	expr, rvs = self.args[0], self.args[1]
	marginalise_out = [i for i in random_symbols(expr) if i not in rvs]
	if isinstance(expr, CompoundDistribution):
	syms = Dummy('x', real=True)
	expr = expr.args[0].pdf(syms)
	elif isinstance(expr, JointDistribution):
	count = len(expr.domain.args)
	x = Dummy('x', real=True, finite=True)
	syms = tuple(Indexed(x, i) for i in count)
	expr = expr.pdf(syms)
	else:
	syms = tuple(rv.pspace.symbol if isinstance(rv, RandomSymbol) else rv.args[0] for rv in rvs)
	return Lambda(syms, self.compute_pdf(expr, marginalise_out))(*x)

Looking into the issue, I see that all the pointed erros are been resolved in current master. I think we can add a test for the same.

[Smit-create]

@Upabjojr @czgdp1807 Please have a look if this can be closed.