support Infty and NaN in has_symbol(...)

[bjodah]

No description provided.

[isuruf]

I'm not sure about this. has_symbol is for symbolic variables, but infty and nan are not symbolic variables.

[bjodah]

Would you feel better with a has_basic(...) function and corresponding visitor? (I'm trying to make symengine.py work more a drop-in replacement for sympy since I'd like to use symengine more)

…

On Mon, Jul 28, 2025, 21:23 Isuru Fernando ***@***.***> wrote: *isuruf* left a comment (symengine/symengine#2105) <#2105 (comment)> I'm not sure about this. has_symbol is for symbolic variables, but infty and nan are not symbolic variables. — Reply to this email directly, view it on GitHub <#2105 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AADWUMFMAOYLS4WRAQBNZ4L3KZ2CVAVCNFSM6AAAAACAT4WQVOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTCMRYHA3TGOBSGE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[isuruf]

What does SymPy do with has_symbol?

[bjodah]

You can take a look at the suggested tests in symengine/symengine.py#515

That behavior matches SymPy. But perhaps we should not use has_symbol to implement the method has?

[isuruf]

Yeah, has is very broad. For eg:

>>> from sympy import *
>>> var("x y z")
(x, y, z)
>>> (2*x*y).has(2*x)
True
>>> (2*x*y).has(2)
True

[bjodah]

I know, in this case, I need to check if a symbolic expression contains infinity. has is useful, but perhaps there's another method?

[certik]

It looks like we already support infinity as a symbol, correct?

If so, adding the support for checking it makes sense.

How is the infinite symbol implemented exactly in SymEngine? That should guide us how to check for it.

[bjodah]

@certik, this is the definition:

symengine/symengine/infinity.h

Lines 20 to 22 in d1eb322

	class Infty : public Number
	{
	RCP<const Number> _direction;

In this case, a "HasBasicVisitor" that allows matching for more Basic subclasses than only Symbol would be useful for me.

If we want symengine.py to be a drop-in replacement for SymPy we need to figure out how to match subtrees I guess? (what I think @isuruf alluded to).

[isuruf]

Yeah, a HasBasicVisitor would be the correct thing to have.

[bjodah]

Alright, I will try to implement a HasBasicVisitor tomorrow (it's getting late here), and hopefully get back with a new PR soon. Thanks!

[bjodah]

Closing in favor of gh-2120