protocols.is_parameterized(sympy.pi) currently returns True

[daxfohl]

is_parameterized needs a check for free symbols, not just isinstance(sympy.Basic)

[NikolaiLong]

Hi again, @daxfohl. I know I just reached out to you regarding issue 7050, but this issue potentially looks like a simple fix. I would like to work on this issue to help give myself more scope of the project. Thanks again, I'm very excited to start working on Cirq!

[NikolaiLong]

I was previously unaware of the triage/discuss tag, I apologize for the confusion.

[mhucka]

Discussed in Cirq Cynq 2025-03-05: @daxfohl could you provide an example of where there is a problem caused by the current behavior? Are there examples where data argument types need to be resolved this way? During discussions, the general sense was that the issue may have merit if the need/scenario could be demonstrated with an example or two.

[mhucka]

I was previously unaware of the triage/discuss tag, I apologize for the confusion.

No worries! Thank you for your interest and patience.

[daxfohl]

I noticed it when working on #7057. I initially tried to use is_parameterized here https://github.com/quantumlib/Cirq/pull/7057/files#diff-a213015092da682cefaaabc63494f1e0b4d627c49b9ffbcbcc642c9ace074930R308, but that failed in cases where the inputs are sympy constants (which occured in already-existing test cases), since is_parameterized says sympy constants are parameterized, so I had to write up my own lambda instead.

[daxfohl]

I just saw the sync notes and wanted to make sure it's clear that currently is_parameterized(sympy.pi) returns true, which seems wrong, and we should change it to return false in that case since there are no free parameters. If I'm interpreting the sync notes correctly, the team thought this issue was saying the opposite.

Updated issue title to better reflect that.

[mhucka]

Add back triage/discuss to bring it to attention to the next Cynq.

@daxfohl Thank you for the clarification!