Formulation of Helicity model in "Modify amplitude model" results in Error

[n-idw]

What happened?

When trying to run the "modify.ipynb" notebook including the code of the "Modify amplitude model" chapter in the documentation, an error is returned when trying to run "original_model = model_builder.formulate()".

Relevant log output

/usr/local/lib/python3.10/dist-packages/ampform/helicity/__init__.py in _generate_kinematic_variable_set(transition, node_id)
    935     child2_mass = get_invariant_mass_symbol(topology, decay.children[1].id)
    936     angular_momentum: int | None = decay.interaction.l_magnitude
--> 937     if angular_momentum is None and decay.parent.particle.spin.is_integer():
    938         angular_momentum = int(decay.parent.particle.spin)
    939     return TwoBodyKinematicVariableSet(

AttributeError: 'Fraction' object has no attribute 'is_integer'

Is it possible to reproduce the bug with a small code snippet?

result = qrules.generate_transitions(
    initial_state=("J/psi(1S)", [-1, +1]),
    final_state=["gamma", "pi0", "pi0"],
    allowed_intermediate_particles=["f(0)(980)", "f(0)(1500)"],
    allowed_interaction_types=["strong", "EM"],
    formalism="helicity",
)
model_builder = get_builder(result)
original_model = model_builder.formulate()

Additional steps to reproduce the bug

No response

Which Python version were you using?

Python 3.10

Python dependencies

[redeboer]

Thanks for posting!
This bug originates from the upgrade to QRules v0.10.4, specifically ComPWA/qrules#288. What I don't get is that the tests didn't catch it – these lines are covered by the unit tests. Would be a good idea to add this channel to the tests.

For you, the easiest fix for now is to downgrade QRules, e.g.

pip install qrules==0.10.3

Does that work?

We'll implement a hotfix in the meantime. @grayson-helmholz?

[n-idw]

Thanks for the quick answer!
I downgraded to ampform==0.14.10, that fixed the problem. Apparently the change was made in 0.14.11.

[redeboer]

I downgraded to ampform==0.14.10, that fixed the problem. Apparently the change was made in 0.14.11.

Also an option 👌 But in 0.14.x you may miss some of the newer functionality. For one, it still uses QRules v0.9.x (which is the reason the bug does not appear here) and there are some other improvements (see releases).

On the other hand, if you upgrade to AmpForm v0.15.x, you might have to spend some time modifying some of your scripts.

[grayson-helmholz]

Alternatively, you can also change to Python 3.12 for now, as Fraction.is_integer has been added since then.
But the bug will still have to be fixed, because we currently still support Python >=3.9

[redeboer]

@grayson-helmholz, as you noted, the bug fix is quite simple. We need a failing test first though.

1. Interestingly, the bug in the notebook was not spotted, because the documentation is built on Python 3.12. To quickly check:

   uv run -p3.10 pytest --nbmake docs/usage/modify.ipynb  # fails

   uv run -p3.12 pytest --nbmake docs/usage/modify.ipynb  # works

   It makes me wonder whether we use the notebooks on integration tests on all Python versions, with pytest --nbmake 🤔 So far, I assumed running them while building the documentation was a sufficient test, but that's only done on one Python version.
2. I don't get though why the unit tests (which are run on all Python versions) didn't catch this. Specifically test_rename_all_parameters_with_stable_final_state(), which uses a parametrized reaction for "helicity" and "canonincal-helicity". The reaction in that test is the same, only the test assigns dynamics, whereas the notebook does not.
   At any rate, that test indeed fails on Python 3.10:

   uv run -p3.10 pytest tests/helicity/test_helicity.py::TestHelicityModel::test_rename_all_parameters_with_stable_final_state

   So why didn't it fail in CI once the lock file updated QRules to v0.10.4 in MAINT: update lock files #445?

[redeboer]

2. I don't get though why the unit tests (which are run on all Python versions) didn't catch this.

See ComPWA/actions#124