effective_area_per_energy_and_fov() creates error in version 0.10.0

[tristangrad]

using this function with below specified (but probably arbitrarily) args creates the error 'AttributeError: 'numpy.ndarray' object has no attribute '_set_unit''

minimal (not) working example:

simulation_info = SimulatedEventsInfo(
    energy_min=0.003*u.TeV,
    energy_max=330*u.TeV,
    spectral_index=-2,
    n_showers=1000000,
    max_impact=500 * u.m,
    viewcone_min=0 * u.deg,
    viewcone_max=10 * u.deg,
)

true_energy_bins= arbitrary 
fov_offset_bins= arbitrary
selected_events = QTable(data=[[0.05*u.TeV] , [1.35*u.deg]],names=['true_energy','true_source_fov_offset'])

[StFroese]

The code for reproducing the error is missing, right? This is not failing for me

[StFroese]

ahh the function is in the title...

[StFroese]

works for me:

In [24]: fov_offset_bins = u.Quantity(sorted(np.random.rand(100)), 'deg')

In [25]: true_energy_bins = u.Quantity(sorted(np.random.rand(100)), 'TeV')

In [26]: 

In [26]: effective_area_per_energy_and_fov(selected_events, simulation_info, true_energy_bins, fov_offset_bins)
Out[26]: 
<Quantity [[0., 0., 0., ..., 0., 0., 0.],
           [0., 0., 0., ..., 0., 0., 0.],
           [0., 0., 0., ..., 0., 0., 0.],
           ...,
           [0., 0., 0., ..., 0., 0., 0.],
           [0., 0., 0., ..., 0., 0., 0.],
           [0., 0., 0., ..., 0., 0., 0.]] m2>

[RuneDominik]

Same here, I've rewritten your example as a test, which passes. Maybe some conflicting versions of astropy and numpy?

def test_viewcone_integral_effective_area_per_energy_and_fov():
    import astropy.units as u

    from astropy.table import QTable

    from pyirf.irf import effective_area_per_energy_and_fov
    from pyirf.simulations import SimulatedEventsInfo

    simulation_info = SimulatedEventsInfo(
        energy_min=0.003 * u.TeV,
        energy_max=330 * u.TeV,
        spectral_index=-2,
        n_showers=1000000,
        max_impact=500 * u.m,
        viewcone_min=0 * u.deg,
        viewcone_max=10 * u.deg,
    )

    fov_bins = [0, 2] * u.deg
    energy_bins = [0, 1] * u.TeV

    selected_events = QTable(
        data=[[0.05 * u.TeV], [1.35 * u.deg]],
        names=["true_energy", "true_source_fov_offset"],
    )

    aeff = effective_area_per_energy_and_fov(
        selected_events=selected_events,
        simulation_info=simulation_info,
        true_energy_bins=energy_bins,
        fov_offset_bins=fov_bins,
    )

    assert aeff >= 0

Tested with the current pyirf main and astropy=5.3.2 numpy=1.25.2 python=3.11.5

Edit: Also works in a fresh python 3.9 env with pyirf v0.10

[StFroese]

Test works for me too with: astropy=5.3.3 numpy=1.25.2 python=3.9.18

installed via conda