stats.rv_discrete is not checking that xk values are integers

scipy.stats.rv_discrete requires that the x input for the values parameters are integers:

values : tuple of two array_like, optional (xk, pk) where xk are integers and pk are the non-zero probabilities between 0 and 1 with sum(pk) = 1. xk and pk must have the same shape.

Reproducing code example:

The example code:

from scipy import stats
xk = np.arange(7)
pk = (0.1, 0.2, 0.3, 0.1, 0.1, 0.0, 0.2)
custm = stats.rv_discrete(name='custm', values=(xk, pk))

import matplotlib.pyplot as plt
fig, ax = plt.subplots(1, 1)
ax.plot(xk, custm.pmf(xk), 'ro', ms=12, mec='r')
ax.vlines(xk, 0, custm.pmf(xk), colors='r', lw=4)
plt.show()

returns

but if I change to xk = np.linspace(1., 6.9, 7) I get:

This is a user error with a silent fail. There should be a validation or at least a warning shown.

Scipy/Numpy/Python version information:

1.6.1 1.19.2 sys.version_info(major=3, minor=8, micro=8, releaselevel='final', serial=0)

[mdhaber]

See gh-2414. I see this is about direct use of rv_discrete though, so maybe not as relevant. Just note that we don’t necessarily want all rv_discrete subclasses to only accept integer input. Not relevant. That was about a shape parameter, not xk.

[mdhaber]

After reading this and gh-3758, it is more clear that they are duplicates. Let’s close this and continue in gh-3758.