BUG: preliz import might break pymc import via scipy

[dirk]

Describe the issue:

This is via a colab

I suspect that preliz install breaks some of the pymc dependencies

Reproduceable code example:

import arviz as az
import bayeux as bx
import jax
import matplotlib.pyplot as plt
import numpy as np
import tensorflow_probability.substrates.jax as tfp
import pymc as pm
from scipy import stats
import preliz

tfd = tfp.distributions

Error message:

ImportError                               Traceback (most recent call last)

<ipython-input-5-6bd83c27ee19> in <cell line: 7>()
      5 import numpy as np
      6 import tensorflow_probability.substrates.jax as tfp
----> 7 import pymc as pm
      8 from scipy import stats
      9 import preliz

4 frames

/usr/local/lib/python3.10/dist-packages/scipy/cluster/vq.py in <module>
     68 import numpy as np
     69 from collections import deque
---> 70 from scipy._lib._array_api import (
     71     as_xparray, array_namespace, size, atleast_nd, copy, cov
     72 )

ImportError: cannot import name 'as_xparray' from 'scipy._lib._array_api' (/usr/local/lib/python3.10/dist-packages/scipy/_lib/_array_api.py)

PyMC version information:

see above

Context for the issue:

No response

[Ricardo Vieira]

Can you share a colab? Perhaps preliz is not pinning scipy well enough? CC @aloctavodia

[dirk]

It seems I cannot reproduce it myself, so maybe we can close it

https://colab.research.google.com/drive/1ZNPl0SM8DgXLK8IVUN9mSl0yr80YVYmw#scrollTo=AhcnyHugbi0o

[Osvaldo A Martin]

Just for reference PyMC asks for scipy>=1.4.1 and PreliZ "scipy>=1.10, <1.13"