Numpy versioning issue with clean install via pip

[lazear]

Steps taken to install

uv venv
source .venv/bin/activate
uv pip install peak-performance

Example code

import pandas
import numpy as np
import arviz as az
from pathlib import Path
from peak_performance import pipeline as pl, models, plots

path = Path(".").absolute() / "example" / "A1t1R1Part2_110_109.9_110.1.npy"
timeseries = np.load(path)


pmodel = models.define_model_skew(
    time=timeseries[0],
    intensity=timeseries[1]
)

idata = pl.sampling(pmodel, tune=6000, draws=2000)

path_result = Path(r"")

plots.plot_posterior(
    identifier="test_plot",
    time=timeseries[0],
    intensity=timeseries[1],
    path=path_result,
    idata=idata,
    discarded=False,
)

Error message

(peak-performance) ➜  peak-performance python p.py
Traceback (most recent call last):
  File "/home/michael/github/peak-performance/p.py", line 5, in <module>
    from peak_performance import pipeline as pl
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/peak_performance/__init__.py", line 3, in <module>
    from . import models, pipeline, plots
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/peak_performance/models.py", line 26, in <module>
    import pymc as pm
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/pymc/__init__.py", line 46, in <module>
    __set_compiler_flags()
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/pymc/__init__.py", line 29, in __set_compiler_flags
    import pytensor
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/pytensor/__init__.py", line 118, in <module>
    from pytensor import scalar, tensor
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/pytensor/tensor/__init__.py", line 103, in <module>
    import pytensor.tensor.rewriting
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/pytensor/tensor/rewriting/__init__.py", line 1, in <module>
    import pytensor.tensor.rewriting.basic
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/pytensor/tensor/rewriting/basic.py", line 43, in <module>
    from pytensor.raise_op import Assert, CheckAndRaise, assert_op
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/pytensor/raise_op.py", line 14, in <module>
    from pytensor.tensor.type import DenseTensorType
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/pytensor/tensor/type.py", line 801, in <module>
    cscalar = TensorType("complex64", ())
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/pytensor/tensor/type.py", line 105, in __init__
    if np.obj2sctype(dtype) is None:
       ^^^^^^^^^^^^^
  File "/home/michael/github/peak-performance/.venv/lib/python3.12/site-packages/numpy/__init__.py", line 400, in __getattr__
    raise AttributeError(
AttributeError: `np.obj2sctype` was removed in the NumPy 2.0 release. Use `np.dtype(obj).type` instead.

Installing numpy==1.26.1 mitigates the issue - not sure if you guys can pin the exact version?

[michaelosthege]

This is a bug I've seen multiple times this week, with different projects. I suspect the NumPy pins by PyTensor are not strict enough for all install methods.
I couldn't reproduce it this time, but will nevertheless pin numpy <2 and remove it once PyTensor becomes compatible.

Thanks for pointing it out!