[BUG] AttributeError: module 'numpy' has no attribute 'math' in 1_feature_overview [5] # Instantiate and fit the SINDy model

[michaelkalz]

I get "AttributeError: module 'numpy' has no attribute 'math'" when I run cell [5] in the binder notebook

Reproducing code example:

model = ps.SINDy()
model.fit(x_train, t=dt)
model.print()

Error message:

---

AttributeError Traceback (most recent call last)
Cell In[5], line 3
1 # Instantiate and fit the SINDy model
2 model = ps.SINDy()
----> 3 model.fit(x_train, t=dt)
4 model.print()

File /srv/conda/envs/notebook/lib/python3.10/site-packages/pysindy/pysindy.py:343, in SINDy.fit(self, x, t, x_dot, u, multiple_trajectories, unbias, quiet, ensemble, library_ensemble, replace, n_candidates_to_drop, n_subset, n_models, ensemble_aggregator)
337 u = validate_control_variables(
338 x,
339 u,
340 trim_last_point=(self.discrete_time and x_dot is None),
341 )
342 self.n_control_features_ = u[0].shape[u[0].ax_coord]
--> 343 x, x_dot = self._process_multiple_trajectories(x, t, x_dot)
345 # Set ensemble variables
346 self.ensemble = ensemble

File /srv/conda/envs/notebook/lib/python3.10/site-packages/pysindy/pysindy.py:665, in SINDy._process_multiple_trajectories(self, x, t, x_dot)
663 x = [xi[:-1] for xi in x]
664 else:
--> 665 x_dot = [
666 self.feature_library.calc_trajectory(
667 self.differentiation_method, xi, ti
668 )
669 for xi, ti in _zip_like_sequence(x, t)
670 ]
671 return x, x_dot

File /srv/conda/envs/notebook/lib/python3.10/site-packages/pysindy/pysindy.py:666, in (.0)
663 x = [xi[:-1] for xi in x]
664 else:
665 x_dot = [
--> 666 self.feature_library.calc_trajectory(
667 self.differentiation_method, xi, ti
668 )
669 for xi, ti in _zip_like_sequence(x, t)
670 ]
671 return x, x_dot

File /srv/conda/envs/notebook/lib/python3.10/site-packages/pysindy/feature_library/base.py:87, in BaseFeatureLibrary.calc_trajectory(self, diff_method, x, t)
85 def calc_trajectory(self, diff_method, x, t):
86 axes = x.dict
---> 87 x_dot = diff_method(x, t=t)
88 return AxesArray(x_dot, axes)

File /srv/conda/envs/notebook/lib/python3.10/site-packages/pysindy/differentiation/base.py:49, in BaseDifferentiation.call(self, x, t)
48 def call(self, x, t=1):
---> 49 return self._differentiate(x, t)

File /srv/conda/envs/notebook/lib/python3.10/site-packages/pysindy/differentiation/finite_difference.py:237, in FiniteDifference._differentiate(self, x, t)
234 if not np.isscalar(t):
235 dt = t[1] - t[0]
--> 237 coeffs = self._constant_coefficients(dt)
238 dims = np.array(x.shape)
239 dims[self.axis] = x.shape[self.axis] - (self.n_stencil - 1)

File /srv/conda/envs/notebook/lib/python3.10/site-packages/pysindy/differentiation/finite_difference.py:199, in FiniteDifference._constant_coefficients(self, dt)
195 matrices = (dt * (np.arange(self.n_stencil) - (self.n_stencil - 1) // 2))[
196 np.newaxis, :
197 ] ** pows
198 b = np.zeros(self.n_stencil)
--> 199 b[self.d] = np.math.factorial(self.d)
200 return np.linalg.solve(matrices, b)

File /srv/conda/envs/notebook/lib/python3.10/site-packages/numpy/init.py:414, in getattr(attr)
411 import numpy.char as char
412 return char.chararray
--> 414 raise AttributeError("module {!r} has no attribute "
415 "{!r}".format(name, attr))

AttributeError: module 'numpy' has no attribute 'math'

PySINDy/Python version information:

Binder Notebook from here.

[humatic]

I have the same issue with the notebook for the original paper.

I'm on Python 3.10, Numpy 2.0, Pysindy 1.7.5

Monkey patching may fix the immediate problem:

# Monkeypatch
from scipy.special import factorial
np.math = type('math', (), {})()
np.math.factorial = factorial

However for me the notebook still failed to execute due to subsequent errors.

BTW, I've noticed that latest dev doesn't seem to have this np.math.factorial code, so maybe a fix is underway..?

[humatic]

Related: #551

[ChidiOkoene]

I had the same issue as @humatic, but after fixing the numpy bug, the major problem was the self differentiation function and that of the optimizer and I don't know how to go about it.

[djpasseyjr]

I'm having the same error.

You can see it in the logs for my unit tests of my interfere package

For what it's worth, this stack exchange makes it sound like numpy.math is actually numpy's import of the builtin package math. (But it is from six years ago)

[djpasseyjr]

Also, most of my unit tests for SINDY passed. Only the tests that use a mix of differentiation methods failed.

[achalmahajan1]

Facing the same issue. Downgrading numpy to "1.26.4" as mentioned here does the trick #551.

[jesusmunozjmj]

Here's what worked for me:

import math
import numpy as np
np.math = math

[minimental]

When using customized libraries (pysindy.CustomLibrary), I also ran into a scikit-learn version mismatch issue. The following installation resolved the above given issue and the scikit-learn related one:

conda create -n pysindy conda-forge:pysindy numpy=1.26.4 scikit-learn=1.5.2