Unable to avoid copy while creating an array as requested

[jclsn]

Hi, I am currently trying to execute the code in your Notebook on Kalman filters. In the Multivariate Gaussians section this error is thrown, which seems to be an allocation problem in filterpy

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[19], line 7
      3 mean = [2., 17.]
      4 cov = [[10., 0.], 
      5        [0., 4.]]
----> 7 mkf_internal.plot_3d_covariance(mean, cov)

File [~/work/Kalman-and-Bayesian-Filters-in-Python/kf_book/mkf_internal.py:283](/user/jan/lab/tree/Kalman-and-Bayesian-Filters-in-Python/Kalman-and-Bayesian-Filters-in-Python/kf_book/mkf_internal.py#line=282), in plot_3d_covariance(mean, cov)
    280 ys = np.arange(miny, maxy, (maxy-miny)[/40.](/40.))
    281 xv, yv = np.meshgrid(xs, ys)
--> 283 zs = np.array([100.* stats.multivariate_gaussian(np.array([x, y]), mean, cov) \
    284                for x, y in zip(np.ravel(xv), np.ravel(yv))])
    285 zv = zs.reshape(xv.shape)
    287 fig = plt.figure()

File [/opt/conda/lib/python3.12/site-packages/filterpy/stats/stats.py:382](/opt/conda/lib/python3.12/site-packages/filterpy/stats/stats.py#line=381), in multivariate_gaussian(x, mu, cov)
    380 # force all to numpy.array type, and flatten in case they are vectors
    381 x = np.array(x, copy=False, ndmin=1).flatten()
--> 382 mu = np.array(mu, copy=False, ndmin=1).flatten()
    384 nx = len(mu)
    385 cov = _to_cov(cov, nx)

ValueError: Unable to avoid copy while creating an array as requested.
If using `np.array(obj, copy=False)` replace it with `np.asarray(obj)` to allow a copy when needed (no behavior change in NumPy 1.x).
For more details, see https://numpy.org/devdocs/numpy_2_0_migration_guide.html#adapting-to-changes-in-the-copy-keyword.
``

[Meikul]

I'm also running into this issue in chapter 5 unfortunately. I installed the project dependencies via the conda environment.yml file. and I've tried changing the versions of some packages, but I've had no luck.

[jclsn]

@Meikul: I patched the mahalanobis and likelihood functions by replacing them with this

 def _safe_mahalanobis(self):
     if self._mahalanobis is None:
         y = np.atleast_2d(self.y)
         if y.shape[0] == 1:
             y = y.T
         SI = np.atleast_2d(self.SI)
         self._mahalanobis = np.sqrt((y.T @ SI @ y).item())
     return self._mahalanobis
 
 def _safe_likelihood(self):
     if self._likelihood is None:
         self._likelihood = exp(self.log_likelihood)
         if self._likelihood == 0:
             self._likelihood = sys.float_info.min
     return self._likelihood

Haven't had any issues since. Since I built myself a JupyterLab Docker container with the book baked in, I used a sitecustomize.py

# sitecustomize.py
import numpy as np
import sys
import warnings

from math import exp
from filterpy.kalman import KalmanFilter

warnings.filterwarnings(
    "ignore",
    category=SyntaxWarning,
    message=r".*invalid escape sequence.*"
)

def _safe_mahalanobis(self):
    if self._mahalanobis is None:
        y = np.atleast_2d(self.y)
        if y.shape[0] == 1:
            y = y.T
        SI = np.atleast_2d(self.SI)
        self._mahalanobis = np.sqrt((y.T @ SI @ y).it
    return self._mahalanobis

def _safe_likelihood(self):
    if self._likelihood is None:
        self._likelihood = exp(self.log_likelihood)
        if self._likelihood == 0:
            self._likelihood = sys.float_info.min
    return self._likelihood

KalmanFilter.mahalanobis = property(_safe_mahalanobis
KalmanFilter.likelihood = property(_safe_likelihood) 

Which I put into /opt/conda/lib/python3.1/site-packages/. This is probably the easiest approach for you as well. You should probably put it into ~/.local/lib/python3.x depending on your setup though.

Unfortunately @rlabbe is not really responding anymore, so I didn't bother to send a PR.