Describe the bug

This issue has been discussed on the Forum here with the suggestion to open a bug report.

I face the following error when running the CSP function using the EEG BCI dataset:

LinAlgError : The leading minor of order 64 of B is not positive definite. The factorization of B could not be completed and no eigenvalues or eigenvectors were computed.

Steps to reproduce

import numpy as np
from mne.decoding import CSP

loaded = np.load('preprocessingoutput.npz')
CSP(n_components=5, reg=1e-4).fit(loaded['x1'], loaded['y1'])

preprocessingoutput.npz and the preprocessing steps run on the EEG data are provided here.

Expected results

running CSP analysis.

Actual results

LinAlgError : The leading minor of order 64 of B is not positive definite. The factorization of B could not be completed and no eigenvalues or eigenvectors were computed.

Additional information

mne.sys_info()
Platform: Windows-10-10.0.19041-SP0
Python: 3.8.3 (default, Jul 2 2020, 17:30:36) [MSC v.1916 64 bit (AMD64)]
Executable: C:\Users????\Anaconda3\python.exe
CPU: Intel64 Family 6 Model 142 Stepping 10, GenuineIntel: 8 cores
Memory: 7.9 GB

mne: 0.22.0
numpy: 1.18.5 {blas=mkl_rt, lapack=mkl_rt}
scipy: 1.5.4
matplotlib: 3.3.3 {backend=module://ipykernel.pylab.backend_inline}

sklearn: 0.23.2
numba: 0.50.1
nibabel: Not found
nilearn: Not found
dipy: Not found
cupy: Not found
pandas: 1.0.5
mayavi: Not found
pyvista: Not found
vtk: Not found