ica.plot_components() causes program to hang

šŸ’¬ Support & Discussions
, ,
1

Hello everyone,

Iā€™m having trouble with MNEā€™s ica.plot_components() function. When I call this function, my program seems to completely hang. Thereā€™s no error message, but thereā€™s also no expected output.

Hereā€™s the code Iā€™m trying to run:

ica = ICA(n_components=40, random_state=97)
ica.fit(raw_filt)
ica.plot_components()

I have tried the following solutions, but the problem persists:

  1. Reducing the number of ICA components
  2. Disabling interactive plotting
  3. Updating MNE and relevant packages to the latest versions

Here are the details of my environment:

  • Operating System: Windows 10 and win11 (I tried on different devices)
  • Python Version: (3.9 ; 3.8)
  • MNE Version: (1.2 ; 1.3)
  • NumPy Version: (1.24.3)
  • SciPy Version: (1.10.1)
  • Matplotlib Version: (-base=3.7.1; -inline=0.1.6)

The data Iā€™m working with is EEG data, and its size is 500mb. Iā€™m wondering if anyone else has encountered a similar problem, or if anyone knows how to solve this issue. Any help would be greatly appreciated.

Thank you for your time and help!

2

Can you save and share your ICA solution (ica.save())?

3

sorry, it seems the forums donā€™t permit this kind of file to be uploaded.

4

Perhaps you can put it on Dropbox/Google Drive/OneDrive/etc. and share a link?

5

[image]
ę–‡ä»¶å¤§å°: 54.4 KB 分äŗ«å†…容: ica
é“¾ęŽ„åœ°å€:https://pan.bnu.edu.cn/l/X1u1Ar

Could you try this one?

6

Works fine on my computer:

import mne

ica = mne.preprocessing.read_ica("ica")
ica.plot_components()

Maybe thereā€™s something wrong with your interactive plotting settings? Do you run the code with python, ipython, within a Jupyter notebook, ā€¦?

7

yes, I ran this on jupyter in pycharm!

8

In contrast to others, I do not recommend to use Jupyter notebooks (you are not the first person to encounter problems with visualizations). In my experience, a plain interactive interpreter session (which includes IPython), works much better for (interactive) data analysis. PyCharm has a very nice integrated Python terminal, please try if this solves your issue.

9

Please do try out @cbrnr 's suggestion, but just FYI, when I am using MNE visualization functions within a Jupyter notebook, I run %matplotlib qt in a cell before doing any plotting

I think mne.preprocessing.plot_components uses a matplotlib backend so this might solve your issue.

EDIT: FYI I think %matplotlib qt requires that you have the pyqt5 python package installed

10

thanks for your advice Scott. I have tried matplotlib qt/ inline. It works well while I doing other plot. Only the plot_component encountered the problem.

11

Can you try ica.plot_components(show=True)?