core dumped in normal code execution or kernel dead in jupyter notebook after calling crop function of raw object..

[abhiramsingh0]

trying to crop the raw object based on some indices. This is killing the kernel in Jupyter notebook. Running in python 2.7 also generates core dump. What is the reason of this?

[agramfort]

how did you install python / jupyter? show us what mne.show_info() tells you

[abhiramsingh0]

python is already installed on ubuntu. Jupyter was installed via pip.
Since program is terminated, after crop(), no information is obtained.
on running mne.show_info(), i got attribute error: module object has no attribute 'show_info'

[agramfort]

sorry i meant mne.sys_info() did you do: sudo pip install jupyter ?

[abhiramsingh0]

Platform: Linux-4.4.0-98-generic-x86_64-with-Ubuntu-16.04-xenial
Python: 2.7.12 (default, Nov 19 2016, 06:48:10) [GCC 5.4.0 20160609]
Executable: /home/gnl/TF/bin/python
CPU: x86_64: 16 cores
Memory: 31.3 GB

mne: 0.15.2
numpy: 1.13.1 {lapack=openblas, blas=openblas}
scipy: 0.19.1
matplotlib: 2.0.2

sklearn: 0.19.0
nibabel: 2.2.1
mayavi: Not found
pycuda: Not found
skcuda: Not found
pandas: Not found

yes i installed jupyter as you have mentioned.

[abhiramsingh0]

also getting same problem of kernel dead while running get_data() method of raw object.

[agramfort]

well that's really dangerous to do sudo pip install I would recommend you install ubuntu version of jupyter with apt-get or use anaconda but do not mix the 2.

[abhiramsingh0]

pip install was done by first creating virtualenv so I think this might not be the problem.

[agramfort]

you have an install problem for sure. it's not an mne problem. try using anaconda

[abhiramsingh0]

One more point I want to mention is that when I am trying to do crop directly on raw object it is not showing any error. But when different raw objects are appended in list and then crop and get_data method on elements of list are giving kernel dead error.

[agramfort]

can you share a code snippet to replicate?

[abhiramsingh0]

working case:

raw_eeg = mne.io.read_raw_eeglab(path_to_set_file, preload='True')
raw_eeg.crop()
raw_eeg.get_data()

non working case:

raw_eeg_obj = []
for index in range(13):
    raw_eeg = mne.io.read_raw_eeglab(path_to_set_file[index], preload='True')     # need .set files
    raw_eeg_obj.append(raw_eeg)
for index in range(13):
    raw_eeg_obj[index].crop()            # this is not working
    raw_eeg_obj[index].get_data()      # this is also not working

[agramfort]

this is not a self-contained example using sample dataset. to try quickly I should just need to "copy and paste". thx

[abhiramsingh0]

Providing full code to replicate the issue.
However you need to download the dataset, extract it and change the initial path in the code where all extracted folders are placed.
You need to download the files MM05.tar.bz2 and so on from the website http://www.cs.toronto.edu/~complingweb/data/karaOne/karaOne.html

import scipy.io as sio
import mne
import os

full_data = []
trial_indices = []

initial_path = '/home/gnl/Abhi/imagined_UoT/kara_one_classify/dataraw/'
dir1 = os.listdir(initial_path)
second_path = '/spoclab/users/szhao/EEG/data/'
for file_ in dir1:
    path = initial_path+file_+second_path
    dir2 = os.listdir(path)
    if dir2[0] != 'P02':
        # collect desired files for data, indices and labels
        files = os.listdir(path+dir2[0])
        set_file = [s for s in files if '.set' in s][0]
        index_file = [s for s in files if 'epoch_inds' in s][0]
        dir3 = os.listdir(path+dir2[0]+'/kinect_data')
        label_file = [s for s in dir3 if 'p.txt' in s][0]
        # now load the data of each subject
        raw_eeg = mne.io.read_raw_eeglab(path+dir2[0]+'/'+set_file, preload='True')
        full_data.append(raw_eeg)
        indices = sio.loadmat(path+dir2[0]+'/'+index_file)
        trial_indices.append(indices)

trial_data_full = []

def data_by_trial():
    for subject in range(13):
        for trial in range(trial_indices[subject]['thinking_inds'].shape[1]):
            # get indices of the trial
            indices = trial_indices[subject]['thinking_inds'][0][trial][0]/1000.0
            # crop data sequence
            tempData = full_data[subject].copy().crop(start=indices[0],stop=indices[1])
            # append to trial_data_full
            trial_data_full.append(tempData)

data_by_trial()

[agramfort]

you don't replicate with eg the sample dataset?