Time-frequency beamforming using LCMV

Compute LCMV source power [1] in a grid of time-frequency windows and display results.

References

[1]Dalal et al. Five-dimensional neuroimaging: Localization of the time-frequency dynamics of cortical activity. NeuroImage (2008) vol. 40 (4) pp. 1686-1700
# Author: Roman Goj <roman.goj@gmail.com>
#
# License: BSD (3-clause)

import mne
from mne import compute_covariance
from mne.datasets import sample
from mne.event import make_fixed_length_events
from mne.beamformer import tf_lcmv
from mne.viz import plot_source_spectrogram

print(__doc__)

data_path = sample.data_path()
raw_fname = data_path + '/MEG/sample/sample_audvis_raw.fif'
noise_fname = data_path + '/MEG/sample/ernoise_raw.fif'
event_fname = data_path + '/MEG/sample/sample_audvis_raw-eve.fif'
fname_fwd = data_path + '/MEG/sample/sample_audvis-meg-eeg-oct-6-fwd.fif'
subjects_dir = data_path + '/subjects'
label_name = 'Aud-lh'
fname_label = data_path + '/MEG/sample/labels/%s.label' % label_name

Read raw data, preload to allow filtering

raw = mne.io.read_raw_fif(raw_fname, preload=True)
raw.info['bads'] = ['MEG 2443']  # 1 bad MEG channel

# Pick a selection of magnetometer channels. A subset of all channels was used
# to speed up the example. For a solution based on all MEG channels use
# meg=True, selection=None and add grad=4000e-13 to the reject dictionary.
# We could do this with a "picks" argument to Epochs and the LCMV functions,
# but here we use raw.pick_types() to save memory.
left_temporal_channels = mne.read_selection('Left-temporal')
raw.pick_types(meg='mag', eeg=False, eog=False, stim=False, exclude='bads',
               selection=left_temporal_channels)
reject = dict(mag=4e-12)
# Re-normalize our empty-room projectors, which should be fine after
# subselection
raw.info.normalize_proj()

# Setting time limits for reading epochs. Note that tmin and tmax are set so
# that time-frequency beamforming will be performed for a wider range of time
# points than will later be displayed on the final spectrogram. This ensures
# that all time bins displayed represent an average of an equal number of time
# windows.
tmin, tmax = -0.55, 0.75  # s
tmin_plot, tmax_plot = -0.3, 0.5  # s

# Read epochs. Note that preload is set to False to enable tf_lcmv to read the
# underlying raw object.
# Filtering is then performed on raw data in tf_lcmv and the epochs
# parameters passed here are used to create epochs from filtered data. However,
# reading epochs without preloading means that bad epoch rejection is delayed
# until later. To perform bad epoch rejection based on the reject parameter
# passed here, run epochs.drop_bad(). This is done automatically in
# tf_lcmv to reject bad epochs based on unfiltered data.
event_id = 1
events = mne.read_events(event_fname)
epochs = mne.Epochs(raw, events, event_id, tmin, tmax, proj=True,
                    baseline=None, preload=False, reject=reject)

# Read empty room noise, preload to allow filtering, and pick subselection
raw_noise = mne.io.read_raw_fif(noise_fname, preload=True)
raw_noise.info['bads'] = ['MEG 2443']  # 1 bad MEG channel
raw_noise.pick_types(meg='mag', eeg=False, eog=False, stim=False,
                     exclude='bads', selection=left_temporal_channels)
raw_noise.info.normalize_proj()

# Create artificial events for empty room noise data
events_noise = make_fixed_length_events(raw_noise, event_id, duration=1.)
# Create an epochs object using preload=True to reject bad epochs based on
# unfiltered data
epochs_noise = mne.Epochs(raw_noise, events_noise, event_id, tmin, tmax,
                          proj=True, baseline=None,
                          preload=True, reject=reject)

# Make sure the number of noise epochs is the same as data epochs
epochs_noise = epochs_noise[:len(epochs.events)]

# Read forward operator
forward = mne.read_forward_solution(fname_fwd, surf_ori=True)

# Read label
label = mne.read_label(fname_label)

Out:

Opening raw data file /home/ubuntu/mne_data/MNE-sample-data/MEG/sample/sample_audvis_raw.fif...
    Read a total of 3 projection items:
        PCA-v1 (1 x 102)  idle
        PCA-v2 (1 x 102)  idle
        PCA-v3 (1 x 102)  idle
    Range : 25800 ... 192599 =     42.956 ...   320.670 secs
Ready.
Current compensation grade : 0
Reading 0 ... 166799  =      0.000 ...   277.714 secs...
72 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Opening raw data file /home/ubuntu/mne_data/MNE-sample-data/MEG/sample/ernoise_raw.fif...
Isotrak not found
    Read a total of 3 projection items:
        PCA-v1 (1 x 102)  idle
        PCA-v2 (1 x 102)  idle
        PCA-v3 (1 x 102)  idle
    Range : 19800 ... 85867 =     32.966 ...   142.965 secs
Ready.
Current compensation grade : 0
Reading 0 ... 66067  =      0.000 ...   109.999 secs...
110 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Loading data for 110 events and 781 original time points ...
1 bad epochs dropped
Reading forward solution from /home/ubuntu/mne_data/MNE-sample-data/MEG/sample/sample_audvis-meg-eeg-oct-6-fwd.fif...
    Reading a source space...
    Computing patch statistics...
    Patch information added...
    Distance information added...
    [done]
    Reading a source space...
    Computing patch statistics...
    Patch information added...
    Distance information added...
    [done]
    2 source spaces read
    Desired named matrix (kind = 3523) not available
    Read MEG forward solution (7498 sources, 306 channels, free orientations)
    Desired named matrix (kind = 3523) not available
    Read EEG forward solution (7498 sources, 60 channels, free orientations)
    MEG and EEG forward solutions combined
    Source spaces transformed to the forward solution coordinate frame
    Converting to surface-based source orientations...
    Average patch normals will be employed in the rotation to the local surface coordinates....
[done]

Time-frequency beamforming based on LCMV

# Setting frequency bins as in Dalal et al. 2008 (high gamma was subdivided)
freq_bins = [(4, 12), (12, 30), (30, 55), (65, 299)]  # Hz
win_lengths = [0.3, 0.2, 0.15, 0.1]  # s

# Setting the time step
tstep = 0.05

# Setting the whitened data covariance regularization parameter
data_reg = 0.001

# Subtract evoked response prior to computation?
subtract_evoked = False

# Calculating covariance from empty room noise. To use baseline data as noise
# substitute raw for raw_noise, epochs.events for epochs_noise.events, tmin for
# desired baseline length, and 0 for tmax_plot.
# Note, if using baseline data, the averaged evoked response in the baseline
# period should be flat.
noise_covs = []
for (l_freq, h_freq) in freq_bins:
    raw_band = raw_noise.copy()
    raw_band.filter(l_freq, h_freq, n_jobs=1)
    epochs_band = mne.Epochs(raw_band, epochs_noise.events, event_id,
                             tmin=tmin_plot, tmax=tmax_plot, baseline=None,
                             proj=True)

    noise_cov = compute_covariance(epochs_band, method='shrunk')
    noise_covs.append(noise_cov)
    del raw_band  # to save memory

# Computing LCMV solutions for time-frequency windows in a label in source
# space for faster computation, use label=None for full solution
stcs = tf_lcmv(epochs, forward, noise_covs, tmin, tmax, tstep, win_lengths,
               freq_bins=freq_bins, subtract_evoked=subtract_evoked,
               reg=data_reg, label=label)

# Plotting source spectrogram for source with maximum activity.
# Note that tmin and tmax are set to display a time range that is smaller than
# the one for which beamforming estimates were calculated. This ensures that
# all time bins shown are a result of smoothing across an identical number of
# time windows.
plot_source_spectrogram(stcs, freq_bins, tmin=tmin_plot, tmax=tmax_plot,
                        source_index=None, colorbar=True)
../../_images/sphx_glr_plot_tf_lcmv_001.png

Out:

Setting up band-pass filter from 4 - 12 Hz
l_trans_bandwidth chosen to be 2.0 Hz
h_trans_bandwidth chosen to be 3.0 Hz
Filter length of 1982 samples (3.300 sec) selected
72 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Loading data for 72 events and 481 original time points ...
0 bad epochs dropped
Estimating covariance using SHRUNK
Done.
Using cross-validation to select the best estimator.
Number of samples used : 34632
[done]
log-likelihood on unseen data (descending order):
   shrunk: -35.028
selecting best estimator: shrunk
Setting up band-pass filter from 12 - 30 Hz
l_trans_bandwidth chosen to be 3.0 Hz
h_trans_bandwidth chosen to be 7.5 Hz
Filter length of 1321 samples (2.199 sec) selected
72 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Loading data for 72 events and 481 original time points ...
0 bad epochs dropped
Estimating covariance using SHRUNK
Done.
Using cross-validation to select the best estimator.
Number of samples used : 34632
[done]
log-likelihood on unseen data (descending order):
   shrunk: -38.827
selecting best estimator: shrunk
Setting up band-pass filter from 30 - 55 Hz
l_trans_bandwidth chosen to be 7.5 Hz
h_trans_bandwidth chosen to be 13.8 Hz
Filter length of 529 samples (0.881 sec) selected
72 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Loading data for 72 events and 481 original time points ...
0 bad epochs dropped
Estimating covariance using SHRUNK
Done.
Using cross-validation to select the best estimator.
Number of samples used : 34632
[done]
log-likelihood on unseen data (descending order):
   shrunk: -39.998
selecting best estimator: shrunk
Setting up band-pass filter from 65 - 3e+02 Hz
l_trans_bandwidth chosen to be 16.2 Hz
h_trans_bandwidth chosen to be 1.3 Hz
Filter length of 3032 samples (5.048 sec) selected
72 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Loading data for 72 events and 481 original time points ...
0 bad epochs dropped
Estimating covariance using SHRUNK
Done.
Using cross-validation to select the best estimator.
Number of samples used : 34632
[done]
log-likelihood on unseen data (descending order):
   shrunk: -46.416
selecting best estimator: shrunk
Loading data for 72 events and 781 original time points ...
0 bad epochs dropped
Setting up band-pass filter from 4 - 12 Hz
72 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Loading data for 72 events and 781 original time points ...
0 bad epochs dropped
Computing time-frequency LCMV beamformer for time window -550 to -250 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.244
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -500 to -200 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.144
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -450 to -150 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.026
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -400 to -100 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.947
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -350 to -50 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.763
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -300 to 0 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.751
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -250 to 49 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.974
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -200 to 99 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.214
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -150 to 149 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.540
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -100 to 199 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.812
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -50 to 249 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -111.077
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 0 to 300 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -111.140
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 50 to 350 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -111.043
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 99 to 399 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.806
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 150 to 450 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.429
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 199 to 499 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.000
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 250 to 550 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.818
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 300 to 600 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.714
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 350 to 649 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.836
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 400 to 700 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.008
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 449 to 750 ms, in frequency range 4 to 12 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 13032
[done]
log-likelihood on unseen data (descending order):
   empirical: -110.190
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Setting up band-pass filter from 12 - 30 Hz
72 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Loading data for 72 events and 781 original time points ...
0 bad epochs dropped
Computing time-frequency LCMV beamformer for time window -550 to -350 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.808
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -500 to -300 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.679
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -450 to -250 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.375
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -400 to -200 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.106
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -350 to -150 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.065
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -300 to -100 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.155
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -250 to -49 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.339
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -200 to 0 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.334
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -150 to 49 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.192
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -100 to 99 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.436
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -50 to 149 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.563
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 0 to 200 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.714
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 50 to 250 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.673
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 99 to 300 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.360
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 150 to 350 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.227
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 199 to 399 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.276
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 250 to 450 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.464
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 300 to 500 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.607
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 350 to 550 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.616
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 400 to 600 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.600
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 449 to 649 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.379
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 500 to 700 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.566
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 550 to 750 ms, in frequency range 12 to 30 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 8712
[done]
log-likelihood on unseen data (descending order):
   empirical: -109.801
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Setting up band-pass filter from 30 - 55 Hz
72 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Loading data for 72 events and 781 original time points ...
0 bad epochs dropped
Computing time-frequency LCMV beamformer for time window -550 to -400 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.461
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -500 to -350 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.516
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -450 to -300 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.452
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -400 to -250 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.221
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -350 to -200 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.103
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -300 to -150 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.174
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -250 to -100 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.388
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -200 to -50 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.363
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -150 to 0 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.296
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -100 to 49 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.193
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -50 to 99 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.357
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 0 to 150 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.545
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 50 to 200 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.427
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 99 to 249 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.391
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 150 to 300 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.354
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 199 to 350 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.308
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 250 to 400 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.190
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 300 to 450 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.145
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 350 to 500 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.140
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 400 to 550 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.190
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 449 to 600 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.120
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 500 to 650 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.129
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 550 to 700 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.234
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 600 to 750 ms, in frequency range 30 to 55 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 6552
[done]
log-likelihood on unseen data (descending order):
   empirical: -106.309
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Setting up band-pass filter from 65 - 3e+02 Hz
72 matching events found
Created an SSP operator (subspace dimension = 3)
3 projection items activated
Loading data for 72 events and 781 original time points ...
0 bad epochs dropped
Computing time-frequency LCMV beamformer for time window -550 to -450 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.443
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -500 to -400 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.483
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -450 to -350 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.342
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -400 to -300 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.207
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -350 to -250 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.290
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -300 to -200 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.216
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -250 to -150 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.235
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -200 to -100 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.308
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -150 to -50 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.337
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -100 to 0 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.508
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window -50 to 49 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.434
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 0 to 100 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.415
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 50 to 150 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.406
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 99 to 199 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.308
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 150 to 250 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.256
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 199 to 299 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.267
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 250 to 350 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.341
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 300 to 400 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.310
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 350 to 449 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.279
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 400 to 500 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.332
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 449 to 549 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.438
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 500 to 600 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.409
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 550 to 650 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.307
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 600 to 700 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.227
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Computing time-frequency LCMV beamformer for time window 650 to 750 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.212
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]
Adding a time window to cover last time points
Computing time-frequency LCMV beamformer for time window 650 to 750 ms, in frequency range 65 to 299 Hz
Estimating covariance using EMPIRICAL
Done.
Using cross-validation to select the best estimator.
Number of samples used : 4392
[done]
log-likelihood on unseen data (descending order):
   empirical: -112.212
selecting best estimator: empirical
    13 out of 366 channels remain after picking
    Created an SSP operator (subspace dimension = 3)
estimated rank (mag): 10
Setting small MEG eigenvalues to zero.
Not doing PCA for MEG.
[done]

Total running time of the script: ( 0 minutes 24.253 seconds)

Generated by Sphinx-Gallery