larsoner

#!/usr/bin/env bash
# Before/after demo for plotly/plotly.py#5701 (sphinx-gallery scraper rewrite).
# Creates ./sgdemo and ./sgdemo-venv, builds the same one-example gallery with
# released plotly and with the PR branch, and prints what to compare.
set -euo pipefail

mkdir -p sgdemo/examples
# a rerun must not reuse sphinx-gallery's cache of executed examples
rm -rf sgdemo/auto_examples sgdemo/_build_before sgdemo/_build_after

larsoner

% FieldTrip reference runs for MNE-Python's mne/stats/tests/test_cluster_equiv.py
% (see mne-tools/mne-python#12663).
%
% The input file cluster_equiv_data.mat is produced by the scenario builders in
% test_cluster_equiv.py (exported with scipy.io.savemat); each scenario holds
% `data` (n_rows x n_chan x n_times), per-row factor labels, and (for
% within-subject designs) 1-based subject indices. This script runs the
% matching FieldTrip cluster-based permutation test for each scenario with
% ft_timelockstatistics (cfg.method = 'montecarlo', cfg.correctm = 'cluster',
% cfg.clusterstatistic = 'maxsum') and saves/prints the observed stat map and

larsoner

"""Minimal reproducer: OpenBLAS 0.3.20 DSYMM is wrong on its Cooperlake kernel.

C = B @ A with A symmetric, column-major, via
    cblas_dsymm(CblasColMajor, CblasRight, CblasUpper,
                m, n, 1.0, A(n x n), n, B(m x n), m, 0.0, C, m)
compared against numpy. (This is the call OpenMEEG makes in
Matrix::operator*(const SymMatrix&).)

Needs only numpy and a libopenblas. To exercise a kernel your CPU does not
have, use Intel SDE -- `-cpx` sets CPUID to Cooper Lake -- together with

larsoner

% Create fsaverage_1020

headshape = ft_read_headshape('~/mne_data/MNE-sample-data/subjects/fsaverage/bem/outer_skin.surf');

% [pos, tri, magic] = read_surf('~/mne_data/MNE-sample-data/subjects/fsaverage/surf/lh.seghead');
% headshape = [];
% headshape.pos = pos;
% headshape.tri = tri;
% ft_plot_mesh(headshape,'facecolor','skin');camlight;

larsoner

larsoner

"""Visual stimulus design.

Tools for drawing shapes and text on the screen.
"""

# Authors: Dan McCloy <drmccloy@uw.edu>
#          Eric Larson <larsoner@uw.edu>
#          Ross Maddox <rkmaddox@uw.edu>
#
# License: BSD (3-clause)

larsoner

import numpy as np
import mne
import time

data_path = mne.datasets.sample.data_path()
subjects_dir = data_path / 'subjects'
n_dips = 10
radius = 0.05
rr = np.random.default_rng(0).normal(size=(n_dips, 3))
rr /= np.linalg.norm(rr, axis=1)[:, np.newaxis] / radius * 0.9

larsoner

import numpy as np

import mne
from mne.datasets import sample
from mne.inverse_sparse import make_stc_from_dipoles, mixed_norm
from mne.minimum_norm import apply_inverse, make_inverse_operator
from mne.viz import (
    plot_dipole_amplitudes,
    plot_dipole_locations,
    plot_sparse_source_estimates,

larsoner

.PHONY: all clean openmeeg_wrap.cxx _example.so test1 test2 test3 test4 test5
all : clean example_wrap.cxx _example.so test1 test2 test3 test4 test5

clean:
	rm -f *.o
	rm -f *.so
	rm -f *wrap*.c*
	rm -f *wrapper.py
	rm -Rf build
	rm -Rf __pycache__

larsoner

import mne
data_path = mne.datasets.sample.data_path()
sample_data_raw_file = data_path / "MEG" / "sample" / "sample_audvis_raw.fif"
subjects_dir = data_path / "subjects"
raw = mne.io.read_raw_fif(sample_data_raw_file, verbose=False).crop(tmax=60)
events = mne.find_events(raw, stim_channel="STI 014")
epochs = mne.Epochs(raw, events, event_id = {"auditory/left": 1}, tmin=-0.3, tmax=0.7)[:10]
inverse_operator_file = data_path / "MEG" / "sample" / "sample_audvis-meg-oct-6-meg-inv.fif"
inv_operator = mne.minimum_norm.read_inverse_operator(inverse_operator_file)
stcs = mne.minimum_norm.apply_inverse_epochs(epochs, inv_operator, lambda2=1/3, pick_ori=None)