def test_dpss(self):
"""
Tests case for dpss. The resulting v tapers are tested. There is
no test for theta or lambda.
"""
v = np.load
datafile = os.path.join(os.path.dirname(__file__), 'data', 'dpss.npz')
v = np.load(datafile)['v']
v2, lamb, theta = dpss(512, 2.5, 2)
# No NaNs are supposed to be in the output.
self.assertEqual(np.isnan(v2).any(), False)
self.assertEqual(np.isnan(lamb).any(), False)
self.assertEqual(np.isnan(theta).any(), False)
# Taper 1, normalize for precision
np.testing.assert_almost_equal(v2[:,0] / v[:,0], v[:,0] / v[:,0])
# Taper 2, normalize for precision
np.testing.assert_almost_equal(v2[:,1] / v[:,1], v[:,1] / v[:,1])
# Do the same but with spline interpolation.
v3, lamb2, thetha2 = dpss(512, 2.5, 2, nmax=400)
# Test both tapers. They are not exactly equal therefore only two
# digits are compared.
np.testing.assert_almost_equal(v3 / v3, v2 / v3, 2)
def test_dpss(self):
"""
Tests case for dpss. The resulting v tapers are tested. There is
no test for theta or lambda.
"""
v = np.load
datafile = os.path.join(os.path.dirname(__file__), 'data', 'dpss.npz')
v = np.load(datafile)['v']
v2, lamb, theta = dpss(512, 2.5, 2)
# No NaNs are supposed to be in the output.
self.assertEqual(np.isnan(v2).any(), False)
self.assertEqual(np.isnan(lamb).any(), False)
self.assertEqual(np.isnan(theta).any(), False)
# Taper 1, normalize for precision
np.testing.assert_almost_equal(v2[:, 0] / v[:, 0], v[:, 0] / v[:, 0])
# Taper 2, normalize for precision
np.testing.assert_almost_equal(v2[:, 1] / v[:, 1], v[:, 1] / v[:, 1])
# Do the same but with spline interpolation.
v3, lamb2, thetha2 = dpss(512, 2.5, 2, nmax=400)
# Test both tapers. They are not exactly equal therefore only two
# digits are compared.
np.testing.assert_almost_equal(v3 / v3, v2 / v3, 2)
def multitaper_2D(EEG,
npts=170,
fw=3,
number_of_tapers=5,
fs=100,
filename=None):
tapers, _, _ = mtspec.dpss(npts=20, fw=3, number_of_tapers=5)
tf = np.stack([
np.mean(np.power(
np.abs([
signal.stft(EEG[j, :, i],
fs=fs,
window=tapers[:, t],
nperseg=tapers.shape[0])[2]
for t in range(tapers.shape[1])
]), 2),
axis=0) for i in range(EEG.shape[2])
] for j in tqdm(range(EEG.shape[0])))
print("MultiTaper shape: ", tf.shape)
if filename is None:
np.save('tf_2D' + str(fs), tf)
print("Saved as:", 'tf_2D' + str(fs))
else:
np.save(filename, tf)
print("Saved as:", filename)
import matplotlib as mpl
mpl.rcParams['font.size'] = 9.0
import matplotlib.pyplot as plt
from mtspec import dpss
tapers, lamb, theta = dpss(512, 2.5, 10)
ax = plt.figure().add_subplot(111)
for i in xrange(10):
ax.plot(tapers[:, i])
ax.set_xlim(0, len(tapers[:, 0]))
import matplotlib.pyplot as plt
plt.style.use("ggplot")
from mtspec import dpss
tapers, _, _ = dpss(npts=512, fw=2.5, number_of_tapers=8)
for i in range(8):
plt.plot(tapers[:, i])
plt.xlim(0, len(tapers[:, 0]))
plt.tight_layout()
plt.show()
import seaborn
import matplotlib.pyplot as plt
from mtspec import dpss
import os
tapers, lamb, theta = dpss(512, 2.5, 10)
output_folder = "_static"
plt.figure(figsize=(8, 4))
for i in range(5):
plt.plot(tapers[:, i], lw=10, alpha=0.4)
plt.xlim(0, len(tapers[:, 0]))
plt.ylim(-0.08, 0.08)
plt.axis('off')
plt.text(x=0, y=1, s="mtspec", transform=plt.gca().transAxes, va="top",
fontsize=130, fontweight=1000, color="0.3", variant="small-caps")
plt.subplots_adjust(left=0.02, right=0.98, bottom=0.02, top=0.98)
plt.savefig(os.path.join(output_folder, "logo.png"), transparent=True)
plt.savefig(os.path.join(output_folder, "logo.svg"), transparent=True)
plt.savefig(os.path.join(output_folder, "logo.pdf"), transparent=True)
