def test_complexity_pdf(self):
targ = np.array([0.92, 0.01, 0.01, 0.06])
complexity = es.complexity_pdf(self.spiketrains, binsize=0.1*pq.s)
assert_array_equal(targ, complexity[:, 0].magnitude)
self.assertEqual(1, complexity[:, 0].magnitude.sum())
self.assertEqual(len(self.spiketrains)+1, len(complexity))
self.assertIsInstance(complexity, neo.AnalogSignalArray)
self.assertEqual(complexity.units, 1*pq.dimensionless)
def test_complexity_pdf(self):
targ = np.array([0.92, 0.01, 0.01, 0.06])
complexity = es.complexity_pdf(self.spiketrains, binsize=0.1*pq.s)
assert_array_equal(targ, complexity.magnitude[:, 0])
self.assertEqual(1, complexity.magnitude[:, 0].sum())
self.assertEqual(len(self.spiketrains)+1, len(complexity))
self.assertIsInstance(complexity, neo.AnalogSignal)
self.assertEqual(complexity.units, 1*pq.dimensionless)
def complexity(all_data, color, trial, bin_size, save_figures=True):
"""
Give all_data, color as a string, trial as a scalar, bins as a scalar, default is save_figures = True.
Plots: raster plot, spike counts per bin (in ms), the complexity distribution for that bin size,
complexity distribution for data vs. surrogates, then substracts the two from each other. Finally, three plots
are created of complexity distributions for data, surrogates, and their difference, scanning across bin sizes
from 0 to 30 ms.
"""
# if color=='blue':
# color_map = 'Blues'
# if color=='green':
# color_map = 'Greens'
# if color=='grey':
# color_map = 'Greys'
# if color=='orange':
# color_map = 'Oranges'
# if color=='purple':
# color_map = 'Purples'
# if color=='red':
# color_map = 'Reds'
# block = ut.load_dataset(color)
block = all_data[color]['neo_block']
train = block.segments[trial].spiketrains # segments is trials
binsize = bin_size * pq.ms
pophist = stats.time_histogram(train, binsize, binary=True)
complexity_cpp = stats.complexity_pdf(train, binsize)
# Plot the results
# plt.xkcd()
fig = plt.figure(figsize=(12, 5))
fig.subplots_adjust(top=.92, bottom=.05, hspace=.5, wspace=.2)
misc.add_raster(fig,
2,
2,
1,
train,
ms=1,
xlabel='time',
ylabel='neuron id')
misc.add_hist(fig,
2,
2,
3,
pophist.times,
pophist,
pophist.sampling_period,
xlabel='time',
ylabel='counts')
ylim = plt.gca().get_ylim()
plt.subplot(1, 2, 2)
# plt.bar(complexity_cpp.times, list(A) + [0]*(n-assembly_size), color='r', label='amplitude distrib.')
#plt.bar(range(len(A)), A, color='r', label='amplitude distrib.')
plt.plot(complexity_cpp.times,
complexity_cpp,
label='complexity distrib.',
color=color)
plt.ylim([0, 0.25])
plt.xlim([0, 30])
plt.xlabel('complexity', size=12)
plt.ylabel('probability', size=12)
plt.suptitle('train', size=14)
plt.legend()
if save_figures:
complexity_png = 'ATplots/complexity_bin' + str(
bin_size) + '_color_' + str(color) + '.png'
complexity_pdf = 'ATplots/complexity_bin' + str(
bin_size) + '_color_' + str(color) + '.pdf'
plt.savefig(complexity_png)
plt.savefig(complexity_pdf)
# generation of surrogates
surr_sts = []
for st in train:
surr_sts.append(surr.randomise_spikes(st)[0])
# Computation of the Complexity Distributions
complexity_surr = stats.complexity_pdf(surr_sts, binsize)
diff_complexity = complexity_cpp - complexity_surr
# Plot the difference of the complexity distributions of the correlated and independent CPP
plt.figure(figsize=(12, 4))
plt.subplot(1, 2, 1)
plt.plot(complexity_cpp.times, complexity_cpp, color=color, label="corr'd")
plt.plot(complexity_surr.times,
complexity_surr,
color=color,
label="surrogate",
ls=':')
plt.xlabel('complexity')
plt.xlim([0, 30])
plt.ylabel('probability')
plt.legend()
plt.subplot(1, 2, 2)
plt.plot(complexity_cpp.times, diff_complexity, color=color)
plt.xlabel('complexity')
plt.xlim([0, 30])
plt.ylabel('probability diff.')
if save_figures:
complexity_surr_png = 'ATplots/complexity_surr' + str(
bin_size) + '_color_' + str(color) + '.png'
complexity_surr_pdf = 'ATplots/complexity_surr' + str(
bin_size) + '_color_' + str(color) + '.pdf'
plt.savefig(complexity_surr_png)
plt.savefig(complexity_surr_pdf)
plt.show()
# computation of the complexity distributions of CPP and surrogates for different binsizes
# and storing the result in matrices
binsizes = range(1, 31) * pq.ms
max2 = []
complexity_cpp_matrix = stats.complexity_pdf(train, binsizes[0]).magnitude
complexity_surr_matrix = stats.complexity_pdf(surr_sts,
binsizes[0]).magnitude
diff_complexity_matrix = complexity_cpp_matrix - complexity_surr_matrix
max2.append(
numpy.argmax(diff_complexity.magnitude[numpy.argmin(diff_complexity.
magnitude):]) +
numpy.argmin(diff_complexity.magnitude))
for i, h in enumerate(binsizes[1:]):
complexity_cpp = stats.complexity_pdf(train, h)
complexity_surr = stats.complexity_pdf(surr_sts, h)
diff_complexity = complexity_cpp - complexity_surr
max2.append(
numpy.argmax(diff_complexity.
magnitude[numpy.argmin(diff_complexity.magnitude):]) +
numpy.argmin(diff_complexity.magnitude))
complexity_cpp_matrix = numpy.hstack(
(complexity_cpp_matrix, complexity_cpp.magnitude))
complexity_surr_matrix = numpy.hstack(
(complexity_surr_matrix, complexity_surr.magnitude))
diff_complexity_matrix = numpy.hstack(
(diff_complexity_matrix, diff_complexity.magnitude))
# Plot the complexity matrices
inch2cm = 0.3937
scale = 2 # figure scale; '1' creates a 8x10 cm figure (half a page width)
label_size = 6 + 1 * scale
text_size = 8 + 1 * scale
tick_size = 4 + 1 * scale
plt.figure(figsize=(8 * scale * inch2cm, 12 * scale * inch2cm),
dpi=100 * scale)
plt.subplots_adjust(right=.97,
top=.95,
bottom=.2 - .08 * scale,
hspace=.55)
plt.subplot(3, 1, 1)
plt.title('CPP complexity')
plt.pcolor(complexity_cpp_matrix.T)
plt.colorbar()
plt.tick_params(length=2, direction='out', pad=0)
plt.yticks(binsizes[0:-1:3].magnitude)
plt.ylabel('Binsize')
plt.xlabel('Complexity')
#plt.xlim([0,complexity_cpp_matrix.T.shape[1]])
plt.xlim([0, 30])
plt.ylim([0, complexity_cpp_matrix.T.shape[0]])
plt.ylim([binsizes[0], binsizes[-1]])
plt.subplot(3, 1, 2)
plt.title('Surrogate complexity')
plt.pcolor(complexity_surr_matrix.T)
plt.colorbar()
plt.ylabel('Binsize')
plt.xlabel('Complexity')
plt.tick_params(length=2, direction='out', pad=0)
plt.yticks(binsizes[0:-1:3].magnitude)
#plt.xlim([0,complexity_cpp_matrix.T.shape[1]])
plt.xlim([0, 30])
plt.ylim([0, complexity_cpp_matrix.T.shape[0]])
plt.ylim([binsizes[0], binsizes[-1]])
plt.subplot(3, 1, 3)
plt.title('Difference of complexities')
plt.pcolor(diff_complexity_matrix.T)
plt.colorbar()
plt.plot(max2, binsizes, 'm')
plt.ylabel('Binsize')
plt.xlabel('Complexity')
#plt.yticks(binsizes[0:-1:3].magnitude)
plt.tick_params(length=2, direction='out', pad=0)
#plt.xlim([0,complexity_cpp_matrix.T.shape[1]])
plt.xlim([0, 30])
plt.ylim([0, complexity_cpp_matrix.T.shape[0]])
plt.ylim([binsizes[0], binsizes[-1]])
if save_figures:
sliding_bin_png = 'ATplots/complexity_sliding_bin_' + str(
color) + '.png'
sliding_bin_pdf = 'ATplots/complexity_sliding_bin_' + str(
color) + '.pdf'
plt.savefig(sliding_bin_png)
plt.savefig(sliding_bin_pdf)
plt.show()