def plot_head_signals_tight_multiple_signals(all_signals, sensor_names=None,
figsize=(10, 8), plot_args=None):
assert sensor_names is not None
assert all([len(signals) == len(all_signals[0]) for signals in all_signals])
if plot_args is None:
plot_args = dict()
figure = plt.figure(figsize=figsize)
sensor_positions = [get_sensor_pos(name, tight_C_positions) for name in sensor_names]
sensor_positions = np.array(sensor_positions) # sensors x 2(row and col)
maxima = np.max(sensor_positions, axis=0)
minima = np.min(sensor_positions, axis=0)
max_row = maxima[0, 0]
max_col = maxima[1, 0]
min_row = minima[0, 0]
min_col = minima[1, 0]
rows = max_row - min_row + 1
cols = max_col - min_col + 1
first_ax = None
# gridspec inside gridspec
outer_grid = gridspec.GridSpec(rows, cols, wspace=0.3, hspace=0.5)
for i in xrange(0, len(all_signals[0])):
sensor_name = sensor_names[i]
sensor_pos = sensor_positions[i]
assert np.all(sensor_pos == get_sensor_pos(sensor_name, tight_C_positions))
row = sensor_pos[0]
col = sensor_pos[1]
inner_grid = gridspec.GridSpecFromSubplotSpec(len(all_signals), 1,
subplot_spec=outer_grid[row - min_row, col - min_col], wspace=0.0, hspace=0.0)
for signal_type in xrange(len(all_signals)):
signal = all_signals[signal_type][i]
if first_ax is None:
ax = plt.Subplot(figure, inner_grid[signal_type, 0])
first_ax = ax
else:
ax = plt.Subplot(figure, inner_grid[signal_type, 0], sharey=first_ax, sharex=first_ax)
if signal_type == 0:
ax.set_title(sensor_name, fontsize=10)
ax.plot(signal, **plot_args)
ax.xaxis.grid(True)
# make line at zero
ax.axhline(y=0, ls=':', color="grey")
figure.add_subplot(ax)
if len(signal) == 600:
plt.xticks([150, 300, 450], [])
else:
plt.xticks([])
plt.yticks([])
return figure
def plot_head_signals(signals,
sensor_names=None,
figsize=(12, 7),
plot_args=None):
assert sensor_names is None or len(signals) == len(sensor_names), (
"need "
"sensor names for all sensor matrices")
if sensor_names is None:
sensor_names = map(str, range(len(signals)))
if plot_args is None:
plot_args = dict()
figure = plt.figure(figsize=figsize)
sensor_positions = [get_sensor_pos(name) for name in sensor_names]
sensor_positions = np.array(sensor_positions) # sensors x 2(row and col)
maxima = np.max(sensor_positions, axis=0)
minima = np.min(sensor_positions, axis=0)
max_row = maxima[0]
max_col = maxima[1]
min_row = minima[0]
min_col = minima[1]
rows = max_row - min_row + 1
cols = max_col - min_col + 1
first_ax = None
for i in xrange(0, len(signals)):
sensor_name = sensor_names[i]
sensor_pos = sensor_positions[i]
assert np.all(sensor_pos == get_sensor_pos(sensor_name))
# Transform to flat sensor pos
row = sensor_pos[0]
col = sensor_pos[1]
subplot_ind = (
row - min_row
) * cols + col - min_col + 1 # +1 as matlab uses based indexing
if first_ax is None:
ax = figure.add_subplot(rows, cols, subplot_ind)
first_ax = ax
else:
ax = figure.add_subplot(rows,
cols,
subplot_ind,
sharey=first_ax,
sharex=first_ax)
signal = signals[i]
ax.plot(signal, **plot_args)
ax.set_title(sensor_name)
ax.set_yticks([])
if len(signal) == 600:
ax.set_xticks([150, 300, 450])
ax.set_xticklabels([])
ax.xaxis.grid(True)
# make line at zero
ax.axhline(y=0, ls=':', color="grey")
return figure
def plot_head_signals(signals, sensor_names=None, figsize=(12, 7),
plot_args=None):
assert sensor_names is None or len(signals) == len(sensor_names), ("need "
"sensor names for all sensor matrices")
if sensor_names is None:
sensor_names = map(str, range(len(signals)))
if plot_args is None:
plot_args = dict()
figure = plt.figure(figsize=figsize)
sensor_positions = [get_sensor_pos(name) for name in sensor_names]
sensor_positions = np.array(sensor_positions) # sensors x 2(row and col)
maxima = np.max(sensor_positions, axis=0)
minima = np.min(sensor_positions, axis=0)
max_row = maxima[0]
max_col = maxima[1]
min_row = minima[0]
min_col = minima[1]
rows = max_row - min_row + 1
cols = max_col - min_col + 1
first_ax = None
for i in xrange(0, len(signals)):
sensor_name = sensor_names[i]
sensor_pos = sensor_positions[i]
assert np.all(sensor_pos == get_sensor_pos(sensor_name))
# Transform to flat sensor pos
row = sensor_pos[0]
col = sensor_pos[1]
subplot_ind = (row - min_row) * cols + col - min_col + 1 # +1 as matlab uses based indexing
if first_ax is None:
ax = figure.add_subplot(rows, cols, subplot_ind)
first_ax = ax
else:
ax = figure.add_subplot(rows, cols, subplot_ind, sharey=first_ax,
sharex=first_ax)
signal = signals[i]
ax.plot(signal, **plot_args)
ax.set_title(sensor_name)
ax.set_yticks([])
if len(signal) == 600:
ax.set_xticks([150, 300, 450])
ax.set_xticklabels([])
ax.xaxis.grid(True)
# make line at zero
ax.axhline(y=0, ls=':', color="grey")
return figure
def plot_chan_matrices(matrices, sensor_names, figname='', figure=None,
figsize=(8, 4.5), yticks=(), yticklabels=(),
correctness_matrices=None, colormap=cm.coolwarm,
sensor_map=cap_positions, vmax=None, vmin=None,
share_y_axes=True):
""" figsize ignored if figure given """
assert len(matrices) == len(sensor_names), "need sensor names for all sensor matrices"
if figure is None:
figure = plt.figure(figsize=figsize)
sensor_positions = [get_sensor_pos(name, sensor_map) for name in sensor_names]
sensor_positions = np.array(sensor_positions) # #sensors x 2(row and col) x1(for some reason:))
maxima = np.max(sensor_positions, axis=0)
minima = np.min(sensor_positions, axis=0)
max_row = maxima[0]
max_col = maxima[1]
min_row = minima[0]
min_col = minima[1]
rows = max_row - min_row + 1
cols = max_col - min_col + 1
mean_abs_weight = np.mean(np.abs(matrices))
if (correctness_matrices is not None):
mean_abs_weight = np.mean(np.abs(matrices * correctness_matrices))
if figname != '':
figure.suptitle(figname, fontsize=14) # + ", pixel abs mean(x100): {:.3f}".format(mean_abs_weight * 100),
first_ax = None
for i in xrange(0, len(matrices)):
sensor_name = sensor_names[i]
sensor_pos = sensor_positions[i]
assert np.all(sensor_pos == get_sensor_pos(sensor_name, sensor_map))
# Transform to flat sensor pos
row = sensor_pos[0]
col = sensor_pos[1]
subplot_ind = (row - min_row) * cols + col - min_col + 1 # +1 as matlab uses based indexing
if vmin is None:
vmin = -2 * mean_abs_weight
if vmax is None:
vmax = 2 * mean_abs_weight
if first_ax is None or not share_y_axes:
ax = figure.add_subplot(rows, cols, subplot_ind)
first_ax = ax
elif share_y_axes:
ax = figure.add_subplot(rows, cols, subplot_ind, sharey=first_ax)
chan_matrix = matrices[i]
ax.set_title(sensor_name)
if (correctness_matrices is None):
# ax.pcolor(chan_matrix.T,
# cmap=cm.bwr, #origin='lower', #interpolation='nearest',#aspect='auto'
# vmin=-mean_abs_weight * 2, vmax= mean_abs_weight * 2)
ax.pcolorfast(chan_matrix.T,
cmap=colormap, # origin='lower', #interpolation='nearest',#aspect='auto'
vmin=vmin, vmax=vmax)
# ax.imshow(chan_matrix.T,
# interpolation='nearest', cmap=cm.bwr, origin='lower',
# vmin=-mean_abs_weight * 2, vmax= mean_abs_weight * 2,
# aspect='auto')#"""
else:
# Show correct and incorrect inputs with different colors
# weighted also by degree of correctness
correctness_mat = correctness_matrices[i]
ax.imshow(
np.ma.masked_where(correctness_mat < 0, chan_matrix * correctness_mat).T,
interpolation='nearest', cmap=cm.bwr, origin='lower',
vmin=vmin, vmax=vmax)
# use yellow/brown for red(positive) incorrect values
# (mask out correct or blue=negative values)
# also take minus the values as otherwise they go from
# minus-something to 0, that makes colormaps
# more complicated to use :)
ax.imshow(
np.ma.masked_where(
np.logical_or(correctness_mat > 0, chan_matrix < 0),
- (chan_matrix * correctness_mat)).T,
interpolation='nearest', cmap=cm.YlOrBr, origin='lower',
vmin=0, vmax=vmax)
# use purple for blue(negative) incorrect values
ax.imshow(
np.ma.masked_where(
np.logical_or(correctness_mat > 0, chan_matrix >= 0),
chan_matrix * correctness_mat).T,
interpolation='nearest', cmap=cm.PuRd, origin='lower',
vmin=0, vmax=vmax)
ax.set_xticks([])
ax.set_yticks(yticks)
ax.set_yticklabels(yticklabels)
ax.tick_params(axis='both', which='major', labelsize=6)
ax.grid(color='k', linewidth=0.1, linestyle=':')
return figure
def plot_chan_matrices(matrices,
sensor_names,
figname='',
figure=None,
figsize=(8, 4.5),
yticks=(),
yticklabels=(),
correctness_matrices=None,
colormap=cm.coolwarm,
sensor_map=cap_positions,
vmax=None,
vmin=None,
share_y_axes=True):
""" Chan matrices in [chan x row x col] order.
figsize ignored if figure given """
assert len(matrices) == len(
sensor_names), "need sensor names for all sensor matrices"
if figure is None:
figure = plt.figure(figsize=figsize)
sensor_positions = [
get_sensor_pos(name, sensor_map) for name in sensor_names
]
sensor_positions = np.array(
sensor_positions) # #sensors x 2(row and col) x1(for some reason:))
maxima = np.max(sensor_positions, axis=0)
minima = np.min(sensor_positions, axis=0)
max_row = maxima[0]
max_col = maxima[1]
min_row = minima[0]
min_col = minima[1]
rows = max_row - min_row + 1
cols = max_col - min_col + 1
mean_abs_weight = np.mean(np.abs(matrices))
if (correctness_matrices is not None):
mean_abs_weight = np.mean(np.abs(matrices * correctness_matrices))
if figname != '':
figure.suptitle(
figname, fontsize=14
) # + ", pixel abs mean(x100): {:.3f}".format(mean_abs_weight * 100),
first_ax = None
for i in xrange(0, len(matrices)):
sensor_name = sensor_names[i]
sensor_pos = sensor_positions[i]
assert np.all(sensor_pos == get_sensor_pos(sensor_name, sensor_map))
# Transform to flat sensor pos
row = sensor_pos[0]
col = sensor_pos[1]
subplot_ind = (
row - min_row
) * cols + col - min_col + 1 # +1 as matlab uses based indexing
if vmin is None:
vmin = -2 * mean_abs_weight
if vmax is None:
vmax = 2 * mean_abs_weight
if first_ax is None or not share_y_axes:
ax = figure.add_subplot(rows, cols, subplot_ind)
first_ax = ax
elif share_y_axes:
ax = figure.add_subplot(rows, cols, subplot_ind, sharey=first_ax)
chan_matrix = matrices[i]
ax.set_title(sensor_name)
if (correctness_matrices is None):
# ax.pcolor(chan_matrix.T,
# cmap=cm.bwr, #origin='lower', #interpolation='nearest',#aspect='auto'
# vmin=-mean_abs_weight * 2, vmax= mean_abs_weight * 2)
ax.pcolorfast(
chan_matrix.T,
cmap=
colormap, # origin='lower', #interpolation='nearest',#aspect='auto'
vmin=vmin,
vmax=vmax)
# ax.imshow(chan_matrix.T,
# interpolation='nearest', cmap=cm.bwr, origin='lower',
# vmin=-mean_abs_weight * 2, vmax= mean_abs_weight * 2,
# aspect='auto')#"""
else:
# Show correct and incorrect inputs with different colors
# weighted also by degree of correctness
correctness_mat = correctness_matrices[i]
ax.imshow(np.ma.masked_where(correctness_mat < 0,
chan_matrix * correctness_mat).T,
interpolation='nearest',
cmap=cm.bwr,
origin='lower',
vmin=vmin,
vmax=vmax)
# use yellow/brown for red(positive) incorrect values
# (mask out correct or blue=negative values)
# also take minus the values as otherwise they go from
# minus-something to 0, that makes colormaps
# more complicated to use :)
ax.imshow(np.ma.masked_where(
np.logical_or(correctness_mat > 0, chan_matrix < 0),
-(chan_matrix * correctness_mat)).T,
interpolation='nearest',
cmap=cm.YlOrBr,
origin='lower',
vmin=0,
vmax=vmax)
# use purple for blue(negative) incorrect values
ax.imshow(np.ma.masked_where(
np.logical_or(correctness_mat > 0, chan_matrix >= 0),
chan_matrix * correctness_mat).T,
interpolation='nearest',
cmap=cm.PuRd,
origin='lower',
vmin=0,
vmax=vmax)
ax.set_xticks([])
ax.set_yticks(yticks)
ax.set_yticklabels(yticklabels)
ax.tick_params(axis='both', which='major', labelsize=6)
ax.grid(color='k', linewidth=0.1, linestyle=':')
return figure
def plot_head_signals_tight_multiple_signals(all_signals,
sensor_names=None,
figsize=(10, 8),
plot_args=None):
assert sensor_names is not None
assert all(
[len(signals) == len(all_signals[0]) for signals in all_signals])
if plot_args is None:
plot_args = dict()
figure = plt.figure(figsize=figsize)
sensor_positions = [
get_sensor_pos(name, tight_C_positions) for name in sensor_names
]
sensor_positions = np.array(sensor_positions) # sensors x 2(row and col)
maxima = np.max(sensor_positions, axis=0)
minima = np.min(sensor_positions, axis=0)
max_row = maxima[0, 0]
max_col = maxima[1, 0]
min_row = minima[0, 0]
min_col = minima[1, 0]
rows = max_row - min_row + 1
cols = max_col - min_col + 1
first_ax = None
# gridspec inside gridspec
outer_grid = gridspec.GridSpec(rows, cols, wspace=0.3, hspace=0.5)
for i in xrange(0, len(all_signals[0])):
sensor_name = sensor_names[i]
sensor_pos = sensor_positions[i]
assert np.all(
sensor_pos == get_sensor_pos(sensor_name, tight_C_positions))
row = sensor_pos[0]
col = sensor_pos[1]
inner_grid = gridspec.GridSpecFromSubplotSpec(
len(all_signals),
1,
subplot_spec=outer_grid[row - min_row, col - min_col],
wspace=0.0,
hspace=0.0)
for signal_type in xrange(len(all_signals)):
signal = all_signals[signal_type][i]
if first_ax is None:
ax = plt.Subplot(figure, inner_grid[signal_type, 0])
first_ax = ax
else:
ax = plt.Subplot(figure,
inner_grid[signal_type, 0],
sharey=first_ax,
sharex=first_ax)
if signal_type == 0:
ax.set_title(sensor_name, fontsize=10)
ax.plot(signal, **plot_args)
ax.xaxis.grid(True)
# make line at zero
ax.axhline(y=0, ls=':', color="grey")
figure.add_subplot(ax)
if len(signal) == 600:
plt.xticks([150, 300, 450], [])
else:
plt.xticks([])
plt.yticks([])
return figure