def create_report_tensors(state, report_id):
T = state.result.T
vname = "" #TODO
caption = '%s learned tensors, %d iterations' % \
(vname, state.total_iterations)
A = create_report_figure_tensors(T, same_scale=False, report_id='tensors',
caption=caption)
B = create_report_figure_tensors(T, same_scale=True, report_id='tensors-ss',
caption='%s (on the same scale)' % caption)
report = Node(report_id)
report.add_child(A)
report.add_child(B)
return report
def main():
"""
Prepare data with:
average_logs_results --dir . --experiment laser_bgds_boot
"""
print "Loading first..."
results = my_pickle_load('out/average_logs_results/laser_bgds_boot.pickle')
report = Node('laser_bgds_boot')
manual = my_figures(results)
write_figures_for_paper(manual)
report.add_child(manual)
k = 1
variants = sorted(results.keys())
# FIXME: this was uncommented (Was I in a hurry?)
# variants = []
for variant in variants:
data = results[variant]
G = data['G']
B = data['B']
# gy_mean = data['gy_mean']
# y_mean = data['y_mean']
# one_over_y_mean = data['one_over_y_mean']
y_dot_var = data['y_dot_var']
y_dot_svar = data['y_dot_svar']
gy_var = data['gy_var']
gy_svar = data['gy_svar']
print 'Considering %s' % variant
n = report.node(variant)
#readings = range(360, G.shape[0])
readings = range(0, G.shape[0])
N = len(readings)
G = G[readings, :]
B = B[readings, :]
y_dot_var = y_dot_var[readings]
y_dot_svar = y_dot_svar[readings]
gy_var = gy_var[readings]
gy_svar = gy_svar[readings]
for k in [0, 2]:
var = 'G[%s]' % k
G_k = n.data(var, G[:, k])
G_k_n = G[:, k] / gy_var
with G_k.data_pylab('original') as pylab:
pylab.plot(G[:, k])
pylab.title(var + ' original')
M = abs(G[:, k]).max()
pylab.axis([0, N, -M, M])
with G_k.data_pylab('normalized') as pylab:
pylab.plot(G_k_n)
pylab.title(var + ' normalized')
M = abs(G_k_n).max()
pylab.axis([0, N, -M, M])
for k in [0, 2]:
var = 'B[%s]' % k
B_k = n.data(var, B[:, k])
B_k_n = B[:, k] / gy_var
with B_k.data_pylab('original') as pylab:
pylab.plot(B[:, k])
pylab.title(var + ' original')
M = abs(B[:, k]).max()
pylab.axis([0, N, -M, M])
with B_k.data_pylab('normalized') as pylab:
pylab.plot(B_k_n)
pylab.title(var + ' normalized')
M = abs(B_k_n).max()
pylab.axis([0, N, -M, M])
n.figure('obtained', sub=['B[0]/normalized', 'B[2]/normalized',
'G[0]/normalized', 'G[2]/normalized', ])
n.figure('G', sub=['G[0]/original', 'G[0]/normalized',
'G[2]/original', 'G[2]/normalized'])
n.figure('B', sub=['B[0]/original', 'B[0]/normalized',
'B[2]/original', 'B[2]/normalized'])
theta = linspace(0, 2 * math.pi, N) - math.pi / 2
with n.data_pylab('B_v') as pylab:
pylab.plot(-cos(theta))
pylab.axis([0, N, -2, 2])
with n.data_pylab('B_omega') as pylab:
pylab.plot(0 * theta)
pylab.axis([0, N, -2, 2])
with n.data_pylab('G_v') as pylab:
pylab.plot(-sin(theta))
pylab.axis([0, N, -2, 2])
with n.data_pylab('G_omega') as pylab:
pylab.plot(numpy.ones((N)))
pylab.axis([0, N, -2, 2])
n.figure('expected', sub=['B_v', 'B_omega', 'G_v', 'G_omega'])
norm = lambda x: x / x.max()
# norm = lambda x : x
with n.data_pylab('y_dot_var') as pylab:
pylab.plot(norm(y_dot_var), 'b', label='y_dot_var')
pylab.plot(norm(y_dot_svar ** 2), 'g', label='y_dot_svar')
pylab.axis([0, N, 0, 1])
pylab.legend()
with n.data_pylab('gy_var') as pylab:
pylab.plot(norm(gy_var) , 'b', label='gy_var')
pylab.plot(norm(gy_svar ** 2), 'g', label='gy_svar')
pylab.axis([0, N, 0, 1])
pylab.legend()
f = n.figure('stats')
f.sub('y_dot_var')
f.sub('gy_var')
for var in ['y_dot_mean', 'gy_mean', 'y_mean', 'y_var', 'one_over_y_mean']:
with n.data_pylab(var) as pylab:
pylab.plot(data[var][readings])
f.sub(var)
with n.data_pylab('y_mean+var') as pylab:
y = data['y_mean'][readings]
var = data['y_var'][readings]
pylab.errorbar(range(0, N), y, yerr=3 * numpy.sqrt(var), fmt='ro')
f.sub('y_mean+var')
print variant
if True: #variant == 'GS_DS':
s = {'variant': variant,
'Gl':G[:, 0] / gy_var,
'Ga':G[:, 2] / gy_var,
'Bl':B[:, 0] / gy_var,
'Ba':B[:, 2] / gy_var }
filename = "out/laser_bgds_boot/%s:GB.pickle" % variant.replace('/', '_')
make_sure_dir_exists(filename)
my_pickle_dump(s, filename)
print 'Written on %s' % filename
node_to_html_document(report, 'out/laser_bgds_boot/report.html')
def analyze_olfaction_covariance(covariance, receptors):
''' Covariance: n x n covariance matrix.
Positions: list of n positions '''
positions = [pose.get_2d_position() for pose, sens in receptors] #@UnusedVariable
positions = array(positions).transpose().squeeze()
require_shape(square_shape(), covariance)
n = covariance.shape[0]
require_shape((2, n), positions)
distances = create_distance_matrix(positions)
correlation = cov2corr(covariance)
flat_distances = distances.reshape(n * n)
flat_correlation = correlation.reshape(n * n)
# let's fit a polynomial
deg = 4
poly = polyfit(flat_distances, flat_correlation, deg=deg)
knots = linspace(min(flat_distances), max(flat_distances), 2000)
poly_int = polyval(poly, knots)
poly_fder = polyder(poly)
fder = polyval(poly_fder, distances)
Ttheta = create_olfaction_Ttheta(positions, fder)
Tx, Ty = create_olfaction_Txy(positions, fder, distances)
report = Node('olfaction-theory')
report.data('flat_distances', flat_distances)
report.data('flat_correlation', flat_correlation)
with report.data_file('dist_vs_corr', 'image/png') as filename:
pylab.figure()
pylab.plot(flat_distances, flat_correlation, '.')
pylab.plot(knots, poly_int, 'r-')
pylab.xlabel('distance')
pylab.ylabel('correlation')
pylab.title('Correlation vs distance')
pylab.legend(['data', 'interpolation deg = %s' % deg])
pylab.savefig(filename)
pylab.close()
with report.data('fder', fder).data_file('fder', 'image/png') as filename:
pylab.figure()
pylab.plot(knots, polyval(poly_fder, knots), 'r-')
pylab.title('f der')
pylab.savefig(filename)
pylab.close()
report.data('distances', distances)
report.data('correlation', correlation)
report.data('covariance', covariance)
report.data('f', polyval(poly, distances))
f = report.figure(id='cor-vs-distnace', caption='Estimated kernels',
shape=(3, 3))
f.sub('dist_vs_corr')
f.sub('fder')
f.sub('f', display='scale')
f.sub('distances', display='scale')
f.sub('correlation', display='posneg')
f.sub('covariance', display='posneg')
T = numpy.zeros(shape=(3, Tx.shape[0], Tx.shape[1]))
T[0, :, :] = Tx
T[1, :, :] = Ty
T[2, :, :] = Ttheta
T_report = create_report_figure_tensors(T, report_id='tensors',
caption="Predicted learned tensors")
report.add_child(T_report)
return report
