def angles_between_discrete(angles):
theta, phi = zip(*angles)
distances = angle_between(0., 0., np.array(theta), np.array(phi))
counts, bins = np.histogram(distances, bins=np.linspace(0, np.pi, 721))
plotd = Plot()
plotd.histogram(counts, np.degrees(bins))
# plotd.set_title('Distance between reconstructed angles for station and cluster')
plotd.set_xlabel('Angle between reconstructions [\si{\degree}]')
plotd.set_ylabel('Counts')
plotd.set_xlimits(min=0, max=90)
plotd.set_ylimits(min=0)
plotd.save_as_pdf('angle_between_Zenith_discrete')
plotd = Plot()
distances = []
for t, p in angles:
distances.extend(angle_between(t, p, np.array(theta), np.array(phi)))
counts, bins = np.histogram(distances, bins=np.linspace(0, np.pi, 361))
plotd.histogram(counts, np.degrees(bins))
# plotd.set_title('Distance between reconstructed angles for station and cluster')
plotd.set_xlabel('Angle between reconstructions [\si{\degree}]')
plotd.set_ylabel('Counts')
plotd.set_xlimits(min=0, max=90)
plotd.set_ylimits(min=0)
plotd.save_as_pdf('angle_between_Zenith_discrete_all')
def test_azimuths(self):
"""Both directions at the horizon"""
zenith = pi / 2
azimuth = random.uniform(-pi, pi, 10000)
angle = utils.angle_between(zenith, azimuth, zenith, 0)
self.assertTrue(all(abs(angle - abs(azimuth)) < 1e-10))
angle = utils.angle_between(zenith, 0, zenith, azimuth)
self.assertTrue(all(abs(angle - abs(azimuth)) < 1e-10))
def test_zeniths(self):
"""One of the directions is the Zenith"""
n = 10000
zenith = random.uniform(0, pi / 2, n)
azimuth1 = random.uniform(-pi, pi, n)
azimuth2 = random.uniform(-pi, pi, n)
angle = utils.angle_between(zenith, azimuth1, 0, azimuth2)
self.assertTrue(all(abs(angle - zenith) < 1e-15))
angle = utils.angle_between(0, azimuth1, zenith, azimuth2)
self.assertTrue(all(abs(angle - zenith) < 1e-15))
def test_no_zenith(self):
"""Azimuths are irrelevant when from the Zenith"""
azimuth1 = random.uniform(-pi, pi, 10000)
azimuth2 = random.uniform(-pi, pi, 10000)
angle = utils.angle_between(0, azimuth1, 0, azimuth2)
self.assertTrue(all(angle == 0))
def test_single_values(self):
"""Other tests use arrays, check if single values also work"""
zenith = random.uniform(0, pi / 2)
azimuth = random.uniform(-pi, pi)
angle = utils.angle_between(zenith, azimuth, zenith, azimuth)
self.assertTrue(angle == 0)
def analyse():
"""Plot results from reconstructions compared to the simulated input"""
with tables.open_file(DATAPATH, 'r') as data:
coincidences = data.get_node('/coincidences/coincidences')
reconstructions = data.get_node(STATION_PATH)
assert coincidences.nrows == reconstructions.nrows
zenith_in = coincidences.col('zenith')
azimuth_in = coincidences.col('azimuth')
zenith_re = reconstructions.col('zenith')
azimuth_re = reconstructions.col('azimuth')
d_angle = angle_between(zenith_in, azimuth_in, zenith_re, azimuth_re)
print sum(isnan(d_angle))
plot = Plot()
counts, bins = histogram(d_angle[invert(isnan(d_angle))], bins=50)
plot.histogram(counts, bins)
plot.set_ylimits(min=0)
plot.set_xlabel(r'Angle between \si{\radian}')
plot.set_ylabel('Counts')
plot.save_as_pdf('angle_between')
plot = Plot()
counts, bins = histogram(zenith_in, bins=50)
plot.histogram(counts, bins, linestyle='red')
counts, bins = histogram(zenith_re, bins=bins)
plot.histogram(counts, bins)
plot.set_ylimits(min=0)
plot.set_xlimits(min=0)
plot.set_xlabel(r'Zenith \si{\radian}')
plot.set_ylabel('Counts')
plot.save_as_pdf('zenith')
plot = Plot()
counts, bins = histogram(azimuth_in, bins=50)
plot.histogram(counts, bins, linestyle='red')
counts, bins = histogram(azimuth_re, bins=bins)
plot.histogram(counts, bins)
plot.set_ylimits(min=0)
plot.set_xlabel(r'Azimuth \si{\radian}')
plot.set_ylabel('Counts')
plot.save_as_pdf('azimuth')
unique_coordinates = list({(z, a) for z, a in zip(zenith_re, azimuth_re)})
zenith_uni, azimuth_uni = zip(*unique_coordinates)
plot = PolarPlot(use_radians=True)
plot.scatter(array(azimuth_uni),
array(zenith_uni),
markstyle='mark size=.75pt')
plot.set_xlabel(r'Azimuth \si{\radian}')
plot.set_ylabel(r'Zenith \si{\radian}')
plot.save_as_pdf('polar')
def plot_reconstructions():
print 'Plotting . . .'
plot = Plot()
bins = linspace(0, 90, 30) # Degrees
plot.set_ylimits(min=0)
plot.set_xlimits(0, 90)
plot.set_ylabel('counts')
plot.set_xlabel(r'Angle between [\si{\degree}]')
colors = ['black', 'red', 'green', 'blue']
for i, c_group in enumerate([
'/coincidences', '/coincidences_original',
'/coincidences_501_original', '/coincidences_510_original'
]):
cq = CoincidenceQuery(DATA, coincidence_group=c_group)
coincidences = cq.all([501, 510], iterator=True)
reconstructions = [cq._get_reconstructions(c) for c in coincidences]
cq.finish()
azi501 = []
zen501 = []
azi510 = []
zen510 = []
for rec1, rec2 in reconstructions:
if rec1[0] == 501:
azi501.append(rec1[1]['azimuth'])
zen501.append(rec1[1]['zenith'])
azi510.append(rec2[1]['azimuth'])
zen510.append(rec2[1]['zenith'])
else:
azi501.append(rec2[1]['azimuth'])
zen501.append(rec2[1]['zenith'])
azi510.append(rec1[1]['azimuth'])
zen510.append(rec1[1]['zenith'])
azi501 = array(azi501)
zen501 = array(zen501)
azi510 = array(azi510)
zen510 = array(zen510)
# Compare angles between old and new
d_angle = angle_between(zen501, azi501, zen510, azi510)
print c_group, r'67\%% within %.1f degrees' % degrees(
percentile(d_angle[isfinite(d_angle)], 67))
plot.histogram(*histogram(degrees(d_angle), bins=bins),
linestyle=colors[i])
plot.save_as_pdf('angle_between_501_510')
def plot_results(data):
rec_paths = ['recs_flat', 'recs_curved', 'recs_xy']
linestyles = ['solid', 'dotted', 'dashed']
plot = Plot()
for i, rec_path in enumerate(rec_paths):
recs = data.get_node('/coincidences/%s' % rec_path)
angles = angle_between(recs.col('zenith'), recs.col('azimuth'),
recs.col('reference_zenith'),
recs.col('reference_azimuth'))
dangles = np.degrees(angles)
counts, bins = np.histogram(dangles, bins=np.arange(0, 25, .5))
plot.histogram(counts, bins + (i / 10.), linestyle=linestyles[i])
plot.set_xlimits(0, 25)
plot.set_ylimits(0)
plot.set_xlabel(r'Angle between [\si{\degree}]')
plot.save_as_pdf('angle_between')
def plot_reconstructions():
with tables.open_file('data.h5', 'r') as data:
rec = data.root.s501.reconstructions
reco = data.root.s501_original.reconstructions
# Compare azimuth distribution
bins = linspace(-pi, pi, 20) # Radians
plot = Plot()
plot.histogram(*histogram(rec.col('azimuth'), bins=bins))
plot.histogram(*histogram(reco.col('azimuth'), bins=bins),
linestyle='red')
plot.set_ylimits(min=0)
plot.set_xlimits(-pi, pi)
plot.set_ylabel('counts')
plot.set_xlabel(r'Azimuth [\si{\radian}]')
plot.save_as_pdf('azimuth')
# Compare zenith distribution
bins = linspace(0, pi / 2, 20) # Radians
plot = Plot()
plot.histogram(*histogram(rec.col('zenith'), bins=bins))
plot.histogram(*histogram(reco.col('zenith'), bins=bins),
linestyle='red')
plot.set_ylimits(min=0)
plot.set_xlimits(0, pi / 2)
plot.set_ylabel('counts')
plot.set_xlabel(r'Zenith [\si{\radian}]')
plot.save_as_pdf('zenith')
# Compare angles between old and new
bins = linspace(0, 20, 20) # Degrees
plot = Plot()
filter = (rec.col('zenith') > .5)
d_angle = angle_between(rec.col('zenith'), rec.col('azimuth'),
reco.col('zenith'), reco.col('azimuth'))
plot.histogram(*histogram(degrees(d_angle), bins=bins))
plot.histogram(*histogram(degrees(d_angle).compress(filter),
bins=bins),
linestyle='red')
plot.histogram(*histogram(degrees(d_angle).compress(invert(filter)),
bins=bins),
linestyle='blue')
plot.set_ylimits(min=0)
plot.set_xlimits(0, 20)
plot.set_ylabel('counts')
plot.set_xlabel(r'Angle between [\si{\degree}]')
plot.save_as_pdf('angle_between')
def plot_reconstruction_accuracy(data, d):
station_path = '/cluster_simulations/station_%d'
cluster = cluster_501_510()
coincidences = data.root.coincidences.coincidences
recs501 = data.root.hisparc.cluster_amsterdam.station_501.reconstructions
recs510 = data.root.hisparc.cluster_amsterdam.station_510.reconstructions
graph = Plot()
ids = set(recs501.col('id')).intersection(recs510.col('id'))
filtered_501 = [(row['zenith'], row['azimuth']) for row in recs501
if row['id'] in ids]
filtered_510 = [(row['zenith'], row['azimuth']) for row in recs510
if row['id'] in ids]
zen501, azi501 = zip(*filtered_501)
zen510, azi510 = zip(*filtered_510)
zen501 = array(zen501)
azi501 = array(azi501)
zen510 = array(zen510)
azi510 = array(azi510)
da = angle_between(zen501, azi501, zen510, azi510)
n, bins = histogram(da, bins=arange(0, pi, .1))
graph.histogram(n, bins)
failed = coincidences.nrows - len(ids)
graph.set_ylimits(min=0)
graph.set_xlimits(min=0, max=pi)
graph.set_ylabel('Count')
graph.set_xlabel('Angle between 501 and 510 [rad]')
graph.set_title('Coincidences between 501 and 510')
graph.set_label('Failed to reconstruct %d events' % failed)
graph.save_as_pdf('coincidences_%s' % d)
graph_recs = PolarPlot()
azimuth = degrees(recs501.col('azimuth'))
zenith = degrees(recs501.col('zenith'))
graph_recs.scatter(azimuth[:5000],
zenith[:5000],
mark='*',
markstyle='mark size=.2pt')
graph_recs.set_ylimits(min=0, max=90)
graph_recs.set_ylabel('Zenith [degrees]')
graph_recs.set_xlabel('Azimuth [degrees]')
graph_recs.set_title('Reconstructions by 501')
graph_recs.save_as_pdf('reconstructions_%s' % d)
def analyse_reconstructions(data):
cq = CoincidenceQuery(data)
c_ids = data.root.coincidences.coincidences.read_where('s501', field='id')
c_recs = cq.reconstructions.read_coordinates(c_ids)
s_recs = data.root.hisparc.cluster_amsterdam.station_501.reconstructions
zenc = c_recs['zenith']
azic = c_recs['azimuth']
zens = s_recs.col('zenith')
azis = s_recs.col('azimuth')
high_zenith = (zenc > .2) & (zens > .2)
for minn in [1, 2, 4, 8, 16]:
filter = (s_recs.col('min_n') > minn)
length = len(azis.compress(high_zenith & filter))
shifts501 = np.random.normal(0, .06, length)
azicounts, x, y = np.histogram2d(azis.compress(high_zenith & filter) +
shifts501,
azic.compress(high_zenith & filter),
bins=np.linspace(-np.pi, np.pi, 73))
plota = Plot()
plota.histogram2d(azicounts,
np.degrees(x),
np.degrees(y),
type='reverse_bw',
bitmap=True)
# plota.set_title('Reconstructed azimuths for events in coincidence (zenith gt .2 rad)')
plota.set_xlabel(r'$\phi_{501}$ [\si{\degree}]')
plota.set_ylabel(r'$\phi_{Science Park}$ [\si{\degree}]')
plota.set_xticks([-180, -90, 0, 90, 180])
plota.set_yticks([-180, -90, 0, 90, 180])
plota.save_as_pdf('azimuth_501_spa_minn%d' % minn)
length = sum(filter)
shifts501 = np.random.normal(0, .04, length)
zencounts, x, y = np.histogram2d(zens.compress(filter) + shifts501,
zenc.compress(filter),
bins=np.linspace(0, np.pi / 3., 41))
plotz = Plot()
plotz.histogram2d(zencounts,
np.degrees(x),
np.degrees(y),
type='reverse_bw',
bitmap=True)
# plotz.set_title('Reconstructed zeniths for station events in coincidence')
plotz.set_xlabel(r'$\theta_{501}$ [\si{\degree}]')
plotz.set_ylabel(r'$\theta_{Science Park}$ [\si{\degree}]')
plotz.set_xticks([0, 15, 30, 45, 60])
plotz.set_yticks([0, 15, 30, 45, 60])
plotz.save_as_pdf('zenith_501_spa_minn%d' % minn)
distances = angle_between(zens.compress(filter), azis.compress(filter),
zenc.compress(filter), azic.compress(filter))
counts, bins = np.histogram(distances, bins=np.linspace(0, np.pi, 91))
plotd = Plot()
plotd.histogram(counts, np.degrees(bins))
sigma = np.degrees(np.percentile(distances[np.isfinite(distances)],
67))
plotd.set_label(r'67\%% within \SI{%.1f}{\degree}' % sigma)
# plotd.set_title('Distance between reconstructed angles for station and cluster')
plotd.set_xlabel('Angle between reconstructions [\si{\degree}]')
plotd.set_ylabel('Counts')
plotd.set_xlimits(min=0, max=90)
plotd.set_ylimits(min=0)
plotd.save_as_pdf('angle_between_501_spa_minn%d' % minn)
def plot_reconstruction_accuracy():
combinations = ['~d1 | ~d2 | ~d3 | ~d4', 'd1 & d2 & d3 & d4']
station_path = '/cluster_simulations/station_%d'
with tables.open_file(RESULT_PATH, 'r') as data:
cluster = data.root.coincidences._v_attrs.cluster
coincidences = data.root.coincidences.coincidences
c_recs = data.root.coincidences.reconstructions
graph = Plot()
da = angle_between(c_recs.col('zenith'), c_recs.col('azimuth'),
c_recs.col('reference_zenith'),
c_recs.col('reference_azimuth'))
ids = c_recs.col('id')
N = coincidences.read_coordinates(ids, field='N')
for k, filter in enumerate([N == 3, N > 3]):
n, bins = histogram(da.compress(filter), bins=arange(0, pi, .1))
graph.histogram(n, bins, linestyle=GRAYS[k % len(GRAYS)])
failed = len(coincidences.get_where_list('N >= 3')) - c_recs.nrows
graph.set_ylimits(min=0)
graph.set_xlimits(min=0, max=pi)
graph.set_ylabel('Count')
graph.set_xlabel('Angle between input and reconstruction [rad]')
graph.set_title('Coincidences')
graph.set_label('Failed to reconstruct %d events' % failed)
graph.save_as_pdf('coincidences_alt')
for station in cluster.stations:
station_group = data.get_node(station_path % station.number)
recs = station_group.reconstructions
rows = coincidences.get_where_list('s%d == True' % station.number)
reference_azimuth = coincidences.read_coordinates(rows,
field='azimuth')
reference_zenith = coincidences.read_coordinates(rows,
field='zenith')
graph = Plot()
for k, combo in enumerate(combinations):
selected_reconstructions = recs.read_where(combo)
filtered_azimuth = array([
reference_azimuth[i]
for i in selected_reconstructions['id']
])
filtered_zenith = array([
reference_zenith[i] for i in selected_reconstructions['id']
])
azimuth = selected_reconstructions['azimuth']
zenith = selected_reconstructions['zenith']
da = angle_between(zenith, azimuth, filtered_zenith,
filtered_azimuth)
n, bins = histogram(da, bins=arange(0, pi, .1))
graph.histogram(n, bins, linestyle=GRAYS[k % len(GRAYS)])
failed = station_group.events.nrows - recs.nrows
graph.set_ylimits(min=0)
graph.set_xlimits(min=0, max=pi)
graph.set_ylabel('Count')
graph.set_xlabel('Angle between input and reconstruction [rad]')
graph.set_title('Station: %d' % station.number)
graph.set_label('Failed to reconstruct %d events' % failed)
graph.save_as_pdf('s_%d' % station.number)
def analyse_reconstructions(path):
seed = os.path.basename(os.path.dirname(path))
cq = CoincidenceQuery(path)
c_ids = cq.coincidences.get_where_list('N >= 3')
if not len(cq.reconstructions) or not len(c_ids):
cq.finish()
return
c_recs = cq.reconstructions.read_coordinates(c_ids)
# Angles
zen_out = c_recs['zenith']
azi_out = c_recs['azimuth']
zen_in = c_recs['reference_zenith']
azi_in = c_recs['reference_azimuth']
# Cores
x_out = c_recs['x']
y_out = c_recs['y']
x_in = c_recs['reference_x']
y_in = c_recs['reference_y']
# Size
size_out = c_recs['size']
energy_out = c_recs['energy']
size_in = c_recs['reference_size']
energy_in = c_recs['reference_energy']
energy = np.log10(energy_in[0])
zenith = np.degrees(zen_in[0])
label = r'$E=10^{%d}$eV, $\theta={%.1f}^{\circ}$' % (energy, zenith)
# Azimuth
bins = np.linspace(-np.pi, np.pi, 21)
acounts_out, bins = np.histogram(azi_out, bins)
acounts_in, bins = np.histogram(azi_in, bins)
plota = Plot()
plota.histogram(acounts_out, bins)
plota.histogram(acounts_in, bins, linestyle='red')
plota.set_xlabel(r'$\phi$ [\si{\radian}]')
plota.set_xlimits(-np.pi, np.pi)
plota.set_ylabel(r'Counts')
plota.set_ylimits(min=0)
plota.save_as_pdf('plots/azimuth_in_out_%s' % seed)
# Zenith
bins = np.linspace(0, np.pi / 2, 21)
zcounts_out, bins = np.histogram(zen_out, bins)
zcounts_in, bins = np.histogram(zen_in, bins)
plotz = Plot()
plotz.histogram(zcounts_out, bins)
plotz.histogram(zcounts_in, bins, linestyle='red')
plotz.set_xlabel(r'$\theta$ [\si{\radian}]')
plotz.set_xlimits(0, np.pi / 2)
plotz.set_ylabel(r'Counts')
plotz.set_ylimits(min=0)
plotz.save_as_pdf('plots/zenith_in_out_%s' % seed)
# Angle between
angle_distances = angle_between(zen_out, azi_out, zen_in, azi_in)
if len(np.isfinite(angle_distances)):
bins = np.linspace(0, np.pi / 2, 91)
counts, bins = np.histogram(angle_distances, bins=bins)
plotd = Plot()
plotd.histogram(counts, np.degrees(bins))
sigma = np.percentile(angle_distances[np.isfinite(angle_distances)],
67)
plotd.set_label(label + r', 67\%% within \SI{%.1f}{\degree}' %
np.degrees(sigma))
plotd.set_xlabel(r'Angle between reconstructions [\si{\degree}]')
plotd.set_ylabel('Counts')
plotd.set_xlimits(np.degrees(bins[0]), np.degrees(bins[-1]))
plotd.set_ylimits(min=0)
plotd.save_as_pdf('plots/angle_between_in_out_%s' % seed)
# Distance beween
filter = size_out != 1e6
core_distances = np.sqrt(
(x_out.compress(filter) - x_in.compress(filter))**2 +
(y_out.compress(filter) - y_in.compress(filter))**2)
if len(np.isfinite(core_distances)):
bins = np.linspace(0, 1000, 100)
counts, bins = np.histogram(core_distances, bins=bins)
plotc = Plot()
plotc.histogram(counts, bins)
sigma = np.percentile(core_distances[np.isfinite(core_distances)], 67)
#
energy = np.log10(energy_in[0])
zenith = np.degrees(zen_in[0])
#
plotc.set_label(r'$E=10^{%d}$eV, $\theta={%.1f}^{\circ}$, 67\%% '
'within \SI{%.1f}{\meter}' % (energy, zenith, sigma))
plotc.set_xlabel(r'Distance between cores [\si{\meter}]')
plotc.set_ylabel('Counts')
plotc.set_xlimits(bins[0], bins[-1])
plotc.set_ylimits(min=0)
plotc.save_as_pdf('plots/core_distance_between_in_out_%s' % seed)
# Core positions
filter = size_out != 1e6
plotc = Plot()
for x0, x1, y0, y1 in zip(x_out, x_in, y_out, y_in):
plotc.plot([x0, x1], [y0, y1])
plotc.set_xlabel(r'x [\si{\meter}]')
plotc.set_ylabel(r'y [\si{\meter}]')
plotc.set_xlimits(bins[0], bins[-1])
plotc.set_ylimits(min=0)
plotc.save_as_pdf('plots/core_positions_in_out_%s' % seed)
# Shower size
relative_size = size_out.compress(filter) / size_in.compress(filter)
counts, bins = np.histogram(relative_size, bins=np.logspace(-2, 2, 21))
plots = Plot('semilogx')
plots.histogram(counts, bins)
plots.set_xlabel('Relative size')
plots.set_ylabel('Counts')
plots.set_xlimits(bins[0], bins[-1])
plots.set_ylimits(min=0)
plots.save_as_pdf('plots/size_in_out_%s' % seed)
# Cleanup
cq.finish()
def oldvsnew_diagram():
"""
Visual accuracy comparisons of old and new transformations.
Compares the correlations between the transformations:
equatorial_to_horizontal and equatorial_to_zenith_azimuth_astropy
horizontal_to_equatorial and horizontal_to_zenith_azimuth_astropy
Makes a histogram of the error differences
between these two functions as well.
The errors seem to be in the order of 1000 arcsec
:return: None
Ethan van Woerkom is responsible for the benchmarking functions;
refer to him for when something is unclear
"""
# make random frames, in correct angle range and from utc time 2000-2020
frames = []
# boxes for the four different transformation results
etoha = []
etoh = []
htoe = []
htoea = []
straight = lambda x : x # straight trendline function
# Create the data sets for eq to az
for i in range(100):
frames.append((r.uniform(-90, 90),
r.uniform(-180,180),
r.randint(946684800,1577836800),
r.uniform(0, 2 * np.pi),
r.uniform(-0.5 * np.pi, 0.5 * np.pi)))
for i in frames:
etoha.append(celestial.equatorial_to_zenithazimuth_astropy(i[0], i[1], i[2], [(i[3], i[4])])[0])
etoh.append(celestial.equatorial_to_zenithazimuth(i[0], i[1], clock.utc_to_gps(i[2]), i[3], i[4]))
# Data sets for hor to eq
for i in frames:
htoe.append(celestial.horizontal_to_equatorial(i[0],
clock.utc_to_lst(datetime.datetime.utcfromtimestamp(i[2]), i[1]), i[4], i[3]))
htoea.extend(celestial.horizontal_to_equatorial_astropy(i[0], i[1], i[2], [(i[3], i[4])]))
# Make figs eq -> zenaz
plt.figure(1)
plt.suptitle('Zen/Az correlation in rads (equatorial_to_zenithazimuth)')
zenrange = [0, np.pi]
plt.subplot(211)
plt.title('Zenith')
plt.axis(zenrange*2)
plt.xlabel('New (Astropy)')
plt.ylabel('Old')
# Make figure and add 1:1 trendline
plt.plot([co[0] for co in etoha], [co[0] for co in etoh], 'r.', zenrange, straight(zenrange), '-')
plt.subplot(212)
plt.title('Azimuth')
azrange = [-np.pi, np.pi]
plt.axis(azrange*2)
plt.xlabel('New (Astropy)')
plt.ylabel('Old')
# Make figure and add 1:1 trendline
plt.plot([co[1] for co in etoha], [co[1] for co in etoh], 'b.', azrange, straight(azrange), '-')
plt.tight_layout() # Prevent titles merging
plt.subplots_adjust(top=0.85)
# Make histogram of differences
plt.figure(2)
# Take diff. and convert to arcsec
nieuw = (np.array(etoh) - np.array(etoha))
nieuw *= 360 * 3600 / (2 * np.pi)
plt.hist([i[0] for i in nieuw], bins=20)
plt.title('Zenith Old-New Error (equatorial_to_zenithazimuth)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
plt.figure(3)
plt.hist([i[1] for i in nieuw], bins=20)
plt.title('Azimuth Old-New Error (equatorial_to_zenithazimuth)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
# Make histogram of differences using the absolute distance in arcsec
# this graph has no wrapping issues
plt.figure(7)
nieuw = np.array([angle_between(etoh[i][0], etoh[i][1], etoha[i][0], etoha[i][1])
for i in range(len(etoh))])
nieuw *= 360 * 3600 / (2 * np.pi)
plt.hist(nieuw, bins=20)
plt.title('ZEN+AZ Old-New Error (equatorial_to_zenithazimuth)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
# Make figs hor - > eq
plt.figure(4)
plt.suptitle('RA/DEC correlation in rads (horizontal_to_equatorial)')
altrange = [-0.5 * np.pi, 0.5 * np.pi]
plt.subplot(211)
plt.title('Declination')
plt.axis(altrange * 2)
plt.xlabel('New (Astropy)')
plt.ylabel('Old')
# Make figure and add 1:1 trendline
plt.plot([co[1] for co in htoea], [co[1] for co in htoe], 'r.', altrange, straight(altrange), '-')
plt.subplot(212)
plt.title('Right Ascension')
azrange = [0, 2 * np.pi]
plt.axis(azrange * 2)
plt.xlabel('New (Astropy)')
plt.ylabel('Old')
# Make figure and add 1:1 trendline
plt.plot([co[0] for co in htoea], [co[0] for co in htoe], 'b.', azrange, straight(azrange), '-')
plt.tight_layout() # Prevent titles merging
plt.subplots_adjust(top=0.85)
# Make histogram of differences
plt.figure(5)
# Take diff. and convert to arcsec
nieuw = (np.array(htoe) - np.array(htoea))
nieuw *= 360 * 3600 / (2 * np.pi)
plt.hist([i[1] for i in nieuw], bins=20)
plt.title('Declination Old-New Error (horizontal_to_equatorial)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
plt.figure(6)
# Take diff. and convert to arcsec
nieuw = (np.array(htoe) - np.array(htoea))
nieuw *= 360 * 3600 / (2 * np.pi)
plt.hist([i[0] for i in nieuw], bins=20)
plt.title('Right Ascension Old-New Error (horizontal_to_equatorial)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
# Make histogram of differences using the absolute distance in arcsec
# this graph has no wrapping issues
plt.figure(8)
nieuw = np.array([angle_between_horizontal(htoe[i][0], htoe[i][1], htoea[i][0], htoea[i][1])
for i in range(len(htoe))])
# Take diff. and convert to arcsec
nieuw /= 2 / np.pi * 360 * 3600
plt.hist(nieuw, bins=20)
plt.title('RA+DEC Old-New Error (horizontal_to_equatorial)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
plt.show()
return
def oldvsnew_diagram():
"""
Visual accuracy comparisons of old and new transformations.
Compares the correlations between the transformations:
equatorial_to_horizontal and equatorial_to_zenith_azimuth_astropy
horizontal_to_equatorial and horizontal_to_zenith_azimuth_astropy
Makes a histogram of the error differences
between these two functions as well.
The errors seem to be in the order of 1000 arcsec
:return: None
Ethan van Woerkom is responsible for the benchmarking functions;
refer to him for when something is unclear
"""
# make random frames, in correct angle range and from utc time 2000-2020
frames = []
# boxes for the four different transformation results
etoha = []
etoh = []
htoe = []
htoea = []
straight = lambda x: x # straight trendline function
# Create the data sets for eq to az
for i in range(100):
frames.append(
(r.uniform(-90,
90), r.uniform(-180,
180), r.randint(946684800, 1577836800),
r.uniform(0, 2 * np.pi), r.uniform(-0.5 * np.pi, 0.5 * np.pi)))
for i in frames:
etoha.append(
celestial.equatorial_to_zenithazimuth_astropy(
i[0], i[1], i[2], [(i[3], i[4])])[0])
etoh.append(
celestial.equatorial_to_zenithazimuth(i[0], i[1],
clock.utc_to_gps(i[2]), i[3],
i[4]))
# Data sets for hor to eq
for i in frames:
htoe.append(
celestial.horizontal_to_equatorial(
i[0],
clock.utc_to_lst(datetime.datetime.utcfromtimestamp(i[2]),
i[1]), i[4], i[3]))
htoea.extend(
celestial.horizontal_to_equatorial_astropy(i[0], i[1], i[2],
[(i[3], i[4])]))
# Make figs eq -> zenaz
plt.figure(1)
plt.suptitle('Zen/Az correlation in rads (equatorial_to_zenithazimuth)')
zenrange = [0, np.pi]
plt.subplot(211)
plt.title('Zenith')
plt.axis(zenrange * 2)
plt.xlabel('New (Astropy)')
plt.ylabel('Old')
# Make figure and add 1:1 trendline
plt.plot([co[0] for co in etoha], [co[0] for co in etoh], 'r.', zenrange,
straight(zenrange), '-')
plt.subplot(212)
plt.title('Azimuth')
azrange = [-np.pi, np.pi]
plt.axis(azrange * 2)
plt.xlabel('New (Astropy)')
plt.ylabel('Old')
# Make figure and add 1:1 trendline
plt.plot([co[1] for co in etoha], [co[1] for co in etoh], 'b.', azrange,
straight(azrange), '-')
plt.tight_layout() # Prevent titles merging
plt.subplots_adjust(top=0.85)
# Make histogram of differences
plt.figure(2)
# Take diff. and convert to arcsec
nieuw = (np.array(etoh) - np.array(etoha))
nieuw *= 360 * 3600 / (2 * np.pi)
plt.hist([i[0] for i in nieuw], bins=20)
plt.title('Zenith Old-New Error (equatorial_to_zenithazimuth)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
plt.figure(3)
plt.hist([i[1] for i in nieuw], bins=20)
plt.title('Azimuth Old-New Error (equatorial_to_zenithazimuth)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
# Make histogram of differences using the absolute distance in arcsec
# this graph has no wrapping issues
plt.figure(7)
nieuw = np.array([
angle_between(etoh[i][0], etoh[i][1], etoha[i][0], etoha[i][1])
for i in range(len(etoh))
])
nieuw *= 360 * 3600 / (2 * np.pi)
plt.hist(nieuw, bins=20)
plt.title('ZEN+AZ Old-New Error (equatorial_to_zenithazimuth)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
# Make figs hor - > eq
plt.figure(4)
plt.suptitle('RA/DEC correlation in rads (horizontal_to_equatorial)')
altrange = [-0.5 * np.pi, 0.5 * np.pi]
plt.subplot(211)
plt.title('Declination')
plt.axis(altrange * 2)
plt.xlabel('New (Astropy)')
plt.ylabel('Old')
# Make figure and add 1:1 trendline
plt.plot([co[1] for co in htoea], [co[1] for co in htoe], 'r.', altrange,
straight(altrange), '-')
plt.subplot(212)
plt.title('Right Ascension')
azrange = [0, 2 * np.pi]
plt.axis(azrange * 2)
plt.xlabel('New (Astropy)')
plt.ylabel('Old')
# Make figure and add 1:1 trendline
plt.plot([co[0] for co in htoea], [co[0] for co in htoe], 'b.', azrange,
straight(azrange), '-')
plt.tight_layout() # Prevent titles merging
plt.subplots_adjust(top=0.85)
# Make histogram of differences
plt.figure(5)
# Take diff. and convert to arcsec
nieuw = (np.array(htoe) - np.array(htoea))
nieuw *= 360 * 3600 / (2 * np.pi)
plt.hist([i[1] for i in nieuw], bins=20)
plt.title('Declination Old-New Error (horizontal_to_equatorial)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
plt.figure(6)
# Take diff. and convert to arcsec
nieuw = (np.array(htoe) - np.array(htoea))
nieuw *= 360 * 3600 / (2 * np.pi)
plt.hist([i[0] for i in nieuw], bins=20)
plt.title('Right Ascension Old-New Error (horizontal_to_equatorial)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
# Make histogram of differences using the absolute distance in arcsec
# this graph has no wrapping issues
plt.figure(8)
nieuw = np.array([
angle_between_horizontal(htoe[i][0], htoe[i][1], htoea[i][0],
htoea[i][1]) for i in range(len(htoe))
])
# Take diff. and convert to arcsec
nieuw = nieuw / 2 / np.pi * 360 * 3600
plt.hist(nieuw, bins=20)
plt.title('RA+DEC Old-New Error (horizontal_to_equatorial)')
plt.xlabel('Error (arcsec)')
plt.ylabel('Counts')
plt.show()
return
def plot_angles(data):
"""Make azimuth and zenith plots to compare the results"""
rec501 = data.get_node('/hisparc/cluster_amsterdam/station_501',
'reconstructions')
rec510 = data.get_node('/hisparc/cluster_amsterdam/station_510',
'reconstructions')
zen501 = rec501.col('zenith')
zen510 = rec510.col('zenith')
azi501 = rec501.col('azimuth')
azi510 = rec510.col('azimuth')
minn501 = rec501.col('min_n')
minn510 = rec510.col('min_n')
sigmas = []
blas = []
minns = [0, 1, 2, 4, 8, 16, 24]
high_zenith = (zen501 > .2) & (zen510 > .2)
for minn in minns:
filter = (minn501 > minn) & (minn510 > minn)
length = len(azi501.compress(high_zenith & filter))
shifts501 = np.random.normal(0, .06, length)
shifts510 = np.random.normal(0, .06, length)
azicounts, x, y = np.histogram2d(
azi501.compress(high_zenith & filter) + shifts501,
azi510.compress(high_zenith & filter) + shifts510,
bins=np.linspace(-pi, pi, 73))
plota = Plot()
plota.histogram2d(azicounts,
degrees(x),
degrees(y),
type='reverse_bw',
bitmap=True)
# plota.set_title('Reconstructed azimuths for events in coincidence (zenith gt .2 rad)')
plota.set_xlabel(r'$\phi_{501}$ [\si{\degree}]')
plota.set_ylabel(r'$\phi_{510}$ [\si{\degree}]')
plota.set_xticks([-180, -90, 0, 90, 180])
plota.set_yticks([-180, -90, 0, 90, 180])
plota.save_as_pdf('azimuth_501_510_minn%d' % minn)
length = len(zen501.compress(filter))
shifts501 = np.random.normal(0, .04, length)
shifts510 = np.random.normal(0, .04, length)
zencounts, x, y = np.histogram2d(zen501.compress(filter) + shifts501,
zen510.compress(filter) + shifts510,
bins=np.linspace(0, pi / 3., 41))
plotz = Plot()
plotz.histogram2d(zencounts,
degrees(x),
degrees(y),
type='reverse_bw',
bitmap=True)
# plotz.set_title('Reconstructed zeniths for station events in coincidence')
plotz.set_xlabel(r'$\theta_{501}$ [\si{\degree}]')
plotz.set_ylabel(r'$\theta_{510}$ [\si{\degree}]')
plotz.set_xticks([0, 15, 30, 45, 60])
plotz.set_yticks([0, 15, 30, 45, 60])
plotz.save_as_pdf('zenith_501_510_minn%d' % minn)
distances = angle_between(zen501.compress(filter),
azi501.compress(filter),
zen510.compress(filter),
azi510.compress(filter))
counts, bins = np.histogram(distances, bins=linspace(0, pi, 100))
plotd = Plot()
plotd.histogram(counts, degrees(bins))
sigma = degrees(percentile(distances[isfinite(distances)], 68))
sigmas.append(sigma)
bla = degrees(percentile(distances[isfinite(distances)], 95))
blas.append(bla)
plotd.set_label(r'67\%% within \SI{%.1f}{\degree}' % sigma)
# plotd.set_title('Distance between reconstructed angles for station events')
plotd.set_xlabel(r'Angle between reconstructions [\si{\degree}]')
plotd.set_ylabel('Counts')
plotd.set_xlimits(min=0, max=90)
plotd.set_ylimits(min=0)
plotd.save_as_pdf('angle_between_501_510_minn%d' % minn)
plot = Plot()
plot.plot(minns, sigmas, mark='*')
plot.plot(minns, blas)
plot.set_ylimits(min=0, max=40)
plot.set_xlabel('Minimum number of particles in each station')
plot.set_ylabel(r'Angle between reconstructions [\si{\degree}]')
plot.save_as_pdf('angle_between_501_510_v_minn')
def plot_angles(data):
"""Make azimuth and zenith plots to compare the results"""
rec = data.root.reconstructions
zenhis = rec.col('reconstructed_theta')
zenkas = rec.col('reference_theta')
azihis = rec.col('reconstructed_phi')
azikas = rec.col('reference_phi')
min_n = rec.col('min_n134')
high_zenith = (zenhis > .2) & (zenkas > .2)
for minn in [0, 1, 2, 4]:
filter = (min_n > minn)
azicounts, x, y = np.histogram2d(azihis.compress(high_zenith & filter),
azikas.compress(high_zenith & filter),
bins=np.linspace(-pi, pi, 73))
plota = Plot()
plota.histogram2d(azicounts,
degrees(x),
degrees(y),
type='reverse_bw',
bitmap=True)
# plota.set_title('Reconstructed azimuths for events in coincidence (zenith gt .2 rad)')
plota.set_xlabel(r'$\phi_\textrm{HiSPARC}$ [\si{\degree}]')
plota.set_ylabel(r'$\phi_\textrm{KASCADE}$ [\si{\degree}]')
plota.set_xticks([-180, -90, 0, 90, 180])
plota.set_yticks([-180, -90, 0, 90, 180])
plota.save_as_pdf('azimuth_his_kas_minn%d' % minn)
zencounts, x, y = np.histogram2d(zenhis.compress(filter),
zenkas.compress(filter),
bins=np.linspace(0, pi / 3., 41))
plotz = Plot()
plotz.histogram2d(zencounts,
degrees(x),
degrees(y),
type='reverse_bw',
bitmap=True)
# plotz.set_title('Reconstructed zeniths for station events in coincidence')
plotz.set_xlabel(r'$\theta_\textrm{HiSPARC}$ [\si{\degree}]')
plotz.set_ylabel(r'$\theta_\textrm{KASCADE}$ [\si{\degree}]')
plotz.set_xticks([0, 15, 30, 45, 60])
plotz.set_yticks([0, 15, 30, 45, 60])
plotz.save_as_pdf('zenith_his_kas_minn%d' % minn)
distances = angle_between(zenhis.compress(filter),
azihis.compress(filter),
zenkas.compress(filter),
azikas.compress(filter))
counts, bins = np.histogram(distances, bins=linspace(0, pi, 100))
plotd = Plot()
plotd.histogram(counts, degrees(bins))
sigma = degrees(scoreatpercentile(distances, 67))
plotd.set_title(r'$N_\textrm{MIP} \geq %d$' % minn)
plotd.set_label(r'67\%% within \SI{%.1f}{\degree}' % sigma)
# plotd.set_title('Distance between reconstructed angles for station events')
plotd.set_xlabel('Angle between reconstructions [\si{\degree}]')
plotd.set_ylabel('Counts')
plotd.set_xlimits(min=0, max=90)
plotd.set_ylimits(min=0)
plotd.save_as_pdf('angle_between_his_kas_minn%d' % minn)