def calculate_muon_parameters(self, tel_id, image, clean_mask, ring):
fov_radius = self.get_fov(tel_id)
x, y = self.get_pixel_coords(tel_id)
# add ring containment, not filled in fit
containment = ring_containment(
ring.radius,
ring.center_x,
ring.center_y,
fov_radius,
)
completeness = ring_completeness(
x,
y,
image,
ring.radius,
ring.center_x,
ring.center_y,
threshold=self.completeness_threshold.tel[tel_id],
)
pixel_width = self.get_pixel_width(tel_id)
intensity_ratio = intensity_ratio_inside_ring(
x[clean_mask],
y[clean_mask],
image[clean_mask],
ring.radius,
ring.center_x,
ring.center_y,
width=self.ratio_width.tel[tel_id] * pixel_width,
)
mse = mean_squared_error(
x[clean_mask],
y[clean_mask],
image[clean_mask],
ring.radius,
ring.center_x,
ring.center_y,
)
return MuonParametersContainer(
containment=containment,
completeness=completeness,
intensity_ratio=intensity_ratio,
mean_squared_error=mse,
)
def analyze_muon_event(subarray, event_id, image, geom,
equivalent_focal_length, mirror_area, plot_rings,
plots_path):
"""
Analyze an event to fit a muon ring
Paramenters
---------
event_id: `int` id of the analyzed event
image: `np.ndarray` number of photoelectrons in each pixel
geom: CameraGeometry
equivalent_focal_length: `float` focal length of the telescope
mirror_area: `float` mirror area of the telescope in square meters
plot_rings: `bool` plot the muon ring
plots_path: `string` path to store the figures
Returns
---------
muonintensityoutput MuonEfficiencyContainer
dist_mask ndarray, pixels used in ring intensity likelihood fit
ring_size float, in p.e. total intensity in ring
size_outside_ring float, in p.e. to check for "shower contamination"
muonringparam MuonParametersContainer
good_ring bool, it determines whether the ring can be used for
analysis or not
radial_distribution dict, return of function radial_light_distribution
mean_pixel_charge_around_ring float, charge "just outside" ring,
to check the possible signal extrator bias
muonparameters MuonParametersContainer
TODO: several hard-coded quantities that can go into a configuration file
"""
lst1_tel_id = 1
lst1_description = subarray.tels[lst1_tel_id]
tailcuts = [10, 5]
cam_rad = (lst1_description.camera.geometry.guess_radius() /
lst1_description.optics.equivalent_focal_length) * u.rad
# some cuts for good ring selection:
min_pix = 148 # (8%) minimum number of pixels in the ring with >0 signal
min_pix_fraction_after_cleaning = 0.1 # minimum fraction of the ring pixels that must be above tailcuts[0]
min_ring_radius = 0.8 * u.deg # minimum ring radius
max_ring_radius = 1.5 * u.deg # maximum ring radius
max_radial_stdev = 0.1 * u.deg # maximum standard deviation of the light distribution along ring radius
max_radial_excess_kurtosis = 1. # maximum excess kurtosis
min_impact_parameter = 0.2 # in fraction of mirror radius
max_impact_parameter = 0.9 # in fraction of mirror radius
ring_integration_width = 0.25 # +/- integration range along ring radius, in fraction of ring radius (was 0.4 until 20200326)
outer_ring_width = 0.2 # in fraction of ring radius, width of ring just outside the integrated muon ring, used to check pedestal bias
x, y = pixel_coords_to_telescope(geom, equivalent_focal_length)
muonringparam, clean_mask, dist, image_clean = fit_muon(
x, y, image, geom, tailcuts)
mirror_radius = np.sqrt(mirror_area / np.pi) # meters
dist_mask = np.abs(dist - muonringparam.radius
) < muonringparam.radius * ring_integration_width
pix_ring = image * dist_mask
pix_outside_ring = image * ~dist_mask
# mask to select pixels just outside the ring that will be integrated to obtain the ring's intensity:
dist_mask_2 = np.logical_and(
~dist_mask,
np.abs(dist - muonringparam.radius) < muonringparam.radius *
(ring_integration_width + outer_ring_width))
pix_ring_2 = image[dist_mask_2]
# nom_dist = np.sqrt(np.power(muonringparam.center_x,2)
# + np.power(muonringparam.center_y, 2))
muonparameters = MuonParametersContainer()
muonparameters.containment = ring_containment(muonringparam.radius,
muonringparam.center_x,
muonringparam.center_y,
cam_rad)
radial_distribution = radial_light_distribution(muonringparam.center_x,
muonringparam.center_y,
x[clean_mask],
y[clean_mask],
image[clean_mask])
# Do complicated calculations (minuit-based max likelihood ring fit) only for selected rings:
candidate_clean_ring = all([
radial_distribution['standard_dev'] < max_radial_stdev,
radial_distribution['excess_kurtosis'] < max_radial_excess_kurtosis,
(pix_ring > tailcuts[0]).sum() >
min_pix_fraction_after_cleaning * min_pix,
np.count_nonzero(pix_ring) > min_pix,
muonringparam.radius < max_ring_radius,
muonringparam.radius > min_ring_radius
])
if candidate_clean_ring:
intensity_fitter = MuonIntensityFitter(subarray)
# Use same hard-coded value for pedestal fluctuations as the previous
# version of ctapipe:
pedestal_stddev = 1.1 * np.ones(len(image))
muonintensityoutput = \
intensity_fitter(1,
muonringparam.center_x,
muonringparam.center_y,
muonringparam.radius,
image,
pedestal_stddev,
dist_mask)
dist_ringwidth_mask = np.abs(dist - muonringparam.radius) < \
muonintensityoutput.width
# We do the calculation of the ring completeness (i.e. fraction of whole circle) using the pixels
# within the "width" fitted using MuonIntensityFitter
muonparameters.completeness = ring_completeness(
x[dist_ringwidth_mask],
y[dist_ringwidth_mask],
image[dist_ringwidth_mask],
muonringparam.radius,
muonringparam.center_x,
muonringparam.center_y,
threshold=30,
bins=30)
# No longer existing in ctapipe 0.8:
# pix_ringwidth_im = image[dist_ringwidth_mask]
# muonintensityoutput.ring_pix_completeness = \
# (pix_ringwidth_im > tailcuts[0]).sum() / len(pix_ringwidth_im)
else:
# just to have the default values with units:
muonintensityoutput = MuonEfficiencyContainer()
muonintensityoutput.width = u.Quantity(np.nan, u.deg)
muonintensityoutput.impact = u.Quantity(np.nan, u.m)
muonintensityoutput.impact_x = u.Quantity(np.nan, u.m)
muonintensityoutput.impact_y = u.Quantity(np.nan, u.m)
# muonintensityoutput.mask = dist_mask # no longer there in ctapipe 0.8
ring_size = np.sum(pix_ring)
size_outside_ring = np.sum(pix_outside_ring * clean_mask)
# This is just mean charge per pixel in pixels just around the ring
# (on the outer side):
mean_pixel_charge_around_ring = np.sum(pix_ring_2) / len(pix_ring_2)
if candidate_clean_ring:
print(
"Impact parameter={:.3f}, ring_width={:.3f}, ring radius={:.3f}, "
"ring completeness={:.3f}".format(
muonintensityoutput.impact,
muonintensityoutput.width,
muonringparam.radius,
muonparameters.completeness,
))
# Now add the conditions based on the detailed muon ring fit:
conditions = [
candidate_clean_ring,
muonintensityoutput.impact < max_impact_parameter * mirror_radius,
muonintensityoutput.impact > min_impact_parameter * mirror_radius,
# TODO: To be applied when we have decent optics.
# muonintensityoutput.width
# < 0.08,
# NOTE: inside "candidate_clean_ring" cuts there is already a cut in
# the std dev of light distribution along ring radius, which is also
# a measure of the ring width
# muonintensityoutput.width
# > 0.04
]
if all(conditions):
good_ring = True
else:
good_ring = False
if (plot_rings and plots_path and good_ring):
focal_length = equivalent_focal_length
ring_telescope = SkyCoord(muonringparam.center_x,
muonringparam.center_y, TelescopeFrame())
ring_camcoord = ring_telescope.transform_to(
CameraFrame(
focal_length=focal_length,
rotation=geom.cam_rotation,
))
centroid = (ring_camcoord.x.value, ring_camcoord.y.value)
radius = muonringparam.radius
width = muonintensityoutput.width
ringrad_camcoord = 2 * radius.to(u.rad) * focal_length
ringwidthfrac = width / radius
ringrad_inner = ringrad_camcoord * (1. - ringwidthfrac)
ringrad_outer = ringrad_camcoord * (1. + ringwidthfrac)
fig, ax = plt.subplots(figsize=(10, 10))
plot_muon_event(ax, geom, image * clean_mask, centroid,
ringrad_camcoord, ringrad_inner, ringrad_outer,
event_id)
plt.figtext(0.15, 0.20, 'radial std dev: {0:.3f}'. \
format(radial_distribution['standard_dev']))
plt.figtext(0.15, 0.18, 'radial excess kurtosis: {0:.3f}'. \
format(radial_distribution['excess_kurtosis']))
plt.figtext(0.15, 0.16, 'fitted ring width: {0:.3f}'.format(width))
plt.figtext(0.15, 0.14, 'ring completeness: {0:.3f}'. \
format(muonparameters.completeness))
fig.savefig('{}/Event_{}_fitted.png'.format(plots_path, event_id))
if (plot_rings and not plots_path):
print("You are trying to plot without giving a path!")
return muonintensityoutput, dist_mask, ring_size, size_outside_ring, \
muonringparam, good_ring, radial_distribution, \
mean_pixel_charge_around_ring, muonparameters
def analyze_muon_event(subarray, tel_id, event_id, image, good_ring_config,
plot_rings, plots_path):
"""
Analyze an event to fit a muon ring
Parameters
----------
subarray: `ctapipe.instrument.subarray.SubarrayDescription`
Telescopes subarray
tel_id : `int`
Id of the telescope used
event_id : `int`
Id of the analyzed event
image : `np.ndarray`
Number of photoelectrons in each pixel
good_ring_config : `dict` or None
Set of parameters used to identify good muon rings to update LST-1 defaults
plot_rings : `bool`
Plot the muon ring
plots_path : `string`
Path to store the figures
Returns
-------
muonintensityoutput : `MuonEfficiencyContainer`
dist_mask : `ndarray`
Pixels used in ring intensity likelihood fit
ring_size : `float`
Total intensity in ring in photoelectrons
size_outside_ring : `float`
Intensity outside the muon ring in photoelectrons
to check for "shower contamination"
muonringparam : `MuonRingContainer`
good_ring : `bool`
It determines whether the ring can be used for analysis or not
radial_distribution : `dict`
Return of function radial_light_distribution
mean_pixel_charge_around_ring : float
Charge "just outside" ring, to check the possible signal extractor bias
muonparameters : `MuonParametersContainer`
"""
tel_description = subarray.tels[tel_id]
cam_rad = (tel_description.camera.geometry.guess_radius() /
tel_description.optics.equivalent_focal_length) * u.rad
geom = tel_description.camera.geometry
equivalent_focal_length = tel_description.optics.equivalent_focal_length
mirror_area = tel_description.optics.mirror_area
# some parameters for analysis and cuts for good ring selection:
params = update_parameters(good_ring_config, geom.n_pixels)
x, y = pixel_coords_to_telescope(geom, equivalent_focal_length)
muonringparam, clean_mask, dist, image_clean = fit_muon(
x, y, image, geom, params['tailcuts'])
mirror_radius = np.sqrt(mirror_area / np.pi) # meters
dist_mask = np.abs(
dist - muonringparam.radius
) < muonringparam.radius * params['ring_integration_width']
pix_ring = image * dist_mask
pix_outside_ring = image * ~dist_mask
# mask to select pixels just outside the ring that will be integrated to obtain the ring's intensity:
dist_mask_2 = np.logical_and(
~dist_mask,
np.abs(dist - muonringparam.radius) < muonringparam.radius *
(params['ring_integration_width'] + params['outer_ring_width']))
pix_ring_2 = image[dist_mask_2]
# nom_dist = np.sqrt(np.power(muonringparam.center_x,2)
# + np.power(muonringparam.center_y, 2))
muonparameters = MuonParametersContainer()
muonparameters.containment = ring_containment(muonringparam.radius,
muonringparam.center_x,
muonringparam.center_y,
cam_rad)
radial_distribution = radial_light_distribution(muonringparam.center_x,
muonringparam.center_y,
x[clean_mask],
y[clean_mask],
image[clean_mask])
# Do complicated calculations (minuit-based max likelihood ring fit) only for selected rings:
candidate_clean_ring = all([
radial_distribution['standard_dev'] < params['max_radial_stdev'],
radial_distribution['excess_kurtosis']
< params['max_radial_excess_kurtosis'],
(pix_ring > params['tailcuts'][0]).sum() >
params['min_pix_fraction_after_cleaning'] * params['min_pix'],
np.count_nonzero(pix_ring) > params['min_pix'],
muonringparam.radius < params['max_ring_radius'],
muonringparam.radius > params['min_ring_radius']
])
if candidate_clean_ring:
intensity_fitter = MuonIntensityFitter(subarray, hole_radius_m=0.308)
# Use same hard-coded value for pedestal fluctuations as the previous
# version of ctapipe:
pedestal_stddev = 1.1 * np.ones(len(image))
muonintensityoutput = \
intensity_fitter(tel_id,
muonringparam.center_x,
muonringparam.center_y,
muonringparam.radius,
image,
pedestal_stddev,
dist_mask)
dist_ringwidth_mask = np.abs(dist - muonringparam.radius) < \
muonintensityoutput.width
# We do the calculation of the ring completeness (i.e. fraction of whole circle) using the pixels
# within the "width" fitted using MuonIntensityFitter
muonparameters.completeness = ring_completeness(
x[dist_ringwidth_mask],
y[dist_ringwidth_mask],
image[dist_ringwidth_mask],
muonringparam.radius,
muonringparam.center_x,
muonringparam.center_y,
threshold=params['ring_completeness_threshold'],
bins=30)
# No longer existing in ctapipe 0.8:
# pix_ringwidth_im = image[dist_ringwidth_mask]
# muonintensityoutput.ring_pix_completeness = \
# (pix_ringwidth_im > tailcuts[0]).sum() / len(pix_ringwidth_im)
else:
# just to have the default values with units:
muonintensityoutput = MuonEfficiencyContainer()
muonintensityoutput.width = u.Quantity(np.nan, u.deg)
muonintensityoutput.impact = u.Quantity(np.nan, u.m)
muonintensityoutput.impact_x = u.Quantity(np.nan, u.m)
muonintensityoutput.impact_y = u.Quantity(np.nan, u.m)
# muonintensityoutput.mask = dist_mask # no longer there in ctapipe 0.8
ring_size = np.sum(pix_ring)
size_outside_ring = np.sum(pix_outside_ring * clean_mask)
# This is just mean charge per pixel in pixels just around the ring
# (on the outer side):
mean_pixel_charge_around_ring = np.sum(pix_ring_2) / len(pix_ring_2)
if candidate_clean_ring:
print(
"Impact parameter={:.3f}, ring_width={:.3f}, ring radius={:.3f}, "
"ring completeness={:.3f}".format(
muonintensityoutput.impact,
muonintensityoutput.width,
muonringparam.radius,
muonparameters.completeness,
))
# Now add the conditions based on the detailed muon ring fit:
conditions = [
candidate_clean_ring,
muonintensityoutput.impact <
params['max_impact_parameter'] * mirror_radius,
muonintensityoutput.impact >
params['min_impact_parameter'] * mirror_radius,
# TODO: To be applied when we have decent optics.
# muonintensityoutput.width
# < 0.08,
# NOTE: inside "candidate_clean_ring" cuts there is already a cut in
# the std dev of light distribution along ring radius, which is also
# a measure of the ring width
# muonintensityoutput.width
# > 0.04
]
if all(conditions):
good_ring = True
else:
good_ring = False
if (plot_rings and plots_path and good_ring):
focal_length = equivalent_focal_length
ring_telescope = SkyCoord(muonringparam.center_x,
muonringparam.center_y, TelescopeFrame())
ring_camcoord = ring_telescope.transform_to(
CameraFrame(
focal_length=focal_length,
rotation=geom.cam_rotation,
))
centroid = (ring_camcoord.x.value, ring_camcoord.y.value)
radius = muonringparam.radius
width = muonintensityoutput.width
ringrad_camcoord = 2 * radius.to(u.rad) * focal_length
ringwidthfrac = width / radius
ringrad_inner = ringrad_camcoord * (1. - ringwidthfrac)
ringrad_outer = ringrad_camcoord * (1. + ringwidthfrac)
fig, ax = plt.subplots(figsize=(10, 10))
plot_muon_event(ax, geom, image * clean_mask, centroid,
ringrad_camcoord, ringrad_inner, ringrad_outer,
event_id)
plt.figtext(0.15, 0.20, 'radial std dev: {0:.3f}'. \
format(radial_distribution['standard_dev']))
plt.figtext(0.15, 0.18, 'radial excess kurtosis: {0:.3f}'. \
format(radial_distribution['excess_kurtosis']))
plt.figtext(0.15, 0.16, 'fitted ring width: {0:.3f}'.format(width))
plt.figtext(0.15, 0.14, 'ring completeness: {0:.3f}'. \
format(muonparameters.completeness))
fig.savefig('{}/Event_{}_fitted.png'.format(plots_path, event_id))
if (plot_rings and not plots_path):
print("You are trying to plot without giving a path!")
return muonintensityoutput, dist_mask, ring_size, size_outside_ring, \
muonringparam, good_ring, radial_distribution, \
mean_pixel_charge_around_ring, muonparameters