def save_callback(event):
# Save pointing model to file
outfile = file(opts.outfilebase + '.csv', 'w')
# The original pointing model description string was comma-separated
outfile.write(new_model.description.replace(" ", ", "))
outfile.close()
logger.debug("Saved %d-parameter pointing model to '%s'" % (len(new_model), opts.outfilebase + '.csv'))
# Turn data recarray into list of dicts and add residuals to the mix
extended_data = []
for n in range(len(data)):
rec_dict = dict(zip(data.dtype.names, data[n]))
rec_dict['keep'] = int(keep[n])
rec_dict['old_residual_xel'] = rad2deg(old.residual_xel[n])
rec_dict['old_residual_el'] = rad2deg(old.residual_el[n])
rec_dict['new_residual_xel'] = rad2deg(new.residual_xel[n])
rec_dict['new_residual_el'] = rad2deg(new.residual_el[n])
extended_data.append(rec_dict)
# Format the data similar to analyse_point_source_scans output CSV file, with four new columns at the end
fields = '%(dataset)s, %(target)s, %(timestamp_ut)s, %(azimuth).7f, %(elevation).7f, ' \
'%(delta_azimuth).7f, %(delta_azimuth_std).7f, %(delta_elevation).7f, %(delta_elevation_std).7f, ' \
'%(data_unit)s, %(beam_height_I).7f, %(beam_height_I_std).7f, %(beam_width_I).7f, ' \
'%(beam_width_I_std).7f, %(baseline_height_I).7f, %(baseline_height_I_std).7f, %(refined_I).0f, ' \
'%(beam_height_HH).7f, %(beam_width_HH).7f, %(baseline_height_HH).7f, %(refined_HH).0f, ' \
'%(beam_height_VV).7f, %(beam_width_VV).7f, %(baseline_height_VV).7f, %(refined_VV).0f, ' \
'%(frequency).7f, %(flux).4f, %(temperature).2f, %(pressure).2f, %(humidity).2f, %(wind_speed).2f, ' \
'%(keep)d, %(old_residual_xel).7f, %(old_residual_el).7f, %(new_residual_xel).7f, %(new_residual_el).7f\n'
field_names = [name.partition(')')[0] for name in fields[2:].split(', %(')]
# Save residual data and flags to file
outfile2 = file(opts.outfilebase + '_data.csv', 'w')
outfile2.write('# antenna = %s\n' % antenna.description)
outfile2.write(', '.join(field_names) + '\n')
outfile2.writelines([fields % rec for rec in extended_data])
outfile2.close()
save_button.color = '0.85'
save_button.hovercolor = '0.95'
def _extract_location_from_katdata(self):
self.metadata["DecRa"] = []
self.metadata["ElAz"] = []
f = self._katdata
f.select(scans="track,scan")
f.select(ants=f.ref_ant)
for i, scan, target in f.scans():
f.select(scans=i)
t = f.catalogue.targets[f.target_indices[0]]
if (t.body_type == 'radec'):
ra, dec = t.radec()
ra, dec = katpoint.rad2deg(ra), katpoint.rad2deg(dec)
self.metadata["DecRa"].append(
"%f, %f" % (dec, katpoint.wrap_angle(ra, 360)))
elif t.body_type == 'azel':
az, el = t.azel()
az, el = katpoint.rad2deg(az), katpoint.rad2deg(el)
if -90 <= el <= 90:
self.metadata["ElAz"].append(
"%f, %f" % (el, katpoint.wrap_angle(az, 360)))
else:
self.metadata["ElAz"].append(
"%f, %f" %
((np.clip(el, -90, 90)), katpoint.wrap_angle(az, 360)))
def save_callback(event):
# Save pointing model to file
outfile = file(opts.outfilebase + '.csv', 'w')
outfile.write(new_model.description)
outfile.close()
logger.debug("Saved %d-parameter pointing model to '%s'" %
(len(new_model.params), opts.outfilebase + '.csv'))
# Turn data recarray into list of dicts and add residuals to the mix
extended_data = []
for n in range(len(data)):
rec_dict = dict(zip(data.dtype.names, data[n]))
rec_dict['keep'] = int(keep[n])
rec_dict['old_residual_xel'] = rad2deg(old.residual_xel[n])
rec_dict['old_residual_el'] = rad2deg(old.residual_el[n])
rec_dict['new_residual_xel'] = rad2deg(new.residual_xel[n])
rec_dict['new_residual_el'] = rad2deg(new.residual_el[n])
extended_data.append(rec_dict)
# Format the data similar to analyse_point_source_scans output CSV file, with four new columns at the end
fields = '%(dataset)s, %(target)s, %(timestamp_ut)s, %(azimuth).7f, %(elevation).7f, ' \
'%(delta_azimuth).7f, %(delta_azimuth_std).7f, %(delta_elevation).7f, %(delta_elevation_std).7f, ' \
'%(data_unit)s, %(beam_height_I).7f, %(beam_height_I_std).7f, %(beam_width_I).7f, ' \
'%(beam_width_I_std).7f, %(baseline_height_I).7f, %(baseline_height_I_std).7f, %(refined_I).0f, ' \
'%(beam_height_HH).7f, %(beam_width_HH).7f, %(baseline_height_HH).7f, %(refined_HH).0f, ' \
'%(beam_height_VV).7f, %(beam_width_VV).7f, %(baseline_height_VV).7f, %(refined_VV).0f, ' \
'%(frequency).7f, %(flux).4f, %(temperature).2f, %(pressure).2f, %(humidity).2f, %(wind_speed).2f, ' \
'%(keep)d, %(old_residual_xel).7f, %(old_residual_el).7f, %(new_residual_xel).7f, %(new_residual_el).7f\n'
field_names = [name.partition(')')[0] for name in fields[2:].split(', %(')]
# Save residual data and flags to file
outfile2 = file(opts.outfilebase + '_data.csv', 'w')
outfile2.write('# antenna = %s\n' % antenna.description)
outfile2.write(', '.join(field_names) + '\n')
outfile2.writelines([fields % rec for rec in extended_data])
outfile2.close()
save_button.color = '0.85'
save_button.hovercolor = '0.95'
def update(self, timestamp):
elapsed_time = timestamp - self._last_update if self._last_update else 0.0
self._last_update = timestamp
if self.mode not in ('POINT', 'SCAN', 'STOW'):
return
az, el = self.pos_actual_scan_azim, self.pos_actual_scan_elev
target = construct_azel_target(deg2rad(az), deg2rad(90.0)) \
if self.mode == 'STOW' else self._target
if not target:
return
requested_az, requested_el = target.azel(timestamp, self.ant)
requested_az = rad2deg(wrap_angle(requested_az))
requested_el = rad2deg(requested_el)
delta_az = wrap_angle(requested_az - az, period=360.)
delta_el = requested_el - el
# Truncate velocities to slew rate limits and update position
max_delta_az = self.max_slew_azim_dps * elapsed_time
max_delta_el = self.max_slew_elev_dps * elapsed_time
az += min(max(delta_az, -max_delta_az), max_delta_az)
el += min(max(delta_el, -max_delta_el), max_delta_el)
# Truncate coordinates to antenna limits
az = min(max(az, self.real_az_min_deg), self.real_az_max_deg)
el = min(max(el, self.real_el_min_deg), self.real_el_max_deg)
# Check angular separation to determine lock
dish = construct_azel_target(deg2rad(az), deg2rad(el))
error = rad2deg(target.separation(dish, timestamp, self.ant))
self.lock = error < self.lock_threshold
# Update position sensors
self.pos_request_scan_azim = requested_az
self.pos_request_scan_elev = requested_el
self.pos_actual_scan_azim = az
self.pos_actual_scan_elev = el
def LoadHDF5(HDF5Filename, header=False):
try:
d = scape.DataSet(HDF5Filename, baseline=opts.baseline)
except ValueError:
print "WARNING:THIS FILE", HDF5Filename.split(
'/'
)[-1], "IS CORRUPTED AND SCAPE WILL NOT PROCESS IT, YOU MAY NEED TO REAUGMENT IT,BUT ITS AN EXPENSIVE TASK..!!"
else:
print "SUCCESSFULLY LOADED: Wellcome to scape Library and scape is busy processing your request"
lo_freq = 4200.0 + d.freqs[len(d.freqs) / 2.0]
# try to check all the rfi channels across all the channels
rfi_chan_across_all = d.identify_rfi_channels()
d = d.select(freqkeep=range(100, 420))
# rfi channels across fringe finder channels ( i.e frequancy range around 100 to 420)
rfi_channels = d.identify_rfi_channels()
freqs = d.freqs
sky_frequency = d.freqs[rfi_channels]
ant = d.antenna.name
data_filename = os.path.splitext(
os.path.basename(HDF5Filename))[0] + '.h5'
# obs_date = os.path.splitext(os.path.basename(HDF5Filename))[0]
#date = time.ctime(float(obs_date))
for compscan in d.compscans:
azimuth = np.hstack(
[scan.pointing['az'] for scan in compscan.scans])
elevation = np.hstack(
[scan.pointing['el'] for scan in compscan.scans])
compscan_times = np.hstack(
[scan.timestamps for scan in compscan.scans])
compscan_start_time = np.hstack(
[scan.timestamps[0] for scan in compscan.scans])
compscan_end_time = np.hstack(
[scan.timestamps[-1] for scan in compscan.scans])
middle_time = np.median(compscan_times, axis=None)
obs_date = katpoint.Timestamp(middle_time)
middle_start_time = np.median(compscan_start_time)
middle_end_time = np.median(compscan_end_time)
end_time = katpoint.Timestamp(middle_end_time)
min_compscan_az = katpoint.rad2deg(azimuth.min())
max_compscan_az = katpoint.rad2deg(azimuth.max())
min_compscan_el = katpoint.rad2deg(elevation.min())
max_compscan_el = katpoint.rad2deg(elevation.max())
start_time = katpoint.Timestamp(middle_start_time)
requested_azel = compscan.target.azel(middle_time)
#ant_az = katpoint.rad2deg(np.array(requested_azel[0]))
#ant_el = katpoint.rad2deg(np.array(requested_azel[1]))
target = compscan.target.name
f = file(opts.outfilebase + '.csv', 'a')
for index in range(0, len(rfi_channels)):
rfi_chan = rfi_channels[index] + 100
rfi_freq = freqs[rfi_channels[index]]
f.write('%s, %s, %s, %s, %s,%f, %f,%f,%f, %f, %d, %f\n' % (data_filename,start_time, end_time, ant,target,min_compscan_az,max_compscan_az,\
min_compscan_el, max_compscan_el,lo_freq, rfi_chan, rfi_freq))
f.close()
def check_target(OVST, target, tmstmp=None, check=True):
'''
:param tar: str or object of class Target
if string: searches for the target in catalogue
:param antennas: list
list of antenna objects of class Antenna
:param catalogue: Catalogue
:param tmstmp: Timestamp
:return: list with position tuples
[(az1, el1), (az2, el2), ...]
'''
antennas = OVST.active_antennas
catalogue = OVST.Catalogue
azel = []
if isinstance(target, Target):
target = target
elif isinstance(
target, str
) and ',' in target: # Check if target has format: e.g. 'azel, 30, 60'
target = Target(target)
elif isinstance(target, str):
target = catalogue[target]
if not target:
raise ValueError("Target not in Catalogue")
if isinstance(tmstmp, str):
if tmstmp and len(tmstmp) == 5:
tmstmp += ':00'
if tmstmp and len(tmstmp) == 8:
tmstmp = str(datetime.now().date()) + ' ' + tmstmp
if isinstance(tmstmp, (int, float)):
tmstmp = Timestamp(tmstmp)
if not tmstmp:
tmstmp = Timestamp()
for antenna in antennas:
ae = target.azel(timestamp=tmstmp, antenna=antenna)
azel.append([rad2deg(ae[0]), rad2deg(ae[1])])
az = [item[0] for item in azel]
el = [item[1] for item in azel]
if check:
if all((OVST.az_limit[1] - 2 < i < OVST.az_limit[0] + 2
for i in az)) or all(i < OVST.el_limit[0] for i in el):
raise LookupError(
'target cannot get focused at % s (target at azimuth %.2f and elevation %.2f).\n '
'Allowed limits: az not in range of 150-173 and elevation > 25'
% (tmstmp.local()[11:19], azel[0][0], azel[0][1]))
return azel # format: [(az1, el1), (az2, el2), ...]
def update(fig):
"""Fit new pointing model and update plots."""
# Perform early redraw to improve interactivity of clicks (which typically change state of target dots)
# Target state: 0 = flagged, 1 = unflagged, 2 = highlighted
target_state = keep * ((target_index == fig.highlighted_target) + 1)
# Specify colours of flagged, unflagged and highlighted dots, respectively, as RGBA tuples
dot_colors = np.choose(target_state, np.atleast_3d(np.vstack([(1,1,1,1), (0,0,1,1), (1,0,0,1)]))).T
for ax in fig.axes[:7]:
ax.dots.set_facecolors(dot_colors)
fig.canvas.draw()
# Fit new pointing model and update results
params, sigma_params = new_model.fit(az[keep], el[keep], measured_delta_az[keep], measured_delta_el[keep],
std_delta_az[keep], std_delta_el[keep], enabled_params)
new.update(new_model)
# Update rest of figure
fig.texts[3].set_text("$\chi^2$ = %.1f" % new.chi2)
fig.texts[4].set_text("all sky rms = %.3f' (robust %.3f')" % (new.sky_rms, new.robust_sky_rms))
new.metrics(target_index == fig.highlighted_target)
fig.texts[5].set_text("target sky rms = %.3f' (robust %.3f')" % (new.sky_rms, new.robust_sky_rms))
new.metrics(keep)
fig.texts[-1].set_text(unique_targets[fig.highlighted_target])
# Update model parameter strings
for p, param in enumerate(display_params):
fig.texts[2*p + 6].set_text(param_to_str(new_model, param) if enabled_params[param] else '')
# HACK to convert sigmas to arcminutes, but not for P9 and P12 (which are scale factors)
# This functionality should really reside inside the PointingModel class
std_param = rad2deg(sigma_params[param]) * 60. if param not in [8, 11] else sigma_params[param]
std_param_str = ("%.2f'" % std_param) if param not in [8, 11] else ("%.0e" % std_param)
fig.texts[2*p + 7].set_text(std_param_str if enabled_params[param] and opts.use_stats else '')
# Turn parameter string bold if it changed significantly from old value
if np.abs(params[param] - old_model.values()[param]) > 3.0 * sigma_params[param]:
fig.texts[2*p + 6].set_weight('bold')
fig.texts[2*p + 7].set_weight('bold')
else:
fig.texts[2*p + 6].set_weight('normal')
fig.texts[2*p + 7].set_weight('normal')
daz_az, del_az, daz_el, del_el, quiver, before, after = fig.axes[:7]
# Update quiver plot
quiver_scale = 0.1 * fig.quiver_scale_slider.val * np.pi / 6 / deg2rad(old.robust_sky_rms / 60.)
quiver.quiv.set_segments(quiver_segments(new.residual_az, new.residual_el, quiver_scale))
quiver.quiv.set_color(np.choose(keep, np.atleast_3d(np.vstack([(0.3,0.3,0.3,0.2), (0.3,0.3,0.3,1)]))).T)
# Update residual plots
daz_az.dots.set_offsets(np.c_[rad2deg(az), rad2deg(new.residual_xel) * 60.])
del_az.dots.set_offsets(np.c_[rad2deg(az), rad2deg(new.residual_el) * 60.])
daz_el.dots.set_offsets(np.c_[rad2deg(el), rad2deg(new.residual_xel) * 60.])
del_el.dots.set_offsets(np.c_[rad2deg(el), rad2deg(new.residual_el) * 60.])
after.dots.set_offsets(np.c_[np.arctan2(new.residual_el, new.residual_xel), new.abs_sky_error])
resid_lim = 1.2 * max(new.abs_sky_error.max(), old.abs_sky_error.max())
daz_az.set_ylim(-resid_lim, resid_lim)
del_az.set_ylim(-resid_lim, resid_lim)
daz_el.set_ylim(-resid_lim, resid_lim)
del_el.set_ylim(-resid_lim, resid_lim)
before.set_ylim(0, resid_lim)
after.set_ylim(0, resid_lim)
# Redraw the figure
fig.canvas.draw()
def LoadHDF5(HDF5Filename, header=False):
try:
d = scape.DataSet(HDF5Filename,baseline=opts.baseline)
except ValueError:
print "WARNING:THIS FILE",HDF5Filename.split('/')[-1], "IS CORRUPTED AND SCAPE WILL NOT PROCESS IT, YOU MAY NEED TO REAUGMENT IT,BUT ITS AN EXPENSIVE TASK..!!"
else:
print "SUCCESSFULLY LOADED: Wellcome to scape Library and scape is busy processing your request"
lo_freq = 4200.0 + d.freqs[len(d.freqs)/2.0]
# try to check all the rfi channels across all the channels
rfi_chan_across_all = d.identify_rfi_channels()
d = d.select(freqkeep=range(100,420))
# rfi channels across fringe finder channels ( i.e frequancy range around 100 to 420)
rfi_channels = d.identify_rfi_channels()
freqs = d.freqs
sky_frequency = d.freqs[rfi_channels]
ant = d.antenna.name
data_filename = os.path.splitext(os.path.basename(HDF5Filename))[0]+'.h5'
# obs_date = os.path.splitext(os.path.basename(HDF5Filename))[0]
#date = time.ctime(float(obs_date))
for compscan in d.compscans:
azimuth = np.hstack([scan.pointing['az'] for scan in compscan.scans])
elevation = np.hstack([scan.pointing['el'] for scan in compscan.scans])
compscan_times = np.hstack([scan.timestamps for scan in compscan.scans])
compscan_start_time = np.hstack([scan.timestamps[0] for scan in compscan.scans])
compscan_end_time = np.hstack([scan.timestamps[-1] for scan in compscan.scans])
middle_time = np.median(compscan_times, axis=None)
obs_date = katpoint.Timestamp(middle_time)
middle_start_time = np.median(compscan_start_time)
middle_end_time = np.median(compscan_end_time)
end_time = katpoint.Timestamp(middle_end_time)
min_compscan_az = katpoint.rad2deg(azimuth.min())
max_compscan_az = katpoint.rad2deg(azimuth.max())
min_compscan_el = katpoint.rad2deg(elevation.min())
max_compscan_el = katpoint.rad2deg(elevation.max())
start_time = katpoint.Timestamp(middle_start_time)
requested_azel = compscan.target.azel(middle_time)
#ant_az = katpoint.rad2deg(np.array(requested_azel[0]))
#ant_el = katpoint.rad2deg(np.array(requested_azel[1]))
target = compscan.target.name
f = file(opts.outfilebase + '.csv', 'a')
for index in range(0,len(rfi_channels)):
rfi_chan = rfi_channels[index] + 100
rfi_freq = freqs[rfi_channels[index]]
f.write('%s, %s, %s, %s, %s,%f, %f,%f,%f, %f, %d, %f\n' % (data_filename,start_time, end_time, ant,target,min_compscan_az,max_compscan_az,\
min_compscan_el, max_compscan_el,lo_freq, rfi_chan, rfi_freq))
f.close()
def metrics(model, az, el, measured_delta_az, measured_delta_el, std_delta_az,
std_delta_el):
"""Determine new residuals and sky RMS from pointing model."""
model_delta_az, model_delta_el = model.offset(az, el)
residual_az = measured_delta_az - model_delta_az
residual_el = measured_delta_el - model_delta_el
residual_xel = residual_az * np.cos(el)
abs_sky_error = rad2deg(np.sqrt(residual_xel**2 + residual_el**2)) * 3600.
###### On the calculation of all-sky RMS #####
# Assume the el and cross-el errors have zero mean, are distributed normally, and are uncorrelated
# They are therefore described by a 2-dimensional circular Gaussian pdf with zero mean and *per-component*
# standard deviation of sigma
# The absolute sky error (== Euclidean length of 2-dim error vector) then has a Rayleigh distribution
# The RMS sky error has a mean value of sqrt(2) * sigma, since each squared error term is the sum of
# two squared Gaussian random values, each with an expected value of sigma^2.
sky_rms = np.sqrt(np.mean(abs_sky_error**2))
# A more robust estimate of the RMS sky error is obtained via the median of the Rayleigh distribution,
# which is sigma * sqrt(log(4)) -> convert this to the RMS sky error = sqrt(2) * sigma
robust_sky_rms = np.median(abs_sky_error) * np.sqrt(2. / np.log(4.))
# The chi^2 value is what is actually optimised by the least-squares fitter (evaluated on the training set)
chi2 = np.sum(
((residual_xel / std_delta_az)**2 + (residual_el / std_delta_el)**2))
text = []
#text.append("$\chi^2$ = %g " % chi2)
text.append("All sky RMS = %.3f\" (robust %.3f\") " %
(sky_rms, robust_sky_rms))
return sky_rms, robust_sky_rms, chi2, text
def update(self, model):
"""Determine new residuals and sky RMS from pointing model."""
model_delta_az, model_delta_el = model.offset(az, el)
self.residual_az = measured_delta_az - model_delta_az
self.residual_el = measured_delta_el - model_delta_el
self.residual_xel = self.residual_az * np.cos(el)
self.abs_sky_error = rad2deg(np.sqrt(self.residual_xel ** 2 + self.residual_el ** 2)) * 60.
self.metrics(keep)
def update(self, model):
"""Determine new residuals and sky RMS from pointing model."""
model_delta_az, model_delta_el = model.offset(az, el)
self.residual_az = measured_delta_az - model_delta_az
self.residual_el = measured_delta_el - model_delta_el
self.residual_xel = self.residual_az * np.cos(el)
self.abs_sky_error = rad2deg(np.sqrt(self.residual_xel ** 2 + self.residual_el ** 2)) * 60.
self.metrics(keep)
def fit_primary_beams(session, data_points):
"""Fit primary beams to receptor gains obtained at various offset pointings.
Parameters
----------
session : :class:`katcorelib.observe.CaptureSession` object
The active capture session
data_points : dict mapping receptor index to (x, y, freq, gain, weight) seq
Complex gains per receptor, as multiple records per offset and frequency
Returns
-------
beams : dict mapping receptor name to list of :class:`BeamPatternFit`
Fitted primary beams, per receptor and per frequency chunk
"""
beams = {}
# Iterate over receptors
for a in data_points:
data = np.rec.fromrecords(data_points[a], names='x,y,freq,gain,weight')
data = data.reshape(-1, NUM_CHUNKS)
ant = session.observers[a]
# Iterate over frequency chunks but discard typically dodgy band edges
for chunk in range(1, NUM_CHUNKS - 1):
chunk_data = data[:, chunk]
is_valid = np.nonzero(~np.isnan(chunk_data['gain'])
& (chunk_data['weight'] > 0.))[0]
chunk_data = chunk_data[is_valid]
if len(chunk_data) == 0:
continue
expected_width = rad2deg(ant.beamwidth * lightspeed /
chunk_data['freq'][0] / ant.diameter)
# Convert power beamwidth to gain / voltage beamwidth
expected_width = np.sqrt(2.0) * expected_width
# XXX This assumes we are still using default ant.beamwidth of 1.22
# and also handles larger effective dish diameter in H direction
expected_width = (0.8 * expected_width, 0.9 * expected_width)
beam = BeamPatternFit((0., 0.), expected_width, 1.0)
x = np.c_[chunk_data['x'], chunk_data['y']].T
y = chunk_data['gain']
std_y = np.sqrt(1. / chunk_data['weight'])
try:
beam.fit(x, y, std_y)
except TypeError:
continue
beamwidth_norm = beam.width / np.array(expected_width)
center_norm = beam.center / beam.std_center
user_logger.debug(
"%s %2d %2d: height=%4.2f width=(%4.2f, %4.2f) "
"center=(%7.2f, %7.2f)%s", ant.name, chunk, len(y),
beam.height, beamwidth_norm[0], beamwidth_norm[1],
center_norm[0], center_norm[1],
' X' if not beam.is_valid else '')
# Store beam per frequency chunk and per receptor
beams_freq = beams.get(ant.name, [None] * NUM_CHUNKS)
beams_freq[chunk] = beam
beams[ant.name] = beams_freq
return beams
def _target_azel(self, target):
"""Get azimuth and elevation co-ordinates for a target at the current time.
Parameters
----------
target: katpoint.Target
The target of interest.
Returns
-------
az: float
The azimuth co-ordinate of the target in degrees.
el: float
The elevation co-ordinate of the target in degrees.
"""
az, el = target.azel(simobserver.date)
az = katpoint.rad2deg(az)
el = katpoint.rad2deg(el)
return az, el
def metrics(model,az,el,measured_delta_az, measured_delta_el ,std_delta_az ,std_delta_el,time_stamps):
"""Determine new residuals and sky RMS from pointing model."""
model_delta_az, model_delta_el = model.offset(az, el)
residual_az = measured_delta_az - model_delta_az
residual_el = measured_delta_el - model_delta_el
residual_xel = residual_az * np.cos(el)
abs_sky_error = rad2deg(np.sqrt(residual_xel ** 2 + residual_el ** 2))
offset_az_ts = pandas.Series(rad2deg(residual_xel), pandas.to_datetime(time_stamps, unit='s'))#.asfreq(freq='1s')
offset_el_ts = pandas.Series(rad2deg(residual_el), pandas.to_datetime(time_stamps, unit='s'))#.asfreq(freq='1s')
offset_total_ts = pandas.Series( abs_sky_error, pandas.to_datetime(time_stamps, unit='s'))#.asfreq(freq='1s')
###### On the calculation of all-sky RMS #####
# Assume the el and cross-el errors have zero mean, are distributed normally, and are uncorrelated
# They are therefore described by a 2-dimensional circular Gaussian pdf with zero mean and *per-component*
# standard deviation of sigma
# The absolute sky error (== Euclidean length of 2-dim error vector) then has a Rayleigh distribution
# The RMS sky error has a mean value of sqrt(2) * sigma, since each squared error term is the sum of
# two squared Gaussian random values, each with an expected value of sigma^2.
sky_rms = np.sqrt(np.mean(abs_sky_error ** 2))
# A more robust estimate of the RMS sky error is obtained via the median of the Rayleigh distribution,
# which is sigma * sqrt(log(4)) -> convert this to the RMS sky error = sqrt(2) * sigma
robust_sky_rms = np.median(abs_sky_error) * np.sqrt(2. / np.log(4.))
# The chi^2 value is what is actually optimised by the least-squares fitter (evaluated on the training set)
#chi2 = np.sum(((residual_xel / std_delta_az) ** 2 + (residual_el / std_delta_el) ** 2))
text = 'All sky RMS = %.3f\" (robust %.3f\") ' % (sky_rms*3600, robust_sky_rms*3600)
fig = plt.figure(figsize=(10,5))
#change_total = np.sqrt(change_el**2 + change_az**2)
#(offset_el_ts*3600.).plot(label='Elevation',legend=True,grid=True,style='*')
#(offset_az_ts*3600.).plot(label='Azimuth',legend=True,grid=True,style='*')
(offset_total_ts*3600.).plot(label='Total pointing Error',legend=True,grid=True,style='*')
dataset_str = ' ,'.join(np.unique(offsetdata['dataset']).tolist() )
#target_str = ' ,'.join(np.unique(offsetdata['target']).tolist() )
plt.title("Offset for Antenna:%s Dataset:%s \n %s " %(ant.name,dataset_str ,text),fontsize=10)
plt.ylabel('Offset (arc-seconds)')
plt.xlabel('Time (UTC)',fontsize=8)
plt.figtext(0.89, 0.18,git_info(), horizontalalignment='right',fontsize=10)
return fig
def select_and_average(filename, average_time):
# Read a file into katdal, and average the data to the prescribed averaging time
# Returns the weather data and timestamps with the correct averaging interval
data = katdal.open(filename)
raw_timestamps = data.sensor.timestamps
raw_wind_speed = data.wind_speed
raw_temperature = data.temperature
raw_dumptime = data.dump_period
# Get azel of each antenna and separation of each antenna
sun = katpoint.Target('Sun, special', antenna=data.ants[0])
alltimestamps = data.timestamps[:]
solar_seps = np.zeros_like(alltimestamps)
for dumpnum, timestamp in enumerate(alltimestamps):
azeltarget = katpoint.construct_azel_target(
katpoint.deg2rad(data.az[dumpnum, 0]),
katpoint.deg2rad(data.el[dumpnum, 0]))
azeltarget.antenna = data.ants[0]
solar_seps[dumpnum] = katpoint.rad2deg(
azeltarget.separation(sun, timestamp))
#Determine number of dumps to average
num_average = max(int(np.round(average_time / raw_dumptime)), 1)
#Array of block indices
indices = list(
range(min(num_average, raw_timestamps.shape[0]),
raw_timestamps.shape[0] + 1,
min(num_average, raw_timestamps.shape[0])))
timestamps = np.average(np.array(np.split(raw_timestamps, indices)[:-1]),
axis=1)
wind_speed = np.average(np.array(np.split(raw_wind_speed, indices)[:-1]),
axis=1)
temperature = np.average(np.array(np.split(raw_temperature, indices)[:-1]),
axis=1)
dump_time = raw_dumptime * num_average
return (timestamps, alltimestamps, wind_speed, temperature, dump_time,
solar_seps, data.ants[0])
def referencemetrics(measured_delta_az, measured_delta_el):
"""Determine and sky RMS from pointing model."""
text = []
measured_delta_xel = measured_delta_az * np.cos(
el) # scale due to sky shape
abs_sky_error = np.ma.array(data=measured_delta_xel, mask=False)
for target in set(offsetdata['target']):
keep = np.ones((len(offsetdata)), dtype=np.bool)
for key, targetv in enumerate(offsetdata['target']):
keep[key] = target == targetv
abs_sky_error[keep] = rad2deg(
np.sqrt(
(measured_delta_xel[keep] - measured_delta_xel[keep][0])**2 +
(measured_delta_el[keep] -
measured_delta_el[keep][0])**2)) * 60.
abs_sky_error.mask[keep.nonzero()[0]
[0]] = True # Mask the reference element
text.append(
"Test Target: '%s' Reference RMS = %.3f' (robust %.3f') (N=%i Data Points)"
% (target, np.sqrt((abs_sky_error[keep]**2).mean()),
np.ma.median(abs_sky_error[keep]) * np.sqrt(2. / np.log(4.)),
keep.sum() - 1))
###### On the calculation of all-sky RMS #####
# Assume the el and cross-el errors have zero mean, are distributed normally, and are uncorrelated
# They are therefore described by a 2-dimensional circular Gaussian pdf with zero mean and *per-component*
# standard deviation of sigma
# The absolute sky error (== Euclidean length of 2-dim error vector) then has a Rayleigh distribution
# The RMS sky error has a mean value of sqrt(2) * sigma, since each squared error term is the sum of
# two squared Gaussian random values, each with an expected value of sigma^2.
sky_rms = np.sqrt(np.ma.mean(abs_sky_error**2))
#print abs_sky_error
# A more robust estimate of the RMS sky error is obtained via the median of the Rayleigh distribution,
# which is sigma * sqrt(log(4)) -> convert this to the RMS sky error = sqrt(2) * sigma
robust_sky_rms = np.ma.median(abs_sky_error) * np.sqrt(2. / np.log(4.))
text.append(
"All Sky Reference RMS = %.3f' (robust %.3f') (N=%i Data Points) R.T.P.4"
% (sky_rms, robust_sky_rms, abs_sky_error.count()))
return text
def test_pointing(self):
az, el = self.target.azel(self.timestamps, self.antennas[1])
assert_array_equal(self.dataset.az[:, 1], rad2deg(az))
assert_array_equal(self.dataset.el[:, 1], rad2deg(el))
ra, dec = self.target.radec(self.timestamps, self.antennas[0])
assert_array_almost_equal(self.dataset.ra[:, 0],
rad2deg(ra),
decimal=5)
assert_array_almost_equal(self.dataset.dec[:, 0],
rad2deg(dec),
decimal=5)
angle = self.target.parallactic_angle(self.timestamps,
self.antennas[0])
# TODO: Check why this is so poor... see SR-1882 for progress on this
assert_array_almost_equal(self.dataset.parangle[:, 0],
rad2deg(angle),
decimal=0)
x, y = self.target.sphere_to_plane(az, el, self.timestamps,
self.antennas[1])
assert_array_equal(self.dataset.target_x[:, 1], rad2deg(x))
assert_array_equal(self.dataset.target_y[:, 1], rad2deg(y))
# track the Moon for a short time
session.label('track')
user_logger.info("Initiating %g-second track on target %s" % (opts.track_duration, target.name,))
session.set_target(target) # Set the target
session.track(target, duration=opts.track_duration, announce=False) # Set the target & mode = point
## 1) Track ephem target behind the moon
user_logger.info("Sleeping for 2 minutes")
time.sleep(120)
user_logger.info("Setting to Ephem target")
observer.date = ephem.now()
moon = ephem.Moon(observer)
moon.compute(observer)
target = katpoint.construct_radec_target(moon.ra, moon.dec)
session.label('track')
user_logger.info("Initiating %g-second track on ephem target (%.2f, %.2f)" % (opts.track_duration, katpoint.rad2deg(float(moon.ra)), katpoint.rad2deg(float(moon.dec)),))
session.set_target(target) # Set the target
session.track(target, duration=opts.track_duration, announce=False) # Set the target & mode = point
## 2) Track moon again
user_logger.info("Sleeping for 2 minutes")
time.sleep(120)
for target in observation_sources.iterfilter(el_limit_deg=opts.horizon):
user_logger.info(target)
# track the Moon for a short time
session.label('track')
user_logger.info("Repeating %g-second track on target %s" % (opts.track_duration, target.name,))
session.set_target(target) # Set the target
session.track(target, duration=opts.track_duration, announce=False) # Set the target & mode = point
markers = []
colors = ['b','g','r','c','m','y','k']
pointtypes = ['o','*','x','^','s','p','h','+','D','d','v','H','d','v']
for point in pointtypes:
for color in colors:
markers.append(str(color+point))
cat = katpoint.Catalogue(file(cat_filename),add_specials=False)
#cat.add('Sun, special') # except maybe the sun
cat.antenna = katpoint.Antenna('ant1, -30:43:17.3, 21:24:38.5, 1038.0, 12.0, 18.4 -8.7 0.0, -0:05:30.6 0 -0:00:03.3 0:02:14.2 0:00:01.6 -0:01:30.6 0:08:42.1, 1.22')
target = cat.targets[0]
t = katpoint.Timestamp().secs + np.arange(0, 24. * 60. * 60., 360.)
lst = katpoint.rad2deg(target.antenna.local_sidereal_time(t)) / 15
fig = plt.figure(1)
plt.clf()
fig.set_size_inches(12, 4)
plt.subplots_adjust(right=0.8)
lines = list()
labels = list()
count = 0
fontP = FontProperties()
fontP.set_size('small')
for target in cat.targets:
count = count + 1
elev = katpoint.rad2deg(target.azel(t)[1])
def select_environment(data, antenna, condition="normal"):
""" Flag data for environmental conditions. Options are:
normal: Wind < 9.8m/s, -5C < Temperature < 40C, DeltaTemp < 3deg in 20 minutes
optimal: Wind < 2.9m/s, -5C < Temperature < 35C, DeltaTemp < 2deg in 10 minutes
ideal: Wind < 1m/s, 19C < Temp < 21C, DeltaTemp < 1deg in 30 minutes
"""
# Convert timestamps to UTCseconds using katpoint
timestamps = np.array(
[katpoint.Timestamp(timestamp) for timestamp in data["timestamp_ut"]],
dtype='float32')
# Fit a smooth function (cubic spline) in time to the temperature and wind data
raw_wind = data["wind_speed"]
raw_temp = data["temperature"]
fit_wind = interpolate.InterpolatedUnivariateSpline(timestamps,
raw_wind,
k=3)
fit_temp = interpolate.InterpolatedUnivariateSpline(timestamps,
raw_temp,
k=3)
#fit_temp_grad = fit_temp.derivative()
# Day/Night
# Night is defined as when the Sun is at -5deg.
# Set up Sun target
sun = katpoint.Target('Sun, special', antenna=antenna)
sun_elevation = katpoint.rad2deg(sun.azel(timestamps)[1])
# Apply limits on environmental conditions
good = [True] * data.shape[0]
# Set up limits on environmental conditions
if condition == 'ideal':
windlim = 1.
temp_low = 19.
temp_high = 21.
deltatemp = 1. / (30. * 60.)
sun_elev_lim = -5.
elif condition == 'optimum':
windlim = 2.9
temp_low = -5.
temp_high = 35.
deltatemp = 2. / (10. * 60.)
sun_elev_lim = -5.
elif condition == 'normal':
windlim = 9.8
temp_low = -5.
temp_high = 40.
deltatemp = 3. / (20. * 60.)
sun_elev_lim = 100. #Daytime
else:
return good
good = good & (fit_wind(timestamps) < windlim)
good = good & ((fit_temp(timestamps) > temp_low) &
(fit_temp(timestamps) < temp_high))
#Get the temperature gradient
temp_grad = [
fit_temp.derivatives(timestamp)[1] for timestamp in timestamps
]
good = good & (np.abs(temp_grad) < deltatemp)
#Day or night?
good = good & (sun_elevation < sun_elev_lim)
return good
plt.subplot(121)
plot_times = np.arange(gain_times[0] - 1000, gain_times[-1] + 1000, 100.)
for n in range(4):
plt.plot(plot_times - gain_times[0], amp_interps[n](plot_times), 'k')
plt.plot(gain_times - gain_times[0],
np.abs(ant_gains[n]),
'o',
label='ant%d' % (n + 1))
plt.xlabel('Time since start (seconds)')
plt.title('Gain amplitude')
plt.legend(loc='upper left')
plt.subplot(122)
for n in range(4):
plt.plot(
plot_times - gain_times[0],
katpoint.rad2deg(scape.stats.angle_wrap(phase_interps[n](plot_times))),
'k')
plt.plot(gain_times - gain_times[0],
katpoint.rad2deg(np.angle(ant_gains[n])),
'o',
label='ant%d' % (n + 1))
plt.xlabel('Time since start (seconds)')
plt.title('Gain phase (degrees)')
plt.legend(loc='lower left')
# Apply both bandpass and gain calibration to cal source visibilities
final_cal_vis_samples = [vis.copy() for vis in cal_vis_samples]
for vis, timestamps in zip(final_cal_vis_samples, cal_timestamps):
# Interpolate antenna gains to timestamps of visibilities
interp_ant_gains = np.zeros((4, len(timestamps)), dtype=np.complex64)
for n in range(4):
def calc_pointing_offsets(session, beams, target, middle_time, temperature,
pressure, humidity):
"""Calculate pointing offsets per receptor based on primary beam fits.
Parameters
----------
session : :class:`katcorelib.observe.CaptureSession` object
The active capture session
beams : dict mapping receptor name to list of :class:`BeamPatternFit`
Fitted primary beams, per receptor and per frequency chunk
target : :class:`katpoint.Target` object
The target on which offset pointings were done
middle_time : float
Unix timestamp at the middle of sequence of offset pointings, used to
find the mean location of a moving target (and reference for weather)
temperature, pressure, humidity : float
Atmospheric conditions at middle time, used for refraction correction
Returns
-------
pointing_offsets : dict mapping receptor name to offset data (10 floats)
Pointing offsets per receptor in degrees, stored as a sequence of
- requested (az, el) after refraction (input to the pointing model),
- full (az, el) offset, including contributions of existing pointing
model, any existing adjustment and newly fitted adjustment
(useful for fitting new pointing models as it is independent),
- full (az, el) adjustment on top of existing pointing model,
replacing any existing adjustment (useful for reference pointing),
- relative (az, el) adjustment on top of existing pointing model and
adjustment (useful for verifying reference pointing), and
- rough uncertainty (standard deviation) of (az, el) adjustment.
"""
pointing_offsets = {}
# Iterate over receptors
for ant in sorted(session.observers):
beams_freq = beams.get(ant.name, [])
beams_freq = [b for b in beams_freq if b is not None and b.is_valid]
if not beams_freq:
user_logger.debug("%s had no valid primary beam fitted", ant.name)
continue
offsets_freq = np.array([b.center for b in beams_freq])
offsets_freq_std = np.array([b.std_center for b in beams_freq])
weights_freq = 1. / offsets_freq_std**2
# Do weighted average of offsets over frequency chunks
results = np.average(offsets_freq,
axis=0,
weights=weights_freq,
returned=True)
pointing_offset = results[0]
pointing_offset_std = np.sqrt(1. / results[1])
user_logger.debug("%s x=%+7.2f'+-%.2f\" y=%+7.2f'+-%.2f\"", ant.name,
pointing_offset[0] * 60,
pointing_offset_std[0] * 3600,
pointing_offset[1] * 60,
pointing_offset_std[1] * 3600)
# Get existing pointing adjustment
receptor = getattr(session.kat, ant.name)
az_adjust = receptor.sensor.pos_adjust_pointm_azim.get_value()
el_adjust = receptor.sensor.pos_adjust_pointm_elev.get_value()
existing_adjustment = deg2rad(np.array((az_adjust, el_adjust)))
# Start with requested (az, el) coordinates, as they apply
# at the middle time for a moving target
requested_azel = target.azel(timestamp=middle_time, antenna=ant)
# Correct for refraction, which becomes the requested value
# at input of pointing model
rc = RefractionCorrection()
def refract(az, el): # noqa: E306, E301
"""Apply refraction correction as at the middle of scan."""
return [az, rc.apply(el, temperature, pressure, humidity)]
refracted_azel = np.array(refract(*requested_azel))
# More stages that apply existing pointing model and/or adjustment
pointed_azel = np.array(ant.pointing_model.apply(*refracted_azel))
adjusted_azel = pointed_azel + existing_adjustment
# Convert fitted offset back to spherical (az, el) coordinates
pointing_offset = deg2rad(np.array(pointing_offset))
beam_center_azel = target.plane_to_sphere(*pointing_offset,
timestamp=middle_time,
antenna=ant)
# Now correct the measured (az, el) for refraction and then apply the
# existing pointing model and adjustment to get a "raw" measured
# (az, el) at the output of the pointing model stage
beam_center_azel = refract(*beam_center_azel)
beam_center_azel = ant.pointing_model.apply(*beam_center_azel)
beam_center_azel = np.array(beam_center_azel) + existing_adjustment
# Make sure the offset is a small angle around 0 degrees
full_offset_azel = wrap_angle(beam_center_azel - refracted_azel)
full_adjust_azel = wrap_angle(beam_center_azel - pointed_azel)
relative_adjust_azel = wrap_angle(beam_center_azel - adjusted_azel)
# Cheap 'n' cheerful way to convert cross-el uncertainty to azim form
offset_azel_std = pointing_offset_std / \
np.array([np.cos(refracted_azel[1]), 1.])
# We store all variants of the pointing offset since we have it all
# at our fingertips here
point_data = np.r_[rad2deg(refracted_azel),
rad2deg(full_offset_azel),
rad2deg(full_adjust_azel),
rad2deg(relative_adjust_azel), offset_azel_std]
pointing_offsets[ant.name] = point_data
return pointing_offsets
sources = katpoint.Catalogue(add_specials=False)
user_logger.info('Performing flux calibration')
ra, dec = target.apparent_radec(timestamp=timenow)
targetName = target.name.replace(" ", "")
print targetName
target.name = targetName + '_O'
sources.add(target)
if opts.cal == 'fluxN':
timenow = katpoint.Timestamp()
sources = katpoint.Catalogue(add_specials=False)
user_logger.info('Performing flux calibration')
ra, dec = target.apparent_radec(timestamp=timenow)
print target
print "ra %f ,dec %f" % (katpoint.rad2deg(ra),
katpoint.rad2deg(dec))
dec2 = dec + katpoint.deg2rad(1)
print dec2, dec
decS = dec - katpoint.deg2rad(1)
targetName = target.name.replace(" ", "")
print targetName
print "newra %f newdec %f" % (katpoint.rad2deg(ra),
katpoint.rad2deg(dec))
Ntarget = katpoint.construct_radec_target(ra, dec2)
Ntarget.antenna = bf_ants
Ntarget.name = targetName + '_N'
sources.add(Ntarget)
if opts.cal == 'fluxS':
timenow = katpoint.Timestamp()
connectionstyle='arc3,rad=-0.2',
fc='w',
zorder=4))
# Set up figure with buttons
plt.ion()
fig = plt.figure(1, figsize=(15, 10))
fig.clear()
# Store highlighted target index on figure object
fig.highlighted_target = 0
# Axes to contain detail residual plots - initialise plots with old residuals
ax = fig.add_axes([0.27, 0.74, 0.2, 0.2])
ax.axhline(0, color='k', zorder=0)
plot_data_and_tooltip(ax, rad2deg(az), rad2deg(old.residual_xel) * 60.)
ax.axis([-180., 180., -resid_lim, resid_lim])
ax.set_xticks([])
ax.yaxis.set_ticks_position('right')
ax.yaxis.set_major_formatter(mpl.ticker.FuncFormatter(arcmin_formatter))
ax.set_ylabel('Cross-EL offset')
ax.set_title('RESIDUALS')
ax = fig.add_axes([0.27, 0.54, 0.2, 0.2])
ax.axhline(0, color='k', zorder=0)
plot_data_and_tooltip(ax, rad2deg(az), rad2deg(old.residual_el) * 60.)
ax.axis([-180., 180., -resid_lim, resid_lim])
ax.set_xlabel('Azimuth (deg)')
ax.yaxis.set_ticks_position('right')
ax.yaxis.set_major_formatter(mpl.ticker.FuncFormatter(arcmin_formatter))
ax.set_ylabel('EL offset')
ra, dec = [], []
for scan in d.scans:
if scan.baseline:
ra_dec = np.array([
katpoint.construct_azel_target(az, el).radec(t, d.antenna)
for az, el, t in zip(scan.pointing['az'], scan.pointing['el'],
scan.timestamps)
])
x, y = target.sphere_to_plane(ra_dec[:, 0],
ra_dec[:, 1],
scan.timestamps,
coord_system='radec')
ra.append(x)
dec.append(y)
# Remove pointing offset (order of a few arcminutes)
ra = katpoint.rad2deg(np.hstack(ra) - d.compscans[0].beam.center[0])
dec = katpoint.rad2deg(np.hstack(dec) - d.compscans[0].beam.center[1])
power = np.hstack([
scan.pol('I').squeeze() - scan.baseline(scan.timestamps)
for scan in d.scans if scan.baseline
])
power = np.abs(power)
# Grid the raster scan to projected plane
min_num_pixels = 201
interp = scape.fitting.Delaunay2DScatterFit(default_val=0.0, jitter=True)
interp.fit([ra, dec], power)
ra_range, dec_range = ra.max() - ra.min(), dec.max() - dec.min()
# Use a square pixel size in projected plane
pixel_size = min(ra_range, dec_range) / min_num_pixels
grid_ra = np.arange(ra.min(), ra.max(), pixel_size)
if compscan.beam is not None and d.data_unit == 'K':
gain_hh = compscan.beam.height / average_flux
baseline_hh = compscan.baseline_height()
if (ant.name + 'V') in h5.inputs:
d.fit_beams_and_baselines(pol='VV', circular_beam=False)
if compscan.beam is not None and d.data_unit == 'K':
gain_vv = compscan.beam.height / average_flux
baseline_vv = compscan.baseline_height()
d.fit_beams_and_baselines(pol='I', circular_beam=True)
beam = compscan.beam
# Obtain middle timestamp of compound scan, where all pointing calculations are done
compscan_times = np.hstack([scan.timestamps for scan in compscan.scans])
middle_time = np.median(compscan_times, axis=None)
# Start with requested (az, el) coordinates, as they apply at the middle time for a moving target
requested_azel = compscan.target.azel(middle_time)
requested_azel = katpoint.rad2deg(np.array(requested_azel))
# The offset is very simplistic and doesn't take into account refraction (see a_p_s_s for more correct way)
offset_azel = katpoint.rad2deg(np.array(beam.center)) if beam else np.zeros(2)
user_logger.info("Antenna %s" % (ant.name,))
user_logger.info("------------")
user_logger.info("Target = '%s', azel around (%.1f, %.1f) deg" %
(compscan.target.name, requested_azel[0], requested_azel[1]))
if beam is None:
user_logger.info("No total power beam found")
else:
user_logger.info("Beam height = %g %s" % (beam.height, d.data_unit))
user_logger.info("Beamwidth = %.1f' (expected %.1f')" %
(60 * katpoint.rad2deg(beam.width), 60 * katpoint.rad2deg(beam.expected_width)))
user_logger.info("Beam offset = (%.1f', %.1f') (expected (0', 0'))" %
(60 * offset_azel[0], 60 * offset_azel[1]))
duration=opts.track_duration,
announce=False) # Set the target & mode = point
## 1) Track ephem target behind the moon
user_logger.info("Sleeping for 2 minutes")
time.sleep(120)
user_logger.info("Setting to Ephem target")
observer.date = ephem.now()
moon = ephem.Moon(observer)
moon.compute(observer)
target = katpoint.construct_radec_target(moon.ra, moon.dec)
session.label('track')
user_logger.info(
"Initiating %g-second track on ephem target (%.2f, %.2f)" % (
opts.track_duration,
katpoint.rad2deg(float(moon.ra)),
katpoint.rad2deg(float(moon.dec)),
))
session.set_target(target) # Set the target
session.track(target, duration=opts.track_duration,
announce=False) # Set the target & mode = point
## 2) Track off target
user_logger.info("Sleeping for 5 minutes")
time.sleep(300)
target = katpoint.construct_radec_target(moon.ra, moon.dec)
session.label('track')
user_logger.info(
"Second set of %g-second track on ephem target (%.2f, %.2f)" %
(
opts.track_duration,
ax.ann = ax.annotate('', xy=(0., 0.), xycoords='data', xytext=(32, 32), textcoords='offset points', size=14,
va='bottom', ha='center', bbox=dict(boxstyle='round4', fc='w'), visible=False, zorder=5,
arrowprops=dict(arrowstyle='-|>', shrinkB=10, connectionstyle='arc3,rad=-0.2',
fc='w', zorder=4))
# Set up figure with buttons
plt.ion()
fig = plt.figure(1, figsize=(15, 10))
fig.clear()
# Store highlighted target index on figure object
fig.highlighted_target = 0
# Axes to contain detail residual plots - initialise plots with old residuals
ax = fig.add_axes([0.27, 0.74, 0.2, 0.2])
ax.axhline(0, color='k', zorder=0)
plot_data_and_tooltip(ax, rad2deg(az), rad2deg(old.residual_xel) * 60.)
ax.axis([-180., 180., -resid_lim, resid_lim])
ax.set_xticks([])
ax.yaxis.set_ticks_position('right')
ax.yaxis.set_major_formatter(mpl.ticker.FuncFormatter(arcmin_formatter))
ax.set_ylabel('Cross-EL offset')
ax.set_title('RESIDUALS')
ax = fig.add_axes([0.27, 0.54, 0.2, 0.2])
ax.axhline(0, color='k', zorder=0)
plot_data_and_tooltip(ax, rad2deg(az), rad2deg(old.residual_el) * 60.)
ax.axis([-180., 180., -resid_lim, resid_lim])
ax.set_xlabel('Azimuth (deg)')
ax.yaxis.set_ticks_position('right')
ax.yaxis.set_major_formatter(mpl.ticker.FuncFormatter(arcmin_formatter))
ax.set_ylabel('EL offset')
def referencemetrics(ant, data, num_samples_limit=1, power_sample_limit=0):
"""Determine and sky RMS from the antenna pointing model."""
"""On the calculation of all-sky RMS
Assume the el and cross-el errors have zero mean, are distributed normally, and are uncorrelated
They are therefore described by a 2-dimensional circular Gaussian pdf with zero mean and *per-component*
standard deviation of sigma
The absolute sky error (== Euclidean length of 2-dim error vector) then has a Rayleigh distribution
The RMS sky error has a mean value of sqrt(2) * sigma, since each squared error term is the sum of
two squared Gaussian random values, each with an expected value of sigma^2.
e.g. sky_rms = np.sqrt(np.mean((abs_sky_error-abs_sky_error.mean()) ** 2))
A more robust estimate of the RMS sky error is obtained via the median of the Rayleigh distribution,
which is sigma * sqrt(log(4)) -> convert this to the RMS sky error = sqrt(2) * sigma
e.g. robust_sky_rms = np.median(np.sqrt((abs_sky_error-abs_sky_error.mean())**2)) * np.sqrt(2. / np.log(4.))
"""
#print type(data.shape[0] ), type(num_samples_limit)
beam = data['beam_height_I'].mean()
good_beam = (data['beam_height_I'] > beam * .8) * (
data['beam_height_I'] < beam * 1.2) * (data['beam_height_I'] >
power_sample_limit)
data = data[good_beam]
if data.shape[0] > 0 and not np.all(good_beam):
print("bad scan", data['target'][0])
if data.shape[0] >= num_samples_limit and (
data['timestamp'][-1] -
data['timestamp'][0]) < 2000: # check all fitted Ipks are valid
condition_str = ['ideal', 'optimal', 'normal', 'other']
condition = 3
text = [
] #azimuth, elevation, delta_azimuth, delta_azimuth_std, delta_elevation, delta_elevation_std,
measured_delta_xel = data['delta_azimuth'] * np.cos(
data['elevation']) # scale due to sky shape
abs_sky_error = measured_delta_xel
model_delta_az, model_delta_el = ant.pointing_model.offset(
data['azimuth'], data['elevation'])
residual_az = data['delta_azimuth'] - model_delta_az
residual_el = data['delta_elevation'] - model_delta_el
residual_xel = residual_az * np.cos(data['elevation'])
delta_xel_std = data['delta_azimuth_std'] * np.cos(data['elevation'])
abs_sky_delta_std = rad2deg(
np.sqrt(delta_xel_std**2 +
data['delta_azimuth_std']**2)) * 3600 # make arc seconds
#for i,val in enumerate(data):
# print ("Test Target: '%s' fit accuracy %.3f\" "%(data['target'][i],abs_sky_delta_std[i]))
abs_sky_error = rad2deg(np.sqrt((residual_xel)**2 +
(residual_el)**2)) * 3600
condition = get_condition(data)
rms = np.std(abs_sky_error)
robust = np.median(
np.abs(abs_sky_error - abs_sky_error.mean())) * np.sqrt(
2. / np.log(4.))
text.append(
"Dataset:%s Test Target: '%s' Reference RMS = %.3f\" {fit-accuracy=%.3f\"} (robust %.3f\") (N=%i Data Points) ['%s']"
% (data['dataset'][0], data['target'][0], rms,
np.mean(abs_sky_delta_std), robust, data.shape[0],
condition_str[condition]))
output_data = data[0].copy() # make a copy of the rec array
for i, x in enumerate(data[0]): # make an average of data
if x.dtype.kind == 'f': # average floats
output_data[i] = data.field(i).mean()
else:
output_data[i] = data.field(i)[0]
sun = Target('Sun,special')
source = Target(
'%s,azel, %f,%f' %
(output_data['target'], np.degrees(
output_data['azimuth']), np.degrees(output_data['elevation'])))
sun_sep = np.degrees(
source.separation(sun,
timestamp=output_data['timestamp'],
antenna=ant))
output_data = recfunctions.append_fields(output_data,
'sun_sep',
np.array([sun_sep]),
dtypes=np.float,
usemask=False,
asrecarray=True)
output_data = recfunctions.append_fields(output_data,
'condition',
np.array([condition]),
dtypes=np.float,
usemask=False,
asrecarray=True)
output_data = recfunctions.append_fields(output_data,
'rms',
np.array([rms]),
dtypes=np.float,
usemask=False,
asrecarray=True)
output_data = recfunctions.append_fields(output_data,
'robust',
np.array([robust]),
dtypes=np.float,
usemask=False,
asrecarray=True)
output_data = recfunctions.append_fields(output_data,
'N',
np.array([data.shape[0]]),
dtypes=np.float,
usemask=False,
asrecarray=True)
#### Debugging
#residual_az = data['delta_azimuth'] - model_delta_az
#residual_el = data['delta_elevation'] - model_delta_el
#residual_xel = residual_az * np.cos(data['elevation'])
output_data = recfunctions.append_fields(
output_data,
'residual_az',
np.array([rad2deg(residual_az.std()) * 3600]),
dtypes=np.float,
usemask=False,
asrecarray=True)
output_data = recfunctions.append_fields(
output_data,
'residual_el',
np.array([rad2deg(residual_el.std()) * 3600]),
dtypes=np.float,
usemask=False,
asrecarray=True)
output_data = recfunctions.append_fields(
output_data,
'residual_xel',
np.array([rad2deg(residual_xel.std()) * 3600]),
dtypes=np.float,
usemask=False,
asrecarray=True)
#print "%10s %i %3.1f, %s"%(data['target'][0],data['timestamp'][-1] - data['timestamp'][0], rms, str(np.degrees(data['delta_elevation']-data['delta_elevation'].mean())*3600) )
output_data['wind_speed'] = data['wind_speed'].max()
return text, output_data
else:
return None, None
def reduce_compscan_inf(h5,
channel_mask=None,
chunks=16,
return_raw=False,
use_weights=False,
compscan_index=None,
debug=False):
"""Break the band up into chunks"""
chunk_size = chunks
rfi_static_flags = np.tile(False, h5.shape[0])
if len(channel_mask) > 0:
pickle_file = open(channel_mask, "rb")
rfi_static_flags = pickle.load(pickle_file)
pickle_file.close()
gains_p = {}
stdv = {}
calibrated = False # placeholder for calibration
h5.select(compscans=compscan_index)
a = []
if len(h5.target_indices) > 1:
print("Warning multiple targets in the compscan")
for scan in h5.scans():
a.append(h5.target_indices[0])
target = h5.catalogue.targets[np.median(a).astype(
np.int)] # Majority Track
compscan_index = h5.compscan_indices[0]
#h5.select(targets=target,compscans=h5.compscan_indices[0]) # Majority Track in compscan
if not return_raw: # Calculate average target flux over entire band
flux_spectrum = h5.catalogue.targets[
h5.target_indices[0]].flux_density(h5.freqs) # include flags
average_flux = np.mean(
[flux for flux in flux_spectrum if not np.isnan(flux)])
temperature = np.mean(h5.temperature)
pressure = np.mean(h5.pressure)
humidity = np.mean(h5.humidity)
wind_speed = np.mean(h5.wind_speed)
wind_direction = np.degrees(
np.angle(np.mean(np.exp(
1j * np.radians(h5.wind_direction))))) # Vector Mean
sun = katpoint.Target('Sun, special')
# Calculate pointing offset
# Obtain middle timestamp of compound scan, where all pointing calculations are done
middle_time = np.median(h5.timestamps[:], axis=None)
# Start with requested (az, el) coordinates, as they apply at the middle time for a moving target
requested_azel = target.azel(middle_time)
# Correct for refraction, which becomes the requested value at input of pointing model
rc = katpoint.RefractionCorrection()
requested_azel = [
requested_azel[0],
rc.apply(requested_azel[1], temperature, pressure, humidity)
]
requested_azel = katpoint.rad2deg(np.array(requested_azel))
gaussian_centre = np.zeros((chunk_size * 2, 2, len(h5.ants)))
gaussian_centre_std = np.zeros((chunk_size * 2, 2, len(h5.ants)))
gaussian_width = np.zeros((chunk_size * 2, 2, len(h5.ants)))
gaussian_width_std = np.zeros((chunk_size * 2, 2, len(h5.ants)))
gaussian_height = np.zeros((chunk_size * 2, len(h5.ants)))
gaussian_height_std = np.zeros((chunk_size * 2, len(h5.ants)))
if debug: #debug_text
debug_text = []
line = []
line.append("#AntennaPol")
line.append("Target")
line.append("Freq(MHz)") #MHz
line.append("Centre Az")
line.append("Centre El")
line.append("Centre Az Std")
line.append("Centre El Std")
line.append("Centre Az Width")
line.append("Centre El Width")
line.append("Centre Az Width Std")
line.append("Centre El Width Std")
line.append("Height")
line.append("Height Std")
debug_text.append(','.join(line))
pols = ["H", "V"] # Put in logic for Intensity
for i, pol in enumerate(pols):
gains_p[pol] = []
pos = []
stdv[pol] = []
h5.select(pol=pol,
corrprods='cross',
ants=h5.antlist,
targets=target,
compscans=compscan_index)
h5.bls_lookup = calprocs.get_bls_lookup(h5.antlist, h5.corr_products)
for scan in h5.scans():
if scan[1] != 'track': continue
valid_index = activity(h5, state='track')
data = h5.vis[valid_index]
if data.shape[0] > 0: # need at least one data point
#g0 = np.ones(len(h5.ants),np.complex)
if use_weights:
weights = h5.weights[valid_index].mean(axis=0)
else:
weights = np.ones(data.shape[1:]).astype(np.float)
gains_p[pol].append(
calprocs.g_fit(data[:].mean(axis=0),
weights,
h5.bls_lookup,
refant=0))
stdv[pol].append(
np.ones(
(data.shape[0], data.shape[1],
len(h5.ants))).sum(axis=0)) #number of data points
# Get coords in (x(time,ants),y(time,ants) coords)
pos.append([
h5.target_x[valid_index, :].mean(axis=0),
h5.target_y[valid_index, :].mean(axis=0)
])
for ant in range(len(h5.ants)):
for chunk in range(chunks):
if np.array(pos).shape[
0] > 4: # Make sure there is enough data for a fit
freq = slice(chunk * (h5.shape[1] // chunks),
(chunk + 1) * (h5.shape[1] // chunks))
rfi = ~rfi_static_flags[freq]
fitobj = fit.GaussianFit(
np.array(pos)[:, :, ant].mean(axis=0), [1., 1.], 1)
x = np.column_stack(
(np.array(pos)[:, 0, ant], np.array(pos)[:, 1, ant]))
y = np.abs(
np.array(gains_p[pol])[:, freq, :][:, rfi,
ant]).mean(axis=1)
y_err = 1. / np.sqrt(
np.array(stdv[pol])[:, freq, :][:, rfi,
ant].sum(axis=1))
gaussian = fitobj.fit(x.T, y, y_err)
#Fitted beam center is in (x, y) coordinates, in projection centred on target
snr = np.abs(np.r_[gaussian.std / gaussian.std_std])
valid_fit = np.all(
np.isfinite(np.r_[gaussian.mean, gaussian.std_mean,
gaussian.std, gaussian.std_std,
gaussian.height, gaussian.std_height,
snr]))
theta = np.sqrt((gaussian.mean**2).sum(
)) # this is to see if the co-ord is out of range
#The valid fit is needed because I have no way of working out if the gain solution was ok.
if not valid_fit or np.any(
theta > np.pi
): # the checks to see if the fit is ok
gaussian_centre[chunk + i * chunk_size, :,
ant] = np.nan
gaussian_centre_std[chunk + i * chunk_size, :,
ant] = np.nan
gaussian_width[chunk + i * chunk_size, :, ant] = np.nan
gaussian_width_std[chunk + i * chunk_size, :,
ant] = np.nan
gaussian_height[chunk + i * chunk_size, ant] = np.nan
gaussian_height_std[chunk + i * chunk_size,
ant] = np.nan
else:
# Convert this offset back to spherical (az, el) coordinates
beam_center_azel = target.plane_to_sphere(
np.radians(gaussian.mean[0]),
np.radians(gaussian.mean[1]), middle_time)
# Now correct the measured (az, el) for refraction and then apply the old pointing model
# to get a "raw" measured (az, el) at the output of the pointing model
beam_center_azel = [
beam_center_azel[0],
rc.apply(beam_center_azel[1], temperature,
pressure, humidity)
]
beam_center_azel = h5.ants[ant].pointing_model.apply(
*beam_center_azel)
beam_center_azel = np.degrees(
np.array(beam_center_azel))
gaussian_centre[chunk + i * chunk_size, :,
ant] = beam_center_azel
gaussian_centre_std[chunk + i * chunk_size, :,
ant] = gaussian.std_mean
gaussian_width[chunk + i * chunk_size, :,
ant] = gaussian.std
gaussian_width_std[chunk + i * chunk_size, :,
ant] = gaussian.std_std
gaussian_height[chunk + i * chunk_size,
ant] = gaussian.height
gaussian_height_std[chunk + i * chunk_size,
ant] = gaussian.std_height
if return_raw:
return np.r_[gaussian_centre, gaussian_centre_std, gaussian_width,
gaussian_width_std, gaussian_height, gaussian_height_std]
else:
ant_pointing = {}
pols = ["HH", "VV", 'I']
pol_ind = {}
pol_ind['HH'] = np.arange(0.0 * chunk_size,
1.0 * chunk_size,
dtype=int)
pol_ind['VV'] = np.arange(1.0 * chunk_size,
2.0 * chunk_size,
dtype=int)
pol_ind['I'] = np.arange(0.0 * chunk_size, 2.0 * chunk_size, dtype=int)
for ant in range(len(h5.ants)):
h_pol = ~np.isnan(
gaussian_centre[pol_ind['HH'], :, ant]) & ~np.isnan(
1. / gaussian_centre_std[pol_ind['HH'], :, ant])
v_pol = ~np.isnan(
gaussian_centre[pol_ind['VV'], :, ant]) & ~np.isnan(
1. / gaussian_centre_std[pol_ind['VV'], :, ant])
valid_solutions = np.count_nonzero(
h_pol & v_pol
) # Note this is twice the number of solutions because of the Az & El parts
print("%i valid solutions out of %s for %s on %s at %s " %
(valid_solutions // 2, chunks, h5.ants[ant].name,
target.name, str(katpoint.Timestamp(middle_time))))
if debug: #debug_text
for pol_i, pol in enumerate(["H", "V"]):
for chunk in range(chunks * pol_i, chunks * (pol_i + 1)):
line = []
freq = h5.channel_freqs[slice(
chunk * (h5.shape[1] // chunks),
(chunk + 1) * (h5.shape[1] // chunks))].mean()
line.append(h5.ants[ant].name + pol)
line.append(target.name)
line.append(str(freq / 1e6)) #MHz
line.append(str(gaussian_centre[chunk, 0, ant]))
line.append(str(gaussian_centre[chunk, 1, ant]))
line.append(str(gaussian_centre_std[chunk, 0, ant]))
line.append(str(gaussian_centre_std[chunk, 1, ant]))
line.append(str(gaussian_width[chunk, 0, ant]))
line.append(str(gaussian_width[chunk, 1, ant]))
line.append(str(gaussian_width_std[chunk, 0, ant]))
line.append(str(gaussian_width_std[chunk, 1, ant]))
line.append(str(gaussian_height[chunk, ant]))
line.append(str(gaussian_height_std[chunk, ant]))
debug_text.append(','.join(line))
if valid_solutions // 2 > 0: # a bit overboard
name = h5.ants[ant].name
ant_pointing[name] = {}
ant_pointing[name]["antenna"] = h5.ants[ant].name
ant_pointing[name]["valid_solutions"] = valid_solutions
ant_pointing[name]["dataset"] = h5.name.split('/')[-1].split(
'.')[0]
ant_pointing[name]["target"] = target.name
ant_pointing[name]["timestamp_ut"] = str(
katpoint.Timestamp(middle_time))
ant_pointing[name][
"data_unit"] = 'Jy' if calibrated else 'counts'
ant_pointing[name]["frequency"] = h5.freqs.mean()
ant_pointing[name]["flux"] = average_flux
ant_pointing[name]["temperature"] = temperature
ant_pointing[name]["pressure"] = pressure
ant_pointing[name]["humidity"] = humidity
ant_pointing[name]["wind_speed"] = wind_speed
ant_pointing[name]["wind_direction"] = wind_direction
# work out the sun's angle
sun_azel = katpoint.rad2deg(
np.array(sun.azel(middle_time, antenna=h5.ants[ant])))
ant_pointing[name]["sun_az"] = sun_azel.tolist()[0]
ant_pointing[name]["sun_el"] = sun_azel.tolist()[1]
ant_pointing[name]["timestamp"] = middle_time.astype(int)
#Work out the Target position and the requested position
# Start with requested (az, el) coordinates, as they apply at the middle time for a moving target
requested_azel = target.azel(middle_time, antenna=h5.ants[ant])
# Correct for refraction, which becomes the requested value at input of pointing model
rc = katpoint.RefractionCorrection()
requested_azel = [
requested_azel[0],
rc.apply(requested_azel[1], temperature, pressure,
humidity)
]
requested_azel = katpoint.rad2deg(np.array(requested_azel))
target_azel = katpoint.rad2deg(
np.array(target.azel(middle_time, antenna=h5.ants[ant])))
ant_pointing[name]["azimuth"] = target_azel.tolist()[0]
ant_pointing[name]["elevation"] = target_azel.tolist()[1]
azel_beam = w_average(
gaussian_centre[pol_ind["I"], :, ant],
axis=0,
weights=1. / gaussian_centre_std[pol_ind["I"], :, ant]**2)
# Make sure the offset is a small angle around 0 degrees
offset_azel = katpoint.wrap_angle(azel_beam - requested_azel,
360.)
ant_pointing[name]["delta_azimuth"] = offset_azel.tolist()[0]
ant_pointing[name]["delta_elevation"] = offset_azel.tolist()[1]
ant_pointing[name]["delta_elevation_std"] = 0.0 #calc
ant_pointing[name]["delta_azimuth_std"] = 0.0 #calc
for pol in pol_ind:
ant_pointing[name]["beam_height_%s" % (pol)] = w_average(
gaussian_height[pol_ind[pol], ant],
axis=0,
weights=1. / gaussian_height_std[pol_ind[pol], ant]**2)
ant_pointing[name]["beam_height_%s_std" % (pol)] = np.sqrt(
np.nansum(1. /
gaussian_height_std[pol_ind[pol], ant]**2))
ant_pointing[name]["beam_width_%s" % (pol)] = w_average(
gaussian_width[pol_ind[pol], :, ant],
axis=0,
weights=1. /
gaussian_width_std[pol_ind[pol], :, ant]**2).mean()
ant_pointing[name]["beam_width_%s_std" % (pol)] = np.sqrt(
np.nansum(1. /
gaussian_width_std[pol_ind[pol], :, ant]**2))
ant_pointing[name]["baseline_height_%s" % (pol)] = 0.0
ant_pointing[name]["baseline_height_%s_std" % (pol)] = 0.0
ant_pointing[name][
"refined_%s" %
(pol)] = 5.0 # I don't know what this means
ant_pointing[name]["azimuth_%s" % (pol)] = w_average(
gaussian_centre[pol_ind[pol], 0, ant],
axis=0,
weights=1. /
gaussian_centre_std[pol_ind[pol], 0, ant]**2)
ant_pointing[name]["elevation_%s" % (pol)] = w_average(
gaussian_centre[pol_ind[pol], 1, ant],
axis=0,
weights=1. /
gaussian_centre_std[pol_ind[pol], 1, ant]**2)
ant_pointing[name]["azimuth_%s_std" % (pol)] = np.sqrt(
np.nansum(
1. / gaussian_centre_std[pol_ind[pol], 0, ant]**2))
ant_pointing[name]["elevation_%s_std" % (pol)] = np.sqrt(
np.nansum(
1. / gaussian_centre_std[pol_ind[pol], 1, ant]**2))
else:
print("No (%i) solutions for %s on %s at %s " %
(valid_solutions, h5.ants[ant].name, target.name,
str(katpoint.Timestamp(middle_time))))
if debug: #debug_text
debug_text.append('')
base = "%s_%s" % (h5.name.split('/')[-1].split('.')[0],
"interferometric_pointing_DEBUG")
g = file('%s:Scan%i:%s' % (base, compscan_index, target.name), 'w')
g.write("\n".join(debug_text))
g.close()
return ant_pointing
def update(fig):
"""Fit new pointing model and update plots."""
# Perform early redraw to improve interactivity of clicks (which typically change state of target dots)
# Target state: 0 = flagged, 1 = unflagged, 2 = highlighted
target_state = keep * ((target_index == fig.highlighted_target) + 1)
# Specify colours of flagged, unflagged and highlighted dots, respectively, as RGBA tuples
dot_colors = np.choose(
target_state,
np.atleast_3d(np.vstack([(1, 1, 1, 1), (0, 0, 1, 1), (1, 0, 0, 1)]))).T
for ax in fig.axes[:7]:
ax.dots.set_facecolors(dot_colors)
fig.canvas.draw()
# Fit new pointing model and update results
params, sigma_params = new_model.fit(az[keep], el[keep],
measured_delta_az[keep],
measured_delta_el[keep],
std_delta_az[keep],
std_delta_el[keep], enabled_params)
new.update(new_model)
# Update rest of figure
fig.texts[3].set_text("$\chi^2$ = %.1f" % new.chi2)
fig.texts[4].set_text("all sky rms = %.3f' (robust %.3f')" %
(new.sky_rms, new.robust_sky_rms))
new.metrics(target_index == fig.highlighted_target)
fig.texts[5].set_text("target sky rms = %.3f' (robust %.3f')" %
(new.sky_rms, new.robust_sky_rms))
new.metrics(keep)
fig.texts[-1].set_text(unique_targets[fig.highlighted_target])
# Update model parameter strings
for p, param in enumerate(display_params):
fig.texts[2 * p + 6].set_text(
new_model.param_str(param +
1, '%.3e') if enabled_params[param] else '')
# HACK to convert sigmas to arcminutes, but not for P9 and P12 (which are scale factors)
# This functionality should really reside inside the PointingModel class
std_param = rad2deg(sigma_params[param]) * 60. if param not in [
8, 11
] else sigma_params[param]
std_param_str = ("%.2f'" %
std_param) if param not in [8, 11] else ("%.0e" %
std_param)
fig.texts[2 * p + 7].set_text(
std_param_str if enabled_params[param] and opts.use_stats else '')
# Turn parameter string bold if it changed significantly from old value
if np.abs(params[param] -
old_model.params[param]) > 3.0 * sigma_params[param]:
fig.texts[2 * p + 6].set_weight('bold')
fig.texts[2 * p + 7].set_weight('bold')
else:
fig.texts[2 * p + 6].set_weight('normal')
fig.texts[2 * p + 7].set_weight('normal')
daz_az, del_az, daz_el, del_el, quiver, before, after = fig.axes[:7]
# Update quiver plot
quiver_scale = 0.1 * fig.quiver_scale_slider.val * np.pi / 6 / deg2rad(
old.robust_sky_rms / 60.)
quiver.quiv.set_segments(
quiver_segments(new.residual_az, new.residual_el, quiver_scale))
quiver.quiv.set_color(
np.choose(
keep,
np.atleast_3d(np.vstack([(0.3, 0.3, 0.3, 0.2),
(0.3, 0.3, 0.3, 1)]))).T)
# Update residual plots
daz_az.dots.set_offsets(np.c_[rad2deg(az),
rad2deg(new.residual_xel) * 60.])
del_az.dots.set_offsets(np.c_[rad2deg(az), rad2deg(new.residual_el) * 60.])
daz_el.dots.set_offsets(np.c_[rad2deg(el),
rad2deg(new.residual_xel) * 60.])
del_el.dots.set_offsets(np.c_[rad2deg(el), rad2deg(new.residual_el) * 60.])
after.dots.set_offsets(np.c_[np.arctan2(new.residual_el, new.residual_xel),
new.abs_sky_error])
resid_lim = 1.2 * max(new.abs_sky_error.max(), old.abs_sky_error.max())
daz_az.set_ylim(-resid_lim, resid_lim)
del_az.set_ylim(-resid_lim, resid_lim)
daz_el.set_ylim(-resid_lim, resid_lim)
del_el.set_ylim(-resid_lim, resid_lim)
before.set_ylim(0, resid_lim)
after.set_ylim(0, resid_lim)
# Redraw the figure
fig.canvas.draw()
err_power = np.dot(resid, resid)
print "Iteration %d: residual = %.2f, beam height = %.3f, width = %s, inner region = %d/%d" % \
(n, (prev_err_power - err_power) / err_power, beam.height, scape.beam_baseline.width_string(beam.width), \
np.sum(~outer), len(outer))
if (err_power == 0.0) or (prev_err_power - err_power) / err_power < 1e-5:
break
prev_err_power = err_power + 0.0
##### PLOT RESULTS #####
start_time = timestamps.min()
t = timestamps - start_time
plt.plot(t, power, 'b')
plt.plot(t, baseline(target_coords), 'g')
plt.plot(t, beam(target_coords) + baseline(target_coords), 'g')
plt.xlabel('Time in seconds since %s' % katpoint.Timestamp(start_time).local())
plt.ylabel('Power')
plt.title('Quick beam and baseline fit')
print "Beam offset is (%f, %f) deg" % (katpoint.rad2deg(
beam.center[0]), katpoint.rad2deg(beam.center[1]))
def set_delay(time_after_now, delay=None):
t = katpoint.Timestamp() + time_after_now
if delay is None:
delay = cable2 - cable1 + tgt.geometric_delay(ant2, t, ant1)[0]
print delay
roach.req.poco_delay('0x', int(t.secs) * 1000, '%.9f' % (delay * 1000))
roach.req.poco_delay('0y', int(t.secs) * 1000, '%.9f' % (delay * 1000))
def next(self, event):
azimuth, elevation = [], []
# Opening the out put file to write the RFI contaminated channels information
if self.ind >= len(datasets):
print "No more data to be reduced, hold on a second while we writting the extracted info in the file."
fout = file(opts.outfilebase + '.csv', 'w')
fout.write("FILENAME, FREQUCENCY[MHz], TIMESTAMPS, ABS_TIME, AZIMUTH, ELEVATION, PEAK POWER [dB]\n")
fout.writelines([('%s, %0.4f, %0.4f, %s, %0.2f, %0.2f,%0.2f \n') % tuple(p) for p in output_data if p])
fout.close()
# Opening the out put file to write only RFI conteminated channels
fout2 = file(opts.outfilebase2 + '.txt', 'w')
fout2.write("CONTAMINATED FREQUCENCY CHANNELS [MHz]\n")
fout2.write("======================================\n")
fout2.writelines([('%0.4f\n') % p for p in set(output_chan) if p])
fout2.close()
# Time vs Frequency for selected channels figure
fig_new = plt.figure()
new_ax1 = fig_new.add_subplot(311,axisbg='#FFFFCC' )
new_ax1.plot(output_ts, output_chan,'k+',lw=3)
new_ax1.set_xlabel("Time [s]")
new_ax1.set_ylabel("Frequency [MHz]")
new_ax2 = fig_new.add_subplot(312, axisbg="#FFFFCC")
new_ax2.plot(output_az,output_chan,'g+', lw=3)
new_ax2.set_xlabel("Azimuth [Deg]")
new_ax2.set_ylabel("Frequency [MHz]")
new_ax3 = fig_new.add_subplot(313, axisbg='#FFFFCC')
new_ax3.plot(output_az, output_el, 'r+', lw=3)
new_ax3.set_xlabel("Azimuth [Deg]")
new_ax3.set_ylabel("Elevation [Deg]")
print "We done writting in to a file,look at the frequency vs time plot to see the RFI mapping as a function of time"
print "The RFI contaminated channels are:", set(output_chan)
plt.show()
sys.exit()
self.filename = datasets[self.ind]
try:
#logger.info("Loading dataset %s , File size is %fMB, This is File number %s" % (os.path.basename(self.filename),os.path.getsize(self.filename)/1e6,self.ind))
logger.info("Loading dataset %s , File size is %fMB, This is File number %s" % (os.path.basename(self.filename),os.path.getsize(self.filename),self.ind))
current_dataset = DataSet(self.filename, baseline=opts.baseline)
out_filename =os.path.basename(self.filename)
start_freq_channel = int(opts.freq_keep.split(',')[0])
end_freq_channel = int(opts.freq_keep.split(',')[1])
current_dataset = current_dataset.select(freqkeep=range(start_freq_channel, end_freq_channel+1))
current_dataset = current_dataset.select(labelkeep='scan', copy=False)
if len(current_dataset.compscans) == 0 or len(current_dataset.scans) == 0:
logger.warning('No scans found in file, skipping data set')
# try to extract antenna target points per each timestamps
for cscan in current_dataset.compscans:
target = cscan.target.name
az = np.hstack([scan.pointing['az'] for scan in cscan.scans])
el = np.hstack([scan.pointing['el'] for scan in cscan.scans])
ts = np.hstack([scan.timestamps for scan in cscan.scans])
azimuth.extend(katpoint.rad2deg(az)),elevation.extend(katpoint.rad2deg(el))
azimuth, elevation = np.array(azimuth), np.array(elevation)
ts,f,amp = extract_xyz_data(current_dataset,'abs_time','freq','amp')
power,freq = amp.data,f.data
t = np.hstack(ts.data)
base_freq = freq[0]
p = np.hstack(power)
T,F = np.meshgrid(t,base_freq)
A,F = np.meshgrid(azimuth,base_freq)
E,F = np.meshgrid(elevation,base_freq)
AA = A.ravel()
EE = E.ravel()
TT = T.ravel()
FF = F.ravel()
PP = p.ravel()
def onselect_next(eclick,erelease):
global output_data, output_chan, output_ts
xmin = min(eclick.xdata, erelease.xdata)
xmax = max(eclick.xdata, erelease.xdata)
ymin = min(eclick.ydata, erelease.ydata)
ymax = max(eclick.ydata, erelease.ydata)
ind = (FF >= xmin) & (FF <= xmax) & (PP >= ymin) & (PP <= ymax)
selected_freq = FF[ind]
selected_amp = 10.0*np.log10(PP[ind])
selected_ts = TT[ind]
selected_az = AA[ind]
selected_el = EE[ind]
print "SUCCESSFUL, CLICK AND DRAG TO SELECT THE NEXT RFI CHANNELS OR NEXT TO LOAD NEW DATASET"
#sorting with increasng X_new
indices = np.lexsort(keys = (selected_ts, selected_freq))
for index in indices:
output_data.append([out_filename, selected_freq[index],selected_ts[index], katpoint.Timestamp(selected_ts[index]).local(), selected_az[index], selected_el[index], selected_amp[index]])
for point in output_data:
output_chan.append(point[1])
output_ts.append(point[2])
output_az.append(point[4])
output_el.append(point[5])
def toggle_selector_next(event):
print ' Key pressed.'
if event.key in ['Q', 'q'] and toggle_selector.RS.active:
print ' RectangleSelector deactivated.'
toggle_selector_next.RS.set_active(False)
if event.key in ['A', 'a'] and not toggle_selector_next.RS.active:
print ' RectangleSelector activated.'
toggle_selector_next.RS.set_active(True)
# New Figure for the current data set
current_ax.clear()
plt.subplots_adjust(bottom=0.2)
current_ax.plot(FF,PP, '+')
current_ax.set_title("CLICK AND DRAG TO SELECT RFI CHAN")
current_ax.set_xlabel('Frequency Channels [MHz]', bbox=dict(facecolor='red'))
current_ax.set_ylabel('Power [Count]', bbox=dict(facecolor='red'))
plt.draw()
print "NEW DATA SET SUCCESSFLY LOADED, CLICK AND DRAG TO SELECT THE RFI CONTAMINATED CHANNELS OR NEXT TO CONTINUE"
toggle_selector_next.RS = RectangleSelector(current_ax, onselect_next, drawtype='box')
plt.connect('key_press_event', toggle_selector_next)
except ValueError:
print os.path.basename(self.filename), "DATA CORUPTED, PLEASE CLICK NEXT TO LOAD ANOTHER DATASET"
self.ind +=1
#
""")
f.writelines(np.sort(outlines))
f.close()
# Test the catalogue
ant = katpoint.Antenna('KAT7, -30:43:16.71, 21:24:35.86, 1055, 12.0')
cat = katpoint.Catalogue(open('bae_optical_pointing_sources.csv'),
add_specials=False,
antenna=ant)
timestamp = katpoint.Timestamp()
ra, dec = np.array([t.radec(timestamp) for t in cat]).transpose()
constellation = [
t.aliases[0].partition(' ')[2][:3] if t.aliases else 'SOL' for t in cat
]
ra, dec = katpoint.rad2deg(ra), katpoint.rad2deg(dec)
az, el = np.hstack([
targ.azel([
katpoint.Timestamp(timestamp + t)
for t in range(0, 24 * 3600, 30 * 60)
]) for targ in cat
])
az, el = katpoint.rad2deg(az), katpoint.rad2deg(el)
plt.figure(1)
plt.clf()
for n, c in enumerate(constellation):
plt.text(ra[n], dec[n], c, ha='left', va='center', size='xx-small')
plt.axis([0, 360, -90, 90])
plt.xlabel('Right Ascension (degrees)')
plt.ylabel('Declination (degrees)')
def main():
# Parse command-line options and arguments
parser = optparse.OptionParser(
usage='%prog [options] <data file> [<data file> ...]',
description='Display a horizon mask from a set of data files.')
parser.add_option(
'-a',
'--baseline',
dest='baseline',
type="string",
metavar='BASELINE',
default='A1A1',
help=
"Baseline to load (e.g. 'A1A1' for antenna 1), default is first single-dish baseline in file"
)
parser.add_option('-o',
'--output',
dest='output',
type="string",
metavar='OUTPUTFILE',
default=None,
help="Write out intermediate h5 file")
parser.add_option('-s',
'--split',
dest='split',
action="store_true",
metavar='SPLIT',
default=False,
help="Whether to split each horizon plot in half")
parser.add_option('-z',
'--azshift',
dest='azshift',
type='float',
metavar='AZIMUTH_SHIFT',
default=45.0,
help="Degrees to rotate azimuth window by.")
parser.add_option(
'--temp-limit',
dest='temp_limit',
type='float',
default=40.0,
help=
"The Tempreture Limit to make the cut-off for the mask. This is calculated "
"as the T_sys at zenith plus the atmospheric noise contrabution at 10 degrees"
"elevation as per R.T. 199 .")
parser.add_option(
"-n",
"--nd-models",
help="Name of optional directory containing noise diode model files")
(opts, args) = parser.parse_args()
# Check arguments
if len(args) < 1:
raise RuntimeError('Please specify the data file to reduce')
# Load data set
gridtemp = []
for filename in args:
print 'Loading baseline', opts.baseline, 'from data file', filename
d = scape.DataSet(filename,
baseline=opts.baseline,
nd_models=opts.nd_models)
if len(d.freqs) > 1:
# Only keep main scans (discard slew and cal scans) a
d = d.select(freqkeep=range(200, 800))
d = remove_rfi(d, width=7, sigma=5)
d = d.convert_power_to_temperature(min_duration=3,
jump_significance=4.0)
d = d.select(flagkeep='~nd_on')
d = d.select(labelkeep='scan', copy=False)
# Average all frequency channels into one band
d.average()
# Extract azimuth and elevation angle from (azel) target associated with scan, in degrees
azimuth, elevation, temp = [], [], []
for s in d.scans:
azimuth.extend(rad2deg(s.pointing['az']))
elevation.extend(rad2deg(s.pointing['el']))
temp.extend(tuple(np.sqrt(s.pol('HH')[:, 0] * s.pol('VV')[:, 0])))
assert len(azimuth) == len(elevation) == len(temp), "sizes don't match"
data = (azimuth, elevation, temp)
np.array(azimuth) < -89
print "Gridding the data"
print "data shape = ", np.shape(data[0] + (
np.array(azimuth)[np.array(azimuth) < -89] + 360.0).tolist())
print np.shape(data[1] +
np.array(elevation)[np.array(azimuth) < -89].tolist())
print np.shape(data[2] +
np.array(temp)[np.array(azimuth) < -89].tolist())
gridtemp.append(
mlab.griddata(
data[0] +
(np.array(azimuth)[np.array(azimuth) < -89] + 360.0).tolist(),
data[1] +
np.array(elevation)[np.array(azimuth) < -89].tolist(),
data[2] + np.array(temp)[np.array(azimuth) < -89].tolist(),
np.arange(-90, 271, 1), np.arange(4, 16, 0.1)))
# The +361 is to ensure that the point are well spaced,
#this offset is not a problem as it is just for sorting out a boundery condition
print "Completed Gridding the data"
print "Making the mask"
mask = gridtemp[0] >= opts.temp_limit
for grid in gridtemp:
mask = mask * (grid >= opts.temp_limit)
maskr = np.zeros((len(np.arange(-90, 271, 1)), 2))
for i, az in enumerate(np.arange(-90, 271, 1)):
print 'at az %f' % (az, )
maskr[i] = az, np.max(elevation)
for j, el in enumerate(np.arange(4, 16, 0.1)):
if ~mask.data[j, i] and ~mask.mask[j, i]:
maskr[i] = az, el
break
np.savetxt('horizon_mask_%s.dat' % (opts.baseline), maskr[1:, :])
if len(observation_sources.filter(el_limit_deg=opts.horizon)) == 0:
user_logger.warning(
"No targets are currently visible - please re-run the script later"
)
else:
# Start capture session, which creates HDF5 file
with start_session(kat, **vars(opts)) as session:
session.standard_setup(**vars(opts))
session.capture_start()
# Iterate through source list, picking the next one that is up
for target in observation_sources.iterfilter(
el_limit_deg=opts.horizon):
user_logger.info(target)
[ra, dec] = target.radec()
(tra, tdec) = (katpoint.rad2deg(float(ra)),
katpoint.rad2deg(float(dec)))
session.label('track')
user_logger.info(
"Initiating %g-second track on target (%.2f, %.2f)" % (
opts.track_duration,
tra,
tdec,
))
session.set_target(target) # Set the target
session.track(target,
duration=opts.track_duration,
announce=False) # Set the target & mode = point
for dra in [-1, 0, 1]:
for ddec in [-1, 0, 1]:
[ra, dec] = target.radec()