def get_corrections(self, seeing_wlen=5400.*u.Angstrom):
# Precompute wlen ratio for wavelength dependent seeing adjustment
wlen_ratio = (self.wlen_grid / seeing_wlen).si
# Initialize acceptance ratio grid
corrections = np.empty(
(self.spec_file.num_exposures, self.num_offset_targets, self.wlen_grid_steps, self.steps_per_exposure),
dtype=float)
guided_centroids = []
# Loop over exposures
for exp_index in range(self.spec_file.num_exposures):
# Estimate the actual offset target paths during the exposure
guided_x, guided_y = self.get_exp_centroids(exp_index)
guided_centroids.append((guided_x, guided_y))
# Calculate centroid offsets for each offset target, relative to its nominal fiber center.
offset = np.sqrt(
(guided_x - self.offset_x0[:, np.newaxis, np.newaxis])**2 +
(guided_y - self.offset_y0[:, np.newaxis, np.newaxis])**2)
# Calculate centroid offsets for each offset target, relative to where its fiber center would
# be if it were designed for the same wavelength as the standard stars.
offset_std = np.sqrt(
(guided_x - self.offset_x0_std[:, np.newaxis, np.newaxis])**2 +
(guided_y - self.offset_y0_std[:, np.newaxis, np.newaxis])**2)
seeing = self.seeing[exp_index]
# psf = sdss_25m.get_atmospheric_psf(seeing_wlen, seeing, gauss=False)
# acceptance_model = sdss_25m.calculate_fiber_acceptance(psf)
seeing_wlen_adjusted = seeing*wlen_ratio**(-0.2)
# acceptance_model_grid = map(tpcorr.acceptance_model.AcceptanceModel, seeing_wlen_adjusted)
max_offset = 1.1/2.0*max(np.max((offset / self.pointing.platescale).to(u.arcsec)).value,
np.max((offset_std / self.pointing.platescale).to(u.arcsec)).value)
for wlen_index in range(self.wlen_grid_steps):
# Build acceptance model for this wavelength
acceptance_model = tpcorr.acceptance_model.AcceptanceModel(seeing_wlen_adjusted[wlen_index], max_offset=max_offset)
# Calculate the acceptance fractions for both sets of centroid offsets.
acceptance = acceptance_model((offset[:,wlen_index,:] / self.pointing.platescale).to(u.arcsec))
acceptance_std = acceptance_model((offset_std[:,wlen_index,:] / self.pointing.platescale).to(u.arcsec))
# Calculate the acceptance fraction ratios, tabulated for each offset target, wavelength and time.
# The ratio calculated this way gives the correction of eqn (13).
corrections[exp_index,:,wlen_index,:] = acceptance_std / acceptance
# Average the correction over each exposure time slice.
avg_corrections = np.mean(np.mean(corrections, axis=-1), axis=0)
return corrections, avg_corrections, guided_centroids
def get_mean_correction(self, extrap_wlen=False):
corrections = np.empty((self.num_offset_targets, self.wlen_grid_steps, 1))
# Estimate the actual offset target paths during the exposure
# (offset_x0, offset_y0), (offset_x0_std, offset_y0_std), (guided_x, guided_y) = self.get_mean_exp_centroids()
guided_x, guided_y = self.get_mean_exp_centroids(extrap_wlen=extrap_wlen)
# Calculate centroid offsets for each offset target, relative to its nominal fiber center.
offset = np.sqrt(
(guided_x - self.offset_x0[:, np.newaxis, np.newaxis])**2 +
(guided_y - self.offset_y0[:, np.newaxis, np.newaxis])**2)
# Calculate centroid offsets for each offset target, relative to where its fiber center would
# be if it were designed for the same wavelength as the standard stars.
offset_std = np.sqrt(
(guided_x - self.offset_x0_std[:, np.newaxis, np.newaxis])**2 +
(guided_y - self.offset_y0_std[:, np.newaxis, np.newaxis])**2)
seeing = np.mean(self.seeing)
max_offset = 1.1/2.0*max(np.max((offset / self.pointing.platescale).to(u.arcsec)).value,
np.max((offset_std / self.pointing.platescale).to(u.arcsec)).value)
acceptance_model = tpcorr.acceptance_model.AcceptanceModel(seeing, max_offset=max_offset)
# Calculate the acceptance fractions for both sets of centroid offsets.
acceptance = acceptance_model((offset / self.pointing.platescale).to(u.arcsec))
acceptance_std = acceptance_model((offset_std / self.pointing.platescale).to(u.arcsec))
# Calculate the acceptance fraction ratios, tabulated for each offset target, wavelength and time.
# The ratio calculated this way gives the correction of eqn (13).
corrections = acceptance_std / acceptance
mean_correction = np.mean(corrections, axis=-1)
return mean_correction, guided_x, guided_y
