def action(self, Data):
# Figure out what map band this corresponds to.
Data.calc_freq()
freq = Data.freq
# We are going to look for an exact match in for the map frequencies.
# This could be made more general since the sub_map function can handle
# partial overlap, but this will be fine for now.
for band_maps in self.maps:
maps_freq = band_maps[0].get_axis('freq')
if sp.allclose(maps_freq, freq):
maps = band_maps
break
else:
raise NotImplementedError('No maps with frequency axis exactly'
' matching data.')
# Now make sure we have the polarizations in the right order.
data_pols = Data.field['CRVAL4'].copy()
for ii in range(len(data_pols)):
if (misc.polint2str(data_pols[ii])
!= self.params['map_polarizations'][ii]):
raise NotImplementedError('Map polarizations not in same order'
' as data polarizations.')
if (not self.params['solve_for_gain'] or
self.params['gain_output_end'] is '') :
sub_map(Data, maps, self.params['solve_for_gain'],
interpolation=self.params['interpolation'])
else :
block_gain = {}
Data.calc_freq()
block_gain['freq'] = sp.copy(Data.freq)
block_gain['time'] = Data.field['DATE-OBS'][0]
block_gain['scan'] = Data.field['SCAN']
block_gain['gain'] = sub_map(Data, maps, True,
interpolation=self.params['interpolation'])
self.gain_list.append(block_gain)
Data.add_history('Subtracted map from data.',
('Map root: ' + ku.abbreviate_file_path(
self.params['map_input_root']),))
return Data
def process_moonscan(moon_dataobj, outfile, avg_width=30):
(ndata, npol, ntoggle, nfreq) = moon_dataobj.dims
moon_dataobj.calc_freq()
moon_dataobj.calc_pointing()
moon_dataobj.calc_time()
freq = moon_dataobj.freq
ra = moon_dataobj.ra
dec = moon_dataobj.dec
date_time = moon_dataobj.field['DATE-OBS']
time = moon_dataobj.time
az = moon_dataobj.field['CRVAL2']
el = moon_dataobj.field['CRVAL3']
print moon_dataobj.field.keys()
print moon_dataobj.field['CAL']
pols = list(moon_dataobj.field['CRVAL4'])
#print moon_dataobj.field['CRVAL4']
pol_names = {}
for pol_idx in range(npol):
pol_names[utils.polint2str(pols[pol_idx])] = pol_idx
print pol_names
# find the peak in XX, YY
maxlist = np.argmax(moon_dataobj.data[:, 0, 0, :], axis=0)
maxlist = maxlist[np.where(maxlist > 0)]
max_xx = np.mean(maxlist)
print "max_xx", max_xx, np.std(maxlist)
maxlist = np.argmax(moon_dataobj.data[:, 3, 0, :], axis=0)
maxlist = maxlist[np.where(maxlist > 0)]
max_yy = np.mean(maxlist)
print "max_yy", max_yy, np.std(maxlist)
time_max = int((max_xx + max_yy) / 2.)
print time_max
# find the max/min in YX (well-defined because of dipole leakage)
maxlist = np.argmax(moon_dataobj.data[:, 2, 0, :], axis=0)
maxlist = maxlist[np.where(maxlist > 0)]
max_yx = int(np.mean(maxlist))
print "max_yx", max_yx, np.std(maxlist)
minlist = np.argmin(moon_dataobj.data[:, 2, 0, :], axis=0)
minlist = minlist[np.where(minlist > 0)]
min_yx = int(np.mean(minlist))
print "min_yx", min_yx, np.std(minlist)
before = np.mean(moon_dataobj.data[0:(2 * avg_width), :, 0, :], axis=0)
during = np.mean(moon_dataobj.data[(time_max - avg_width): \
(time_max + avg_width), :, 0, :], axis=0)
during_u = np.mean(moon_dataobj.data[(max_yx - avg_width): \
(max_yx + avg_width), :, 0, :], axis=0)
during_d = np.mean(moon_dataobj.data[(min_yx - avg_width): \
(min_yx + avg_width), :, 0, :], axis=0)
after = np.mean(moon_dataobj.data[-(2 * avg_width):-1, :, 0, :], axis=0)
offsrc = (before + after) / 2.
during -= offsrc
during_u -= offsrc
during_d -= offsrc
outfp = open(outfile, "w")
for freq_idx in range(nfreq):
compilation = [freq_idx]
#compilation.extend(before[:, freq_idx])
compilation.extend(during[:, freq_idx])
compilation.extend(during_u[:, freq_idx])
compilation.extend(during_d[:, freq_idx])
#compilation.extend(after[:, freq_idx])
outfp.write(("%.5g " * 13) % tuple(compilation) + "\n")
outfp.close()
def calibrate_file(self, middle, gain, freq):
# This function is largely cut and pasted from process file. I should
# really combine the code into an iterator but that's a lot of work.
# Alternativly, I could make a meta function and pass a function to it.
params = self.params
file_name = (params['input_root'] + middle
+ params['input_end'])
# Output parameters.
Writer = core.fitsGBT.Writer(feedback=self.feedback)
out_filename = (params['output_root'] + middle
+ params['output_end'])
band_inds = params["IFs"]
Reader = core.fitsGBT.Reader(file_name, feedback=self.feedback)
n_bands = len(Reader.IF_set)
if not band_inds:
band_inds = range(n_bands)
# Number of bands we acctually process.
n_bands_proc = len(band_inds)
if not band_inds:
band_inds = range(n_bands)
# Number of bands we acctually process.
n_bands_proc = len(band_inds)
# Get the key that will group this file with other files.
key = get_key(middle)
# Read one block to figure out how many polarizations and channels
# there are.
Data = Reader.read(0,0)
n_pol = Data.dims[1]
n_cal = Data.dims[2]
n_chan = Data.dims[3]
# Allowcate memory for the outputs.
corr = np.zeros((n_bands_proc, n_pol, n_cal, n_chan),
dtype=float)
norm = np.zeros(corr.shape, dtype=int)
freq = np.empty((n_bands_proc, n_chan))
for ii in range(n_bands_proc):
Blocks = Reader.read((), ii)
Blocks[0].calc_freq()
freq[ii,:] = Blocks[0].freq
# We are going to look for an exact match in for the map
# frequencies. This could be made more general since the sub_map
# function can handle partial overlap, but this will be fine for
# now.
for band_maps in self.maps:
maps_freq = band_maps[0].get_axis('freq')
if np.allclose(maps_freq, freq[ii,:]):
maps = band_maps
break
else:
raise NotImplementedError('No maps with frequency axis exactly'
' matching data.')
# Check the polarization axis. If the same number of maps where
# passed, check that the polarizations are in order. If only one
# map was passed, correlate all data polarizations against it.
data_pols = Blocks[0].field['CRVAL4'].copy()
if len(band_maps) == 1:
maps_to_correlate = band_maps * len(data_pols)
else:
for ii in range(len(data_pols)):
if (misc.polint2str(data_pols[ii])
!= self.params['map_polarizations'][ii]):
msg = ('Map polarizations not in same order'
' as data polarizations.')
raise NotImplementedError(map)
maps_to_correlate = band_maps
# Now process each block.
for Data in Blocks:
if params['diff_gain_cal_only']:
if tuple(Data.field['CRVAL4']) != (-5, -7, -8, -6):
msg = ("Expected polarizations to be ordered "
"(XX, XY, YX, YY).")
raise NotImplementedError(msg)
Data.data[:,0,:,:] /= gain[ii,0,:,:]
Data.data[:,3,:,:] /= gain[ii,3,:,:]
cross_gain = np.sqrt(gain[ii,0,:,:] * gain[ii,3,:,:])
Data.data[:,1,:,:] /= cross_gain
Data.data[:,2,:,:] /= cross_gain
else:
Data.data /= gain[ii,...]
Writer.add_data(Data)
# Write the data back out.
utils.mkparents(out_filename)
Writer.write(out_filename)
def process_file(self, middle, Pipe=None):
params = self.params
file_name = (params['input_root'] + middle
+ params['input_end'])
band_inds = params["IFs"]
Reader = core.fitsGBT.Reader(file_name, feedback=self.feedback)
n_bands = len(Reader.IF_set)
if not band_inds:
band_inds = range(n_bands)
# Number of bands we acctually process.
n_bands_proc = len(band_inds)
if not band_inds:
band_inds = range(n_bands)
# Number of bands we acctually process.
n_bands_proc = len(band_inds)
# Get the key that will group this file with other files.
key = get_key(middle)
# Read one block to figure out how many polarizations and channels
# there are.
Data = Reader.read(0,0)
n_pol = Data.dims[1]
n_cal = Data.dims[2]
n_chan = Data.dims[3]
# Allowcate memory for the outputs.
corr = np.zeros((n_bands_proc, n_pol, n_cal, n_chan),
dtype=float)
norm = np.zeros(corr.shape, dtype=float)
freq = np.empty((n_bands_proc, n_chan))
for ii in range(n_bands_proc):
Blocks = Reader.read((), ii)
Blocks[0].calc_freq()
freq[ii,:] = Blocks[0].freq
# We are going to look for an exact match in for the map
# frequencies. This could be made more general since the sub_map
# function can handle partial overlap, but this will be fine for
# now.
for band_maps in self.maps:
maps_freq = band_maps[0].get_axis('freq')
if np.allclose(maps_freq, freq[ii,:]):
maps = band_maps
break
else:
raise NotImplementedError('No maps with frequency axis exactly'
' matching data.')
# Check the polarization axis. If the same number of maps where
# passed, check that the polarizations are in order. If only one
# map was passed, correlate all data polarizations against it.
data_pols = Blocks[0].field['CRVAL4'].copy()
if len(band_maps) == 1:
maps_to_correlate = band_maps * len(data_pols)
else:
for ii in range(len(data_pols)):
if (misc.polint2str(data_pols[ii])
!= self.params['map_polarizations'][ii]):
msg = ('Map polarizations not in same order'
' as data polarizations.')
raise NotImplementedError(map)
maps_to_correlate = band_maps
# Now process each block.
for Data in Blocks:
this_corr, this_norm = get_correlation(Data,
maps_to_correlate,
interpolation=params['interpolation'],
modes_subtract=params['smooth_modes_subtract'],
filter_type=params['filter_type'])
# Check that the answer we got is sane, if not, throw away the
# this set.
tmp_corr = this_corr.copy()
tmp_norm = this_norm.copy()
tmp_corr[tmp_norm == 0] = 1.
tmp_norm[tmp_norm == 0] = 1.
tmp_gains = tmp_corr / tmp_norm
if np.all(tmp_gains < 2) and np.all(tmp_gains > 0.5):
corr[ii,...] += this_corr
norm[ii,...] += this_norm
else:
pass
if Pipe is None:
return key, corr, norm, freq
else:
Pipe.send((key, corr, norm, freq))
def process_moonscan(moon_dataobj, outfile, avg_width=30):
(ndata, npol, ntoggle, nfreq) = moon_dataobj.dims
moon_dataobj.calc_freq()
moon_dataobj.calc_pointing()
moon_dataobj.calc_time()
freq = moon_dataobj.freq
ra = moon_dataobj.ra
dec = moon_dataobj.dec
date_time = moon_dataobj.field['DATE-OBS']
time = moon_dataobj.time
az = moon_dataobj.field['CRVAL2']
el = moon_dataobj.field['CRVAL3']
print moon_dataobj.field.keys()
print moon_dataobj.field['CAL']
pols = list(moon_dataobj.field['CRVAL4'])
#print moon_dataobj.field['CRVAL4']
pol_names = {}
for pol_idx in range(npol):
pol_names[utils.polint2str(pols[pol_idx])] = pol_idx
print pol_names
# find the peak in XX, YY
maxlist = np.argmax(moon_dataobj.data[:, 0, 0, :], axis=0)
maxlist = maxlist[np.where(maxlist > 0)]
max_xx = np.mean(maxlist)
print "max_xx", max_xx, np.std(maxlist)
maxlist = np.argmax(moon_dataobj.data[:, 3, 0, :], axis=0)
maxlist = maxlist[np.where(maxlist > 0)]
max_yy = np.mean(maxlist)
print "max_yy", max_yy, np.std(maxlist)
time_max = int((max_xx + max_yy) / 2.)
print time_max
# find the max/min in YX (well-defined because of dipole leakage)
maxlist = np.argmax(moon_dataobj.data[:, 2, 0, :], axis=0)
maxlist = maxlist[np.where(maxlist > 0)]
max_yx = int(np.mean(maxlist))
print "max_yx", max_yx, np.std(maxlist)
minlist = np.argmin(moon_dataobj.data[:, 2, 0, :], axis=0)
minlist = minlist[np.where(minlist > 0)]
min_yx = int(np.mean(minlist))
print "min_yx", min_yx, np.std(minlist)
before = np.mean(moon_dataobj.data[0:(2 * avg_width), :, 0, :], axis=0)
during = np.mean(moon_dataobj.data[(time_max - avg_width): \
(time_max + avg_width), :, 0, :], axis=0)
during_u = np.mean(moon_dataobj.data[(max_yx - avg_width): \
(max_yx + avg_width), :, 0, :], axis=0)
during_d = np.mean(moon_dataobj.data[(min_yx - avg_width): \
(min_yx + avg_width), :, 0, :], axis=0)
after = np.mean(moon_dataobj.data[-(2 * avg_width):-1, :, 0, :], axis=0)
offsrc = (before + after) / 2.
during -= offsrc
during_u -= offsrc
during_d -= offsrc
outfp = open(outfile, "w")
for freq_idx in range(nfreq):
compilation = [freq_idx]
#compilation.extend(before[:, freq_idx])
compilation.extend(during[:, freq_idx])
compilation.extend(during_u[:, freq_idx])
compilation.extend(during_d[:, freq_idx])
#compilation.extend(after[:, freq_idx])
outfp.write(("%.5g " * 13) % tuple(compilation) + "\n")
outfp.close()
