def median_spectrum(redshift, wave_grid, flux_rms, flux_std):
medianSpec = spectrum()
medianSpec.z = redshift
medianSpec.npix = wave_grid.size
medianSpec.wave = wave_grid
medianSpec.flux = flux_rms['BOSS']
medianSpec.flux_error = flux_std['BOSS']
medianSpec.__fit_powerlaw__(type='BOSS')
medianSpec.alpha_BOSS = medianSpec.alpha
medianSpec.flux_corr = flux_rms['CORR']
medianSpec.flux_error = flux_std['CORR']
medianSpec.__fit_powerlaw__(type='CORR')
medianSpec.alpha_CORR = medianSpec.alpha
medianSpec.flux_corr = flux_rms['MARG']
medianSpec.flux_error = flux_std['MARG']
medianSpec.__fit_powerlaw__(type='CORR')
medianSpec.alpha_MARG = medianSpec.alpha
print '{0:>45s}'.format('SPECTRAL INDEX')
print '{0:>33s} '.format('SI_medspec')
print '{0:>20s} {1:15.12f}'.format('UNCORRECTED:', medianSpec.alpha_BOSS)
print '{0:>20s} {1:15.12f}'.format('CORRECTED (Ours):',
medianSpec.alpha_CORR)
print '{0:>20s} {1:15.12f}'.format('CORRECTED (Margala):',
medianSpec.alpha_MARG)
return medianSpec
def main():
# ======================= SETTINGS =======================
settings = program_settings()
settings.plot = True
settings.smoothness = 5
settings.resample = True
settings.z_min = 0
settings.z_max = 4
settings.z_delta = 0.1
# ======================= PARSER =======================
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument(
'--spec-dir',
type=str,
default=None,
help='full path to the directory containing the FITS files, required')
args = parser.parse_args()
# ======================= FILES =======================
# from the input argument, fetch file names
if args.spec_dir:
spectral_list = return_spectral_list(args.spec_dir)
numspec = len(spectral_list[0])
uncorrected_files = spectral_list[0]
corrected_files = spectral_list[1]
corrected_files_m = spectral_list[2]
# list that saves the number of files in each subfolder
numfiles = [
len(uncorrected_files),
len(corrected_files),
len(corrected_files_m)
]
labels = []
dir_basename = os.path.dirname(args.spec_dir)
thingid = dir_basename.split('/')[7]
# read the data from the uncorrected FITS files
for fileList in spectral_list:
ids = []
for file in fileList:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field) for field in os.path.splitext(spec_basename)
[0].split('-')[1:]
]
ids.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
labels.append(ids)
#data, header = fitsio.read(file,ext=0,header=True)
# ======================= ======================== ======================
flux_boss_list = []
flux_corr_list = []
flux_marg_list = []
# ======================= UNIVERSAL WAVELENGTH GRID =======================
loglam_grid = np.arange(start=3.55, stop=4.02, step=0.0001)
wave_grid = np.power(10, loglam_grid)
wave_npix = wave_grid.size
wave_shape = wave_grid.shape
row = np.dtype([('BOSS', 'd8', 1), ('CORR', 'd8', 1), ('MARG', 'd8', 1)])
si_data = np.zeros(shape=(max(numfiles)), dtype=row)
pl_flux = np.zeros(shape=(max(numfiles), wave_npix), dtype=row)
fl_data = np.zeros(shape=(max(numfiles), wave_npix), dtype=row)
print pl_flux[0].size
# ======================= CALCULATIONS =======================
for i in xrange(len(spectral_list)):
for j, file in enumerate(spectral_list[i]):
# ======================= READ SPECTRA =======================
spec = spectrum()
spec_dirname = os.path.dirname(file)
spec_basename = os.path.basename(file)
if 'corr' not in spec_basename:
spec.__read_BOSS__(file)
elif 'corr' in spec_basename:
spec.__read_CORR__(file)
minwav = spec.wave.min()
maxwav = spec.wave.max()
index_start = np.argmin(abs(wave_grid - minwav))
index_stop = np.argmin(abs(wave_grid - maxwav))
newpix = index_stop - index_start
# ======================= GET SPECTRAL INDEX =======================
if 'corr' not in spec_basename:
spec.__fit_powerlaw__(type='BOSS')
flux_old = spec.flux
si_data[j]['BOSS'] = spec.alpha
fl_data[j]['BOSS'][index_start:index_stop] = congrid.rebin_1d(
spec.flux, newpix)
pl_flux[j]['BOSS'][index_start:index_stop] = congrid.rebin_1d(
spec.powerlaw,
newpix) #10.0**(spec.delta + spec.beta*loglam_grid)
if 'corr' in spec_basename:
spec.__fit_powerlaw__(type='CORR')
flux_old = spec.flux_corr
if 'margala' not in spec_dirname:
si_data[j]['CORR'] = spec.alpha
fl_data[j]['CORR'][
index_start:index_stop] = congrid.rebin_1d(
spec.flux_corr, newpix)
pl_flux[j]['CORR'][
index_start:index_stop] = congrid.rebin_1d(
spec.powerlaw, newpix)
elif 'margala' in spec_dirname:
si_data[j]['MARG'] = spec.alpha
fl_data[j]['MARG'][
index_start:index_stop] = congrid.rebin_1d(
spec.flux_corr, newpix)
pl_flux[j]['MARG'][
index_start:index_stop] = congrid.rebin_1d(
spec.powerlaw, newpix)
ra = spec.ra
dec = spec.dec
z = spec.z
del (spec)
print np.median(si_data['BOSS']), np.std(si_data['BOSS'])
print np.median(si_data['CORR']), np.std(si_data['CORR'])
print np.median(si_data['MARG']), np.std(si_data['MARG'])
def main():
np.set_printoptions(threshold=6000)
# PARSER
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument(
'--tpcorr',
type=str,
default=
'/Volumes/T2TB/Isotropy/spectra/BOSScorrections/tpcorr-0.1/tpcorr.hdf5',
help='Throughput correction filename, required')
parser.add_argument(
'--spec-dir',
type=str,
default=None,
help='Path to directory containing individual spec files')
parser.add_argument(
'--all',
action='store_true',
help=
'Save corrected spectra for all available objects of the given data sample'
)
parser.add_argument('--data',
choices=['QSO', 'FQSO', 'STAR'],
nargs=1,
type=str,
default='QSO',
help='Object type of the chosen spectrum')
parser.add_argument(
'--nobalmer',
action='store_true',
default=True,
help=
'Balmer corrections flag. If the flag is true, no Balmer corrections are applied.'
)
args = parser.parse_args()
# data sample name
sample = args.data[0]
# MARGALA CORRECTION FILE
tpcorr = h5py.File(args.tpcorr, 'r')
tpcorr_wave = tpcorr['wave'].value
# BALMER CORRECTION FILE
if args.nobalmer == True:
resid_corr = None
elif args.nobalmer == False:
resid_corr = read_resid_corr(
'/Volumes/T2TB/Isotropy/PySDSS/residcorr_v5_4_45.dat')
print resid_corr
# Parse arguments: extract plate, mjd, fiberid as integers and the spec filename
ids = []
sid_tot = []
if args.spec_dir:
spectral_list = return_spectral_list(args.spec_dir)
uncorrected_files = spectral_list[0]
total_files = uncorrected_files
for file in uncorrected_files:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
ids.append((plate, mjd, fiberid))
sid_tot.append('%04d%5d%04d' % (plate, mjd, fiberid))
dtypedict = {'QSO': 'QUASARS', 'FQSO': 'FAILED QUASARS', 'STAR': 'STARS'}
if args.all:
print "CORRECTING ALL AVAILABLE SPECTRA OF {data}".format(
data=dtypedict[sample])
PathToSpectra = os.path.join(
"/Volumes/T2TB/Isotropy/spectra/SDSS_DR12", sample)
total_files = glob(os.path.join(PathToSpectra, "*.fits"))
print "Spectra available: {0:8d}".format(len(total_files))
for file in total_files:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
ids.append((plate, mjd, fiberid))
sid_tot.append('%04d%5d%04d' % (plate, mjd, fiberid))
#else:
# parser.error('Must specify either a spec file using the --spec-file option or path to directory containing spec files and a target.')
#sid_tot = np.array(sid_tot)
sid_tot_array = np.array(sid_tot)
sid_tot = set(sid_tot)
if args.all:
if args.nobalmer == False:
dirpath = os.path.join(
'/Volumes/T2TB/Isotropy/spectra/SDSS_DR12',
'{data}_{name}'.format(data=sample, name='Margala'))
elif args.nobalmer == True:
dirpath = os.path.join(
'/Volumes/T2TB/Isotropy/spectra/SDSS_DR12',
'{data}_{name}_nobalmer'.format(data=sample, name='Margala'))
else:
dirpath = os.path.join(os.path.normpath(args.spec_dir), 'corrected',
'margala')
# remove already corrected spectra from the list
corrected_files = glob(os.path.join(dirpath, "*.fits"))
print "Spectra previously corrected: {0:8d}".format(len(corrected_files))
sid_corr = []
for file in corrected_files:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
sid_corr.append('%04d%5d%04d' % (plate, mjd, fiberid))
sid_corr = set(sid_corr)
sid_unc = sid_tot.difference(sid_corr)
uncorrected_files = []
for i, file in enumerate(total_files):
if sid_tot_array[i] in sid_unc:
uncorrected_files.append(file)
print "Spectra to be corrected: {0:8d}".format(len(uncorrected_files))
file_array = np.array(uncorrected_files)
for file in uncorrected_files:
print file
HDU = fitsio.FITS(file)
h = HDU[0].read_header()
flux = HDU[1]['flux'][:]
ivar = HDU[1]['ivar'][:]
wavelength = np.power(10, HDU[1]['loglam'][:])
ra = HDU[2]['ra'][0][0]
dec = HDU[2]['dec'][0][0]
z = HDU[2]['z'][0][0]
HDU.close()
plate, mjd, fiberid = ids[np.argwhere(file_array == file)[0][0]]
# Read the target's throughput correction vector
tpcorr_key = '%s/%s/%s' % (plate, mjd, fiberid)
try:
correction = tpcorr[tpcorr_key].value
except:
print "No data on object found"
continue
# Create an interpolated correction function
correction_interp = interp1d(tpcorr_wave, correction, kind='linear')
# Sample the interpolated correction using the observation's wavelength grid
resampled_correction = correction_interp(wavelength)
# Apply the correction to the observed flux and ivar
# Save the data into a fits file
spec = spectrum()
spec.plateid = plate
spec.mjd = mjd
spec.fiberid = fiberid
spec.ra = ra
spec.dec = dec
spec.z = z
spec.beginwl = h['COEFF0']
spec.npix = np.size(wavelength)
spec.wave = wavelength
spec.flux = flux
spec.ivar = ivar
spec.flux_error = np.sqrt(1. / ivar)
apply_resid_corr(spec, resid_corr)
corrected_flux = spec.flux * resampled_correction
corrected_ivar = spec.ivar / resampled_correction**2
spec.flux_corr = corrected_flux
spec.flux_error = np.sqrt(1. / corrected_ivar)
spec.corr = resampled_correction
spec.basename = os.path.basename(file)
spec.dirname = os.path.dirname(file)
spec.__save_fitsio__(dirpath=dirpath)
del (spec)
tpcorr.close()
def main(args):
# files
pathToFile = args.list
# read Balmer corrections
from SDSSmodules.SDSSfiles import read_resid_corr
resid_corr = read_resid_corr(
'/Volumes/T2TB/Isotropy/PySDSS/residcorr_v5_4_45.dat')
objects = []
with open(pathToFile, 'r') as file:
for line in file:
plate, mjd, fiber = [
field for field in line.rstrip('\n').split(',')
]
objects.append((plate, mjd, fiber))
k = 0
for object in objects:
# define a spectrum instance:
spec = spectrum()
# read plate, mjd, fiber:
plate, mjd, fiber = [int(value) for value in object]
spec.plate = plate
spec.MJD = mjd
spec.fiberid = fiber
spec.dirname = savedir
spec.basename = 'spec-{plate:4d}-{mjd:5d}-{fiber:04d}.fits'.format(
plate=plate, mjd=mjd, fiber=fiber)
i = fiber - 1 #systematic offset
# Pick your plate/mjd and read the data:
spfile = '{top}{plate}/spPlate-{plate}-{mjd}.fits'.format(top=topdir,
plate=plate,
mjd=mjd)
zbfile = '{top}{plate}/v5_7_0/spZbest-{plate}-{mjd}.fits'.format(
top=topdir, plate=plate, mjd=mjd)
zafile = '{top}{plate}/v5_7_0/spZall-{plate}-{mjd}.fits'.format(
top=topdir, plate=plate, mjd=mjd)
hdulist = pf.open(spfile)
c0 = hdulist[0].header['coeff0']
c1 = hdulist[0].header['coeff1']
npix = hdulist[0].header['naxis1']
wave = 10.**(c0 + c1 * n.arange(npix))
spec.beginwl = c0
spec.npix = npix
spec.wave = wave
spec.lambda_eff = hdulist[5].data['LAMBDA_EFF'][i]
# Following commented-out bit was needed for some of the early redux:
#bzero = hdulist[0].header['bzero']
bunit = hdulist[0].header['bunit']
flux = hdulist[0].data
ivar = hdulist[1].data
# put Balmer _uncorrected_ flux and errors into the spectrum
spec.flux = flux[i, :] * (ivar[i, :] > 0)
spec.flux[spec.flux == 0] = n.nan
spec.ivar = ivar[i, :] * (ivar[i, :] > 0)
spec.ivar[spec.ivar == 0] = n.nan
spec.flux_error = 1 / n.sqrt(spec.ivar) * (ivar[i, :] > 0)
# apply Balmer corrections to the spectrum
if args.balmer == True:
apply_resid_corr(spec, resid_corr)
elif args.balmer == False:
pass
hdulist.close()
hdulist = 0
hdulist = pf.open(zbfile)
metadata = hdulist[0].header
synflux = hdulist[2].data
zstruc = hdulist[1].data
hdulist.close()
hdulist = 0
k += 1
spec.alt = metadata['ALT']
spec.airtemp = metadata['AIRTEMP']
spec.seeing50 = metadata['SEEING50']
spec.humidity = metadata['HUMIDITY']
spec.pressure = metadata['PRESSURE']
spec.az = metadata['AZ']
spec.z = zstruc[i].field('z')
spec.ra = zstruc[i].field('plug_ra')
spec.dec = zstruc[i].field('plug_dec')
#for key, val in vars(spec).items():
# print '{key} = {val}'.format(key=key, val=val)
#p.figure()
# Set starting fiber point (above), then copy and paste
# the following repeatedly to loop over spectra:
#i+=1
# Following commented-out bit was needed for some of the early redux:
#p.plot(wave, (flux[i,:]-bzero) * (ivar[i,:] > 0), 'k', hold=False)
#auxflx = flux[i,:] * (ivar[i,:] > 0)
#p.plot(wave, spec.flux, 'k', hold=False)
#p.plot(wave, synflux[i,:], 'g', hold=True)
#p.xlabel('Angstroms')
#p.ylabel(bunit)
#p.title(zstruc[i].field('class') + ', z = ' + str(zstruc[i].field('z')))
#p.ylim(n.percentile(auxflx,1), 1.2*n.percentile(auxflx, 99))
#p.show()
spec.__save_fitsio__(dirpath=savedir, filetype='valid')
def read_spec(filepath):
"""
Reads a FITS file containing the spectrum of a Quasar from the input path
and returns a 'spectrum' class object into the main program.
Parameters:
-----------
filepath : Path to the FITS file
Returns:
--------
spectrum : A 'spectrum' class object
"""
spec_dirname = os.path.dirname(filepath)
spec_basename = os.path.basename(filepath)
#initialise spectrum
spec = spectrum()
spec.filename = spec_basename
if 'corr' not in spec_basename:
#print 'uncorrected'
fits = fitsio.FITS(filepath)
spec.ra = fits[2]['RA'][0]
spec.dec = fits[2]['DEC'][0]
spec.z = fits[2]['Z'][0]
spec.loglam = fits[1]['LOGLAM'][:]
spec.wav = np.power(10, spec.loglam)
spec.flx = fits[1]['FLUX'][:]
spec.err = fits[1]['IVAR'][:]
spec.ivar = fits[1]['IVAR'][:]
spec.npix = np.size(spec.loglam)
spec.numexp = fits[-1].get_extnum(
) - 3 #(-4 because of the 4 standard HDUs + 1 because we start at index 0, i.e. fits[0])
spec.loglam_start = np.min(spec.loglam)
spec.loglam_end = np.max(spec.loglam)
else:
#print 'corrected'
fits = fitsio.FITS(filepath)
header = fits[0].read_header()
spec.ra = header['RA']
spec.dec = header['DEC']
spec.z = header['Z']
spec.npix = header['NPIX']
spec.wav = fits[1]['WAV'][:]
spec.flx = fits[1]['FLUX'][:]
spec.flx_corr = fits[1]['FLUX_CORR'][:]
spec.err = fits[1]['FLUX_ERR'][:]
if 'margala' not in spec_dirname:
spec.corr5400 = fits[1]['CORR5400'][:]
spec.corr4000 = fits[1]['CORR4000'][:]
spec.corr = spec.corr5400 / spec.corr4000
spec.loglam_start = np.log10(spec.wav[0])
spec.loglam_end = np.log10(spec.wav[spec.npix - 1])
return spec
def read_spectra(path):
np.set_printoptions(threshold=6000)
outspec = spectrum()
if os.path.isdir(path):
filetype = 'dir'
elif os.path.isfile(path):
filetype = 'file'
labels = []
if os.path.isdir(path):
spectral_list = return_spectral_list(path)
numspec = len(spectral_list[0])
uncorrected_files = spectral_list[0]
corrected_files = spectral_list[1]
corrected_files_m = spectral_list[2]
# list that saves the number of files in each subfolder
numfiles = [
len(uncorrected_files),
len(corrected_files),
len(corrected_files_m)
]
dir_basename = os.path.dirname(path)
thingid = dir_basename.split('/')[7]
# read the data from the uncorrected FITS files
for file in uncorrected_files:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
labels.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
if os.path.isfile(path):
spectral_list = []
spectral_list.append([path])
numspec = 1
dir_basename = os.path.dirname(path)
thingid = dir_basename.split('/')[7]
spec_basename = os.path.basename(path).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
labels.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
# ======================= UNIVERSAL WAVELENGTH GRID =======================
loglam_grid = np.arange(start=3.55, stop=4.02, step=0.0001)
wave_grid = np.power(10, loglam_grid)
wave_npix = wave_grid.size
wave_shape = wave_grid.shape
npix = wave_npix
# ======================= ======================== ======================
keys = ['BOSS', 'CORR', 'MARG']
row_flux = np.dtype([('BOSS', np.float32, 1), ('CORR', np.float32, 1),
('MARG', np.float32, 1)])
row_data = np.dtype([('WAVE', np.float32, 1), ('FLUX', row_flux, 1),
('FLUX_ERR', row_flux, 1), ('POWERLAW', row_flux, 1),
('POWERLAW_ERR', row_flux, 1), ('CORR', row_flux, 1)])
row_si = np.dtype([('BOSS', np.float32, 3), ('CORR', np.float32, 3),
('MARG', np.float32, 3)])
data = np.zeros(shape=(numspec, wave_npix), dtype=row_data)
data.fill(np.nan)
data['WAVE'] = wave_grid
si_data = np.ones(shape=(numspec, ), dtype=row_si)
# ======================= ======================== ======================
outspec.keys = keys
outspec.npix = wave_npix
outspec.wave = wave_grid
# ======================= READ DATA FROM FILES =======================
for i in xrange(len(spectral_list)):
for j, file in enumerate(spectral_list[i]):
# ======================= READ SPECTRA =======================
spec = spectrum()
spec_dirname = os.path.dirname(file)
spec_basename = os.path.basename(file)
filetype, fileclass = check_filetype(file)
if filetype == 'BOSS':
spec.__read_BOSS__(file)
flux_boss = np.zeros(shape=wave_shape)
elif filetype == 'CORR':
spec.__read_CORR__(file)
flux_corr = np.zeros(shape=wave_shape)
# __IMPORTANT__: use corrected flux as 'flux'
spec.flux = spec.flux_corr
# fit a powerlaw to the spectrum
spec.__fit_powerlaw__(type=filetype)
# ======================= REBINNING =======================
minwav = spec.wave.min()
maxwav = spec.wave.max()
# find where in the rebinned wavelength grid to put the rebinned flux values
index_start = np.argmin(abs(wave_grid - minwav))
index_stop = np.argmin(abs(wave_grid - maxwav))
# the spectra is rebinned to 'newpix' pixels
newpix = index_stop - index_start
# put new flux values & the powerlaw into the data structure
data[j]['FLUX'][fileclass][
index_start:index_stop] = congrid.rebin_1d(spec.flux, newpix)
data[j]['FLUX_ERR'][fileclass][
index_start:index_stop] = congrid.rebin_1d(
spec.flux_error, newpix)
data[j]['POWERLAW'][fileclass][
index_start:index_stop] = congrid.rebin_1d(
spec.powerlaw, newpix)
data[j]['POWERLAW_ERR'][fileclass][
index_start:index_stop] = congrid.rebin_1d(
spec.powerlaw_error, newpix)
if len(spec.corr) > 0:
data[j]['CORR'][fileclass][
index_start:index_stop] = congrid.rebin_1d(
spec.corr, newpix)
#for k in xrange(spec.npix):
# print j, k, fileclass, spec.flux_error[k], data[j]['FLUX_ERR'][fileclass][k]
# save the data on the spectral index & error into the si_data structure
si_data[j][fileclass][0] = spec.alpha
si_data[j][fileclass][1] = spec.alpha_error
si_data[j][fileclass][2] = spec.delta
outspec.ra = spec.ra
outspec.dec = spec.dec
outspec.z = spec.z
del (spec)
# ======================= ======================== ======================
# ======================= ======================== ======================
# calculate the RMS of flux for each spectrum class
dataT = data.transpose()
rmsFlux = np.zeros(shape=wave_shape, dtype=row_flux)
rmsFlux.fill(np.nan)
rmsR2 = np.zeros(shape=wave_shape, dtype=row_flux)
rmsR2.fill(np.nan)
rmsErr = np.zeros(shape=wave_shape, dtype=row_flux)
rmsErr.fill(np.nan)
rmsCorr = np.zeros(shape=wave_shape, dtype=row_si)
rmsCorr.fill(np.nan)
#rmsCorrE = np.zeros(shape=wave_shape, dtype=row_flux); rmsCorrE.fill(np.nan)
rmsPL = np.zeros(shape=wave_shape, dtype=row_flux)
rmsPL.fill(np.nan)
rmsPLe = np.zeros(shape=wave_shape, dtype=row_flux)
rmsPLe.fill(np.nan)
for key in keys:
for pix in xrange(npix):
# calculate the variance weigthed mean and deviation from numspec values of flux
rmsFlux[pix][key] = rms(dataT[pix]['FLUX'][key],
variance=dataT[pix]['FLUX_ERR'][key]**2)
rmsR2[pix][key] = np.sum(
(dataT[pix]['FLUX'][key] - rmsFlux[pix][key])**2 /
(np.nanstd(dataT[pix]['FLUX'][key]))**2)
rmsErr[pix][key] = np.nanstd(dataT[pix]['FLUX'][key])
rmsCorr[pix][key][0] = np.nanpercentile(dataT[pix]['CORR'][key],
50) # MEDIAN
rmsCorr[pix][key][1] = np.nanpercentile(dataT[pix]['CORR'][key],
36) # LOWER RANGE
rmsCorr[pix][key][2] = np.nanpercentile(dataT[pix]['CORR'][key],
84) # UPPER RANGE
rmsPL[pix][key] = rms(dataT[pix]['POWERLAW'][key],
variance=dataT[pix]['POWERLAW_ERR'][key]**2)
rmsPLe[pix][key] = np.nanstd(dataT[pix]['POWERLAW'][key])
outspec.flux = rmsFlux # has three keys: ['BOSS', 'CORR', 'MARG']
outspec.flux_error = rmsErr # has three keys: ['BOSS', 'CORR', 'MARG']
outspec.corr = rmsCorr # has three keys: ['BOSS', 'CORR', 'MARG']
outspec.R2 = rmsR2 # has three keys: ['BOSS', 'CORR', 'MARG']
outspec.powerlaw = rmsPL # has three keys: ['BOSS', 'CORR', 'MARG']
outspec.powerlaw_error = rmsPLe # has three keys: ['BOSS', 'CORR', 'MARG']
# ======================= ======================== ======================
auxspec = copy.deepcopy(outspec)
rmsSIalpha = np.zeros(shape=(1, ), dtype=row_si)
rmsSIdelta = np.zeros(shape=(1, ), dtype=row_si)
for key in keys:
auxspec.flux = rmsFlux[key]
auxspec.flux_error = rmsErr[key]
auxspec.__fit_powerlaw__(type='BOSS')
rmsSIalpha[key][0][0] = auxspec.alpha
rmsSIalpha[key][0][1] = auxspec.alpha_error
rmsSIdelta[key][0][0] = auxspec.delta
del auxspec
outspec.alpha = np.array([rmsSIalpha[key][0][0] for key in keys])
outspec.alpha_error = np.array([rmsSIalpha[key][0][1] for key in keys])
outspec.delta = np.array([rmsSIdelta[key][0][0] for key in keys])
# spectrum metadata to return to the main program
outspec.nspectra = numspec
# spectral indices of _INDIVIDUAL_(!) spectra
outspec.alpha_BOSS = si_data[
'BOSS'] # has three values: [alpha, alpha_error, delta]
outspec.alpha_CORR = si_data[
'CORR'] # has three values: [alpha, alpha_error, delta]
outspec.alpha_MARG = si_data[
'MARG'] # has three values: [alpha, alpha_error, delta]
outspec.labels = labels
outspec.thingid = thingid
#del(spec)
return data, outspec
def main():
# ======================= PARSER =======================
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument(
'--spec-dir',
type=str,
default=None,
help='full path to the directory containing the FITS files, required')
args = parser.parse_args()
# ======================= FILES =======================
if args.spec_dir:
specList = glob(os.path.join(args.spec_dir, '*.fits'))
numfiles = len(specList)
labels = []
dir_basename = os.path.dirname(args.spec_dir)
thingid = dir_basename.split('/')[6]
for file in specList:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
#spec_filename.append(os.path.join(args.dir, 'spec-%04d-%5d-%04d.fits' % (plate, mjd, fiberid)))
labels.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
#data, header = fitsio.read(file,ext=0,header=True)
# ======================= UNIVERSAL WAVELENGTH GRID =======================
loglam_grid = np.arange(start=3.55, stop=4.02, step=0.0001)
wave_grid = np.power(10, loglam_grid)
wave_npix = wave_grid.size
wave_shape = wave_grid.shape
# ======================= ======================== ======================
flux_boss_list = []
flux_corr_list = []
flux_marg_list = []
keys = ['BOSS', 'CORR', 'MARG']
row = np.dtype([('WAVE', 'f8', 1), ('FLUX', 'f8', 1),
('FLUX_ERR', 'f8', 1)])
data = np.zeros(shape=(numfiles, wave_npix), dtype=row)
# ======================= READ DATA FROM FILES =======================
for i, file in enumerate(specList):
spec = spectrum()
spec_dirname = os.path.dirname(file)
spec_basename = os.path.basename(file)
if 'corr' not in spec_basename:
#print 'uncorrected'
spec.__read_BOSS__(file)
key = 'BOSS'
outfile = '../paper/plots/' + thingid + '_spectrum.pdf'
else:
#print 'corrected'
spec.__read_CORR__(file)
if 'margala' not in spec_dirname:
key = 'CORR'
outfile = '../paper/plots/' + thingid + '_corr_spectrum.pdf'
else:
key = 'MARG'
outfile = '../paper/plots/' + thingid + '_corr_spectrum_margala.pdf'
# ======================= REBINNING =======================
minwav = spec.wave.min()
maxwav = spec.wave.max()
index_start = np.argmin(abs(wave_grid - minwav))
index_stop = np.argmin(abs(wave_grid - maxwav))
newpix = index_stop - index_start
data[i]['WAVE'][index_start:index_stop] = congrid.rebin_1d(
spec.wave, newpix)
if key == 'BOSS':
data[i]['FLUX'][index_start:index_stop] = congrid.rebin_1d(
spec.flux, newpix)
elif (key == 'CORR' or key == 'MARG'):
data[i]['FLUX'][index_start:index_stop] = congrid.rebin_1d(
spec.flux, newpix)
ra = spec.ra
dec = spec.dec
z = spec.z
majorLocator = MultipleLocator(500)
majorFormatter = FormatStrFormatter('%d')
minorLocator = MultipleLocator(100)
# ======================= PLOT =======================
# PLOT
xmin = wave_grid.min
xmax = wave_grid.max
xplotlim = (xmin, xmax)
cmap = plt.get_cmap('Paired')
line_colors = cmap(np.linspace(0, 1, numfiles))
# ======================= DEFINING THE CANVAS =======================
fsize = (20, 5)
fig, axes = plt.subplots(2, 1, figsize=fsize, sharey=True)
#plt.suptitle('PLATEID=%04d, MJD=%5d, FIBREID=%04d, RA = %10.6f, DEC = %10.6f , z = %5.3f' % (plate, mjd, fiberid, ra, dec, z))
fig.subplots_adjust(hspace=0.15)
plt.suptitle('RA = %10.6f, DEC = %10.6f , z = %5.3f' % (ra, dec, z))
# ======================= LABELS =======================
#fig.text(0.5,0.0, r'Observed wavelength $[\AA]$', ha='center')
fig.text(0.09,
0.5,
r'Flux $[10^{-17} \rm{erg/s/cm^2/\AA}]$',
va='center',
rotation='vertical')
ax1, ax2 = axes.flat
xmins = []
xmaxs = []
ymins = []
ymaxs = []
j = 0
for i in xrange(0, numfiles):
auxwav = wave_grid
auxflx = data['FLUX'][i]
# ======================= SMOOTHING =======================
box = 5
auxflx = smooth_array(auxflx, box)
if np.size(auxflx) != np.size(auxwav):
auxwav = auxwav[:np.size(auxflx)]
#print np.size(auxwav), np.size(auxflx)
# get the limits in y-axis
xmin = np.min(auxwav)
xmax = np.max(auxwav)
xmins.append(xmin)
xmaxs.append(xmax)
ymin = np.percentile(auxflx, 0.5)
ymins.append(ymin)
ymax = 2.0 * np.percentile(auxflx, 99)
ymaxs.append(ymax)
ax1.plot(auxwav, auxflx, color=line_colors[j], label=labels[i])
ax2.plot(auxwav, auxflx, color=line_colors[j], label=labels[i])
j = j + 1
xplotlim = (min(xmins), max(xmaxs), 0.5 * (min(xmins) + max(xmaxs)))
yplotlim = (min(ymins), max(ymaxs))
#print 'NEW GLOBAL MIN & MAX = ', yplotlim[0], yplotlim[1]
ax1.set_xlim(xplotlim[0], xplotlim[2])
ax2.set_xlim(xplotlim[2], xplotlim[1])
ax1.set_ylim(yplotlim[0], yplotlim[1])
ax2.set_xlabel(r'Observed wavelength $[\AA]$')
ax1.legend(loc='upper right', prop={'size': 10})
plt.savefig(outfile, format='pdf')
plt.show()
def main():
# ======================= PARSER =======================
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('--spec',
type=str,
default=None,
help='full path to the corrected FITS file, required')
args = parser.parse_args()
specpath = args.spec
save_dirname = os.path.dirname(os.path.dirname(specpath))
sdss_flag = os.path.isdir(os.path.join(save_dirname, 'sdss'))
if sdss_flag:
sdss_specpath = glob(os.path.join(save_dirname, 'sdss', '*.fits'))
# spec_sdss = spectrum()
# spec_sdss.__read_BOSS__(sdss_specpath[0])
# spec_sdss.__fit_powerlaw__('BOSS')
# ======================= UNIVERSAL WAVELENGTH GRID =======================
loglam_grid = np.arange(start=3.55, stop=4.02, step=0.0001)
wave_grid = np.power(10, loglam_grid)
wave_npix = wave_grid.size
wave_shape = wave_grid.shape
npix = wave_npix
# ======================= ======================== ======================
keys = ['SDSS', 'BOSS', 'CORR']
row_flux = np.dtype([('BOSS', 'f8', 1), ('CORR', 'f8', 1),
('SDSS', 'f8', 1)])
row_data = np.dtype([('WAVE', 'f8', 1), ('FLUX', row_flux, 1),
('FLUX_ERR', row_flux, 1), ('POWERLAW', row_flux, 1),
('CORR', row_flux, 1)])
row_si = np.dtype([('BOSS', 'f8', 3), ('CORR', 'f8', 3),
('SDSS', 'f8', 3)])
data = np.zeros(shape=(1, wave_npix), dtype=row_data)
data.fill(np.nan)
data['WAVE'] = wave_grid
si_data = np.ones(shape=(1, ), dtype=row_si)
for key in keys:
spec = spectrum()
if key == 'SDSS':
spec.__read_BOSS__(sdss_specpath[0])
filetype = 'BOSS'
else:
spec.__read_CORR__(specpath)
filetype = key
if key == 'CORR':
spec.flux = spec.flux_corr
spec.__fit_powerlaw__(filetype)
# ======================= REBINNING =======================
minwav = spec.wave.min()
maxwav = spec.wave.max()
# find where in the rebinned wavelength grid to put the rebinned flux values
index_start = np.argmin(abs(wave_grid - minwav))
index_stop = np.argmin(abs(wave_grid - maxwav))
# the spectra is rebinned to 'newpix' pixels
newpix = index_stop - index_start
print key, data['FLUX'][key]
# put new flux values & the powerlaw into the data structure
data[0]['FLUX'][key][index_start:index_stop] = congrid.rebin_1d(
spec.flux, newpix)
si_data[0][key][0] = spec.alpha
si_data[0][key][1] = spec.alpha_error
si_data[0][key][2] = spec.chisq
print save_dirname
# READ SPECTRUM FROM FILE & FIT POWERLAWS TO THE UNCORRECTED AND CORRECTED SPECTRA
# PRINT SI DATA
print '{0:-^80s}'.format(' POWER-LAW FIT ')
print '{0:>8}{1:^14s}{2:>10s}'.format('', 'SI', 'CHI2')
for key in keys:
print '{key} : {0:+6.4f}+-{1:6.4f}{2:>10.4f}'.format(key=key,
*si_data[0][key])
# CALCULATE THE CHI SQUARE OF BOSS AND CORR DATA (USING SDSS DATA AS EXPECTED VALUES)
if sdss_flag:
print '{0:-^80s}'.format(' AGREEMENT WITH SDSS DATA ')
print '{0:>7s}{1:>10s}'.format('', 'CHI2')
for key in keys:
if key != 'SDSS':
chisq = np.nansum(
(data[0]['FLUX'][key] - data[0]['FLUX']['SDSS'])**2 /
data[0]['FLUX']['SDSS']) / npix
print '{key} : {0:>10.4f}'.format(chisq, key=key)
sys.exit()
# ======================= BASTI'S SMOOTHING FUNCTION =======================
basti = True
if basti:
box = 5
flx = smooth_array(array=spec.flux, smooth_width=box)
flx_sdss = smooth_array(array=spec_sdss.flux, smooth_width=box)
flx_corr = smooth_array(array=spec.flux_corr, smooth_width=box)
# ======================= DEFINING THE CANVAS =======================
fsize = (12, 5)
fig = plt.figure(figsize=fsize)
plt.suptitle(
r'PLATE = %04d, MJD = %5d, FIBRE = %04d, RA = %6.3f, Dec = %6.3f , z = %5.3f'
% (spec.plateid, spec.MJD, spec.fiberid, spec.ra, spec.dec, spec.z))
fig.subplots_adjust(hspace=0.15)
# ======================= LABELS =======================
#fig.text(0.5,0.0, r'Observed wavelength $[\AA]$', ha='center')
#fig.text(0.09, 0.5, r'Flux $[10^{-17} \rm{erg/s/cm^2/\AA}]$', va='center', rotation='vertical')
# ======================= FIRST SUBPLOT =======================
majorLocator = MultipleLocator(500)
majorFormatter = FormatStrFormatter('%d')
minorLocator = MultipleLocator(250)
#### PLOT
ax2 = plt.subplot(223)
ax3 = plt.subplot(224)
ax1 = plt.subplot(211)
ax1.plot(spec_sdss.wave,
flx_sdss,
color='orange',
label='SDSS',
linewidth=1.4)
ax1.plot(spec.wave, flx, color='b', label='BOSS', linewidth=1.4)
ax1.plot(spec.wave,
flx_corr,
color='g',
label='Corrected BOSS',
linewidth=1.4)
#### SET LIMITS
xmin = np.min(spec.wave)
xmax = np.max(spec.wave)
xhalf = (xmin + xmax) / 2
ymin = min(0, np.percentile(spec.flux, 1))
ymax = 1.1 * np.percentile(spec.flux, 99.7)
xplotlim = (xmin, xmax)
yplotlim = (ymin, ymax)
ax1.set_xlim(xplotlim[0], xplotlim[1])
ax1.set_ylim(yplotlim[0], yplotlim[1])
ax1.xaxis.set_major_locator(majorLocator)
ax1.xaxis.set_minor_locator(minorLocator)
fig.text(0.5,
0.0,
r'Observed wavelength $[\AA]$',
ha='center',
va='bottom',
rotation='horizontal')
ax1.set_ylabel(r'Flux $[10^{-17} \rm{erg/s/cm^2/\AA}]$')
ax1.legend(loc='upper right', prop={'size': 10})
# ======================= SECOND SUBPLOT =======================
majorLocator = MultipleLocator(2000)
majorFormatter = FormatStrFormatter('%d')
minorLocator = MultipleLocator(1000)
#### REDEFINE PLOTTING ARRAYS
#flxarray = np.zeros(np.size(flx))
#flxcorrarray_margala = flx_corr_margala - flx
#flxcorrarray = flx_corr - flx
#### PLOT
ax2.plot(spec.wave,
spec.corr,
color='r',
label='Correction function',
linewidth=2)
ax2.set_ylabel(r'$C(\lambda)$')
ax2.set_xlim(xplotlim[0], xplotlim[1])
#ax2.set_xlabel(r'Observed wavelength $[\AA]$')
ax2.xaxis.set_major_locator(majorLocator)
ax2.xaxis.set_minor_locator(minorLocator)
#axes[1].legend(loc='lower right',prop={'size':12})
# ======================= THIRD SUBPLOT =======================
majorLocator = MultipleLocator(2000)
majorFormatter = FormatStrFormatter('%d')
minorLocator = MultipleLocator(1000)
#### REDEFINE PLOTTING ARRAYS
#flxarray = np.zeros(np.size(flx))
#flxcorrarray_margala = flx_corr_margala - flx
#flxcorrarray = flx_corr - flx
#### PLOT
ax3.plot(spec_sdss.wave,
spec_sdss.powerlaw,
color='orange',
linestyle='-',
linewidth=2.)
ax3.plot(spec.wave,
spec.powerlaw_BOSS,
color='b',
linestyle='-',
linewidth=2.)
ax3.plot(spec.wave,
spec.powerlaw_CORR,
color='g',
linestyle='-',
linewidth=2.)
colors = ['orange', 'blue', 'green']
alpha = [spec_sdss.alpha_BOSS, spec.alpha_BOSS, spec.alpha_CORR]
alpha_err = [
spec_sdss.alpha_error_BOSS, spec.alpha_error_BOSS,
spec.alpha_error_CORR
]
ymin, ymax = ax3.get_ylim()
dy = (ymax - ymin) / 10
for i in xrange(3):
text = r'$\alpha_{{\nu}} = {0:=+{w}.{p}f}\pm{1:{w}.{p}f} $'.format(
alpha[i], alpha_err[i], w=7, p=4)
ax3.text(7700, 0.8 * ymax - dy * i, text, color=colors[i])
ax3.set_xlim(xplotlim[0], xplotlim[1])
#ax3.set_xlabel(r'Observed wavelength $[\AA]$')
ax3.xaxis.set_major_locator(majorLocator)
ax3.xaxis.set_minor_locator(minorLocator)
ax3.set_ylabel(r'Flux $[10^{-17} \rm{erg/s/cm^2/\AA}]$')
#ax3.legend(loc='upper right',prop={'size':9})
plt.savefig(os.path.join(
save_dirname, '%04d-%05d-%04d-corrected_spectrum_2.pdf' %
(spec.plateid, spec.MJD, spec.fiberid)),
format='pdf')
plt.show()
def main():
# ======================= SETTINGS =======================
settings = program_settings()
settings.plot = True
settings.smoothness = 5
settings.resample = True
# ======================= PARSER =======================
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument(
'--spec-dir',
type=str,
default=None,
help='full path to the directory containing the FITS files, required')
args = parser.parse_args()
# ======================= FILES =======================
# from the input argument, fetch file names
if args.spec_dir:
spectral_list = return_spectral_list(args.spec_dir)
uncorrected_files = spectral_list[0]
corrected_files = spectral_list[1]
corrected_files_m = spectral_list[2]
# list that saves the number of files in each subfolder
numfiles = [
len(uncorrected_files),
len(corrected_files),
len(corrected_files_m)
]
labels = []
dir_basename = os.path.dirname(args.spec_dir)
thingid = dir_basename.split('/')[7]
# read the data from the uncorrected FITS files
ids = []
for file in uncorrected_files:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
ids.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
labels.append(ids)
# read the data from the corrected FITS files
ids = []
for file in corrected_files:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
ids.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
labels.append(ids)
# read the data from the corrected FITS files (Margala)
ids = []
for file in corrected_files_m:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
ids.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
labels.append(ids)
#data, header = fitsio.read(file,ext=0,header=True)
# ======================= ======================== ======================
wav_all = []
wav_ucorr = []
wav_corr = []
wav_corr_m = []
flx_all = []
flx_ucorr = []
flx_corr = []
flx_corr_m = []
npix_all = []
npix_ucorr = []
npix_corr = []
npix_corr_m = []
loglam_start_all = []
loglam_start_ucorr = []
loglam_start_corr = []
loglam_start_corr_m = []
loglam_end_all = []
loglam_end_ucorr = []
loglam_end_corr = []
loglam_end_corr_m = []
# ======================= READ DATA FROM FILES =======================
# uncorrected spectra
npix = []
for file in uncorrected_files:
spec = spectrum()
spec.__read_BOSS__(file)
wav_ucorr.append(spec.wave)
flx_ucorr.append(spec.flux)
npix_ucorr.append(spec.npix)
loglam_start_ucorr.append(spec.beginwl)
loglam_end_ucorr.append(spec.beginwl + (spec.npix * spec.deltawl))
ra = spec.ra
dec = spec.dec
z = spec.z
wav_all.append(wav_ucorr)
flx_all.append(flx_ucorr)
npix_all.append(npix_ucorr)
loglam_start_all.append(loglam_start_ucorr)
loglam_end_all.append(loglam_end_ucorr)
del spec
# corrected spectra
for file in corrected_files:
spec = spectrum()
spec.__read_CORR__(file)
wav_corr.append(spec.wave)
#flx_corr.append(spec.corr_flux)
flx_corr.append(spec.flux_corr)
npix_corr.append(spec.npix)
loglam_start_corr.append(np.log10(np.min(spec.wave)))
loglam_end_corr.append(np.log10(np.max(spec.wave)))
del spec
wav_all.append(wav_corr)
flx_all.append(flx_corr)
npix_all.append(npix_corr)
loglam_start_all.append(loglam_start_corr)
loglam_end_all.append(loglam_end_corr)
# corrected margala spectra
for file in corrected_files_m:
spec = spectrum()
spec.__read_CORR__(file)
wav_corr_m.append(spec.wave)
flx_corr_m.append(spec.flux_corr)
npix_corr_m.append(spec.npix)
loglam_start_corr_m.append(np.log10(np.min(spec.wave)))
loglam_end_corr_m.append(np.log10(np.max(spec.wave)))
del spec
wav_all.append(wav_corr_m)
flx_all.append(flx_corr_m)
npix_all.append(npix_corr_m)
loglam_start_all.append(loglam_start_corr_m)
loglam_end_all.append(loglam_end_corr_m)
# ======================= RESAMPLING THE FLUX ARRAYS =======================
# Each spectrum has a different number of pixels, 'starting' and 'end' wavelengths. We need to resample them a common pixel grid.
if settings.resample == True:
flx_all_resampled = []
# reshaped = transposed
flx_all_resampled_reshaped = []
# create a resampled array in wavelengths: take minimum of 'starting' and maximum of 'end' wavelengths
loglam_start = np.min(np.array(loglam_start_all))
loglam_end = np.max(np.array(loglam_end_all))
loglam_step = 0.0001
loglam_resampled = np.arange(loglam_start, loglam_end, loglam_step)
wav_resampled = ma.power(10, loglam_resampled)
## # RESAMPLING THE FLUXES
for i in xrange(np.size(numfiles)):
# i goes over 3 indices: uncorrected, corrected, corrected_margala
flx_kind = []
for j in xrange(numfiles[i]):
# j goes over the spectra
# locate the minimum and maximum of the wavelength array for this spectrum
minwav = np.min(wav_all[i][j])
maxwav = np.max(wav_all[i][j])
# define auxiliarry arrays
index = np.where((wav_resampled >= minwav)
& (wav_resampled <= maxwav))[0]
# find the interpolation function from data in the FITS file
intfunc = interp1d(wav_all[i][j], flx_all[i][j])
# apply the interpolation (but only within the proper boundaries)
# defining a masked array flx_resampled with the same mask as the wavelength array
flx_resampled = ma.array(np.full(wav_resampled.size,
np.nan)) #,mask = wav_mask)
# changing the values of the flx_resampled array, but only for indices that are not masked
flx_resampled[index] = intfunc(wav_resampled[index])
# define a mask in wavelengths not covered by the j-th spectrum
wav_mask = ma.masked_outside(wav_resampled, minwav,
maxwav).mask
# define a mask of outliers (further than nsigma)
nsigma = 8
outliers_boolean = is_outlier(flx_resampled, nsigma)
outliers_mask = ma.make_mask(outliers_boolean)
# combine the masks into a new mask and attach it to the flx_resampled
mask = ma.mask_or(wav_mask, outliers_mask)
flx_resampled.mask = mask
#for k in xrange(np.size(wav_resampled)):
# print i,j,k, wav_resampled[k], flx_resampled[k]
flx_kind.append(flx_resampled)
# append the arrays
flx_kind = ma.array(flx_kind)
flx_all_resampled.append(flx_kind)
flx_kind = ma.array(flx_kind).T
flx_all_resampled_reshaped.append(flx_kind)
# ======================= CALCULATE THE VARIANCE =======================
flx_resampled_array = ma.array(flx_all_resampled_reshaped)
print 'Calculating the variance'
medFlux = []
devFlux = []
for i in xrange(np.size(numfiles)):
# i goes over the uncorrected, corrected, corrected_margala
median_i = []
dev_i = []
for j in xrange(np.size(flx_resampled_array[i]) / numfiles[i]):
# j goes over the pixels
fluxes = []
for k in xrange(numfiles[i]):
# k goes over individual exposures (spectra)
fluxes.append(flx_resampled_array[i][j][k])
median = np.nanmedian(fluxes)
dev = np.nanstd(fluxes)
median_i.append(median)
dev_i.append(dev)
medFlux.append(median_i)
devFlux.append(dev_i)
medFluxArray = np.array(medFlux)
devFluxArray = np.array(devFlux)
print '{0:>50s}'.format('STANDARD DEVIATION OF FLUX')
print '{0:>31s} {1:>15s}'.format('MEDIAN', 'STD')
print '{0:>20s} {1:15.12f} {2:15.12f}'.format(
'UNCORRECTED:', np.nanmedian(devFluxArray[0]),
np.nanstd(devFluxArray[0]))
print '{0:>20s} {1:15.12f} {2:15.12f}'.format(
'CORRECTED (Ours):', np.nanmedian(devFluxArray[1]),
np.nanstd(devFluxArray[1]))
print '{0:>20s} {1:15.12f} {2:15.12f}'.format(
'CORRECTED (Margala):', np.nanmedian(devFluxArray[2]),
np.nanstd(devFluxArray[2]))
# RMS IN THE LYMAN ALPHA FOREST
index = np.where(wav_resampled <= 1215.67 * (1 + z))[0]
medFluxArray_LyA = medFluxArray[:, index]
devFluxArray_LyA = devFluxArray[:, index]
if medFluxArray_LyA.size != 0:
print '{0:>50s}'.format(
'STANDARD DEVIATION OF FLUX IN THE LYMAN ALPHA FOREST')
print '{0:>31s} {1:>15s} {2:>15s} {3:>15s}'.format(
'MEDIAN', 'STD', 'MIN', 'MAX')
print '{0:>20s} {1:15.12f} {2:15.12f} {3:15.12f} {4:15.12f}'.format(
'UNCORRECTED:', np.nanmedian(devFluxArray_LyA[0]),
np.nanstd(devFluxArray_LyA[0]), np.nanmin(devFluxArray_LyA[0]),
np.nanmax(devFluxArray_LyA[0]))
print '{0:>20s} {1:15.12f} {2:15.12f} {3:15.12f} {4:15.12f}'.format(
'CORRECTED (Ours):', np.nanmedian(devFluxArray_LyA[1]),
np.nanstd(devFluxArray_LyA[1]), np.nanmin(devFluxArray_LyA[1]),
np.nanmax(devFluxArray_LyA[1]))
print '{0:>20s} {1:15.12f} {2:15.12f} {3:15.12f} {4:15.12f}'.format(
'CORRECTED (Margala):', np.nanmedian(devFluxArray_LyA[2]),
np.nanstd(devFluxArray_LyA[2]), np.nanmin(devFluxArray_LyA[2]),
np.nanmax(devFluxArray_LyA[2]))
settings.plot = True
if settings.plot == True:
# ======================= PLOT =======================
majorLocator = MultipleLocator(500)
majorFormatter = FormatStrFormatter('%d')
minorLocator = MultipleLocator(100)
# PLOT NAME
outfile = os.path.join(args.spec_dir,
'%s_rms_triple_panel.pdf' % thingid)
# PLOT PARAMETERS
cmap = plt.cm.Spectral
line_colors = cmap(np.linspace(0, 1, len(uncorrected_files)))
#colors = ['Black','Crimson','Gold','Green','Navy','DarkMagenta','CornflowerBlue','Sienna','Green','Purple','Teal','MediumSpringGreen','Turquoise','FireBrick','Brown','Bisque','Yellow','Coral','Linen']
# ======================= DEFINING THE CANVAS =======================
# PLOT SIZE
fsize = (12, 8)
# MAKE SUBPLOTS
fig, axes = plt.subplots(3, 1, figsize=fsize, sharey=True)
fig.subplots_adjust(hspace=0.02)
ax1, ax2, ax3 = axes.flat
# SET TITLE
plt.suptitle('THING ID = %9s' % (thingid))
# plt.suptitle('RA = %10.6f, DEC = %10.6f , z = %5.3f' %(ra,dec,z))
# ======================= AXES LABELS =======================
#fig.text(0.5,0.0, r'Observed wavelength $[\AA]$', ha='center')
fig.text(0.08,
0.5,
r'Flux $[10^{-17} \rm{erg/s/cm^2/\AA}]$',
va='center',
rotation='vertical')
xmins = []
xmaxs = []
ymins = []
ymaxs = []
# PLOTTING DATA
box = settings.smoothness
for i in xrange(np.size(numfiles)):
# i goes over the uncorrected, corrected, corrected_margala
k = 0
auxmedian = smooth_array(np.array(medFlux[i]), box)
auxdev = smooth_array(np.array(devFlux[i]), box)
for j in xrange(numfiles[i]):
# j goes over the spectra
auxwav = ma.array(wav_resampled,
mask=flx_all_resampled[i][j].mask)
auxflx = flx_all_resampled[i][j]
# ======================= SMOOTHING =======================
auxflx = smooth_array(auxflx, box)
#if np.size(auxflx)!=np.size(auxwav):
#auxwav = auxwav[:np.size(auxflx)]
# get the limits in y-axis
xmin = np.min(auxwav)
xmax = np.max(auxwav)
xmins.append(xmin)
xmaxs.append(xmax)
ymin = np.percentile(auxflx, 0.5)
ymins.append(ymin)
ymax = 1.2 * np.percentile(auxflx, 99)
ymaxs.append(ymax)
if i == 0:
ax1.plot(auxwav,
auxflx,
color=line_colors[k],
label=labels[i][j],
linestyle=':',
alpha=0.6)
elif i == 1:
ax2.plot(auxwav,
auxflx,
color=line_colors[k],
label=labels[i][j],
linestyle=':',
alpha=0.6)
elif i == 2:
ax3.plot(auxwav,
auxflx,
color=line_colors[k],
label=labels[i][j],
linestyle=':',
alpha=0.6)
k = k + 1
if i == 0:
sigma = ma.array(devFluxArray[0],
mask=flx_all_resampled[i][j].mask)
# PLOT MEDIAN FLUX AND RMS
ax1.plot(auxwav, auxmedian, color='Green', linewidth=1.3)
ax1.plot(auxwav, auxdev, color='Red', linewidth=1)
# FILL AREA OF RMS
ax1.fill_between(auxwav,
auxmedian - sigma,
auxmedian + sigma,
color="PaleGreen")
# FILL AREA OF LY A FOREST
ax1.axvspan(10**loglam_start,
1215.67 * (1 + z),
alpha=0.3,
color="SkyBlue")
print 10**loglam_start, 1215.67 * (1 + z)
elif i == 1:
sigma = ma.array(devFluxArray[1],
mask=flx_all_resampled[i][j].mask)
ax2.plot(auxwav, auxmedian, color='Green', linewidth=1.3)
ax2.plot(auxwav, auxdev, color='Red', linewidth=1)
ax2.fill_between(auxwav,
auxmedian - sigma,
auxmedian + sigma,
color="PaleGreen")
ax2.axvspan(10**loglam_start,
1215.67 * (1 + z),
alpha=0.5,
color="SkyBlue")
elif i == 2:
sigma = ma.array(devFluxArray[2],
mask=flx_all_resampled[i][j].mask)
ax3.plot(auxwav, auxmedian, color='Green', linewidth=1.3)
ax3.plot(auxwav, auxdev, color='Red', linewidth=1)
ax3.fill_between(auxwav,
auxmedian - sigma,
auxmedian + sigma,
color="PaleGreen")
ax3.axvspan(10**loglam_start,
1215.67 * (1 + z),
alpha=0.5,
color="SkyBlue")
xplotlim = (min(xmins), max(xmaxs), 0.5 * (min(xmins) + max(xmaxs)))
yplotlim = (min(ymins), max(ymaxs))
# AXES FINE TUNING
ax1.text(6000, 0.85 * yplotlim[1], 'UNCORRECTED spectra')
ax1.text(6000, 0.70 * yplotlim[1], 'RMS in the LyA forest')
ax1.text(
6000, 0.55 * yplotlim[1],
'{0:>5s} {1:5.2f} {2:2s} {3:5.2f}'.format(
'$\sigma$ = ', np.nanmedian(devFluxArray_LyA[0]), '$\pm$',
np.nanstd(devFluxArray_LyA[0])))
ax2.text(6000, 0.85 * yplotlim[1], 'CORRECTED spectra (this paper)')
ax2.text(6000, 0.70 * yplotlim[1], 'RMS in the LyA forest')
ax2.text(
6000, 0.55 * yplotlim[1],
'{0:>5s} {1:5.2f} {2:2s} {3:5.2f}'.format(
'$\sigma$ = ', np.nanmedian(devFluxArray_LyA[1]), '$\pm$',
np.nanstd(devFluxArray_LyA[1])))
ax3.text(6000, 0.85 * yplotlim[1], 'CORRECTED spectra (Margala)')
ax3.text(6000, 0.70 * yplotlim[1], 'RMS in the LyA forest')
ax3.text(
6000, 0.55 * yplotlim[1],
'{0:>5s} {1:5.2f} {2:2s} {3:5.2f}'.format(
'$\sigma$ = ', np.nanmedian(devFluxArray_LyA[2]), '$\pm$',
np.nanstd(devFluxArray_LyA[2])))
ax1.set_xlim(xplotlim[0], xplotlim[1])
ax2.set_xlim(xplotlim[0], xplotlim[1])
ax3.set_xlim(xplotlim[0], xplotlim[1])
ax1.set_ylim(yplotlim[0], yplotlim[1])
ax3.set_xlabel(r'Observed wavelength $[\AA]$')
# LEGEND
ax1.legend(loc='upper right', prop={'size': 9})
plt.savefig(outfile, format='pdf')
plt.show()
def main():
# PARSER
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument(
'--tpcorr',
type=str,
default=
'/Volumes/Transcend/Isotropy/spectra/BOSScorrections/tpcorr-0.1/tpcorr.hdf5',
help='throughput correction filename, required')
parser.add_argument('--all',
action='store_true',
help='save corrected spectra for all available QSOs')
args = parser.parse_args()
# MARGALA CORRECTION FILE
tpcorr = h5py.File(args.tpcorr, 'r')
tpcorr_wave = tpcorr['wave'].value
# DIRECTORY TO SAVE THE CORRECTED SPECTRA TO, USER-DEFINED
dirpath = os.path.normpath(
'/Volumes/Transcend/Isotropy/spectra/SDSS_DR12/QSO_Margala')
# -------------------------------------------------------------------------------------
# Parse arguments: extract plate, mjd, fiberid as integers and the spec filename
# ids = list of tuples : (plate,mjd,fiberid)
# sid = list of strings : string(plate+mjd+fiberid) ---> used to check already corrected spectra
ids = []
sid_tot = []
if args.all:
print "CORRECTING ALL AVAILABLE SPECTRA"
# CREATE A LIST OF ALL FILES TO BE CORRECTED : total_files
PathToSpectra = os.path.normpath(
"/Volumes/Transcend/Isotropy/spectra/SDSS_DR12/QSO")
total_files = glob(os.path.join(PathToSpectra, "*.fits"))
print "Spectra available: {0:8d}".format(len(total_files))
for file in total_files:
# APPEND THE UNIQUE IDENTIFIER TO sid_tot
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
ids.append((plate, mjd, fiberid))
sid_tot.append('%04d%5d%04d' % (plate, mjd, fiberid))
# CREATE A PYTHON SET OBJECT FOR ALL SIDs (ALLOWS INTERSECTION, DIFERENCE)
sid_tot_array = np.array(sid_tot)
sid_tot = set(sid_tot)
# CREATE A LIST OF ALL FILES ALREADY CORRECTED AND PRESENT IN THE OUTPUT DIRECTORY: corrected_files
corrected_files = glob(os.path.join(dirpath, "*.fits"))
print "Spectra corrected: {0:8d}".format(len(corrected_files))
sid_corr = []
for file in corrected_files:
# APPEND THE UNIQUE IDENTIFIERS TO sid_corr
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [
int(field)
for field in os.path.splitext(spec_basename)[0].split('-')[1:]
]
sid_corr.append('%04d%5d%04d' % (plate, mjd, fiberid))
# CREATE A PYTHON SET OBJECT FOR CORRECTED SIDs
sid_corr = set(sid_corr)
sid_unc = sid_tot.difference(sid_corr)
# FIND THE DIFFERENCE OF THE TWO SETS (sid_tot - sid_corr)
# APPEND SIDs TO uncorrected_files ---> used later
uncorrected_files = []
for i, file in enumerate(total_files):
if sid_tot_array[i] in sid_unc:
uncorrected_files.append(file)
# ITERATE OVER FILES IN uncorrected_files AND APPLY THE CORRECTION
print "To be corrected: {0:8d}".format(len(uncorrected_files))
file_array = np.array(uncorrected_files)
for file in uncorrected_files:
HDU = fitsio.FITS(file)
flux = HDU[1]['flux'][:]
ivar = HDU[1]['ivar'][:]
wavelength = np.power(10, HDU[1]['loglam'][:])
ra = HDU[2]['ra'][0][0]
dec = HDU[2]['dec'][0][0]
z = HDU[2]['z'][0][0]
HDU.close()
plate, mjd, fiberid = ids[np.argwhere(file_array == file)[0][0]]
# Read the target's throughput correction vector
tpcorr_key = '%s/%s/%s' % (plate, mjd, fiberid)
try:
correction = tpcorr[tpcorr_key].value
except:
print "No data on object found"
continue
# Create an interpolated correction function
correction_interp = interp1d(tpcorr_wave, correction, kind='linear')
# Sample the interpolated correction using the observation's wavelength grid
resampled_correction = correction_interp(wavelength)
# Apply the correction to the observed flux and ivar
corrected_flux = flux * resampled_correction
corrected_ivar = ivar / resampled_correction**2
# Save the data into a fits file
spec = spectrum()
spec.plateid = plate
spec.mjd = mjd
spec.fiberid = fiberid
spec.ra = ra
spec.dec = dec
spec.z = z
spec.npix = np.size(wavelength)
spec.wave = wavelength
spec.flux = flux
spec.ivar = ivar
spec.flux_corr = corrected_flux
spec.flux_error = np.sqrt(1. / corrected_ivar)
spec.corr = resampled_correction
spec.basename = os.path.basename(file)
spec.dirname = os.path.dirname(file)
spec.__save_fitsio__(dirpath=dirpath)
del (spec)
tpcorr.close()
def main():
# ======================= SETTINGS =======================
settings = program_settings()
settings.plot = True
settings.smoothness = 5
settings.resample = True
# ======================= PARSER =======================
parser = argparse.ArgumentParser(
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('--spec-dir', type=str, default=None,
help='full path to the directory containing the FITS files, required')
parser.add_argument('--panels', type=int, default=3,
help='number of panels, default = 3')
args = parser.parse_args()
# ======================= FILES =======================
# from the input argument, fetch file names
if args.spec_dir:
spectral_list = return_spectral_list(args.spec_dir)
numspec = len(spectral_list[0])
uncorrected_files = spectral_list[0]
corrected_files = spectral_list[1]
corrected_files_m = spectral_list[2]
# list that saves the number of files in each subfolder
numfiles = [len(uncorrected_files),len(corrected_files),len(corrected_files_m)]
labels = []
dir_basename = os.path.dirname(args.spec_dir)
thingid = dir_basename.split('/')[7]
# read the data from the uncorrected FITS files
ids = []
for file in uncorrected_files:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [int(field) for field in os.path.splitext(spec_basename)[0].split('-')[1:]]
ids.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
labels.append(ids)
# read the data from the corrected FITS files
ids = []
for file in corrected_files:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [int(field) for field in os.path.splitext(spec_basename)[0].split('-')[1:]]
ids.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
labels.append(ids)
# read the data from the corrected FITS files (Margala)
ids = []
for file in corrected_files_m:
spec_basename = os.path.basename(file).split('.fits')[0]
plate, mjd, fiberid = [int(field) for field in os.path.splitext(spec_basename)[0].split('-')[1:]]
ids.append('%04d-%5d-%04d' % (plate, mjd, fiberid))
labels.append(ids)
#data, header = fitsio.read(file,ext=0,header=True)
# ======================= UNIVERSAL WAVELENGTH GRID =======================
loglam_grid = np.arange(start=3.55, stop = 4.02, step = 0.0001)
wave_grid = np.power(10, loglam_grid)
wave_npix = wave_grid.size
wave_shape = wave_grid.shape
# ======================= ======================== ======================
flux_boss_list = []; flux_corr_list = []; flux_marg_list = []
keys = ['BOSS','CORR','MARG']
row = np.dtype([('BOSS','f8',1),('CORR','f8',1),('MARG','f8',1)])
flux_data = np.zeros(shape = (max(numfiles),wave_npix), dtype=row)
si_data = np.zeros(shape = (max(numfiles)), dtype=row)
pwrlaw_data = np.zeros(shape = (max(numfiles),wave_npix), dtype=row)
# ======================= READ DATA FROM FILES =======================
for i in xrange(len(spectral_list)):
for j,file in enumerate(spectral_list[i]):
# ======================= READ SPECTRA =======================
spec = spectrum()
spec_dirname = os.path.dirname(file)
spec_basename = os.path.basename(file)
if 'corr' not in spec_basename:
spec.__read_BOSS__(file)
flux_boss = np.zeros(shape=wave_shape)
elif 'corr' in spec_basename:
spec.__read_CORR__(file)
flux_corr = np.zeros(shape=wave_shape)
# ======================= REBINNING =======================
minwav = spec.wave.min() ; maxwav = spec.wave.max()
index_start = np.argmin(abs(wave_grid-minwav))
index_stop = np.argmin(abs(wave_grid-maxwav))
newpix = index_stop - index_start
if 'corr' not in spec_basename:
spec.__fit_powerlaw__(type='BOSS')
si_data[j]['BOSS'] = spec.alpha
flux_data[j]['BOSS'][index_start:index_stop] = congrid.rebin_1d(spec.flux, newpix)
pwrlaw_data[j]['BOSS'][index_start:index_stop] = congrid.rebin_1d(spec.powerlaw, newpix)
if 'corr' in spec_basename:
spec.__fit_powerlaw__(type='CORR')
if 'margala' not in spec_dirname:
si_data[j]['CORR'] = spec.alpha
flux_data[j]['CORR'][index_start:index_stop] = congrid.rebin_1d(spec.flux_corr, newpix)
pwrlaw_data[j]['CORR'][index_start:index_stop] = congrid.rebin_1d(spec.powerlaw, newpix)
elif 'margala' in spec_dirname:
si_data[j]['MARG'] = spec.alpha
flux_data[j]['MARG'][index_start:index_stop] = congrid.rebin_1d(spec.flux_corr, newpix)
pwrlaw_data[j]['MARG'][index_start:index_stop] = congrid.rebin_1d(spec.powerlaw, newpix)
ra = spec.ra ; dec = spec.dec; z = spec.z
del(spec)
# TRANSPOSE THE FLUXES TO CALCULATE RMS
flux = np.transpose(flux_data)
# ======================= CALCULATE THE VARIANCE =======================
print 'Calculating the variance'
flux_rms = np.zeros(shape = wave_shape, dtype = row)
flux_std = np.zeros(shape = wave_shape, dtype = row)
for pix in xrange(wave_npix):
flux_rms[pix]['BOSS'] = np.sqrt(np.mean(np.square(flux['BOSS'][pix]))); flux_std[pix]['BOSS'] = np.std(flux['BOSS'][pix])
flux_rms[pix]['CORR'] = np.sqrt(np.mean(np.square(flux['CORR'][pix]))); flux_std[pix]['CORR'] = np.std(flux['CORR'][pix])
flux_rms[pix]['MARG'] = np.sqrt(np.mean(np.square(flux['MARG'][pix]))); flux_std[pix]['MARG'] = np.std(flux['MARG'][pix])
medianSpec = median_spectrum(z, wave_grid, flux_rms, flux_std)
print '{0:>50s}'.format('STANDARD DEVIATION OF FLUX')
print '{0:>31s} {1:>15s}'.format('RMS','STD')
print '{0:>20s} {1:15.12f} {2:15.12f}'.format('UNCORRECTED:', np.nanmedian(flux_rms['BOSS']), np.nanstd(flux_rms['BOSS']))
print '{0:>20s} {1:15.12f} {2:15.12f}'.format('CORRECTED (Ours):', np.nanmedian(flux_rms['CORR']), np.nanstd(flux_rms['CORR']))
print '{0:>20s} {1:15.12f} {2:15.12f}'.format('CORRECTED (Margala):', np.nanmedian(flux_rms['MARG']), np.nanstd(flux_rms['MARG']))
settings.plot=True
if settings.plot==True:
npanels = args.panels
# ======================= PLOT =======================
majorLocator = MultipleLocator(500)
majorFormatter = FormatStrFormatter('%d')
minorLocator = MultipleLocator(100)
# PLOT NAME
if npanels == 1:
outfile = os.path.join(args.spec_dir,'%s_rms_single_panel.pdf' %thingid)
fsize = (12,3)
elif npanels == 2:
outfile = os.path.join(args.spec_dir,'%s_rms_double_panel.pdf' %thingid)
fsize = (12,5)
elif npanels == 3:
outfile = os.path.join(args.spec_dir,'%s_rms_triple_panel.pdf' %thingid)
fsize = (12,7)
# PLOT PARAMETERS
cmap = plt.get_cmap('Paired')
line_colors = cmap(np.linspace(0,1,len(uncorrected_files)))
# ======================= DEFINING THE CANVAS =======================
# PLOT SIZE
fsize = (12,5)
# MAKE SUBPLOTS
fig, axs = plt.subplots(npanels,1,figsize=fsize,sharey=True)
ax = np.array(axs)
fig.subplots_adjust(hspace=0.02)
# SET TITLE
plt.suptitle('THING ID = %9s' %(thingid))
# plt.suptitle('RA = %10.6f, DEC = %10.6f , z = %5.3f' %(ra,dec,z))
# ======================= AXES LABELS =======================
#fig.text(0.5,0.0, r'Observed wavelength $[\AA]$', ha='center')
fig.text(0.08, 0.5, r'Flux $[10^{-17} \rm{erg/s/cm^2/\AA}]$', va='center', rotation='vertical')
xmins = []
xmaxs = []
ymins = []
ymaxs = []
# PLOTTING DATA
box=settings.smoothness
auxwav = wave_grid
for i, key in zip(xrange(npanels),keys[0:npanels]):
print 'PLOTTING {0:s}'.format(key)
# i goes over the uncorrected, corrected, corrected_margala
k = 0
# ======================= SMOOTHING =======================
auxmedian = smooth_array(flux_rms[key],box)
auxdev = smooth_array(flux_std[key],box)
for j in xrange(numfiles[i]):
# j goes over the spectra
auxflx = smooth_array(flux_data[j][key], box)
if np.size(auxflx)!=np.size(auxwav):
print 'error'
print auxwav.shape
print auxflx.shape
sys.exit()
#auxwav = auxwav[:np.size(auxflx)]
# get the limits in y-axis
ymin = np.percentile(auxflx, 0.5); ymins.append(ymin)
ymax = 1.2*np.percentile(auxflx, 99.5); ymaxs.append(ymax)
ax[i].plot(auxwav,auxflx,color=line_colors[k],label=labels[i][j],linestyle=':', alpha=0.6)
k = k+1
# PLOT MEDIAN FLUX AND RMS
sigma = flux_std[key]
ax[i].plot(auxwav,auxmedian,color='Green',linewidth=1.3)
ax[i].plot(auxwav,auxdev,color='Red',linewidth=1)
# FILL AREA OF RMS
ax[i].fill_between(auxwav, auxmedian - sigma, auxmedian + sigma, color="Green", alpha=0.3)
# FILL AREA OF LY A FOREST
#ax[i].axvspan(wave_grid.min(), 1215.67*(1+z), alpha=0.3, color="SkyBlue")
xplotlim = (wave_grid.min(), wave_grid.max())
yplotlim = (min(ymins),max(ymaxs))
for i in xrange(npanels):
ax[i].set_xlim(xplotlim[0],xplotlim[1])
ax[i].set_ylim(yplotlim[0],yplotlim[1])
ax[npanels-1].set_xlabel(r'Observed wavelength $[\AA]$')
#0.08, 0.5, r'Flux $[10^{-17} \rm{erg/s/cm^2/\AA}]$', va='center', rotation='vertical')
ax[0].text(0.96,0.5,'UNCORRECTED spectra', va='center', rotation='vertical')
fig.text(6000,0.85*yplotlim[1],'CORRECTED spectra (this paper)')
fig.text(6000,0.85*yplotlim[1],'CORRECTED spectra (Margala)')
plt.savefig(outfile, format = 'pdf')
print 'PLOT SAVED TO {0:s}'.format(outfile)
plt.show()
