def prospector_LGAL_sourceSpec_i(galid, mask=False, infer_method='dynesty'):
''' run prospector on L-Gal source spectra
'''
# read in source spectra
f_name = 'gal_spectrum_'+str(galid)+'_BGS_template_BC03_Stelib.fits'
f_inspec = fits.open(''.join([UT.dat_dir(), 'Lgal/templates/', f_name]))
specin = f_inspec[1].data
zred = f_inspec[0].header['REDSHIFT']
wave = specin['wave']
flux = specin['flux_nodust_nonoise'] * 1e3 * 1e17 # in units of erg/s/A/cm2
prosp = Fitters.Prospector()
str_mask = ''
if mask: str_mask = '.masked'
f_name = ''.join([UT.dat_dir(), 'Lgal/templates/',
'prospector.', infer_method, str_mask,
'.gal_spectrum_', str(galid), '_BGS_template_BC03_Stelib.h5'])
if infer_method == 'dynesty': # run dynamic nested sampling
prosp.dynesty_spec(wave, flux, None, zred, nested=True, maxcall_init=50000, maxcall=100000,
write=True, output_file=f_name)
elif infer_method == 'emcee': # emcee
prosp.emcee_spec(wave, flux, None, zred, mask=mask,
write=True, output_file=f_name, silent=False)
return None
def prospector_LGAL_desiSpec_i(galid, mask=False, infer_method='dynesty'):
''' run prospector on L-Gal DESI-like spectra
'''
# read in source spectra (only to get the redshift)
f_name = 'gal_spectrum_'+str(galid)+'_BGS_template_BC03_Stelib.fits'
f_inspec = fits.open(''.join([UT.dat_dir(), 'Lgal/templates/', f_name]))
zred = f_inspec[0].header['REDSHIFT']
# read desi-like spectra
f_name = 'gal_spectrum_'+str(galid)+'_BGS_template_BC03_Stelib.fits'
f_outspec = ''.join([UT.dat_dir(), 'Lgal/spectra/', 'desi_out_', f_name])
spec_desi = desiIO.read_spectra(f_outspec)
wave = np.concatenate([spec_desi.wave[b] for b in ['b', 'r', 'z']])
flux = np.concatenate([spec_desi.flux[b][0] for b in ['b', 'r', 'z']]) # 10-17 ergs/s/cm2/AA
flux_unc = np.concatenate([spec_desi.ivar[b][0]**-0.5 for b in ['b', 'r', 'z']])
prosp = Fitters.Prospector()
str_mask = ''
if mask: str_mask = '.masked'
f_name = ''.join([UT.dat_dir(), 'Lgal/spectra/',
'prospector.', infer_method, str_mask,
'.desi_out_gal_spectrum_', str(galid), '_BGS_template_BC03_Stelib.h5'])
if infer_method == 'dynesty': # run dynamic nested sampling
prosp.dynesty_spec(wave, flux, flux_unc, zred, mask=mask,
nested=True, maxcall_init=25000, maxcall=50000,
write=True, output_file=f_name)
elif infer_method == 'emcee': # emcee
prosp.emcee_spec(wave, flux, flux_unc, zred, mask=mask,
write=True, output_file=f_name, silent=False)
return None
def firefly_LGAL_sourceSpec_i(galid,
model='m11',
model_lib='MILES',
imf='cha',
hpf_mode='on'):
''' run firefly on L-Gal source spectra
'''
# read in source spectra
f_name = 'gal_spectrum_' + str(galid) + '_BGS_template_BC03_Stelib.fits'
f_inspec = fits.open(''.join([UT.dat_dir(), 'Lgal/templates/', f_name]))
specin = f_inspec[1].data
spec_in = {}
spec_in['redshift'] = f_inspec[0].header['REDSHIFT']
spec_in['wave'] = specin['wave']
spec_in['flux'] = specin[
'flux_dust_nonoise'] * 1e-4 * 1e7 * 1e17 #from W/A/m2 to 10e-17 erg/s/A/cm2
spec_in[
'flux_dust_nonoise'] = specin['flux_dust_nonoise'] * 1e-4 * 1e7 * 1e17
spec_in['flux_nodust_nonoise'] = specin[
'flux_nodust_nonoise'] * 1e-4 * 1e7 * 1e17
# structure source spectra for firefly read in
gspec = Spec.GSfirefly()
gspec.generic(spec_in['wave'],
spec_in['flux_dust_nonoise'],
redshift=spec_in['redshift'])
gspec.path_to_spectrum = UT.dat_dir()
# output firefly file
f_name = ''.join([
'firefly.', model, '.', model_lib, '.imf_', imf, '.dust_', hpf_mode,
'.', 'gal_spectrum_',
str(galid), '_BGS_template_BC03_Stelib.hdf5'
])
f_firefly = ''.join([UT.dat_dir(), 'Lgal/templates/', f_name])
firefly = Fitters.Firefly(
gspec,
f_firefly, # output file
co.Planck13, # comsology
models=model, # model ('m11', 'bc03', 'm09')
model_libs=[model_lib], # model library for M11
imfs=[imf], # IMF used ('ss', 'kr', 'cha')
hpf_mode=
hpf_mode, # uses HPF to dereden the spectrum
age_limits=[0, 15],
Z_limits=[-3., 5.],
wave_limits=[3350., 9000.],
suffix=None,
downgrade_models=False,
data_wave_medium='vacuum',
use_downgraded_models=False,
write_results=True)
bestfit = firefly.fit_models_to_data()
return None
def obs_condition(sampling='spacefill', overwrite=False):
''' Sample mock BGS observing conditions
'''
if sampling == 'spacefill':
f_obscond = ''.join([UT.dat_dir(), 'spectral_challenge/',
'bgs_mockexp_obscond_testset.', sampling, '.txt'])
else:
raise NotImplementedError("maybe implement random later")
if os.path.isfile(f_obscond) and not overwrite:
out_obscond = np.loadtxt(f_obscond, unpack=True, skiprows=1)
else:
# read in bgs mock exposure observing conditions generated from
# https://github.com/changhoonhahn/feasiBGS/blob/master/notebook/local_bgs_mockexp.ipynb
f_exp = ''.join([UT.dat_dir(), 'spectral_challenge/',
'bgs_survey_exposures.withsun.hdf5'])
f = h5py.File(f_exp, 'r')
mock_exps = {}
for k in f.keys():
mock_exps[k] = f[k].value
f.close()
nexps = len(mock_exps['RA'])
# we want to sample main meta-data from the mock exposures
meta_keys = ['MOONFRAC', 'MOONALT', 'SUNALT'] #'AIRMASS', 'SUNSEP', 'MOONSEP'
meta = np.zeros((nexps, len(meta_keys)))
for i, k in enumerate(meta_keys):
meta[:,i] = mock_exps[k]
if sampling == 'spacefill':
histmd, edges = np.histogramdd(meta, 2)
_hasexp = histmd > 0.
has_exp = np.where(_hasexp)
print('%i exposures' % np.sum(_hasexp))
iexp_sample = []
for i in range(np.sum(_hasexp)):
in_bin = np.ones(nexps).astype(bool)
for i_dim in range(len(meta_keys)):
in_bin = (in_bin &
(meta[:,i_dim] > edges[i_dim][has_exp[i_dim]][i]) &
(meta[:,i_dim] <= edges[i_dim][has_exp[i_dim]+1][i]))
assert np.sum(in_bin) > 0
iexp_sample.append(np.random.choice(np.arange(nexps)[in_bin], 1)[0])
iexp_sample = np.array(iexp_sample)
out_keys = ['AIRMASS', 'SEEING', 'EXPTIME', 'MOONFRAC', 'MOONALT', 'MOONSEP', 'SUNALT', 'SUNSEP']
hdr = ', '.join([k.lower() for k in out_keys])
out_obscond = np.zeros((len(iexp_sample), len(out_keys)))
for i, k in enumerate(out_keys):
out_obscond[:,i] = mock_exps[k][iexp_sample]
np.savetxt(f_obscond, out_obscond, header=hdr)
return out_obscond.T
def firefly_LGAL_desiSpec_i(galid,
model='m11',
model_lib='MILES',
imf='cha',
hpf_mode='on'):
''' run firelfy on L-Gal DESI-like spectra
'''
t0 = time.time()
# read in source spectra (only to get the redshift)
f_name = 'gal_spectrum_' + str(galid) + '_BGS_template_BC03_Stelib.fits'
f_inspec = fits.open(''.join([UT.dat_dir(), 'Lgal/templates/', f_name]))
redshift = f_inspec[0].header['REDSHIFT']
# read desi-like spectra
f_name = 'gal_spectrum_' + str(galid) + '_BGS_template_BC03_Stelib.fits'
f_outspec = ''.join([UT.dat_dir(), 'Lgal/spectra/', 'desi_out_', f_name])
spec_desi = desiIO.read_spectra(f_outspec)
gspec = Spec.GSfirefly()
gspec.DESIlike(spec_desi, redshift=redshift)
gspec.path_to_spectrum = UT.dat_dir()
# output firefly file
f_out = ''.join([
'firefly.', model, '.', model_lib, '.imf_', imf, '.dust_', hpf_mode,
'.', 'desi_out_',
f_name.split('.fits')[0], '.hdf5'
])
f_firefly = ''.join([UT.dat_dir(), 'Lgal/spectra/', f_out])
firefly = Fitters.Firefly(
gspec,
f_firefly, # output file
co.Planck13, # comsology
models=model, # model ('m11', 'bc03', 'm09')
model_libs=[model_lib], # model library for M11
imfs=[imf], # IMF used ('ss', 'kr', 'cha')
hpf_mode=
hpf_mode, # uses HPF to dereden the spectrum
age_limits=[0, 15],
Z_limits=[-3., 5.],
wave_limits=[3350., 9000.],
suffix=None,
downgrade_models=False,
data_wave_medium='vacuum',
use_downgraded_models=False,
write_results=True)
bestfit = firefly.fit_models_to_data()
print('galid %i took %f' % (galid, (time.time() - t0) / 60.))
return None
def singleSP(iz, age=1., mtot=1e8, zred=0.01):
''' construct spectra of single stellar population with single metallicity, no dust,
no nonsense from firefly SSPs (for consistency). metallicity set by z_arr[iz] where z_arr is
'''
z_arr = np.array([0.5, 1.0, 2.0, 10**-1.301, 10**-1.302, 10**-2.301, 10**-2.302, 10**-0.6,
10**-0.9, 10**-1.2, 10**-1.6, 10**-1.9])
z_strs = np.array(['z001', 'z002', 'z004', 'z0001.bhb', 'z0001.rhb', 'z10m4.bhb', 'z10m4.rhb',
'z-0.6', 'z-0.9', 'z-1.2', 'z-1.6', 'z-1.9'])
z_metal = z_arr[iz]
f_ssp = os.environ['STELLARPOPMODELS_DIR']+'/data/SSP_M11_MILES/ssp_M11_MILES.cha'+z_strs[iz]
model_age, model_wave, model_flux = np.loadtxt(f_ssp, unpack=True, usecols=[0,2,3])
isage = (model_age == age)
wave = model_wave[isage] # wavelength
flux = model_flux[isage] # flux [ergs/s/A]
cosmo = FlatLambdaCDM(70., 0.3)
wave *= (1. + zred)
flux *= mtot / (4.*np.pi * cosmo.luminosity_distance(zred).to(U.cm).value**2)
fig = plt.figure(figsize=(10,4))
sub = fig.add_subplot(111)
sub.plot(wave, flux, c='k', lw=1)
sub.set_xlabel('observed-frame wavelength [$A$]', fontsize=15)
sub.set_xlim([3500., 1e4])
sub.set_ylabel('flux [$ergs/s/cm^2/A$]', fontsize=15)
sub.set_ylim([0., flux[(wave > 3000.) & (wave < 1e4)].max()])
fig.savefig(''.join([UT.fig_dir(), 'SSP.iz', str(iz), '.age', str(age), '.z', str(zred), '.png']), bbox_inches='tight')
# save spectra
f_spec = ''.join([UT.dat_dir(), 'SSP.iz', str(iz), '.age', str(age), '.z', str(zred), '.dat'])
np.savetxt(f_spec, np.vstack([wave, flux]).T)
return None
def f_BGSspec(galid, iobs, lib='bc03', obs_sampling='spacefill', nobs=8):
''' DESI BGS-like spectra
'''
fsource = f_Source(galid, lib=lib)
fsource = fsource.rsplit('/',1)[1].rsplit('.', 1)[0]
return ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'BGSsim.', fsource, '.obscond_', obs_sampling, '.', str(iobs+1), 'of', str(nobs), '.fits'])
def f_Source(galid, lib='bc03'):
''' source spectra
'''
if lib == 'bc03': lib_str = 'BC03_Stelib'
elif lib == 'fsps': lib_str = 'FSPS_uvmiles'
return ''.join([UT.dat_dir(), 'Lgal/templates/',
'gal_spectrum_'+str(galid)+'_BGS_template_', lib_str, '.fits'])
def myFirefly_lgal_sourceSpec(galid, dust=False, lib='bc03', model='m11', model_lib='MILES', imf='kr'):
''' run firefly on simulated source spectra from testGalIDs LGal
'''
# read in source spectra
specin = lgal_sourceSpectra(galid, lib=lib)
redshift = specin['meta']['REDSHIFT']
print('z = %f' % redshift)
wave = specin['wave']
wave_rest = wave / (1.+redshift)
if not dust:
flux = specin['flux_nodust_nonoise']
# output firefly file
f_specin = f_Source(galid, lib=lib)
if not dust:
f_firefly = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'myFirefly.', model, '.', model_lib, '.imf_', imf, '.nodust__',
f_specin.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.nodust.hdf5'])
ffly = Fitters.myFirefly(
Planck13, # comsology
model=model,
model_lib=model_lib,
imf=imf,
dust_corr=False, # no dust correction
age_lim=[0., Planck13.age(redshift).value], # can't have older populations
logZ_lim=[-2.,1.])
mask = ffly.emissionlineMask(wave_rest)
ff_fit, ff_prop = ffly.Fit(wave, flux, flux*0.001, redshift, mask=mask, flux_unit=1e-17, f_output=f_firefly, silent=False)
print ff_prop['logM_total']
return None
def testGalIDs():
''' get gal IDs for test set of LGal SAM objects
'''
# read in spectral challenge test set filenames set by Rita
f_test = ''.join([UT.dat_dir(), 'spectral_challenge/', 'lgal_filenames_testset_BC03_Stellib.txt'])
fnames_test = np.loadtxt(f_test, unpack=True, dtype='S', skiprows=1)
# get gal ID's from the filename
galid_test = [int(fname.split('_')[2]) for fname in fnames_test]
return galid_test
def firefly_lgal_bgsSpec(galid, iobs, lib='bc03', obs_sampling='spacefill',
model='m11', model_lib='MILES', imf='cha', hpf_mode='on'):
''' run firefly on simulated DESI BGS spectra from testGalIDs LGal
'''
obscond = obs_condition(sampling=obs_sampling)
if iobs >= obscond.shape[0]: raise ValueError
# read in source spectra
f_inspec = fits.open(f_Source(galid, lib=lib))
redshift = f_inspec[0].header['REDSHIFT']
# read in simulated DESI bgs spectra
f_bgsspec = f_BGSspec(galid, iobs, lib=lib, obs_sampling=obs_sampling, nobs=obscond.shape[0])
if not os.path.isfile(f_bgsspec): raise ValueError('spectra file does not exist')
spec_desi = read_spectra(f_bgsspec)
gspec = Spec.GSfirefly()
gspec.DESIlike(spec_desi, redshift=redshift)
gspec.path_to_spectrum = UT.dat_dir()
# output firefly file
f_bgs = f_BGSspec(galid, iobs, lib=lib, obs_sampling=obs_sampling, nobs=obscond.shape[0])
f_firefly = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'firefly.', model, '.', model_lib, '.imf_', imf, '.dust_', hpf_mode, '__',
f_bgs.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.hdf5'])
firefly = Fitters.Firefly(gspec,
f_firefly, # output file
co.Planck13, # comsology
models = model, # model ('m11', 'bc03', 'm09')
model_libs = [model_lib], # model library for M11
imfs = [imf], # IMF used ('ss', 'kr', 'cha')
hpf_mode = hpf_mode, # uses HPF to dereden the spectrum
age_limits = [0, np.log10(co.Planck13.age(redshift).value * 1e9)],
Z_limits = [0.001, 4.],
wave_limits = [3350., 9000.],
suffix=None,
downgrade_models = False,
data_wave_medium = 'vacuum',
use_downgraded_models = False,
write_results = True)
bestfit = firefly.fit_models_to_data()
return None
def firefly_lgal_sourceSpec(galid, dust=True, lib='bc03', model='m11', model_lib='MILES', imf='cha', hpf_mode='on'):
''' run firefly on simulated source spectra from testGalIDs LGal
'''
# read in source spectra
specin = lgal_sourceSpectra(galid, lib=lib)
redshift = specin['meta']['REDSHIFT']
gspec = Spec.GSfirefly()
if dust:
gspec.generic(specin['wave'], specin['flux_dust_nonoise'], redshift=redshift)
else:
gspec.generic(specin['wave'], specin['flux_nodust_nonoise'], redshift=redshift)
gspec.path_to_spectrum = UT.dat_dir()
# output firefly file
f_specin = f_Source(galid, lib=lib)
if dust:
f_firefly = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'firefly.', model, '.', model_lib, '.imf_', imf, '.dust_', hpf_mode, '__',
f_specin.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.hdf5'])
else:
f_firefly = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'firefly.', model, '.', model_lib, '.imf_', imf, '.dust_', hpf_mode, '__',
f_specin.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.nodust.hdf5'])
firefly = Fitters.Firefly(gspec,
f_firefly, # output file
co.Planck13, # comsology
models = model, # model ('m11', 'bc03', 'm09')
model_libs = [model_lib], # model library for M11
imfs = [imf], # IMF used ('ss', 'kr', 'cha')
hpf_mode = hpf_mode, # uses HPF to dereden the spectrum
age_limits = [0, np.log10(co.Planck13.age(redshift).value * 1e9)],
wave_limits = [3350., 9000.],
suffix=None,
downgrade_models = False,
data_wave_medium = 'vacuum',
Z_limits = [0.001, 4.],
use_downgraded_models = False,
write_results = True)
bestfit = firefly.fit_models_to_data()
return None
def f_nonoise(galid, lib='bc03'):
''' retrun noiseless spectra template file name
:param galid:
galaxy id number
:param lib: (default: 'bc03')
specify stellar population synthesis library. Options are
'bc03' and 'fsps'
'''
if lib == 'bc03': lib_str = 'BC03_Stelib'
elif lib == 'fsps': lib_str = 'FSPS_uvmiles'
return os.path.join(UT.dat_dir(), 'Lgal', 'templates', 'gal_spectrum_'+str(galid)+'_BGS_template_'+lib_str+'.fits')
def prospector_mockdata(infer_method='dynesty'):
'''
'''
prosp = Fitters.Prospector()
# mock data
tt = np.array([-0.5, 0.3, 3., 12., 1e10])
zred = 0.1
mock_lam = np.linspace(3000, 1e4, 1e3)
mock_flux_maggies, _, _ = prosp.model(mock_lam, tt, zred)
# convert to 10^-17 ergs/s/cm^2/Ang
mock_flux = mock_flux_maggies * 1e17 * UT.c_light() / mock_lam**2 * (3631. * UT.jansky_cgs())
if infer_method == 'dynesty':
# run dynamic nested sampling
prosp.dynesty_spec(mock_lam, mock_flux, None, zred,
nested=True, maxcall_init=50000, maxcall=50000,
write=True, output_file=''.join([UT.dat_dir(), 'prospector_dynesty_mock_test.h5']))
elif infer_method == 'emcee':
# emcee
prosp.emcee_spec(mock_lam, mock_flux, None, zred,
write=True, output_file=''.join([UT.dat_dir(), 'prospector_emcee_mock_test.h5']),
silent=False)
return None
def firefly_LGAL_desiSpec(model='m11',
model_lib='MILES',
imf='cha',
hpf_mode='on'):
''' run firefly on all LGal DESI-like spectra
'''
# read in galids
f_ids = ''.join([UT.dat_dir(), 'Lgal/galid_BGS_BC03_templates.txt'])
galids = np.loadtxt(f_ids, unpack=True, dtype='int')
for id in galids:
firefly_LGAL_desiSpec_i(id,
model=model,
model_lib=model_lib,
imf=imf,
hpf_mode=hpf_mode)
return None
def singleSP_myfirefly(iz, age=1., mtot=1e8, zred=0.01):
''' fit the single stellar population file from `singleSP` using firefly
'''
z_arr = np.array([0.5, 1.0, 2.0, 10**-1.301, 10**-1.302, 10**-2.301, 10**-2.302, 10**-0.6,
10**-0.9, 10**-1.2, 10**-1.6, 10**-1.9])
f_spec = ''.join([UT.dat_dir(), 'SSP.iz', str(iz), '.age', str(age), '.z', str(zred), '.dat'])
wave, flux = np.loadtxt(f_spec, unpack=True, usecols=[0,1])
wave_rest = wave/(1.+zred)
err = flux * 0.001
cosmo = FlatLambdaCDM(70., 0.3)
ffly = Fitters.myFirefly(cosmo,
model='m11',
model_lib='MILES',
imf='cha',
dust_corr=False, # no dust correction
age_lim=[0., cosmo.age(zred).value], # can't have older populations
logZ_lim=[np.log10(z_arr[iz])-0.1, np.log10(z_arr[iz])+0.1])
mask = ffly.emissionlineMask(wave_rest)
ff_fit, ff_prop = ffly.Fit(wave, flux, err, zred, mask=mask, flux_unit=1.)
print('total mass = %f' % ff_prop['logM_total'])
for i in range(ff_prop['n_ssp']):
print('--- SSP %i; weight=%f ---' % (i, ff_prop['weightLight_ssp_'+str(i)]))
print('age = %f'% ff_prop['age_ssp_'+str(i)])
print('log Z = %f'% ff_prop['logZ_ssp_'+str(i)])
print('log M_tot = %f' % ff_prop['logM_total_ssp_'+str(i)])
fig = plt.figure()
sub = fig.add_subplot(111)
sub.plot(ff_fit['wavelength'], ff_fit['flux'], label='SSP')
sub.plot(ff_fit['wavelength'], ff_fit['flux_model'], label='myFirefly best-fit')
sub.plot(ff_fit['wavelength'], ff_fit['flux'] - ff_fit['flux_model'], label='residual')
#sub.plot(wave_rest, flux, c='k', ls=':')
sub.set_xlim([3e3, 1e4])
fig.savefig(''.join([UT.fig_dir(), 'myfirefly.SSP.iz', str(iz), '.age', str(age), '.z', str(zred), '.png']),
bbox_inches='tight')
return None
def prospector_lgal_bgsSpec(galid, iobs, lib='bc03', obs_sampling='spacefill', mask=False, infer_method='dynesty'):
''' run prospector on simulated DESI BGS spectra from testGalIDs LGal
'''
obscond = obs_condition(sampling=obs_sampling)
if iobs >= obscond.shape[0]: raise ValueError
# read in source spectra
f_inspec = fits.open(f_Source(galid, lib=lib))
redshift = f_inspec[0].header['REDSHIFT']
# read in simulated DESI bgs spectra
f_bgsspec = f_BGSspec(galid, iobs, lib=lib, obs_sampling=obs_sampling, nobs=obscond.shape[0])
if not os.path.isfile(f_bgsspec): raise ValueError('spectra file does not exist')
spec_desi = read_spectra(f_bgsspec)
wave = np.concatenate([spec_desi.wave[b] for b in ['b', 'r', 'z']])
flux = np.concatenate([spec_desi.flux[b][0] for b in ['b', 'r', 'z']]) # 10-17 ergs/s/cm2/AA
flux_unc = np.concatenate([spec_desi.ivar[b][0]**-0.5 for b in ['b', 'r', 'z']])
# output firefly file
str_mask = ''
if mask: str_mask = '.masked'
f_bgs = f_BGSspec(galid, iobs, lib=lib, obs_sampling=obs_sampling, nobs=obscond.shape[0])
f_prosp = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'prospector.', infer_method, str_mask, '__',
f_bgs.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.h5'])
prosp = Fitters.Prospector()
if infer_method == 'dynesty': # run dynamic nested sampling
prosp.dynesty_spec(wave, flux, flux_unc, redshift, mask=mask,
nested=True, maxcall_init=25000, maxcall=50000,
write=True, output_file=f_prosp)
elif infer_method == 'emcee': # emcee
prosp.emcee_spec(wave, flux, flux_unc, redshift, mask=mask,
write=True, output_file=f_prosp, silent=False)
return None
def prospector_lgal_bgsSpec_validate(galid, iobs, lib='bc03', obs_sampling='spacefill', mask=False, infer_method='dynesty'):
obscond = obs_condition(sampling=obs_sampling)
if iobs >= obscond.shape[0]: raise ValueError
# read in source spectra and meta data
specin = lgal_sourceSpectra(galid, lib=lib)
zred = specin['meta']['REDSHIFT']
# read in simulated DESI bgs spectra
_spec_desi = lgal_bgsSpec(galid, iobs, lib=lib, obs_sampling=obs_sampling)
spec_desi = {}
spec_desi['wave'] = np.concatenate([_spec_desi.wave[b] for b in ['b', 'r', 'z']])
spec_desi['flux'] = np.concatenate([_spec_desi.flux[b][0] for b in ['b', 'r', 'z']]) # 10-17 ergs/s/cm2/AA
spec_desi['flux_unc'] = np.concatenate([_spec_desi.ivar[b][0]**-0.5 for b in ['b', 'r', 'z']])
# output firefly file
str_mask = ''
if mask: str_mask = '.masked'
f_bgs = f_BGSspec(galid, iobs, lib=lib, obs_sampling=obs_sampling, nobs=obscond.shape[0])
f_prosp = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'prospector.', infer_method, str_mask, '__',
f_bgs.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.h5'])
chain, lnp = UT.readProspector(f_prosp)
imax = np.argmax(lnp)
i, j = np.unravel_index(imax, lnp.shape)
prosp = Fitters.Prospector()
tt_map_d = chain[i,j,:]
flux_map, _, _ = prosp.model(spec_desi['wave'], tt_map_d, zred)
# spectra comparison
fig = plt.figure(figsize=(15,5))
sub = fig.add_subplot(111)
sub.plot(spec_desi['wave'], spec_desi['flux'], c='k', lw=0.5, label='DESI-like')
sub.plot(specin['wave'], specin['flux'], c='C0', lw=1, label='Source')
sub.plot(spec_desi['wave'], flux_map, c='C1', lw=1, label='Prospector best-fit')
sub.legend(loc='upper right', fontsize=20)
sub.set_xlabel('Wavelength [$\AA$]', fontsize=25)
sub.set_xlim([spec_desi['wave'].min(), spec_desi['wave'].max()])
sub.set_ylabel('flux [$10^{-17} erg/s/cm^2/\AA$]', fontsize=25)
sub.set_ylim([0., 2.*specin['flux'][(specin['wave'] > spec_desi['wave'].min()) & (specin['wave'] < spec_desi['wave'].max())].max()])
fig.savefig(''.join([f_prosp.rsplit('/', 1)[0], '/',
'_spectra.', f_prosp.rsplit('/', 1)[1].rsplit('.', 1)[0], '.png']), bbox_inches='tight')
# compare inferred properties to input properties
gal_input = LGalInput(galid)
fig = plt.figure(figsize=(7,4))
sub = fig.add_subplot(111)
sub.plot(gal_input['sfh_t'], gal_input['sfh_disk'], color='C0', label='disk')
sub.plot(gal_input['sfh_t'], gal_input['sfh_bulge'], color='red', label='bulge')
sub.plot(gal_input['sfh_t'], gal_input['sfh_bulge'] + gal_input['sfh_disk'], color='black', label='total')
chain_flat = chain.reshape(chain.shape[0] * chain.shape[1], chain.shape[2])
mmed = np.percentile(chain_flat[:,4], [50])
sub.plot([0., 15.], [mmed, mmed], c='k', ls='--', label='Firefly')
sub.set_xlabel('Lookback time (Gyr)', fontsize=25)
sub.set_xlim([1e-2, 13.])
sub.set_ylabel(r'Mass formed (M$_\odot$)', fontsize=25)
sub.set_yscale('log')
sub.set_ylim([1e8, 5e10])
fig.savefig(''.join([f_prosp.rsplit('/', 1)[0], '/',
'_mstar.', f_prosp.rsplit('/', 1)[1].rsplit('.', 1)[0], '.png']), bbox_inches='tight')
return None
def obs_SkyBrightness(sampling='spacefill', overwrite=False):
''' sky brightness of the sampled observing condition
'''
# observing condition
obscond = obs_condition(sampling=sampling)
f_skybright = ''.join([
UT.dat_dir(), 'spectral_challenge/',
'sky_brightness.bgs_mockexp_obscond_testset.', sampling, '.p'
])
if os.path.isfile(f_skybright) and not overwrite:
wave, skybrights = pickle.load(open(f_skybright, 'rb'))
return wave, skybrights
else:
from feasibgs import skymodel as Sky
specsim_sky = Sky.specsim_initialize('desi')
specsim_wave = specsim_sky._wavelength # Ang
for i in range(obscond.shape[0]):
airmass, seeing, exptime, moonfrac, moonalt, moonsep, sunalt, sunsep = obscond[
i, :]
# re-scaled KS contribution
specsim_sky.airmass = airmass
specsim_sky.moon.moon_phase = np.arccos(2. * moonfrac - 1) / np.pi
specsim_sky.moon.moon_zenith = (90. - moonalt) * U.deg
specsim_sky.moon.separation_angle = moonsep * U.deg
# re-calibrated KS coefficients
specsim_sky.moon.KS_CR = 458173.535128
specsim_sky.moon.KS_CM0 = 5.540103
specsim_sky.moon.KS_CM1 = 178.141045
I_ks_rescale = specsim_sky.surface_brightness
if sunalt > -20.:
import pandas as pd
# there is twilight contributions
# read in coefficients from Parker
fmoon = ''.join(
[UT.dat_dir(), 'spectral_challenge/', 'MoonResults.csv'])
coeffs = pd.DataFrame.from_csv(fmoon)
coeffs.columns = [
'wl', 'model', 'data_var', 'unexplained_var', ' X2', 'rX2',
'c0', 'c_am', 'tau', 'tau2', 'c_zodi', 'c_isl', 'sol', 'I',
't0', 't1', 't2', 't3', 't4', 'm0', 'm1', 'm2', 'm3', 'm4',
'm5', 'm6', 'feb', 'mar', 'apr', 'may', 'jun', 'jul',
'aug', 'sep', 'oct', 'nov', 'dec', 'c2', 'c3', 'c4', 'c5',
'c6'
]
# keep moon models
twi_coeffs = coeffs[coeffs['model'] == 'twilight']
coeffs = coeffs[coeffs['model'] == 'moon']
# order based on wavelengths for convenience
wave_sort = np.argsort(np.array(coeffs['wl']))
for k in coeffs.keys():
coeffs[k] = np.array(coeffs[k])[wave_sort]
for k in twi_coeffs.keys():
twi_coeffs[k] = np.array(twi_coeffs[k])[wave_sort]
I_twi = (
twi_coeffs['t0'] * np.abs(sunalt) + # CT2
twi_coeffs['t1'] * np.abs(sunalt)**2 + # CT1
twi_coeffs['t2'] * np.abs(sunsep)**2 + # CT3
twi_coeffs['t3'] * np.abs(sunsep) # CT4
) * np.exp(-twi_coeffs['t4'] * airmass) + twi_coeffs['c0']
I_twi = np.array(I_twi) / np.pi
I_twi_interp = sp.interpolate.interp1d(
10. * np.array(coeffs['wl']),
I_twi,
fill_value='extrapolate')
skybright = I_ks_rescale.value + I_twi_interp(
specsim_wave.value)
else:
skybright = I_ks_rescale.value
if i == 0:
skybrights = np.zeros((obscond.shape[0], len(skybright)))
skybrights[i, :] = skybright
pickle.dump([specsim_wave.value, skybrights], open(f_skybright, 'wb'))
return specsim_wave.value, skybrights
def mFF_BGSspectra_LGal_FSPS_nodust(galid,
iobs,
imf='chabrier',
obs_sampling='spacefill'):
''' generate spectra from LGal_FSPS_nodust
'''
# read source spectra for meta data
f_lgal = ''.join([
UT.dat_dir(), 'spectral_challenge/bgs/', 'LGAL.',
str(galid), '.FSPS.nodust.imf_', imf, '.hdf5'
])
f_source = h5py.File(f_lgal, 'r')
zred = f_source.attrs['zred']
# read bgs spectra
f_bgs = ''.join([
f_lgal.rsplit('.hdf5', 1)[0], '.BGS.', obs_sampling, '_obs',
str(iobs), '.fits'
])
bgs_spec = UT.readDESIspec(f_bgs)
wave_bgs = np.concatenate([bgs_spec['wave_' + b]
for b in ['b', 'r', 'z']]) # observed frame
flux_bgs = np.concatenate([
bgs_spec['flux_' + b][0] for b in ['b', 'r', 'z']
]) # 10-17 ergs/s/cm2/AA
flux_err_bgs = np.concatenate(
[bgs_spec['ivar_' + b][0]**-0.5 for b in ['b', 'r', 'z']])
wavesort = np.argsort(wave_bgs) # sort by wavelenght
wave_bgs = wave_bgs[wavesort]
flux_bgs = flux_bgs[wavesort]
flux_err_bgs = flux_err_bgs[wavesort]
ffly = Fitters.myFirefly(
Planck13, # comsology
model='m11',
model_lib='MILES',
imf='cha',
dust_corr=False, # no dust correction
age_lim=[0., 13.], # can't have older populations
logZ_lim=[-3., 1.])
mask = ffly.emissionlineMask(wave_bgs / (1. + zred))
f_mff = ''.join([
f_bgs.rsplit('/', 1)[0], '/mFF.',
f_bgs.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.hdf5'
])
ff_fit, ff_prop = ffly.Fit(wave_bgs,
flux_bgs,
flux_err_bgs,
zred,
mask=mask,
f_output=f_mff,
flux_unit=1e-17)
print('input: total_mass = %f' % f_source.attrs['logM_total'])
print('firefly: total mass = %f' % ff_prop['logM_total'])
_ssp_age, _ssp_z, _ssp_mtot = [], [], []
for i in range(ff_prop['n_ssp']):
print('--- SSP %i; weight=%f ---' %
(i, ff_prop['weightLight_ssp_' + str(i)]))
print('age = %f' % ff_prop['age_ssp_' + str(i)])
print('log Z = %f' % ff_prop['logZ_ssp_' + str(i)])
print('log M_tot = %f' % ff_prop['logM_total_ssp_' + str(i)])
_ssp_age.append(ff_prop['age_ssp_' + str(i)])
_ssp_z.append(10**ff_prop['logZ_ssp_' + str(i)])
_ssp_mtot.append(10**ff_prop['logM_total_ssp_' + str(i)])
agesort = np.argsort(_ssp_age)
ssp_age = np.array(_ssp_age)[agesort]
ssp_z = np.array(_ssp_z)[agesort]
ssp_mtot = np.array(_ssp_mtot)[agesort]
# validation plot
fig = plt.figure(figsize=(12, 10))
gs = mpl.gridspec.GridSpec(2, 2, figure=fig)
sub = plt.subplot(gs[0, :]) # flux plot
sub.plot(wave_bgs, flux_bgs, c='C0', lw=1, label='LGal BGS spectrum')
sub.plot(ff_fit['wavelength'] * (1. + zred),
ff_fit['flux_model'],
c='C1',
label='FIREFLY best-fit')
sub.plot(ff_fit['wavelength'] * (1. + zred),
ff_fit['flux'] - ff_fit['flux_model'],
c='r',
label='residual')
sub.text(0.05,
0.95,
('$M_\mathrm{tot}=10^{%.2f}; M_\mathrm{firefly} = 10^{%.2f}$' %
(f_source.attrs['logM_total'], ff_prop['logM_total'])),
ha='left',
va='top',
transform=sub.transAxes,
fontsize=20)
sub.set_xlabel('observed-frame wavelength', fontsize=25)
sub.set_xlim([3e3, 1e4])
sub.set_ylabel('flux [$erg/s/cm^2/A$]', fontsize=25)
sub.legend(loc='upper right', fontsize=20)
sub = plt.subplot(gs[1, 0]) # SFH
sub.plot(f_source.attrs['tage'],
f_source.attrs['sfh_bulge'],
c='C0',
ls=':',
label='Bulge')
sub.plot(f_source.attrs['tage'],
f_source.attrs['sfh_disk'],
c='C0',
ls='--',
label='Disk')
sub.plot(f_source.attrs['tage'],
f_source.attrs['sfh_bulge'] + f_source.attrs['sfh_disk'],
c='C0',
label='Total')
sub.plot(ssp_age, ssp_mtot, c='C1')
sub.set_xlabel('Lookback Time [$Gyr$]', fontsize=20)
sub.set_xlim([0., 13.])
sub.set_ylabel("$M_\mathrm{formed}$", fontsize=25)
sub.set_yscale("log")
sub.set_ylim([5e6, 5e10])
sub = plt.subplot(gs[1, 1]) # ZH
sub.plot(f_source.attrs['tage'],
f_source.attrs['Z_bulge'],
c='C0',
ls=':',
label='Bulge')
sub.plot(f_source.attrs['tage'],
f_source.attrs['Z_disk'],
c='C0',
ls='--',
label='Disk')
sub.plot(ssp_age, ssp_z, c='C1')
sub.set_xlabel('Lookback Time [$Gyr$]', fontsize=20)
sub.set_xlim([0., 13.])
sub.set_ylabel("metallicity, $Z$", fontsize=25)
sub.set_yscale('log')
sub.set_ylim([1e-3, 1e-1])
fig.subplots_adjust(wspace=0.25, hspace=0.25)
fig.savefig(''.join([f_mff.rsplit('.hdf5', 1)[0], '.png']),
bbox_inches='tight')
plt.close()
return None
def BGSspectra_LGal_FSPS_nodust(galid,
iobs=0,
imf='chabrier',
obs_sampling='spacefill',
validate=False):
'''
'''
from feasibgs import forwardmodel as FM
# read in galaxy source spectrum and metadata
f_source = ''.join([
UT.dat_dir(), 'spectral_challenge/bgs/', 'LGAL.',
str(galid), '.FSPS.nodust.imf_', imf, '.hdf5'
])
fh5 = h5py.File(f_source, 'r')
wave = fh5['wavelength'].value
flux_total = fh5['flux'].value
flux_disk = fh5['flux_disk'].value
flux_bulge = fh5['flux_bulge'].value
zred = fh5.attrs['zred']
logM_disk = fh5.attrs['logM_disk']
logM_bulge = fh5.attrs['logM_bulge']
logM_total = fh5.attrs['logM_total']
zred = fh5.attrs['zred']
t_lb = fh5.attrs['tage'] # lookback time (age)
sfh_disk = fh5.attrs['sfh_disk']
sfh_bulge = fh5.attrs['sfh_bulge']
Z_disk = fh5.attrs['Z_disk']
Z_bulge = fh5.attrs['Z_bulge']
fh5.close()
# BGS -like spectrum
wmin, wmax = 3523.0, 9923.0
wave_source = np.arange(wmin, wmax, 0.2)
flux_source_interp = sp.interpolate.interp1d(wave,
flux_total * 1e17,
fill_value='extrapolate')
flux_source = flux_source_interp(wave_source)
# read in observing conditions
obscond = obs_condition(sampling=obs_sampling)
airmass, seeing, exptime, _, _, _, _, _ = obscond[iobs, :]
# read in sky surface brightness
w_sky, skybrights = obs_SkyBrightness(sampling=obs_sampling)
skybright = np.clip(
skybrights[iobs, :], 0,
None) * 1e-17 * U.erg / U.angstrom / U.arcsec**2 / U.cm**2 / U.second
f_bgs = ''.join([
f_source.rsplit('.hdf5', 1)[0], '.BGS.', obs_sampling, '_obs',
str(iobs), '.fits'
])
# simulate BGS spectra
fdesi = FM.fakeDESIspec()
# BGS spectra output file
bgs_spec = fdesi.simExposure(wave_source,
np.atleast_2d(flux_source),
exptime=exptime,
airmass=airmass,
seeing=seeing,
skycondition={
'name': 'input',
'sky': skybright,
'wave': w_sky
},
filename=f_bgs)
if not validate: return None
# validation plot
fig = plt.figure(figsize=(12, 10))
gs = mpl.gridspec.GridSpec(2, 2, figure=fig)
sub = plt.subplot(gs[0, :]) # flux plot
for band in ['b', 'r', 'z']:
sub.plot(bgs_spec.wave[band],
1e-17 * bgs_spec.flux[band][0],
color='gray',
lw=0.25)
for m, fi in zip(sfh_bulge, flux_bulge):
if m > 0.: sub.plot(wave, fi, ls='--', lw=0.1)
sub.plot(wave, np.sum(flux_bulge, axis=0), c='C0', ls='--', label='Bulge')
for m, fi in zip(sfh_disk, flux_disk):
if m > 0.: sub.plot(wave, fi, ls='--', lw=0.1)
sub.plot(wave, fi, lw=0.1)
sub.plot(wave, np.sum(flux_disk, axis=0), c='C1', label='Disk')
sub.plot(wave, flux_total, c='k', ls=':', label='Total')
sub.set_xlabel('observed-frame wavelength', fontsize=25)
sub.set_xlim([3e3, 1e4])
sub.set_ylabel('flux [$erg/s/cm^2/A$]', fontsize=25)
sub.set_ylim([-1e-18, 1e-15])
sub.legend(loc='upper right', fontsize=20)
sub = plt.subplot(gs[1, 0]) # SFH
sub.plot(t_lb, sfh_bulge, c='C0', label='Bulge')
sub.plot(t_lb, sfh_disk, c='C1', label='Disk')
sub.text(
0.05,
0.95,
('$M_\mathrm{tot}=10^{%.1f}, M_\mathrm{disk}=10^{%.1f}, M_\mathrm{bulge}=10^{%.1f}$'
% (logM_total, logM_disk, logM_bulge)),
ha='left',
va='top',
transform=sub.transAxes,
fontsize=15)
sub.set_xlabel('Lookback Time [$Gyr$]', fontsize=20)
sub.set_xlim([0., 13.])
sub.set_ylabel("$M_\mathrm{formed}$", fontsize=25)
sub.set_yscale("log")
sub.set_ylim([5e6, 5e10])
sub = plt.subplot(gs[1, 1]) # ZH
sub.plot(t_lb, Z_bulge, c='C0', label='Bulge')
sub.plot(t_lb, Z_disk, c='C1', label='Disk')
sub.set_xlabel('Lookback Time [$Gyr$]', fontsize=20)
sub.set_xlim([0., 13.])
sub.set_ylabel("metallicity, $Z$", fontsize=25)
sub.set_yscale('log')
sub.set_ylim([1e-3, 1e-1])
fig.subplots_adjust(wspace=0.25, hspace=0.25)
fig.savefig(f_bgs.rsplit('.hdf5', 1)[0] + '.png', bbox_inches='tight')
plt.close()
return None
def singleSP_SFH_myfirefly(iz, zred=0.01):
''' construct spectra with single metallicity, with eagle SFH, no dust, no nonsense
'''
z_arr = np.array([0.5, 1.0, 2.0, 10**-1.301, 10**-1.302, 10**-2.301, 10**-2.302, 10**-0.6,
10**-0.9, 10**-1.2, 10**-1.6, 10**-1.9])
z_strs = np.array(['z001', 'z002', 'z004', 'z0001.bhb', 'z0001.rhb', 'z10m4.bhb', 'z10m4.rhb',
'z-0.6', 'z-0.9', 'z-1.2', 'z-1.6', 'z-1.9'])
z_metal = z_arr[iz]
# read in SFH and M_*
f_eagle = h5py.File(''.join([UT.dat_dir(), '0EAGLE_SFRHs.hdf5']), 'r')
i_zmid = np.abs(z_arr[iz] - 10**np.array(logZ_mid)).argmin()
logZ = logZ_mid[i_zmid]
t = age_bounds[:,1]
dt = age_bounds[:,2] - age_bounds[:,0]
sfh = f_eagle['SFRH'].value[0,:,i_zmid]
f_ssp = os.environ['STELLARPOPMODELS_DIR']+'/data/SSP_M11_MILES/ssp_M11_MILES.cha'+z_strs[iz]
model_age, model_wave, model_flux = np.loadtxt(f_ssp, unpack=True, usecols=[0,2,3])
age_uniq = np.sort(np.unique(model_age))
age_bin = np.concatenate([[0], 0.5*(age_uniq[1:] + age_uniq[:-1])])
mtot_bin, flux_bin = [], []
for i_a in range(len(age_bin)-1):
in_agebin = ((14.-t >= age_bin[i_a]) & (14-t < age_bin[i_a+1]))
mtot_i = np.sum(dt[in_agebin] * sfh[in_agebin] * 1e9)
mtot_bin.append(mtot_i)
assert np.sum(mtot_bin) == np.sum(sfh * dt * 1e9)
print('log M_tot = %f' % np.log10(np.sum(mtot_bin)))
# read in spectra
f_spec = ''.join([UT.dat_dir(), 'SSP_SFH.iz', str(iz), '.z', str(zred), '.dat'])
wave, flux = np.loadtxt(f_spec, unpack=True, usecols=[0,1])
wave_rest = wave/(1.+zred)
err = flux * 0.001
cosmo = FlatLambdaCDM(70., 0.3)
ffly = Fitters.myFirefly(cosmo,
model='m11',
model_lib='MILES',
imf='cha',
dust_corr=False, # no dust correction
age_lim=[0., cosmo.age(zred).value], # can't have older populations
logZ_lim=[np.log10(z_arr[iz])-0.1, np.log10(z_arr[iz])+0.1])
mask = ffly.emissionlineMask(wave_rest)
ff_fit, ff_prop = ffly.Fit(wave, flux, err, zred, mask=mask, flux_unit=1.)
print('total mass = %f' % ff_prop['logM_total'])
ssp_age_bin0, ssp_age_bin1 = [], []
ssp_mtot = []
for i in range(ff_prop['n_ssp']):
print('--- SSP %i; weight=%f ---' % (i, ff_prop['weightLight_ssp_'+str(i)]))
print('age = %f'% ff_prop['age_ssp_'+str(i)])
print('log Z = %f'% ff_prop['logZ_ssp_'+str(i)])
print('log M_tot = %f' % ff_prop['logM_total_ssp_'+str(i)])
i_age_bin = np.abs(age_uniq - ff_prop['age_ssp_'+str(i)]).argmin()
ssp_age_bin0.append(age_bin[i_age_bin])
ssp_age_bin1.append(age_bin[i_age_bin+1])
ssp_mtot.append(10**ff_prop['logM_total_ssp_'+str(i)])
fig = plt.figure(figsize=(12,4))
sub = fig.add_subplot(121)
sub.bar(age_bin[:-1], mtot_bin, width=age_bin[:-1]-age_bin[1:], alpha=0.75)
sub.bar(ssp_age_bin0, ssp_mtot, width=np.array(ssp_age_bin0)-np.array(ssp_age_bin1), alpha=0.75)
sub.set_xlabel('Lookback Time [Gyr]', fontsize=25)
sub.set_xlim([0., 14])
sub.set_ylabel("$M_\mathrm{form}$", fontsize=25)
sub.set_yscale("log")
sub.set_ylim([1e7, 5e9])
sub = fig.add_subplot(122)
sub.plot(ff_fit['wavelength'], ff_fit['flux'], label='SSP')
sub.plot(ff_fit['wavelength'], ff_fit['flux_model'], label='myFirefly best-fit')
sub.plot(ff_fit['wavelength'], ff_fit['flux'] - ff_fit['flux_model'], label='residual')
#sub.plot(wave_rest, flux, c='k', ls=':')
sub.set_xlim([3e3, 1e4])
fig.savefig(''.join([UT.fig_dir(), 'myfirefly.SSP_SFH.iz', str(iz), '.z', str(zred), '.png']),
bbox_inches='tight')
return None
def LGalInput(galid):
f_input = ''.join([UT.dat_dir(), 'Lgal/gal_inputs/',
'gal_input_' + str(galid) + '_BGS_template_FSPS_uvmiles.csv'])
gal_input = Table.read(f_input, delimiter=' ')
return gal_input
def singleSP_SFH(iz, zred=0.01):
''' construct spectra with single metallicity, with eagle SFH, no dust, no nonsense
'''
z_arr = np.array([0.5, 1.0, 2.0, 10**-1.301, 10**-1.302, 10**-2.301, 10**-2.302, 10**-0.6,
10**-0.9, 10**-1.2, 10**-1.6, 10**-1.9])
z_strs = np.array(['z001', 'z002', 'z004', 'z0001.bhb', 'z0001.rhb', 'z10m4.bhb', 'z10m4.rhb',
'z-0.6', 'z-0.9', 'z-1.2', 'z-1.6', 'z-1.9'])
z_metal = z_arr[iz]
# read in SFH and M_*
f_eagle = h5py.File(''.join([UT.dat_dir(), '0EAGLE_SFRHs.hdf5']), 'r')
i_zmid = np.abs(z_arr[iz] - 10**np.array(logZ_mid)).argmin()
logZ = logZ_mid[i_zmid]
t = age_bounds[:,1]
dt = age_bounds[:,2] - age_bounds[:,0]
sfh = f_eagle['SFRH'].value[0,:,i_zmid]
f_ssp = os.environ['STELLARPOPMODELS_DIR']+'/data/SSP_M11_MILES/ssp_M11_MILES.cha'+z_strs[iz]
model_age, model_wave, model_flux = np.loadtxt(f_ssp, unpack=True, usecols=[0,2,3])
age_uniq = np.sort(np.unique(model_age))
age_bin = np.concatenate([[0], 0.5*(age_uniq[1:] + age_uniq[:-1])])
mtot_bin, flux_bin = [], []
for i_a in range(len(age_bin)-1):
in_agebin = ((14.-t >= age_bin[i_a]) & (14-t < age_bin[i_a+1]))
mtot_i = np.sum(dt[in_agebin] * sfh[in_agebin] * 1e9)
mtot_bin.append(mtot_i)
isage = (model_age == age_uniq[i_a])
flux_i = model_flux[isage] * mtot_i
flux_bin.append(flux_i)
if i_a == 0:
wave = model_wave[isage] # wavelength
flux = flux_i
else:
flux += flux_i
flux_bin = np.array(flux_bin)
cosmo = FlatLambdaCDM(70., 0.3)
wave *= (1. + zred)
flux /= (4.*np.pi * cosmo.luminosity_distance(zred).to(U.cm).value**2)
flux_bin /= (4.*np.pi * cosmo.luminosity_distance(zred).to(U.cm).value**2)
assert np.sum(mtot_bin) == np.sum(sfh * dt * 1e9)
print('log M_tot = %f' % np.log10(np.sum(mtot_bin)))
fig = plt.figure(figsize=(12,4))
sub = fig.add_subplot(121)
sub.bar(age_bin[:-1], mtot_bin, width=age_bin[:-1]-age_bin[1:])
sub.bar(14.-age_bounds[:,0], sfh * dt * 1e9, width=dt)
sub.set_xlabel('Lookback Time [Gyr]', fontsize=25)
sub.set_xlim([0., 14])
sub.set_ylabel("$M_\mathrm{form}$", fontsize=25)
sub.set_yscale("log")
sub.set_ylim([1e7, 5e9])
sub = fig.add_subplot(122)
for i in range(flux_bin.shape[0]):
sub.plot(wave, flux_bin[i])
sub.plot(wave, flux, c='k', lw=3)
sub.set_xlabel('observed-frame wavelength [$A$]', fontsize=20)
sub.set_xlim([3500., 1e4])
sub.set_ylabel('flux [$ergs/s/cm^2/A$]', fontsize=20)
fig.savefig(''.join([UT.fig_dir(), 'SSP_SFH.iz', str(iz), '.z', str(zred), '.png']), bbox_inches='tight')
# save spectra
f_spec = ''.join([UT.dat_dir(), 'SSP_SFH.iz', str(iz), '.z', str(zred), '.dat'])
np.savetxt(f_spec, np.vstack([wave, flux]).T)
return None
def myFirefly_lgal_sourceSpec_validate(galid, dust=False, lib='bc03', obs_sampling='spacefill',
model='m11', model_lib='MILES', imf='kr'):
obscond = obs_condition(sampling=obs_sampling)
if iobs >= obscond.shape[0]: raise ValueError
# read in source spectra and meta data
specin = lgal_sourceSpectra(galid, lib=lib)
zred = specin['meta']['REDSHIFT']
# read firefly file
f_specin = f_Source(galid, lib=lib)
if not dust:
f_firefly = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'myFirefly.', model, '.', model_lib, '.imf_', imf, '.nodust__',
f_specin.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.nodust.hdf5'])
fh = h5py.File(f_firefly, 'r')
ff_fit = {}
for k in fh.keys():
ff_fit[k] = fh[k].value
ff_prop = {}
for k in fh.attrs.keys():
ff_prop[k] = fh.attrs[k]
# spectra comparison
fig = plt.figure(figsize=(15,5))
sub = fig.add_subplot(111)
if not dust:
sub.plot(specin['wave'], specin['flux_nodust_nonoise'], c='C0', lw=1, label='Source')
sub.plot(ff_fit['wavelength'] * (1. + zred), ff_fit['flux_model'], c='C1', label='FIREFLY best-fit')
sub.plot(ff_fit['wavelength'] * (1. + zred), ff_fit['flux'] - ff_fit['flux_model'], c='r', label='residual')
sub.legend(loc='upper right', fontsize=20)
sub.set_xlabel('Wavelength [$\AA$]', fontsize=25)
sub.set_xlim([3.3e3, 9.8e3])
sub.set_ylabel('flux [$10^{-17} erg/s/cm^2/\AA$]', fontsize=25)
sub.set_ylim([0., 1.5*specin['flux_nodust_nonoise'][(specin['wave'] > 3e3) & (specin['wave'] < 1e4)].max()])
fig.savefig(''.join([f_firefly.rsplit('/', 1)[0], '/',
'_spectra.', f_firefly.rsplit('/', 1)[1].rsplit('.', 1)[0], '.png']), bbox_inches='tight')
# compare inferred properties to input properties
gal_input = LGalInput(galid)
print('input: total_mass = %f' % np.sum(gal_input['sfh_disk'] + gal_input['sfh_bulge']))
print('firefly: total mass = %f' % ff_prop['logM_total'])
ssp_age, ssp_z, ssp_mtot = [], [], []
for i in range(ff_prop['n_ssp']):
print('--- SSP %i; weight=%f ---' % (i, ff_prop['weightLight_ssp_'+str(i)]))
print('age = %f'% ff_prop['age_ssp_'+str(i)])
print('log Z = %f'% ff_prop['logZ_ssp_'+str(i)])
print('log M_tot = %f' % ff_prop['logM_total_ssp_'+str(i)])
ssp_age.append(ff_prop['age_ssp_'+str(i)])
ssp_z.append(10**ff_prop['logZ_ssp_'+str(i)])
ssp_mtot.append(10**ff_prop['logM_total_ssp_'+str(i)])
fig = plt.figure(figsize=(12,4))
sub = fig.add_subplot(121)
sub.plot(gal_input['sfh_t'], gal_input['sfh_disk'], color='C0', label='disk')
sub.plot(gal_input['sfh_t'], gal_input['sfh_bulge'], color='red', label='bulge')
sub.plot(gal_input['sfh_t'], gal_input['sfh_bulge'] + gal_input['sfh_disk'], color='black', label='total')
sub.plot(ssp_age, ssp_mtot)
print ssp_age
print np.log10(ssp_mtot)
sub.set_xlabel('Lookback time (Gyr)', fontsize=25)
sub.set_xlim([1e-2, 13.])
sub.set_ylabel(r'Mass formed (M$_\odot$)', fontsize=25)
sub.set_yscale('log')
sub.set_ylim([1e7, 5e10])
sub = fig.add_subplot(122)
sub.plot(gal_input['sfh_t'], gal_input['Z_disk'], label='disk')
sub.plot(gal_input['sfh_t'], gal_input['Z_bulge'], color='red', label='bulge')
sub.plot([0., 0.], [0., 0.], c='k', ls='-', label='total')
sub.plot(ssp_age, ssp_z)
sub.legend(fontsize=15, ncol=2)
sub.set_xlabel('Lookback time (Gyr)', fontsize=25)
sub.set_xlim([1e-2, 13.])
sub.set_ylabel('Metallicity', fontsize=25)
sub.set_yscale('log')
sub.set_ylim([0., 0.1])
fig.savefig(''.join([f_firefly.rsplit('/', 1)[0], '/',
'_mstar.', f_firefly.rsplit('/', 1)[1].rsplit('.', 1)[0], '.png']), bbox_inches='tight')
return None
def fit_test(galid):
cosmo = co.Planck13
#spec = gs.GalaxySpectrumFIREFLY(None, milky_way_reddening=True)
# read in source spectra
f_s = ''.join([
UT.dat_dir(), 'Lgal/templates/', 'gal_spectrum_',
str(galid), '_BGS_template_BC03_Stelib.fits'
])
f_inspec = fits.open(f_s)
hdr = f_inspec[0].header
specin = f_inspec[1].data
meta = {}
for k in hdr.keys():
meta[k] = hdr[k]
zred = meta['REDSHIFT']
f_input = ''.join([
UT.dat_dir(), 'Lgal/gal_inputs/', 'gal_input_',
str(galid), '_BGS_template_FSPS_uvmiles.csv'
])
gal_input = Table.read(f_input, delimiter=' ')
spec_in = {}
spec_in['wave'] = specin['wave']
#flux = specin['flux_dust_nonoise'] * u.Watt / u.AA / u.m**2 #from W/A/m2 to 10e-17 erg/s/cm2/A
#flux = flux.to(1e-17*u.erg/u.s/u.cm**2/u.AA)
spec_in[
'flux_dust_nonoise'] = specin['flux_dust_nonoise'] * 1e-4 * 1e7 * 1e17
spec_in['flux_nodust_nonoise'] = specin[
'flux_nodust_nonoise'] * 1e-4 * 1e7 * 1e17
gspec = Spec.GSfirefly()
gspec.generic(spec_in['wave'],
spec_in['flux_nodust_nonoise'],
redshift=zred)
gspec.path_to_spectrum = UT.dat_dir()
outputfile = 'test.fits'
bestfit = spm.StellarPopulationModel(gspec,
outputfile,
cosmo,
models='m11',
model_libs=['MILES'],
imfs=['cha'],
hpf_mode='on',
age_limits=[0., 15.],
downgrade_models=False,
data_wave_medium='vacuum',
Z_limits=[0.001, 4.],
wave_limits=[3350., 9000.],
use_downgraded_models=False,
write_results=True)
bestfit.fit_models_to_data()
keys = [
'age_lightW', 'age_massW', 'metallicity_lightW', 'metallicity_massW',
'stellar_mass'
] #, 'HIERARCH stellar_mass']
units = [
'Age_unit', 'Age_unit', 'Metallicity_unit', 'Metallicity_unit',
'Mass_unit'
] #, 'Mass_unit']
#for k, un in zip(keys, units):
# print('%s, %s = %f' % (k, bestfit.thdulist[1].header[un], bestfit.thdulist[1].header[k]))
print('redshift = %f' % zred)
print('input log M*= %f' % np.log10(
np.sum(gal_input['sfh_disk']) + np.sum(gal_input['sfh_bulge'])))
print(np.log10(bestfit.averages['stellar_mass']))
print(bestfit.thdulist[1].header['stellar_mass'])
print(bestfit.thdulist[1].header['total_mass'])
plt.plot(bestfit.thdulist[1].data['wavelength'],
bestfit.thdulist[1].data['original_data'])
plt.plot(bestfit.thdulist[1].data['wavelength'],
bestfit.thdulist[1].data['firefly_model'],
ls='--')
plt.ylim([0., 2.])
plt.savefig('test' + str(galid) + '.png')
plt.close()
return None
def firefly_Mstar(lib='bc03', obs_sampling='spacefill', model='m11', model_lib='MILES', imf='cha', hpf_mode='on'):
''' see how well Mstar inferred from firefly reproduces the
input mstar
'''
obscond = obs_condition(sampling=obs_sampling)
galids = testGalIDs()
obs = [0]#range(8)
mtot_ffly_source, mtot_ffly_source_nodust, mtot_ffly_obs = [], [], []
mstar_ffly_source, mstar_ffly_source_nodust, mstar_ffly_obs = [], [], []
mtot_input, mdisk_input, mbulg_input = [], [], []
for iobs in obs:
mstar_ffly, mtot_ffly = [], []
for galid in galids:
# read firefly fitting output
f_bgs = f_BGSspec(galid, iobs, lib=lib, obs_sampling=obs_sampling, nobs=obscond.shape[0])
f_firefly = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'firefly.', model, '.', model_lib, '.imf_', imf, '.dust_', hpf_mode, '__',
f_bgs.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.hdf5'])
ffly_out, ffly_prop = UT.readFirefly(f_firefly)
mstar_ffly.append(10**ffly_prop['stellar_mass'])
if iobs == 0:
gal_input = LGalInput(galid)
mdisk_input.append(np.sum(gal_input['sfh_disk']))
mbulg_input.append(np.sum(gal_input['sfh_bulge']))
mtot_input.append(np.sum(gal_input['sfh_disk'])+np.sum(gal_input['sfh_bulge']))
# source firefly file
f_specin = f_Source(galid, lib=lib)
f_firefly_s = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'firefly.', model, '.', model_lib, '.imf_', imf, '.dust_', hpf_mode, '__',
f_specin.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.hdf5'])
ffly_out, ffly_prop = UT.readFirefly(f_firefly_s)
mtot_ffly_source.append(10**ffly_prop['total_mass'])
mstar_ffly_source.append(10**ffly_prop['stellar_mass'])
# source no dust firefly file
f_firefly_s = ''.join([UT.dat_dir(), 'spectral_challenge/bgs/',
'firefly.', model, '.', model_lib, '.imf_', imf, '.dust_', hpf_mode, '__',
f_specin.rsplit('/', 1)[1].rsplit('.fits', 1)[0], '.nodust.hdf5'])
ffly_out, ffly_prop = UT.readFirefly(f_firefly_s)
mtot_ffly_source_nodust.append(10**ffly_prop['total_mass'])
mstar_ffly_source_nodust.append(10**ffly_prop['stellar_mass'])
mtot_ffly_obs.append(mstar_ffly)
mstar_ffly_obs.append(mstar_ffly)
fig = plt.figure(figsize=(7,7))
sub = fig.add_subplot(111)
sub.hist(np.log10(mtot_input), range=(9,12), bins=20, histtype='step', color='k', linewidth=2, label='input')
sub.hist(np.log10(mstar_ffly_source), range=(9,12), bins=20, histtype='step',
color='k', linewidth=1, linestyle=':', label='source')
sub.hist(np.log10(mstar_ffly_source_nodust), range=(9,12), bins=20, histtype='step',
color='r', linewidth=1, linestyle=':', label='source (no dust)')
for iobs in obs:
sub.hist(np.log10(mstar_ffly_obs[iobs]), range=(9,12), bins=20, histtype='step',
color='C'+str(iobs), linewidth=1, label=r'bgs; $i_\mathrm{obs}='+str(iobs+1)+'$')
sub.set_xlabel(r'$\log(\,M_*$ [$M_\odot$]\,)', fontsize=25)
sub.set_xlim([9, 12])
sub.legend(loc='upper right', frameon=True, fontsize=20)
fig.savefig(''.join([UT.fig_dir(),
'mstar_hist.', f_firefly.rsplit('/', 1)[1].rsplit('__', 1)[0], '.png']), bbox_inches='tight')
fig = plt.figure(figsize=(7,7))
sub = fig.add_subplot(111)
sub.scatter(mtot_input, mtot_ffly_source, s=10, color='k', label='source')
sub.scatter(mtot_input, mtot_ffly_source_nodust, s=10, color='r', label='source (no dust)')
for iobs in obs:
sub.scatter(mtot_input, mtot_ffly_obs[iobs], s=5, color='C'+str(iobs), label=r'bgs; $i_\mathrm{obs}='+str(iobs+1)+'$')
sub.plot([1e8, 1e12], [1e8,1e12], c='k', ls='--')
sub.set_xlabel(r'$M_*^\mathrm{(input)}$ [$M_\odot$]', fontsize=25)
sub.set_xscale('log')
sub.set_xlim([1e8, 1e12])
sub.set_ylabel(r'$M_*^\mathrm{(firefly)}$ [$M_\odot$]', fontsize=25)
sub.set_yscale('log')
sub.set_ylim([1e8, 1e12])
sub.legend(loc='upper left', markerscale=5, handletextpad=0, fontsize=20)
fig.savefig(''.join([UT.fig_dir(),
'mstar_comparison.', f_firefly.rsplit('/', 1)[1].rsplit('__', 1)[0], '.png']), bbox_inches='tight')
fig = plt.figure(figsize=(7,7))
sub = fig.add_subplot(111)
sub.hist(np.log10(mtot_input) - np.log10(mstar_ffly_source), range=(-1,5), bins=20, color='k', histtype='stepfilled', alpha=0.5, label='source')
sub.hist(np.log10(mtot_input) - np.log10(mstar_ffly_source_nodust), range=(-1,5), bins=20, color='r', histtype='stepfilled', alpha=0.5, label='source (no dust)')
for iobs in obs:
sub.hist(np.log10(mtot_input) - np.log10(mstar_ffly_obs[iobs]), range=(-1,5), bins=20, histtype='stepfilled', alpha=0.5,
color='C'+str(iobs), label=r'bgs; $i_\mathrm{obs}='+str(iobs+1)+'$')
sub.set_xlabel(r'$\log(\,M_*^\mathrm{(input)}\,)-\,\log(\,M_*^\mathrm{(firefly)}\,)$ ', fontsize=25)
sub.set_xlim([-1, 5])
sub.legend(loc='upper right', fontsize=20)
fig.savefig(''.join([UT.fig_dir(),
'dmstar_hist.', f_firefly.rsplit('/', 1)[1].rsplit('__', 1)[0], '.png']), bbox_inches='tight')
return None
def LGal_FSPS_nodust(galid, imf='chabrier', validate=False):
''' simplest spectra. No dust. No emission lines. No nothing.
:params galid:
id of galaxy in the GQP spectral challenge test set
:params iobs:
index of observing condition
:params imf: (default: chabrier)
imf of the SSP in FSPS forward model
:params obs_sampling: (default: obs_sampling)
sampling strategy of the observing parameters
:params validate: (default: False)
if true generate plots that validate the spectra
'''
# get Lgal galaxy SF and Z histories
f_input = ''.join([
UT.dat_dir(), 'Lgal/gal_inputs/', 'gal_input_',
str(galid), '_BGS_template_FSPS_uvmiles.csv'
])
lgal = np.loadtxt(f_input, skiprows=1, unpack=True, delimiter=' ')
t_lb = lgal[0] # lookback time [Gyr]
sfh_disk = lgal[2] # M* formed in disk at t_lb
Z_disk = lgal[4] # Z of disk at t_lb
sfh_bulge = lgal[3] # M* formed in bulge at t_lb
Z_bulge = lgal[5] # Z of bulge at t_lb
logM_disk = np.log10(np.sum(sfh_disk))
logM_bulge = np.log10(np.sum(sfh_bulge))
logM_total = np.log10(np.sum(sfh_disk) + np.sum(sfh_bulge))
print('disk: log(Mformed) = %f' % logM_disk)
print('bulge: log(Mformed) = %f' % logM_bulge)
print('total: log(Mformed) = %f' % logM_total)
# get redshift
f_lgal = fits.open(''.join([
UT.dat_dir(), 'Lgal/templates/',
'gal_spectrum_' + str(galid) + '_BGS_template_FSPS_uvmiles.fits'
]))
zred = f_lgal[0].header['REDSHIFT']
# disk contribution
wave_rest, lum_disk = FOMO.FSPS_nodust(t_lb, sfh_disk, Z_disk, imf=imf)
wave, flux_disk = FOMO.zSpectrum(wave_rest, lum_disk, zred, cosmo=Planck13)
# bulge contribution
_, lum_bulge = FOMO.FSPS_nodust(t_lb, sfh_bulge, Z_bulge, imf=imf)
wave, flux_bulge = FOMO.zSpectrum(wave_rest,
lum_bulge,
zred,
cosmo=Planck13)
flux_total = np.sum(flux_disk, axis=0) + np.sum(flux_bulge, axis=0)
# write out galaxy source spectrum
f_output = ''.join([
UT.dat_dir(), 'spectral_challenge/bgs/', 'LGAL.',
str(galid), '.FSPS.nodust.imf_', imf, '.hdf5'
])
fh5 = h5py.File(f_output, 'w')
# store metadata
fh5.attrs['logM_disk'] = logM_disk
fh5.attrs['logM_bulge'] = logM_bulge
fh5.attrs['logM_total'] = logM_total
fh5.attrs['tage'] = t_lb # lookback time (age)
fh5.attrs['sfh_disk'] = sfh_disk
fh5.attrs['sfh_bulge'] = sfh_bulge
fh5.attrs['Z_disk'] = Z_disk
fh5.attrs['Z_bulge'] = Z_bulge
fh5.attrs['zred'] = zred
fh5.create_dataset('wavelength_rest', data=wave_rest)
fh5.create_dataset('wavelength', data=wave) # observe-frame wavelength
fh5.create_dataset('flux',
data=flux_total) # flux of spectrum [erg/s/cm^2/A]
fh5.create_dataset('flux_disk', data=flux_disk) # flux of disk SSPs
fh5.create_dataset('flux_bulge', data=flux_bulge) # flux of bulge SSPs
fh5.close()
if not validate: return None
# validation plot
fig = plt.figure(figsize=(12, 10))
gs = mpl.gridspec.GridSpec(2, 2, figure=fig)
sub = plt.subplot(gs[0, :]) # flux plot
for m, fi in zip(sfh_bulge, flux_bulge):
if m > 0.: sub.plot(wave, fi, ls='--', lw=0.1)
sub.plot(wave, np.sum(flux_bulge, axis=0), c='C0', ls='--', label='Bulge')
for m, fi in zip(sfh_disk, flux_disk):
if m > 0.: sub.plot(wave, fi, ls='--', lw=0.1)
sub.plot(wave, fi, lw=0.1)
sub.plot(wave, np.sum(flux_disk, axis=0), c='C1', label='Disk')
sub.plot(wave, flux_total, c='k', ls=':', label='Total')
sub.set_xlabel('observed-frame wavelength', fontsize=25)
sub.set_xlim([3e3, 1e4])
sub.set_ylabel('flux [$erg/s/cm^2/A$]', fontsize=25)
sub.set_ylim([-1e-18, 1e-15])
sub.legend(loc='upper right', fontsize=20)
sub = plt.subplot(gs[1, 0]) # SFH
sub.plot(t_lb, sfh_bulge, c='C0', label='Bulge')
sub.plot(t_lb, sfh_disk, c='C1', label='Disk')
sub.text(
0.05,
0.95,
('$M_\mathrm{tot}=10^{%.1f}, M_\mathrm{disk}=10^{%.1f}, M_\mathrm{bulge}=10^{%.1f}$'
% (logM_total, logM_disk, logM_bulge)),
ha='left',
va='top',
transform=sub.transAxes,
fontsize=15)
sub.set_xlabel('Lookback Time [$Gyr$]', fontsize=20)
sub.set_xlim([0., 13.])
sub.set_ylabel("$M_\mathrm{formed}$", fontsize=25)
sub.set_yscale("log")
sub.set_ylim([5e6, 5e10])
sub = plt.subplot(gs[1, 1]) # ZH
sub.plot(t_lb, Z_bulge, c='C0', label='Bulge')
sub.plot(t_lb, Z_disk, c='C1', label='Disk')
sub.set_xlabel('Lookback Time [$Gyr$]', fontsize=20)
sub.set_xlim([0., 13.])
sub.set_ylabel("metallicity, $Z$", fontsize=25)
sub.set_yscale('log')
sub.set_ylim([1e-3, 1e-1])
fig.subplots_adjust(wspace=0.25, hspace=0.25)
fig.savefig(f_output.rsplit('.hdf5', 1)[0] + '.png', bbox_inches='tight')
plt.close()
return None
def Lgal_nonoisePhoto(lib='bc03', dust=False, overwrite=False, validate=False):
''' generate noiseless photometry from noiseless spectral_challenge Lgal spectra
generated by Rita. Super simple code that convolves the spectra with the bandpass
decam g, r, z, W1, W2, W3, W4 filtes.
'''
fsource = f_nonoise(4, lib=lib)
fsource = os.path.basename(fsource).rsplit('.', 1)[0].rsplit('_BGS_template_')[1]
if dust: str_dust = 'dust'
else: str_dust = 'nodust'
fphot = os.path.join(UT.dat_dir(), 'spectral_challenge', 'bgs','photo.%s.%s.nonoise.dat' % (fsource, str_dust))
if os.path.isfile(fphot) and not overwrite:
_mags = np.loadtxt(fphot, unpack=True, skiprows=1, usecols=range(1,8))
mags = _mags.T
else:
galids = np.unique(testGalIDs()) # IDs of spectral_challenge galaxies
fluxes = np.zeros((len(galids), 8))
fluxes[:,0] = galids
for i, gid in enumerate(galids):
# read ins ource spectra
spec_source = Lgal_nonoiseSpectra(gid, lib=lib)
if dust:
flux = spec_source['flux_dust_nonoise']
else:
flux = spec_source['flux_nodust_nonoise']
# apply filters
filter_response = specFilter.load_filters('decam2014-g', 'decam2014-r', 'decam2014-z','wise2010-W1', 'wise2010-W2', 'wise2010-W3', 'wise2010-W4')
fluxes_i = filter_response.get_ab_maggies(np.atleast_2d(flux)*U.Watt/U.m**2/U.Angstrom, spec_source['wave']*U.Angstrom)
fluxes[i,1:] = 1e9 * np.array([fluxes_i[0][0], fluxes_i[0][1], fluxes_i[0][2], fluxes_i[0][3], fluxes_i[0][4], fluxes_i[0][5], fluxes_i[0][6]]) # nanomaggies
fmt = ['%.5e' for i in range(8)]
fmt[0] = '%i'
np.savetxt(fphot, fluxes, header='galid, g, r, z, W1, W2, W3, W4 fluxes in nanomaggies', fmt=fmt)
mags = fUT.flux2mag(fluxes[:,1:], method='log')
if validate: # compare the Lgal nonoise photometry to photometry from GAMA-Legacy
cats = Cats.GamaLegacy()
gleg = cats.Read('g15')
fig = plt.figure(figsize=(20,4))
for i_b, band in enumerate(['g', 'r', 'z', 'w1', 'w2']):
sub = fig.add_subplot(1,5,i_b+1)
sub.hist(fUT.flux2mag(gleg['legacy-photo']['flux_%s' % band]), color='k', range=(14,25), density=True, label='Legacy')
sub.hist(mags[:,i_b], color='C1', range=(14, 25), density=True, alpha=0.5, label='LGal\nSpectral\nChallenge')
sub.set_xlabel('%s magnitude' % band, fontsize=20)
sub.set_xlim(14, 25)
sub.legend(loc='upper right', fontsize=15)
fig.savefig(os.path.join(UT.fig_dir(), os.path.basename(fphot).replace('dat', 'hist.png')), bbox_inches='tight')
plt.close()
# g-r vs r-z
fig = plt.figure(figsize=(6,6))
sub = fig.add_subplot(111)
g_r = fUT.flux2mag(gleg['legacy-photo']['flux_g']) - fUT.flux2mag(gleg['legacy-photo']['flux_r'])
r_z = fUT.flux2mag(gleg['legacy-photo']['flux_r']) - fUT.flux2mag(gleg['legacy-photo']['flux_z'])
sub.scatter(g_r, r_z, c='k', s=1, label='Legacy')
g_r = mags[:,0] - mags[:,1]
r_z = mags[:,1] - mags[:,2]
sub.scatter(g_r, r_z, c='C1', s=4, zorder=10, label='LGal\nSpectral Challenge')
sub.set_xlabel('$g - r$', fontsize=20)
sub.set_xlim(-1., 3.)
sub.set_ylabel('$r - z$', fontsize=20)
sub.set_ylim(-1., 3.)
sub.legend(loc='upper right', handletextpad=0.2, markerscale=3, fontsize=15)
fig.savefig(os.path.join(UT.fig_dir(), os.path.basename(fphot).replace('dat', 'color.png')), bbox_inches='tight')
plt.close()
return mags
def Lgal_noisePhoto(lib='bc03', dust=False, overwrite=False, validate=False):
''' generate photometry from noiseless spectral_challenge Lgal spectra
generated by Rita then add realistic noise from Legacy survey.
'''
fsource = f_nonoise(4, lib=lib)
fsource = os.path.basename(fsource).rsplit('.', 1)[0].rsplit('_BGS_template_')[1]
if dust: str_dust = 'dust'
else: str_dust = 'nodust'
fphot = os.path.join(UT.dat_dir(), 'spectral_challenge', 'bgs',
'mag.%s.%s.legacy_noise.dat' % (fsource, str_dust))
if os.path.isfile(fphot) and not overwrite:
_mags = np.loadtxt(fphot, unpack=True, skiprows=1, usecols=[-7, -6, -5, -4, -3, -2, -1])
mags = _mags.T
else:
galids = np.unique(testGalIDs()) # IDs of unique spectral_challenge spectra
mags = np.zeros((len(galids), 7))
for i, gid in enumerate(galids):
# read ins ource spectra
spec_source = Lgal_nonoiseSpectra(gid, lib=lib)
if dust:
flux = spec_source['flux_dust_nonoise']
else:
flux = spec_source['flux_nodust_nonoise']
# apply filters
filter_response = specFilter.load_filters('decam2014-g', 'decam2014-r', 'decam2014-z',
'wise2010-W1', 'wise2010-W2', 'wise2010-W3', 'wise2010-W4')
mags_i = filter_response.get_ab_magnitudes(flux*U.Watt/U.m**2/U.Angstrom, spec_source['wave']*U.Angstrom)
mags[i,:] = np.array([mags_i[0][0], mags_i[0][1], mags_i[0][2], mags_i[0][3],
mags_i[0][4], mags_i[0][5], mags_i[0][6]]) # this is because mags_i is an astropy table. ugh
fluxs = fUT.mag2flux(mags, method='log')
# add in noise from legacy
cats = Cats.GamaLegacy()
gleg = cats.Read('g15')
for ib, band in enumerate(['g', 'r', 'z', 'w1', 'w2', 'w3', 'w4']):
_flux = gleg['legacy-photo']['flux_%s' % band]
_ivar = gleg['legacy-photo']['flux_ivar_%s' % band]
if ib == 0:
gleg_fluxs = np.zeros((len(_flux), 7))
gleg_ivars = np.zeros((len(_flux), 7))
gleg_fluxs[:,ib] = _flux
gleg_ivars[:,ib] = _ivar
tree = KDTree(gleg_fluxs[:,:5])
dist, indx = tree.query(fluxs[:,:5])
ivars = np.zeros_like(mags)
ivars = gleg_ivars[indx,:]
# now lets apply noise to thse mags
sig = ivars**-0.5
fluxs += sig * np.random.randn(sig.shape[0], sig.shape[1]) # noisy flux
noisy_mags = fUT.flux2mag(fluxs)
hdr = 'galid, flux_g, flux_r, flux_z, flux_w1, flux_w2, flux_w3, flux_w4, flux_ivar_g, flux_ivar_r, flux_ivar_z, flux_ivar_w1, flux_ivar_w2, flux_ivar_w3, flux_ivar_w4, mag_g, mag_r, mag_z, mag_w1, mag_w2, mag_w3, mag_w4'
fmt = ['%.5e' for i in range(22)]
fmt[0] = '%i'
np.savetxt(fphot, np.vstack([galids, fluxs.T, ivars.T, noisy_mags.T]).T, header=hdr,
fmt=fmt)
if validate: # compare the Lgal nonoise photometry to photometry from GAMA-Legacy
cats = Cats.GamaLegacy()
gleg = cats.Read('g15')
# histogram of magnitudes
fig = plt.figure(figsize=(20,4))
for i_b, band in enumerate(['g', 'r', 'z', 'w1', 'w2']):
sub = fig.add_subplot(1,5,i_b+1)
sub.hist(fUT.flux2mag(gleg['legacy-photo']['flux_%s' % band]), color='k', range=(14,25), density=True, label='Legacy')
sub.hist(mags[:,i_b], color='C1', range=(14, 25), density=True, alpha=0.5, label='LGal\nSpectral\nChallenge')
sub.set_xlabel('%s magnitude' % band, fontsize=20)
sub.set_xlim(14, 25)
sub.legend(loc='upper right', fontsize=15)
fig.savefig(os.path.join(UT.fig_dir(), os.path.basename(fphot).replace('dat', 'hist.png')), bbox_inches='tight')
plt.close()
# g-r vs r-z
fig = plt.figure(figsize=(6,6))
sub = fig.add_subplot(111)
g_r = fUT.flux2mag(gleg['legacy-photo']['flux_g']) - fUT.flux2mag(gleg['legacy-photo']['flux_r'])
r_z = fUT.flux2mag(gleg['legacy-photo']['flux_r']) - fUT.flux2mag(gleg['legacy-photo']['flux_z'])
sub.scatter(g_r, r_z, c='k', s=1, label='Legacy')
g_r = mags[:,0] - mags[:,1]
r_z = mags[:,1] - mags[:,2]
sub.scatter(g_r, r_z, c='C1', s=4, zorder=10, label='LGal\nSpectral Challenge')
sub.set_xlabel('$g - r$', fontsize=20)
sub.set_xlim(-1., 3.)
sub.set_ylabel('$r - z$', fontsize=20)
sub.set_ylim(-1., 3.)
sub.legend(loc='upper right', handletextpad=0.2, markerscale=3, fontsize=15)
fig.savefig(os.path.join(UT.fig_dir(), os.path.basename(fphot).replace('dat', 'color.png')), bbox_inches='tight')
plt.close()
return mags
