def get_line_fluxes(z0=1.0, mag=21):
"""
Get emission line fluxes for a given continuum magnitude.
"""
print z0
xflux, yline = np.loadtxt(unicorn.GRISM_HOME +
'/templates/dobos11/SF0_0.emline.txt',
unpack=True)
xflux *= (1 + z0)
#### Add continuum, here with level 0.1*max(line)
ycont = yline.max() * 0.1
yflux = ycont + yline
#### Normalize to F140W passband
x_filt, y_filt = np.loadtxt(os.getenv('iref') + '/F140W.dat', unpack=True)
y_filt_int = utils_c.interp_c(xflux, x_filt, y_filt)
filt_norm = np.trapz(y_filt_int * yflux, xflux) / np.trapz(
y_filt_int, xflux)
yflux /= filt_norm
yline /= filt_norm
ycont /= filt_norm
fnu = 10**(-0.4 * (mag + 48.6))
flam = fnu * 3.e18 / (6564. * (1 + z0))**2 / 1.e-17
ha = np.abs(xflux - 6564 * (1 + z0)) < 100
ha_flux = np.trapz(yline[ha], xflux[ha])
s2 = np.abs(xflux - 6731 * (1 + z0)) < 100
s2_flux = np.trapz(yline[s2], xflux[s2])
return ha_flux * flam, s2_flux * flam
def get_line_fluxes(z0=1.0, mag=21):
"""
Get emission line fluxes for a given continuum magnitude.
"""
print z0
xflux, yline = np.loadtxt(unicorn.GRISM_HOME+'/templates/dobos11/SF0_0.emline.txt', unpack=True)
xflux *= (1+z0)
#### Add continuum, here with level 0.1*max(line)
ycont = yline.max()*0.1
yflux = ycont+yline
#### Normalize to F140W passband
x_filt, y_filt = np.loadtxt(os.getenv('iref')+'/F140W.dat', unpack=True)
y_filt_int = utils_c.interp_c(xflux, x_filt, y_filt)
filt_norm = np.trapz(y_filt_int*yflux, xflux) / np.trapz(y_filt_int, xflux)
yflux /= filt_norm
yline /= filt_norm
ycont /= filt_norm
fnu = 10**(-0.4*(mag+48.6))
flam = fnu*3.e18/(6564.*(1+z0))**2/1.e-17
ha = np.abs(xflux-6564*(1+z0)) < 100
ha_flux = np.trapz(yline[ha], xflux[ha])
s2 = np.abs(xflux-6731*(1+z0)) < 100
s2_flux = np.trapz(yline[s2], xflux[s2])
return ha_flux*flam, s2_flux*flam
# #### Trying to figure out units
# plt.plot(gris.twod.im['WAVE'].data, gris.twod.im['SENS'].data)
# plt.plot(unicorn.reduce.sens_files['A'].field('WAVELENGTH'), unicorn.reduce.sens_files['A'].field('SENSITIVITY')*1.e-17*np.median(np.diff(gris.twod.im['WAVE'].data))/2**2)
#
# # test, FLT errors
# flt = pyfits.open('ibhm47gwq_flt.fits')
# err_flt = np.random.normal(size=flt[1].data.shape)*flt[2].data
# mask_flt = (flt[1].data < 0.1) & (err_flt != 0)
# threedhst.utils.biweight(flt[1].data[mask_flt].flatten())
# threedhst.utils.biweight(err_flt[mask_flt].flatten())
def simspec(root='COSMOS-19'):
"""
Root is the base image where the noise and direct images come from.
"""
#### Simple model of a gaussian Ha emission line
xflux = np.arange(1.e4, 1.8e4)
dv = 100 # km/s
z0 = 1.0
l0 = 6564.61 * (1 + z0)
dlam = dv / 3.e5 * l0
yline = 1. / np.sqrt(2 * np.pi * dlam**2) * np.exp(
-(xflux - l0)**2 / 2 / dlam**2)
### Use template emission lines rather than a single gaussian
xflux, yline = np.loadtxt(unicorn.GRISM_HOME +
'/templates/dobos11/SF0_0.emline.txt',
unpack=True)
xflux *= (1 + z0)
#### Add continuum, here with level 0.1*max(line)
ycont = yline.max() * 0.1
yflux = ycont + yline
#### Normalize to F140W passband
x_filt, y_filt = np.loadtxt(os.getenv('iref') + '/F140W.dat', unpack=True)
y_filt_int = utils_c.interp_c(xflux, x_filt, y_filt)
filt_norm = np.trapz(y_filt_int * yflux, xflux) / np.trapz(
y_filt_int, xflux)
yflux /= filt_norm
yline /= filt_norm
ycont /= filt_norm
ids = [290]
model = unicorn.reduce.GrismModel(root)
ids = model.cat.id[model.cat.mag < 24]
#ids = [245]
#### Generate model where every spectrum is the line template but the mag/shape of the galaxies
#### is as observed
for i, id in enumerate(ids):
print unicorn.noNewLine + '%d (%d/%d)' % (id, i + 1, len(ids))
model.compute_object_model(id, lam_spec=xflux, flux_spec=yflux)
model.model += model.object
#### Get error array from the error extension
err = np.random.normal(size=model.model.shape) * model.gris[2].data
mask = (err != 0) & (model.segm[0].data == 0)
#### Compare background flux distributions
#plt.hist(model.gris[1].data[mask].flatten(), range=(-0.1,0.1), bins=100, alpha=0.5)
#plt.hist(err[mask].flatten(), range=(-0.1,0.1), bins=100, alpha=0.5)
#### Store the new model in the grism image data extension so that we can fit it with the
#### various tools (z, line strength, etc)
#old = model.gris[1].data*1.
model.gris[1].data = model.model * (err != 0) + err
model.get_corrected_wcs(verbose=True)
model.init_object_spectra()
model.model *= 0
##### Try extracting a spectrum and fitting it
#id=685
#id=343
#id=ids[0]
for id in ids:
obj = '%s_%05d' % (root, id)
print '%s.linefit.png' % (obj)
if os.path.exists('%s.linefit.png' % (obj)):
print 'skip'
continue
flam = np.sum(model.flux[model.segm[0].data == id])
fnu = np.sum(model.flux_fnu * (model.segm[0].data == id))
### *Input* line flux, should be able to get this directly from the input spectrum and the
### observed magnitude, but check units.
#plt.plot(xflux, yflux/filt_norm*flam*1.e-17)
ha = np.abs(xflux - 6564 * (1 + z0)) < 100
ha_flux = np.trapz(yline[ha] * flam * 1.e-17, xflux[ha])
ha_eqw = np.trapz(yline[ha] / ycont, xflux[ha])
s2 = np.abs(xflux - 6731 * (1 + z0)) < 100
s2_flux = np.trapz(yline[s2] * flam * 1.e-17, xflux[s2])
s2_eqw = np.trapz(yline[s2] / ycont, xflux[s2])
model.twod_spectrum(id, refine=True, verbose=True)
if not model.twod_status:
continue
model.show_2d(savePNG=True)
spec = unicorn.reduce.Interlace1D(root + '_%05d.1D.fits' % (id),
PNG=True)
#### Redshift fit, set template to flat and the redshift prior to a broad gaussian centered
#### on the input value, z0
zgrid = np.arange(0, 4, 0.005)
pz = np.exp(-(zgrid - z0)**2 / 2 / 0.5**2)
lnprob = np.log(pz)
gris = unicorn.interlace_fit.GrismSpectrumFit(root=obj,
lowz_thresh=0.01,
FIGURE_FORMAT='png')
if not gris.status:
continue
gris.zout.z_spec = gris.zout.z_spec * 0. + z0
gris.zout.l99 = gris.zout.l99 * 0. + z0 - 0.1
gris.zout.u99 = gris.zout.l99 + 0.2
gris.z_peak = 1
gris.best_fit = gris.best_fit * 0 + 1
gris.phot_zgrid = zgrid
gris.phot_lnprob = lnprob
try:
gris.fit_in_steps(dzfirst=0.005,
dzsecond=0.0005,
zrfirst=(z0 - 0.2, z0 + 0.2))
except:
continue
if not gris.status:
continue
#### Emission line fit
try:
gris.fit_free_emlines(ztry=gris.z_max_spec,
verbose=True,
NTHREADS=1,
NWALKERS=50,
NSTEP=100,
FIT_REDSHIFT=False,
FIT_WIDTH=False,
line_width0=100)
except:
continue
status = os.system('cat %s.linefit.dat' % (obj))
print '\n -- input --\nSII %6.2f %6.2f' % (s2_flux / 1.e-17, s2_eqw)
print ' Ha %6.2f %6.2f' % (ha_flux / 1.e-17, ha_eqw)
def simspec(root='COSMOS-19'):
"""
Root is the base image where the noise and direct images come from.
"""
#### Simple model of a gaussian Ha emission line
xflux = np.arange(1.e4,1.8e4)
dv = 100 # km/s
z0 = 1.0
l0 = 6564.61*(1+z0)
dlam = dv/3.e5*l0
yline = 1./np.sqrt(2*np.pi*dlam**2)*np.exp(-(xflux-l0)**2/2/dlam**2)
### Use template emission lines rather than a single gaussian
xflux, yline = np.loadtxt(unicorn.GRISM_HOME+'/templates/dobos11/SF0_0.emline.txt', unpack=True)
xflux *= (1+z0)
#### Add continuum, here with level 0.1*max(line)
ycont = yline.max()*0.1
yflux = ycont+yline
#### Normalize to F140W passband
x_filt, y_filt = np.loadtxt(os.getenv('iref')+'/F140W.dat', unpack=True)
y_filt_int = utils_c.interp_c(xflux, x_filt, y_filt)
filt_norm = np.trapz(y_filt_int*yflux, xflux) / np.trapz(y_filt_int, xflux)
yflux /= filt_norm
yline /= filt_norm
ycont /= filt_norm
ids = [290]
model = unicorn.reduce.GrismModel(root)
ids = model.cat.id[model.cat.mag < 24]
#ids = [245]
#### Generate model where every spectrum is the line template but the mag/shape of the galaxies
#### is as observed
for i,id in enumerate(ids):
print unicorn.noNewLine+'%d (%d/%d)' %(id, i+1, len(ids))
model.compute_object_model(id, lam_spec=xflux, flux_spec=yflux)
model.model += model.object
#### Get error array from the error extension
err = np.random.normal(size=model.model.shape)*model.gris[2].data
mask = (err != 0) & (model.segm[0].data == 0)
#### Compare background flux distributions
#plt.hist(model.gris[1].data[mask].flatten(), range=(-0.1,0.1), bins=100, alpha=0.5)
#plt.hist(err[mask].flatten(), range=(-0.1,0.1), bins=100, alpha=0.5)
#### Store the new model in the grism image data extension so that we can fit it with the
#### various tools (z, line strength, etc)
#old = model.gris[1].data*1.
model.gris[1].data = model.model*(err != 0) + err
model.get_corrected_wcs(verbose=True)
model.init_object_spectra()
model.model*=0
##### Try extracting a spectrum and fitting it
#id=685
#id=343
#id=ids[0]
for id in ids:
obj='%s_%05d' %(root, id)
print '%s.linefit.png' %(obj)
if os.path.exists('%s.linefit.png' %(obj)):
print 'skip'
continue
flam = np.sum(model.flux[model.segm[0].data == id])
fnu = np.sum(model.flux_fnu*(model.segm[0].data == id))
### *Input* line flux, should be able to get this directly from the input spectrum and the
### observed magnitude, but check units.
#plt.plot(xflux, yflux/filt_norm*flam*1.e-17)
ha = np.abs(xflux-6564*(1+z0)) < 100
ha_flux = np.trapz(yline[ha]*flam*1.e-17, xflux[ha])
ha_eqw = np.trapz(yline[ha]/ycont, xflux[ha])
s2 = np.abs(xflux-6731*(1+z0)) < 100
s2_flux = np.trapz(yline[s2]*flam*1.e-17, xflux[s2])
s2_eqw = np.trapz(yline[s2]/ycont, xflux[s2])
model.twod_spectrum(id, refine=True, verbose=True)
if not model.twod_status:
continue
model.show_2d(savePNG=True)
spec = unicorn.reduce.Interlace1D(root+'_%05d.1D.fits' %(id), PNG=True)
#### Redshift fit, set template to flat and the redshift prior to a broad gaussian centered
#### on the input value, z0
zgrid = np.arange(0,4,0.005)
pz = np.exp(-(zgrid-z0)**2/2/0.5**2)
lnprob = np.log(pz)
gris = unicorn.interlace_fit.GrismSpectrumFit(root=obj, lowz_thresh=0.01, FIGURE_FORMAT='png')
if not gris.status:
continue
gris.zout.z_spec = gris.zout.z_spec*0.+z0
gris.zout.l99 = gris.zout.l99*0.+z0-0.1
gris.zout.u99 = gris.zout.l99+0.2
gris.z_peak = 1
gris.best_fit = gris.best_fit*0+1
gris.phot_zgrid = zgrid
gris.phot_lnprob = lnprob
try:
gris.fit_in_steps(dzfirst=0.005, dzsecond=0.0005, zrfirst=(z0-0.2,z0+0.2))
except:
continue
if not gris.status:
continue
#### Emission line fit
try:
gris.fit_free_emlines(ztry=gris.z_max_spec, verbose=True, NTHREADS=1, NWALKERS=50, NSTEP=100, FIT_REDSHIFT=False, FIT_WIDTH=False, line_width0=100)
except:
continue
status = os.system('cat %s.linefit.dat' %(obj))
print '\n -- input --\nSII %6.2f %6.2f' %(s2_flux/1.e-17, s2_eqw)
print ' Ha %6.2f %6.2f' %(ha_flux/1.e-17, ha_eqw)
