def get_normalized_flux(w, f, e):
#Redden/Un-redden the spectra.
aux_flux = pyasl.unred(w, f, ebv=e, R_V=3.1)
#Wavelength window where the mean flux is computed.
window_condition = ((w >= 4000.) & (w <= 9000.))
flux_window = aux_flux[window_condition]
normalization_factor = np.mean(flux_window)
aux_flux /= normalization_factor
return aux_flux, normalization_factor
def lnlike(params, lbd, logF, dlogF, logF_mod, include_rv, model):
"""
Returns the likelihood probability function.
# p0 = mass
# p1 = oblat
# p2 = age
# p3 = inclination
# p4 = ebmv
# p5 = distance
"""
if model == 'befavor' or model == 'befavor_new':
dist = params[4]
ebmv = params[5]
if model == 'aara' or model == 'acol' or\
model == 'bcmi':
dist = params[7]
ebmv = params[8]
if model == 'beatlas':
dist = params[5]
ebmv = params[6]
if include_rv is True:
RV = params[6]
else:
RV = 3.1
if model == 'bcmi_pol' or model == 'aara_pol':
# print(np.mean(logF), np.mean(logF_mod), np.mean(dlogF))
chi2 = np.sum((logF - logF_mod)**2 / (dlogF)**2.)
else:
norma = (10 / dist)**2
uplim = dlogF == 0
keep = np.logical_not(uplim)
logF_mod += np.log10(norma)
# print(logF_mod)
tmp_flux = 10**logF_mod
flux_mod = pyasl.unred(lbd * 1e4, tmp_flux, ebv=-1 * ebmv, R_V=RV)
# print(flux_mod)
logF_mod = np.log10(flux_mod)
# print(logF_mod)
chi2 = np.sum(((logF[keep] - logF_mod[keep])**2 / (dlogF[keep])**2.))
# chi2 = np.sum((logF - logF_mod)**2 / (dlogF)**2.)
if chi2 is np.nan:
chi2 = np.inf
return -0.5 * chi2
def lnlike(params, lbd, logF, dlogF, logF_mod, ranges, include_rv, model):
"""
Returns the likelihood probability function.
# p0 = mass
# p1 = oblat
# p2 = age
# p3 = inclination
# p4 = ebmv
# p5 = parallax
"""
if model == 'befavor':
plx = params[4]
ebmv = params[5]
if model == 'aara' or model == 'acol' or model == 'bcmi':
plx = params[7]
ebmv = params[8]
if model == 'beatlas':
plx = params[5]
ebmv = params[6]
if include_rv is True:
RV = params[6]
else:
RV = 3.1
dist = 1e3 / plx
norma = (10 / dist)**2
uplim = dlogF == 0
keep = np.logical_not(uplim)
logF_mod += np.log10(norma)
tmp_flux = 10**logF_mod
flux_mod = pyasl.unred(lbd * 1e4, tmp_flux, ebv=-1 * ebmv, R_V=RV)
logF_mod = np.log10(flux_mod)
if uplim.any():
if (logF_mod[uplim] > logF[uplim]).any():
chi2 = np.inf
else:
chi2 = np.sum(((logF[keep] - logF_mod[keep])**2 / (dlogF[keep])**2.))
# chi2 = np.sum((logF - logF_mod)**2 / (dlogF)**2.)
if chi2 is np.nan:
chi2 = np.inf
return -0.5 * chi2
def sanity_example(self):
"""
Check the example
"""
from PyAstronomy import pyasl
import numpy as np
import matplotlib.pylab as plt
# Approximate a solar spectrum using a Planck
# function with a temperature of 5778 K between
# 3000 A and 8000 A.
wvl = np.arange(3000., 8000., 1.0)
flux = pyasl.planck(T=5778., lam=wvl*1e-10)
# Deredden the spectrum assuming ebv=0.1
fluxUnred = pyasl.unred(wvl, flux, ebv=0.1, R_V=3.1)
# Plot the result
plt.title("Reddened flux (red) and dereddened flux (blue)")
plt.plot(wvl, flux, 'r--')
plt.plot(wvl, fluxUnred, 'b--')
def lnlike(params, logF_mod):
"""
Returns the likelihood probability function (-0.5 * chi2).
Usage
prob = lnlike(params, logF_mod)
"""
if flag.SED:
u = np.where(info.lista_obs == 'UV')
index = u[0][0]
lbd_UV = info.wave[index]
logF_UV = info.logF[index]
dlogF_UV = info.dlogF[index]
if flag.model == 'befavor' or flag.model == 'aeri':
dist = params[4]
ebmv = params[5]
if flag.model == 'aara' or flag.model == 'acol':
dist = params[7]
ebmv = params[8]
if flag.model == 'beatlas':
dist = params[5]
ebmv = params[6]
if flag.include_rv and not flag.binary_star:
RV = params[-1]
elif flag.binary_star:
RV = params[-2]
else:
RV = 3.1
#print(logF_UV)
dist = 1e3 / dist
norma = (10 / dist)**2
uplim = dlogF_UV == 0.0
keep = np.logical_not(uplim)
logF_mod[index] += np.log10(norma)
tmp_flux = 10**logF_mod[index]
flux_mod = pyasl.unred(info.wave[index] * 1e4,
tmp_flux,
ebv=-1 * ebmv,
R_V=RV)
logF_mod_UV = np.log10(flux_mod)
rms = np.array([
jy2cgs(1e-3 * 0.1, 20000),
jy2cgs(1e-3 * 0.1, 35000),
jy2cgs(1e-3 * 0.1, 63000)
])
#rms = np.array([1e-3*0.1, 1e-3*0.1, 1e-3*0.1])
#
upper_lim = jy2cgs(10**logF_UV[uplim], lbd_UV[uplim], inverse=True)
mod_upper = jy2cgs(10**logF_mod_UV[uplim], lbd_UV[uplim], inverse=True)
#a parte dos uplims não é em log!
chi2_UV = np.sum(
((logF_UV[keep] - logF_mod_UV[keep])**2. / (dlogF_UV[keep])**2.))
#chi2_uplim = - 2. * np.sum( np.log( (np.pi/2.)**0.5 * rms * (1. + erf(((upper_lim - mod_upper)/((2**0.5)*rms))))))
#chi2_UV = chi2_UV + chi2_uplim
N_UV = len(logF_UV[keep])
chi2_UV_red = chi2_UV / N_UV
#print('chi2_UV = {:.2f}'.format(chi2_UV))
#print(upper_lim-mod_upper)
#print('chi2_uplim = {:.2f}'.format(chi2_uplim))
else:
chi2_UV_red = 0.
N_UV = 0.
#if flag.Ha:
chi2_Ha_red, N_Ha = get_line_chi2(flag.Ha, 'Ha', logF_mod)
#if flag.Hb:
chi2_Hb_red, N_Hb = get_line_chi2(flag.Hb, 'Hb', logF_mod)
#if flag.Hd:
chi2_Hd_red, N_Hd = get_line_chi2(flag.Hd, 'Hd', logF_mod)
#if flag.Hg:
chi2_Hg_red, N_Hg = get_line_chi2(flag.Hg, 'Hg', logF_mod)
if flag.pol:
chi2 = np.sum((info.logF[0] - logF_mod)**2 / (info.dlogF[0])**2.)
else:
chi2 = (chi2_UV_red + chi2_Ha_red + chi2_Hb_red+ chi2_Hd_red+ chi2_Hg_red)\
*(N_UV + N_Ha + N_Hb + N_Hd+ N_Hg)
#print(chi2)
if chi2 is np.nan:
chi2 = np.inf
return -0.5 * chi2
def plot_fit(par, lbd, logF, dlogF, minfo, listpar, lbdarr, logF_grid, isig,
dims, Nwalk, Nmcmc, par_list, ranges, include_rv, npy, log_scale):
'''
Plots best model fit over data
Usage:
plot_fit(par, lbd, logF, dlogF, minfo, logF_grid, isig, Nwalk, Nmcmc)
where
par = np.array([Mstar, oblat, Sig0, Rd, n, cosi, dist])
'''
# model parameters
if include_rv is True:
Mstar, oblat, Hfrac, cosi, dist, ebv, rv = par
lim = 3
lim2 = 2
else:
Mstar, oblat, Hfrac, cosi, dist, ebv = par
lim = 2
lim2 = 1
rv = 3.1
# Rpole, Lstar, Teff = vkg.geneve_par(Mstar, oblat, Hfrac, folder_tables)
# t = np.max(np.array([hfrac2tms(hfrac), 0.]))
t = np.max(np.array([hfrac2tms(Hfrac), 0.]))
Rpole, logL = geneva_interp_fast(Mstar,
oblat,
t,
neighbours_only=True,
isRpole=False)
norma = (10. / dist)**2 # (Lstar*Lsun) / (4. * pi * (dist*pc)**2)
uplim = dlogF == 0
keep = np.logical_not(uplim)
# chain = np.load(npy)
# interpolate models
logF_mod = griddataBA(minfo, logF_grid, par[:-lim], listpar, dims)
logF_list = np.zeros([len(par_list), len(logF_mod)])
chi2 = np.zeros(len(logF_list))
for i in range(len(par_list)):
logF_list[i] = griddataBA(minfo, logF_grid, par_list[i, :-lim],
listpar, dims)
# convert to physical units
logF_mod += np.log10(norma)
logF_list += np.log10(norma)
for j in range(len(logF_list)):
chi2[j] = np.sum(
(logF[keep] - logF_list[j][keep])**2 / (dlogF[keep])**2)
# chib = chi2[np.argsort(chi2)[-30:]] / max(chi2)
flux = 10.**logF
dflux = dlogF * flux
flux_mod = 10.**logF_mod
flux_mod = pyasl.unred(lbd * 1e4, flux_mod, ebv=-1 * ebv, R_V=rv)
# Plot definitions
bottom, left = 0.75, 0.48
width, height = 0.96 - left, 0.97 - bottom
plt.axes([left, bottom, width, height])
# plot fit
if log_scale is True:
for i in range(len(par_list)):
if i % 80 == 0:
ebv_temp = np.copy(par_list[i][-lim2])
F_temp = pyasl.unred(lbd * 1e4,
10**logF_list[i],
ebv=-1 * ebv_temp,
R_V=rv)
plt.plot(lbd, lbd * F_temp, color='gray', alpha=0.1)
plt.errorbar(lbd,
lbd * flux,
yerr=lbd * dflux,
ls='',
marker='o',
alpha=0.5,
ms=10,
color='blue',
linewidth=2)
plt.plot(lbd,
lbd * flux_mod,
color='red',
ls='-',
lw=3.5,
alpha=0.4,
label='$\mathrm{Best\, fit}$')
plt.xlabel('$\lambda\,\mathrm{[\mu m]}$', fontsize=20)
plt.ylabel(r'$\lambda F_{\lambda}\,\mathrm{[erg\, s^{-1}\, cm^{-2}]}$',
fontsize=20)
plt.yscale('log')
else:
for i in range(len(par_list)):
if i % 80 == 0:
ebv_temp = np.copy(par_list[i][-lim2])
F_temp = pyasl.unred(lbd * 1e4,
10**logF_list[i],
ebv=-1 * ebv_temp,
R_V=rv)
plt.plot(lbd, F_temp, color='gray', alpha=0.1)
plt.errorbar(lbd,
flux,
yerr=lbd * dflux,
ls='',
marker='o',
alpha=0.5,
ms=10,
color='blue',
linewidth=2)
plt.plot(lbd,
flux_mod,
color='red',
ls='-',
lw=3.5,
alpha=0.4,
label='$\mathrm{Best\, fit}$')
plt.xlabel('$\lambda\,\mathrm{[\mu m]}$', fontsize=20)
plt.ylabel(r'$F_{\lambda}\,\mathrm{[erg\, s^{-1}\, cm^{-2} \mu m]}$',
fontsize=20)
plt.xlim(min(lbd), max(lbd))
plt.tick_params(direction='in',
length=6,
width=2,
colors='gray',
which='both')
plt.legend(loc='upper right')
return
def plot_residuals(par, lbd, logF, dlogF, minfo, listpar, lbdarr, logF_grid,
isig, dims, Nwalk, Nmcmc, ranges, include_rv, npy,
log_scale, model):
'''
Plots best model fit over data
Usage:
plot_fit(par, lbd, logF, dlogF, minfo, logF_grid, isig, Nwalk, Nmcmc)
where
par = np.array([Mstar, oblat, Sig0, Rd, n, cosi, dist])
'''
# model parameters
if include_rv is True:
Mstar, oblat, Hfrac, cosi, dist, ebv, rv = par
lim = 3
lim2 = 2
else:
if model == 'befavor':
Mstar, oblat, Hfrac, cosi, dist, ebv = par
if model == 'aara' or model == 'acol' or model == 'bcmi':
Mstar, oblat, Hfrac, Sig0, Rd, n, cosi, dist, ebv = par
if model == 'beatlas':
Mstar, oblat, Sig0, n, cosi, dist, ebv = par
Hfrac = 0.3
lim = 2
lim2 = 1
rv = 3.1
# Rpole, Lstar, Teff = vkg.geneve_par(Mstar, oblat, Hfrac, folder_tables)
# t = np.max(np.array([hfrac2tms(hfrac), 0.]))
t = np.max(np.array([hfrac2tms(Hfrac), 0.]))
Rpole, logL = geneva_interp_fast(Mstar,
oblat,
t,
neighbours_only=True,
isRpole=False)
norma = (10. / dist)**2 # (Lstar*Lsun) / (4. * pi * (dist*pc)**2)
uplim = dlogF == 0
keep = np.logical_not(uplim)
# ***
chain = np.load(npy)
par_list = chain[:, -1, :]
# interpolate models
if model == 'beatlas':
logF_mod = griddataBAtlas(minfo, logF_grid, par[:-lim], listpar, dims,
isig)
else:
logF_mod = griddataBA(minfo, logF_grid, par[:-lim], listpar, dims)
logF_list = np.zeros([len(par_list), len(logF_mod)])
chi2 = np.zeros(len(logF_list))
for i in range(len(par_list)):
if model == 'beatlas':
logF_list[i] = griddataBAtlas(minfo, logF_grid, par_list[i, :-lim],
listpar, dims, isig)
else:
logF_list[i] = griddataBA(minfo, logF_grid, par_list[i, :-lim],
listpar, dims)
# convert to physical units
logF_mod += np.log10(norma)
logF_list += np.log10(norma)
for j in range(len(logF_list)):
chi2[j] = np.sum(
(logF[keep] - logF_list[j][keep])**2 / (dlogF[keep])**2)
# chib = chi2[np.argsort(chi2)[-30:]] / max(chi2)
flux = 10.**logF
dflux = dlogF * flux
flux_mod = 10.**logF_mod
flux_mod = pyasl.unred(lbd * 1e4, flux_mod, ebv=-1 * ebv, R_V=rv)
# alphas = (1. - chi2 / max(chi2)) / 50.
# Plot definitions
bottom, left = 0.71, 0.48
width, height = 0.96 - left, 0.785 - bottom
plt.axes([left, bottom, width, height])
# plot fit
for i in range(len(par_list)):
ebv_temp = np.copy(par_list[i][-lim2])
F_temp = pyasl.unred(lbd * 1e4,
10**logF_list[i],
ebv=-1 * ebv_temp,
R_V=rv)
plt.plot(lbd, (flux - F_temp) / dflux, 'bs', alpha=0.2)
# plt.plot(lbd, (flux - flux_mod) / dflux, 'bs', markersize=5, alpha=0.1)
# plt.ylabel('$(\mathrm{F}-F_\mathrm{model})/\sigma$', fontsize=20)
plt.ylabel('$(F-F_\mathrm{m})/\sigma$', fontsize=20)
plt.xlabel('$\lambda\,\mathrm{[\mu m]}$', fontsize=20)
# plt.ylim(-100, 100)
plt.hlines(y=0,
xmin=min(lbd),
xmax=max(lbd),
linestyles='--',
color='black')
plt.xlim(min(lbd), max(lbd))
if model == 'aara' or model == 'beatlas' or\
model == 'acol' or model == 'bcmi':
plt.xscale('log')
plt.tick_params(direction='in',
length=6,
width=2,
colors='gray',
which='both')
plt.legend(loc='upper right')
return
def main():
# 1. Reading the Paris DR12Q catalog
X = cfg.ConfigMap('MakeCat')
if (X == 'FAIL'):
sys.exit("Errors! Couldn't locate section in .ini file")
data = fitsio.read(X['drqcat'], ext=int(X['drq_dataext']))
plate, mjd, fiberid = data['PLATE'], data['MJD'], data['FIBERID']
# Which redshift to use?
z_use = data[cfg.drq_z]
# Getting the extinction
g_ext = data['EXTINCTION'][:, 1]
ebv = g_ext / 3.793
# ref: Schlegel, Finkbeiner & Davis 1998 (ApJ 500, 525, 1998; SFD98)
print("The total number of objects in the catalog are %d" % len(ebv))
# -------------------------------------------------------------------------
# BAL flagged from visual inspection
# Ref: http://www.sdss.org/dr12/algorithms/boss-dr12-quasar-catalog/
balFlag = data['BAL_FLAG_VI']
# Ref: http://www.sdss.org/dr12/algorithms/bitmasks/#BOSSTILE_STATUS
# targetFlag = data['BOSS_TARGET1'] ----- currently using everything
ind = np.where((balFlag == 0) & (data['ZWARNING'] == 0)
& (z_use > float(cfg.zmin)))[0]
# --------------------------------------------------------------------------
# 2b. Remove DLA's - Ref: (Noterdaeme et al. 2012c)
# http://adsabs.harvard.edu/abs/2012A%26A...547L...1N
dla_data = np.loadtxt(X['dla_file'],
skiprows=2,
usecols=([1]),
dtype='str').T
dla_list = np.array([dla_data[i].split('-') for i in range(len(dla_data))])
myinfo = np.array([mjd[ind], plate[ind], fiberid[ind]]).T
target = dict((tuple(k), i) for i, k in enumerate(myinfo))
candidate = [tuple(i) for i in dla_list.astype(int)]
# Way faster than before - http://stackoverflow.com/questions/1388818/
# how-can-i-compare-two-lists-in-python-and-return-matches
inter = set(target).intersection(candidate)
indices = [target[x] for x in inter]
ind[indices] = -1
print("The total number of DLAs removed are %d" % len(indices))
ind = ind[ind != -1]
# In case a test run is needed
if cfg.run != 0:
ind = np.random.choice(ind, size=cfg.run, replace=False)
# --------------------------------------------------------------------------
# Load in the sky-lines file - There are a lot of skylines!!!
spec_dir = X['spec_dir']
skylines = np.loadtxt(X['skyline_file']).T
# These will be written to the final catalogue
calFlux, calInvar = np.zeros((len(ind), cfg.NPIX)), np.zeros(
(len(ind), cfg.NPIX))
nObj = len(ind)
print("Processing data. The total number of objects in"
"the final catalog are % d" % nObj)
# --------------------------------------------------------------------------
start_time = timeit.default_timer()
for k in range(nObj):
x = ind[k]
# Fetch the spectra
SpecName = os.path.join(
spec_dir, str(plate[x]),
'spec-%04d-%5d-%04d.fits' % (plate[x], mjd[x], fiberid[x]))
try:
a = fitsio.read(SpecName, ext=1)
except IOError:
continue
else:
corr_flux, corr_ivar, loglam = a['flux'], a['ivar'], a['loglam']
# Deredden the flux vector using the Fitzpatrick parameterization
corr_flux, corr_ivar = pyasl.unred(10**loglam,
corr_flux,
corr_ivar,
ebv=ebv[x])
# 6. Shift to rest-frame of the quasar
log2rest = loglam - np.log10(1 + z_use[x])
# Nearest pixel assignment for flux and invar
s = np.floor((log2rest - cfg.COEFF0) / cfg.COEFF1).astype(int)
temp1 = np.where((s >= 0) & (s < cfg.NPIX))[0]
calFlux[k, s[temp1]], calInvar[
k, s[temp1]] = corr_flux[temp1], corr_ivar[temp1]
# More efficient code for masking of skylines - halves the time as used before
minSky = np.log10(skylines - 2) - np.log10(1 + z_use[x])
maxSky = np.log10(skylines + 2) - np.log10(1 + z_use[x])
minT = np.floor((minSky - cfg.COEFF0) / cfg.COEFF1).astype(int)
maxT = np.ceil((maxSky - cfg.COEFF0) / cfg.COEFF1).astype(int)
for t in range(len(minT)):
calInvar[k, minT[t]:maxT[t]] = 0
elapsed_time = timeit.default_timer() - start_time
print(elapsed_time)
# 7. Write everything to a fits file
cat_file = X['cat_file']
if os.path.exists(cat_file):
os.remove(cat_file)
fits = fitsio.FITS(cat_file, 'rw')
mydict = OrderedDict()
keys = X['paramstowrite'].split(', ')
for key in keys:
mydict[key] = data[key][ind]
# Write images and table to fits file
hdict = {
'COEFF0': cfg.COEFF0,
'COEFF1': cfg.COEFF1,
'NPIX': cfg.NPIX,
'AUTHOR': X['author']
}
fits.write(calFlux, header=hdict)
fits.write(calInvar)
fits.write(mydict)
fits.close()
print("Creation of catalogue complete. The file is %s." % cat_file)
def plot_fit_last(par, lbd, logF, dlogF, minfo, listpar, lbdarr, logF_grid,
isig, dims, Nwalk, Nmcmc, ranges, include_rv, npy, log_scale,
model):
'''
Plots best model fit over data
Usage:
plot_fit(par, lbd, logF, dlogF, minfo, logF_grid, isig, Nwalk, Nmcmc)
where
par = np.array([Mstar, oblat, Sig0, Rd, n, cosi, dist])
'''
# model parameters
if include_rv is True:
if model == 'befavor':
Mstar, oblat, Hfrac, cosi, dist, ebv, rv = par
if model == 'befavor_new':
Mstar, oblat, t, cosi, dist, ebv, rv = par
if model == 'aara' or model == 'acol' or\
model == 'bcmi':
Mstar, oblat, Hfrac, Sig0, Rd, n, cosi, dist, ebv, rv = par
if model == 'bcmi_pol' or model == 'aara_pol':
Mstar, oblat, Hfrac, Sig0, Rd, n, cosi = par
if model == 'beatlas':
Mstar, oblat, Sig0, n, cosi, dist, ebv, rv = par
Hfrac = 0.30
lim = 3
lim2 = 2
else:
if model == 'befavor':
Mstar, oblat, Hfrac, cosi, dist, ebv = par
if model == 'befavor_new':
Mstar, oblat, t, cosi, dist, ebv = par
if model == 'aara' or model == 'acol' or\
model == 'bcmi':
Mstar, oblat, Hfrac, Sig0, Rd, n, cosi, dist, ebv = par
if model == 'bcmi_pol' or model == 'aara_pol':
Mstar, oblat, Hfrac, Sig0, Rd, n, cosi = par
if model == 'beatlas':
Mstar, oblat, Sig0, n, cosi, dist, ebv = par
Hfrac = 0.30
lim = 2
lim2 = 1
rv = 3.1
# Rpole, Lstar, Teff = vkg.geneve_par(Mstar, oblat, Hfrac, folder_tables)
# t = np.max(np.array([hfrac2tms(hfrac), 0.]))
if model != 'befavor_new':
t = np.max(np.array([hfrac2tms(Hfrac), 0.]))
Rpole, logL = geneva_interp_fast(Mstar,
oblat,
t,
neighbours_only=True,
isRpole=False)
if model != 'bcmi_pol' and model != 'aara_pol':
norma = (10. / dist)**2 # (Lstar*Lsun) / (4. * pi * (dist*pc)**2)
uplim = dlogF == 0
keep = np.logical_not(uplim)
# ***
chain = np.load(npy)
par_list = chain[:, -1, :]
# interpolate model
if model == 'beatlas':
logF_mod = griddataBAtlas(minfo, logF_grid, par[:-lim], listpar, dims,
isig)
elif model == 'befavor_new':
# print(isig, par[:-lim])
logF_mod = bat.griddataBA(minfo,
logF_grid,
par[:-lim],
isig=isig,
silent=True)
elif model == 'bcmi_pol' or model == 'aara_pol':
logF_mod = griddataBA(minfo, logF_grid, par, listpar, dims)
else:
logF_mod = griddataBA(minfo, logF_grid, par[:-lim], listpar, dims)
logF_list = np.zeros([len(par_list), len(logF_mod)])
chi2 = np.zeros(len(logF_list))
for i in range(len(par_list)):
if model == 'beatlas':
logF_list[i] = griddataBAtlas(minfo, logF_grid, par_list[i, :-lim],
listpar, dims, isig)
elif model == 'befavor_new':
logF_list[i] = bat.griddataBA(minfo,
logF_grid,
par_list[i, :-lim],
isig=isig,
silent=True)
elif model == 'bcmi_pol' or model == 'aara_pol':
logF_list[i] = griddataBA(minfo, logF_grid, par_list[i, :],
listpar, dims)
else:
logF_list[i] = griddataBA(minfo, logF_grid, par_list[i, :-lim],
listpar, dims)
# convert to physical units
if model != 'bcmi_pol' and model != 'aara_pol':
logF_mod += np.log10(norma)
logF_list += np.log10(norma)
for j in range(len(logF_list)):
chi2[j] = np.sum(
(logF[keep] - logF_list[j][keep])**2 / (dlogF[keep])**2)
flux = 10.**logF
dflux = dlogF * flux
flux_mod = 10.**logF_mod
if model != 'bcmi_pol' and model != 'aara_pol':
flux_mod = pyasl.unred(lbd * 1e4, flux_mod, ebv=-1 * ebv, R_V=rv)
# Plot definitions
bottom, left = 0.79, 0.51 #0.80, 0.48 # 0.75, 0.48
width, height = 0.96 - left, 0.97 - bottom
plt.axes([left, bottom, width, height])
# plot fit
if log_scale is True:
for i in range(len(par_list)):
if model != 'bcmi_pol' and model != 'aara_pol':
ebv_temp = np.copy(par_list[i][-lim2])
F_temp = pyasl.unred(lbd * 1e4,
10**logF_list[i],
ebv=-1 * ebv_temp,
R_V=rv)
plt.plot(lbd, lbd * F_temp, color='gray', alpha=0.1)
else:
F_temp = pyasl.unred(lbd * 1e4,
10**logF_list[i],
ebv=0,
R_V=rv)
plt.plot(lbd, F_temp, color='gray', alpha=0.1)
if model != 'bcmi_pol' and model != 'aara_pol':
plt.plot(lbd,
lbd * flux_mod,
color='red',
ls='-',
lw=3.5,
alpha=0.4,
label='$\mathrm{Best\, fit}$')
plt.errorbar(lbd,
lbd * flux,
yerr=lbd * dflux,
ls='',
marker='o',
alpha=0.5,
ms=10,
color='blue',
linewidth=2)
else:
plt.plot(lbd,
flux_mod,
color='red',
ls='-',
lw=3.5,
alpha=0.4,
label='$\mathrm{Best\, fit}$')
plt.errorbar(lbd,
flux,
yerr=dflux,
ls='',
marker='o',
alpha=0.5,
ms=10,
color='blue',
linewidth=2)
if model != 'bcmi_pol' and model != 'aara_pol':
plt.ylabel(
r'$\lambda F_{\lambda}\,\mathrm{[erg\, s^{-1}\, cm^{-2}]}$',
fontsize=17)
plt.xscale('log')
else:
plt.ylabel(r'$P [\%]$', fontsize=17)
plt.tick_params(labelbottom='off')
else:
for i in range(len(par_list)):
if model != 'bcmi_pol' and model != 'aara_pol':
ebv_temp = np.copy(par_list[i][-lim2])
F_temp = pyasl.unred(lbd * 1e4,
10**logF_list[i],
ebv=-1 * ebv_temp,
R_V=rv)
plt.plot(lbd, lbd * F_temp, color='gray', alpha=0.1)
else:
F_temp = pyasl.unred(lbd * 1e4,
10**logF_list[i],
ebv=0,
R_V=rv)
plt.plot(lbd, F_temp, color='gray', alpha=0.1)
if model != 'bcmi_pol' and model != 'aara_pol':
plt.plot(lbd,
lbd * flux_mod,
color='red',
ls='-',
lw=3.5,
alpha=0.4,
label='$\mathrm{Best\, fit}$')
plt.errorbar(lbd,
lbd * flux,
yerr=lbd * dflux,
ls='',
marker='o',
alpha=0.5,
ms=10,
color='blue',
linewidth=2)
plt.ylabel(
r'$\lambda F_{\lambda}\,\mathrm{[erg\, s^{-1}\, cm^{-2}]}$',
fontsize=20)
else:
plt.plot(lbd,
flux_mod,
color='red',
ls='-',
lw=3.5,
alpha=0.4,
label='$\mathrm{Best\, fit}$')
plt.errorbar(lbd,
flux,
yerr=dflux,
ls='',
marker='o',
alpha=0.5,
ms=10,
color='blue',
linewidth=2)
plt.ylabel(r'$P [\%]$', fontsize=20)
plt.tick_params(labelbottom='off')
plt.xscale('linear')
plt.xlim(min(lbd), max(lbd))
if model != 'bcmi_pol' and model != 'aara_pol':
plt.yscale('log')
if model == 'aara' or model == 'beatlas' or\
model == 'acol' or model == 'bcmi':
plt.xscale('log')
if log_scale is True:
plt.xscale('log')
else:
plt.xscale('linear')
if model == 'bcmi_pol' or model == 'aara_pol':
plt.xlim(0.9 * min(lbd), 1.1 * max(lbd))
plt.tick_params(direction='in',
length=6,
width=2,
colors='gray',
which='both')
plt.legend(loc='upper right')
return
def plot_residuals_new(par, lbd, logF, dlogF, minfo, listpar, lbdarr,
logF_grid, isig, dims, Nwalk, Nmcmc, npy, box_lim,
lista_obs, current_folder, fig_name):
'''
Create residuals plot separated from the corner
'''
chain = np.load(npy)
par_list = []
flat_samples = chain.reshape((-1, Ndim))
inds = np.random.randint(len(flat_samples), size=100)
for ind in inds:
params = flat_samples[ind]
par_list.append(params)
#oblat = 1 + 0.5*(W**2) # Rimulo 2017
#Rpole, logL, _ = geneva_interp_fast(Mstar, oblat, tms, Zstr='014')
#print(cosi)
#cosi = np.cos(cosi * np.pi/180.)
#par[3] = np.cos(par[3] * np.pi/180.)
#print(cosi)
# ***
#chain = np.load(npy)
#par_list = chain[:, -1, :]
if flag.SED:
if flag.model == 'aeri':
dist = par[4][0]
ebv = par[5][0]
if flag.include_rv:
rv = par[6][0]
else:
rv = 3.1
elif flag.model == 'acol':
dist = par[7][0]
ebv = par[8][0]
if flag.include_rv:
rv = par[9][0]
else:
rv = 3.1
elif flag.model == 'beatlas':
dist = par[5][0]
ebv = par[6][0]
if flag.include_rv:
rv = par[7][0]
else:
rv = 3.1
lim = find_lim()
# Finding position
u = np.where(lista_obs == 'UV')
index = u[0][0]
# Finding the corresponding flag.model (interpolation)
#logF_mod_UV = griddataBA(minfo, logF_grid[index], par[:-lim],
# listpar, dims)
# Observations
logF_UV = logF[index]
flux_UV = 10.**logF_UV
dlogF_UV = dlogF[index]
lbd_UV = lbd[index]
dist = 1e3 / dist
norma = (10. / dist)**2 # (Lstar*Lsun) / (4. * pi * (dist*pc)**2)
uplim = dlogF[index] == 0
keep = np.logical_not(uplim)
# convert to physical units
#logF_mod_UV += np.log10(norma)
#flux_mod_UV = 10.**logF_mod_UV
dflux = dlogF_UV * flux_UV
#flux_mod_UV = pyasl.unred(lbd_UV * 1e4, flux_mod_UV, ebv=-1 * ebv, R_V=rv)
#chi2_UV = np.sum((flux_UV - flux_mod_UV)**2. / dflux**2.)
N_UV = len(logF_UV)
#chi2_UV = chi2_UV/N_UV
logF_list = np.zeros([len(par_list), len(logF_UV)])
#chi2 = np.zeros(len(logF_list))
#for i in range(len(par_list)):
# logF_list[i] = griddataBA(minfo, logF_grid[index], par_list[i, :-lim],
# listpar, dims)
for i, params in enumerate(par_list):
if flag.binary_star:
logF_mod_UV_1_list = griddataBA(minfo, logF_grid[index],
params[:-lim], listpar, dims)
logF_mod_UV_2_list = griddataBA(
minfo, logF_grid[index],
np.array([params[-1], 0.1, params[2], params[3]]),
info.listpar, info.dims)
logF_list[i] = np.log10(10.**np.array(logF_mod_UV_1_list) +
10.**np.array(logF_mod_UV_2_list))
else:
if flag.model != 'beatlas':
logF_list[i] = griddataBA(minfo, logF_grid[index],
params[:-lim], listpar, dims)
else:
logF_list[i] = griddataBAtlas(minfo,
logF_grid[index],
params[:-lim],
listpar,
dims,
isig=dims["sig0"])
#par_list[i] = params
#logF_list += np.log10(norma)
# Plot
#fig, (ax1,ax2) = plt.subplots(2,1,gridspec_kw={'height_ratios': [3, 1]})
logF_list += np.log10(norma)
bottom, left = 0.84, 0.51 #0.80, 0.48 # 0.75, 0.48
width, height = 0.96 - left, 0.97 - bottom
ax1 = plt.axes([left, bottom, width, height])
#for j in range(len(logF_list)):
# chi2[j] = np.sum((logF_UV[keep] - logF_list[j][keep])**2 / (dlogF_UV[keep])**2)
# Plot Models
for i in range(len(par_list)):
if flag.model == 'aeri':
ebv_temp = par_list[i][5]
elif flag.model == 'beatlas':
ebv_temp = par_list[i][6]
else:
ebv_temp = par_list[i][8]
#ebv_temp = np.copy(ebv)
F_temp = pyasl.unred(lbd_UV * 1e4,
10**logF_list[i],
ebv=-1 * ebv_temp,
R_V=rv)
plt.plot(lbd_UV, F_temp, color='gray', alpha=0.1)
# Residuals --- desta forma plota os residuos de todos os modelos, mas acho que nao eh o que quero
#ax2.plot(lbd_UV, (flux_UV - F_temp) / dflux, 'bs', alpha=0.2)
# Applying reddening to the best model
# Best fit
#ax1.plot(lbd_UV, flux_mod_UV, color='red', ls='-', lw=3.5, alpha=0.4,
# label=r'Best Fit' '\n' '$\chi^2$ = {0:.2f}'.format(chi2_UV))
#ax2.plot(lbd_UV, (flux_UV - flux_mod_UV) / dflux, 'bs', alpha=0.2)
#ax2.set_ylim(-10,10)
# Plot Data
keep = np.where(flux_UV > 0) # avoid plot zero flux
ax1.errorbar(lbd_UV[keep],
flux_UV[keep],
yerr=dflux[keep],
ls='',
marker='o',
alpha=0.5,
ms=4,
color='k',
linewidth=1)
ax1.set_xlabel('$\lambda\,\mathrm{[\mu m]}$', fontsize=14)
ax1.set_ylabel(
r'$F_{\lambda}\,\mathrm{[erg\, s^{-1}\, cm^{-2} \mu m^{-1}}$',
fontsize=14)
#ax2.set_ylabel('$(F-F_\mathrm{m})/\sigma$', fontsize=14)
#plt.tick_params(labelbottom='off')
ax1.set_xlim(min(lbd_UV), max(lbd_UV))
#ax2.set_xlim(min(lbd_UV), max(lbd_UV))
#plt.tick_params(direction='in', length=6, width=2, colors='gray',
# which='both')
#ax1.legend(loc='upper right', fontsize=13)
ax1.ticklabel_format(style='sci', axis='y', scilimits=(0, 0))
#plt.tight_layout()
#plt.savefig(current_folder + fig_name + '_new_residuals-UV-REDONE' + '.png', dpi=100)
#plt.close()
if flag.Ha:
# Finding position
line = 'Ha'
plot_line(line, lista_obs, minfo, logF_grid, par, listpar, dims, logF,
dlogF, lbd, par_list, current_folder, fig_name)
if flag.Hb:
line = 'Hb'
plot_line(line, lista_obs, minfo, logF_grid, par, listpar, dims, logF,
dlogF, lbd, par_list, current_folder, fig_name)
if flag.Hd:
line = 'Hd'
plot_line(line, lista_obs, minfo, logF_grid, par, listpar, dims, logF,
dlogF, lbd, par_list, current_folder, fig_name)
if flag.Hg:
line = 'Hg'
plot_line(line, lista_obs, minfo, logF_grid, par, listpar, dims, logF,
dlogF, lbd, par_list, current_folder, fig_name)
#ebv = 0.99 # 3D map Bayestar17
R_V = 3.1 # (Fitzpatrick and Massa 2007)
#R_V = 3.1
A_V = 2.4 #Claire
ebv = A_V / R_V
somme_file = "/Users/azuri/daten/uni/HKU/Pa30/_pa30_somme6_scaled.fits"
somme_hdulist = pyfits.open(somme_file)
somme_header = somme_hdulist[0].header
somme_wavelength = getWavelength(somme_header, 1)
somme_spectrum = fits.getdata(somme_file)
print('somme_wavelength = ', somme_wavelength)
somme_spectrum_smoothed = ndimage.median_filter(somme_spectrum, size=21)
somme_spectrum_smoothed_dereddened = pyasl.unred(somme_wavelength,
somme_spectrum_smoothed,
ebv=ebv,
R_V=R_V)
plt.plot(somme_wavelength,
np.log10(somme_spectrum_smoothed),
'm-',
label='Buil scaled and smoothed')
plt.plot(somme_wavelength, np.log10(somme_spectrum_smoothed_dereddened),
'm-') #, label='Somme scaled, smoothed, and dereddened')
plt.legend()
plotname = '/Users/azuri/daten/uni/HKU/Pa30/pa30_buil_uvuex.pdf'
plt.savefig(plotname,
format='pdf',
frameon=False,
bbox_inches='tight',
pad_inches=0.1)
plt.show()
somme_header = somme_hdulist[0].header
somme_wavelength = getWavelength(somme_header, 1)
somme_spectrum = fits.getdata(somme_file)
# somme_spectrum_absolute = calibratedFluxToAbsoluteFlux(somme_spectrum, 3370.0)
print('somme_wavelength = ', somme_wavelength)
leDu_spectrum_smoothed = scipy.ndimage.median_filter(
leDu_spectrum, 21) #boxCarMeanSmooth(leDu_spectrum, 0, 21)
somme_spectrum_smoothed = scipy.ndimage.median_filter(
somme_spectrum, 21) #boxCarMeanSmooth(somme_spectrum, 0, 21)
#leDu_spectrum_absolute_smoothed = scipy.ndimage.median_filter(leDu_spectrumboxCarMeanSmooth(leDu_spectrum_absolute, 0, 21)
#somme_spectrum_absolute_smoothed = boxCarMeanSmooth(somme_spectrum_absolute, 0, 21)
leDu_spectrum_smoothed_dereddened = pyasl.unred(leDu_wavelength,
leDu_spectrum_smoothed,
ebv=ebv,
R_V=R_V)
plt.plot(leDu_wavelength,
np.log10(leDu_spectrum_smoothed),
'-',
color='#64a030',
label='$\mathrm{LeD\^u\,scaled\,and\,smoothed}$')
plt.plot(leDu_wavelength,
np.log10(leDu_spectrum_smoothed_dereddened),
'-',
color='#64a030') #, label='Le Du scaled, smoothed, and dereddened')
somme_spectrum_smoothed_dereddened = pyasl.unred(somme_wavelength,
somme_spectrum_smoothed,
ebv=ebv,
R_V=R_V)
def plot_residuals(par, npy, current_folder, fig_name):
'''
Create residuals plot separated from the corner
For the SED and the lines
Usage:
plot_residuals(par, npy, current_folder, fig_name)
'''
plt.rc('xtick', labelsize='medium')
plt.rc('ytick', labelsize='medium')
chain = np.load(npy)
par_list = []
flat_samples = chain.reshape((-1, info.Ndim))
inds = np.random.randint(len(flat_samples), size=100)
for ind in inds:
params = flat_samples[ind]
par_list.append(params)
if flag.SED:
if flag.model == 'aeri':
dist = par[4][0]
ebv = par[5][0]
if flag.include_rv:
rv = par[6][0]
else:
rv = 3.1
elif flag.model == 'acol':
dist = par[7][0]
ebv = par[8][0]
if flag.include_rv:
rv = par[9][0]
else:
rv = 3.1
elif flag.model == 'beatlas':
dist = par[5][0]
ebv = par[6][0]
if flag.include_rv:
rv = par[7][0]
else:
rv = 3.1
#print(dist)
u = np.where(info.lista_obs == 'UV')
index = u[0][0]
# Observations
logF_UV = info.logF[index]
flux_UV = 10.**logF_UV
dlogF_UV = info.dlogF[index]
lbd_UV = info.wave[index]
dist = 1e3 / dist
norma = (10. / dist)**2 # (Lstar*Lsun) / (4. * pi * (dist*pc)**2)
uplim = info.dlogF[index] == 0.0
keep = np.logical_not(uplim)
dflux = dlogF_UV * flux_UV
logF_list = np.zeros([len(par_list), len(logF_UV)])
#par_list = chain[:, -1, :]
#inds = np.random.randint(len(flat_samples), size=100)
for i, params in enumerate(par_list):
if flag.binary_star:
logF_mod_UV_1_list = griddataBA(info.minfo,
info.logF_grid[index],
params[:-info.lim],
info.listpar, info.dims)
logF_mod_UV_2_list = griddataBA(
info.minfo, info.logF_grid[index],
np.array([params[-1], 0.1, params[2], params[3]]),
info.listpar, info.dims)
logF_list[i] = np.log10(10.**np.array(logF_mod_UV_1_list) +
10.**np.array(logF_mod_UV_2_list))
else:
if flag.model != 'beatlas':
logF_list[i] = griddataBA(info.minfo,
info.logF_grid[index],
params[:-info.lim], info.listpar,
info.dims)
else:
logF_list[i] = griddataBAtlas(info.minfo,
info.logF_grid[index],
params[:-info.lim],
info.listpar,
info.dims,
isig=info.dims["sig0"])
#par_list[i] = params
logF_list += np.log10(norma)
fig, (ax1, ax2) = plt.subplots(2,
1,
sharex=True,
gridspec_kw={'height_ratios': [3, 1]})
# Plot Models
for i in range(len(par_list)):
if flag.model == 'aeri':
ebv_temp = par_list[i][5]
elif flag.model == 'beatlas':
ebv_temp = par_list[i][6]
else:
ebv_temp = par_list[i][8]
F_temp = pyasl.unred(lbd_UV * 1e4,
10**logF_list[i],
ebv=-1 * ebv_temp,
R_V=rv)
ax1.plot(lbd_UV, F_temp, color='gray', alpha=0.1, lw=0.6)
ax2.plot(lbd_UV, (flux_UV - F_temp) / dflux, 'ks', ms=5, alpha=0.2)
#ax2.set_ylim(-10,10)
if flag.votable or flag.data_table:
#ax2.set_xscale('log')
ax1.set_xscale('log')
ax1.set_yscale('log')
# Plot Data
keep = np.where(flux_UV > 0) # avoid plot zero flux
ax1.errorbar(lbd_UV[keep],
flux_UV[keep],
yerr=dflux[keep],
ls='',
marker='o',
alpha=0.5,
ms=5,
color='k',
linewidth=1)
ax2.set_xlabel('$\lambda\,\mathrm{[\mu m]}$') #, fontsize=12)
ax1.set_ylabel(
r'$F_{\lambda}\,\mathrm{[erg\, s^{-1}\, cm^{-2}\, \mu m^{-1}]}$')
#fontsize=12)
ax2.set_ylabel('$(F-F_\mathrm{m})/\sigma$') #, fontsize=12)
plt.tight_layout()
plt.savefig(current_folder + fig_name + '_new_residuals-UV' + '.png',
dpi=100)
plt.close()
if flag.Ha:
line = 'Ha'
plot_line(line, par, par_list, current_folder, fig_name)
if flag.Hb:
line = 'Hb'
plot_line(line, par, par_list, current_folder, fig_name)
if flag.Hd:
line = 'Hd'
plot_line(line, par, par_list, current_folder, fig_name)
if flag.Hg:
line = 'Hg'
plot_line(line, par, par_list, current_folder, fig_name)
if flag.pol:
index = 0
lbd_pol = info.wave[index]
flux = info.logF[index]
sigma = info.dlogF[index]
logF_list = np.zeros([len(par_list), len(flux)])
for i, params in enumerate(par_list):
logF_list[i] = griddataBA(info.minfo, info.logF_grid[index],
params, info.listpar, info.dims)
fig, (ax1, ax2) = plt.subplots(2,
1,
sharex=True,
gridspec_kw={'height_ratios': [3, 1]})
for i in range(len(par_list)):
ax1.plot(lbd_pol, logF_list[i], color='gray', alpha=0.1, lw=0.6)
ax2.plot(lbd_pol, (flux - logF_list[-1]) / sigma,
'ks',
ms=5,
alpha=0.2)
ax1.errorbar(lbd_pol,
flux,
yerr=sigma,
ls='',
marker='o',
alpha=0.5,
ms=5,
color='k',
linewidth=1)
ax1.set_xlim(0.3, 0.9)
ax2.set_xlabel('$\lambda\,\mathrm{[\mu m]}$')
ax1.set_ylabel(r'$P_{\%}$')
ax2.set_ylabel('$(P-P_\mathrm{m})/\sigma$')
plt.tight_layout()
plt.savefig(current_folder + fig_name + '_new_residuals-POL' + '.png',
dpi=100)
plt.close()
return
#for chosen_bin in range(5):
fig = plt.figure(figsize=(15, 5))
img_spectrum = img[:,0,0]*0
for index, bin_num in enumerate(bins):
if bin_num == chosen_bin:
img_spectrum = img_spectrum+img[:,y_pix[index],x_pix[index]]
spaxel_x = x_pix[index]+1
spaxel_y = y_pix[index]+1
spectrum = img_spectrum
if max(spectrum) < 0.00001:
continue
#Correction for reddening
#E_B_minus_V = 1.97*np.log10(balmer_dec/2.86)
if len(sys.argv) > 5:
RsubV = 3.1
fluxUnred = pyasl.unred(wavelength_angstroms, spectrum, ebv=float(color_excess), R_V=RsubV)
spectrum = fluxUnred
noise_level = np.std(spectrum[310:390])*2
#noise_out.write(str(noise_level)+'\n')
if iteration != 0:
#print("Measured noise level: "+str(noise_level))
noise_array = [random.normalvariate(0.0, noise_level) for _ in xrange(len(spectrum))]
spectrum = spectrum+noise_array
fit_window_size = 500.0
mask_window_size = 40.0
first_cut = np.all([wavelength_angstroms>(wavelengths[1]-fit_window_size), wavelength_angstroms<(wavelengths[1]+fit_window_size)], axis=0)
fit_cut = first_cut
max_mask = np.all([wavelength_angstroms<3000, wavelength_angstroms>7000], axis=0)
for line in wavelengths:
