def plot_dla_mass_metallicity(self, color="blue"):
"""Plot host halo mass vs metallicity for DLAs"""
(halo_mass, _, _) = self._load_halo(0)
self._get_sigma_DLA(0,2)
ind = np.where(self.dla_halo >= 0)
masses = halo_mass[self.dla_halo[ind]]
met = self.get_dla_metallicity()[ind]
ind2 = np.where((met > -2.7)*(met < 0.5)*(masses > 10**9))
met = met[ind2]
masses = np.log10(masses[ind2])
(H, xedges, yedges) = np.histogram2d(masses, met,bins=10,normed=True)
xbins=np.array([(xedges[i+1]+xedges[i])/2 for i in xrange(0,np.size(xedges)-1)])
ybins=np.array([(yedges[i+1]+yedges[i])/2 for i in xrange(0,np.size(yedges)-1)])
plt.contourf(10**xbins,ybins,H.T,[0.1,1],colors=(color,"black"),alpha=0.5)
#Get fit
amed=halohi.calc_binned_median(xedges, masses, met)
aupq=halohi.calc_binned_percentile(xedges, masses, met,75)-amed
#Addition to avoid zeros
aloq=amed - halohi.calc_binned_percentile(xedges, masses, met,100-75)
err = (aupq+aloq)/2.
#Arbitrary large values if err is zero
ap = powerfit(xbins, amed, np.log10(err), breakpoint=10)
mdiff=xbins-ap[0]
fit=(ap[2]*mdiff+ap[1])
# asfit=broken_fit(ap, np.log10(mass))
print "Fit: ",ap
#Plot median sigma DLA
plt.errorbar(10**xbins, amed,yerr=[aloq,aupq],fmt='^',color=color,ms=15,elinewidth=4)
plt.semilogx(10**xbins, fit,color=color,ls="-",lw=3)
plt.ylabel(r"log $(Z / Z_\odot)$")
plt.xlabel(r"Halo mass ($M_\odot$)")
return (masses,met)
def get_sDLA_fit(self):
"""Fit an broken power law profile based function to sigma_DLA as binned."""
ind = np.where(self.real_sub_mass > 0)
minM = np.min(self.real_sub_mass[ind])
maxM = np.max(self.real_sub_mass)
bins=30
mass=np.logspace(np.log10(minM),np.log10(maxM),num=bins)
bin_mass = np.array([(mass[i]+mass[i+1])/2. for i in xrange(0,np.size(mass)-1)])
(sDLA,loq,upq)=self.get_sigma_DLA_binned(mass,sigma=68)
ind = np.where((sDLA > 0)*(loq+upq > 0)*(bin_mass > 10**8.5))
err = (upq[ind]+loq[ind])/2.
#Arbitrary large values if err is zero
return powerfit(np.log10(bin_mass[ind]), np.log10(sDLA[ind]), np.log10(err), breakpoint=10)
def get_sDLA_fit(self):
"""Fit an broken power law profile based function to sigma_DLA as binned."""
ind = np.where(self.real_sub_mass > 0)
minM = np.min(self.real_sub_mass[ind])
maxM = np.max(self.real_sub_mass)
bins=30
mass=np.logspace(np.log10(minM),np.log10(maxM),num=bins)
bin_mass = np.array([(mass[i]+mass[i+1])/2. for i in xrange(0,np.size(mass)-1)])
(sDLA,loq,upq)=self.get_sigma_DLA_binned(mass,sigma=68)
ind = np.where((sDLA > 0)*(loq+upq > 0)*(bin_mass > 10**8.5))
err = (upq[ind]+loq[ind])/2.
#Arbitrary large values if err is zero
return powerfit(np.log10(bin_mass[ind]), np.log10(sDLA[ind]), np.log10(err), breakpoint=10)
