def findAGNnumber(limits, QSOTemplate):
AGN = createOzDESAGN(QSOTemplate)
rFilter = AGN.getFilter('rFilter')
iFilter = AGN.getFilter('iFilter')
# Extract data from file
name = '/Users/natsomme/Dropbox/Uni/Thesis/Run007_MasterInput.cat'
data = np.loadtxt(name)
mag = data[:, 3]
z = data[:, 4]
# Extract physically reasonable AGN
mag = mag[z>0]
z = z[z>0]
n = np.size(z)
Mag = np.zeros(n)
# Calculate absMag using QSO-functions
for i in range(n):
Lbol = AGN.getQSObol(mag[i], z[i], rFilter)
Mag[i] = AGN.getQSOabsMag(Lbol, 0.0, iFilter)
# End for-loop
# Extract limits
zMin = limits[0]
zMax = limits[1]
MagMin = limits[2]
MagMax = limits[3]
num = 0
for i in range(n):
if (z[i] >= zMin and z[i] < zMax and Mag[i] > MagMin and Mag[i] <= MagMax):
num = num + 1
# End if-statement
# End for-loop
return num
def compareToTheory():
# Do theoretical calculations
QSOTemplate = ascii.read('/Users/natsomme/Dropbox/Uni/Thesis/fake_data_code/sdssqsocomposite.txt')
create = createOzDESAGN(QSOTemplate)
z = 0.0 # Redshift at which AGN are to be created (zero due to filter)
n = 100 # Number of points to create a smooth theoretical curve
lag = np.zeros(n) # Initialising array for all the lags we'll be considering
absmag = np.linspace(-28, -18.5, n) # Initialising array with magnitudes we're interested in
# Calculate a continuus time lag curve for the magnitudes we have
for i in range(n):
Lbol = create.getQSObolFromAbsMag(absmag[i], z, create.iFilter)
_, lag[i], _ = create.getLag(Lbol, z)
# End for-loop
# Check how the time lag changes over time
gradlag = np.gradient(lag) # Calculate gradient of time lag
graddiv = gradlag/lag # Calculate the gradient of the lag divided by the lag
# Get hold of time lag estimates to compare theory to
matr = np.loadtxt('/Volumes/mimsy/Obelix/Mhas50binsKcorr/test.dat')
nbin = matr[:,0] # Which bin the current bin starts on (no longer a thing)
mag = matr[:,1] # The lower end of the magnitude bin
nagn = matr[:,2] # Number of AGN in bin
tmean = matr[:,3] # Average mean time lag for AGN in bin
tstd = matr[:,4] # Standard deviation of lags for AGN in bin
tlo = matr[:,5] # Difference between average lag and shortest lag
thi = matr[:,6] # Difference between average lag and longest lag
sgmean = matr[:,7] # Time lag estimate based on fitted skewed gaussian
sglo = matr[:,8] # Lower uncertainty estimate based on fitted skewed gaussian
sghi = matr[:,9] # Upper uncertainty estimate based on fitted skewed gaussian
llhmean = matr[:,10] # Time lag estimate using maximum likelihood
llhlo = matr[:,11] # Lower uncertainty estimate for maximum likelihood
llhhi = matr[:,12] # Upper uncertainty estimate for maximum likelihood
med1 = matr[:,13] # Time lag estimate based on median
med1lo = matr[:,14] # Lower uncertainty estimate for median
med1hi = matr[:,15] # Upper uncertainty estimate for median
med2 = matr[:,16] # Time lag estimate based on median of maximum likelihoods
med2lo = matr[:,17] # Lower uncertainty estimate for median of max llhs
med2hi = matr[:,18] # Upper uncertainty estimate for median of max llhs
# Find average uncertainty where there is an upper and lower limit
sgstd = 0.5*(sglo+sghi)
llstd = 0.5*(llhlo+llhhi)
med1std = 0.5*(med1lo+med1hi)
med2std = 0.5*(med2lo+med2hi)
# Perform a chi2 fit
N = 1000 # Number of values to check for with chi2 test
fitconst = np.linspace(0.0002,0.0004,N) # Values to check for with chi2 test
chi2 = np.zeros(N) # Initialise
for i in range(N):
# Print some information for values close to the the lowest chi2 value
if (i > 360 and i < 375):
print 'mag -- C*10^{-0.22 mag} - tau_D_stacked - std(tau_D_stacked) -- chi2'
# End if-statement
for j in range(len(tmean)):
# Calculate chi2 value using all data points
chi2[i] = chi2[i] + ( ( (fitconst[i]*10**(-0.22*mag[j])) - tmean[j]) / tstd[j] )**2
# Print some information for values close to the the lowest chi2 value
if (i > 360 and i < 375):
print mag[j], '--', fitconst[i]*10**(-0.22*mag[j]), '-', tmean[j], '-', tstd[j], \
'--', ( ( (fitconst[i]*10**(-0.22*mag[j])) - tmean[j]) / tstd[j] )**2
# End if-statement
# End for-loop
# Print some information for values close to the the lowest chi2 value
if (i > 360 and i < 375):
print '\n'
if (i == 368 or i == 369):
print '-------------'
# End if-statements
# End for-loop
chi2dof = np.min(chi2)/9.0 # Calculate chi2 pr dof
a = fitconst[np.argmin(chi2)] # Calculate the best constant based on chi2 test
Dchi2 = chi2 - np.min(chi2) # Calculate difference between all chi2 values and min(chi2)
# Print information about chi2 test to screen
print 'Chi2 best fit slope: a =', a
print 'Chi2DOF =', chi2dof
# Define remaining parameters to draw best fit line
b = 0.
m = np.linspace(-30.0, -18.5, 100)
tau = a * 10**( -0.22*m ) + b # Best fit line based on chi2 test on estimates
# Create figure showing the comparison between theory and data
fig = plt.figure()
ax = plt.axes()
ratio = float((np.max(absmag) - np.min(absmag))/(np.max(lag) - np.min(lag)))
# Plot theory and data
plt.plot([0,0], [0,0], 'w', label=' ') # Plot nothing just to get a prettier legend
plt.errorbar(mag, tmean, yerr=[tlo, thi], fmt='o', color='grey')
plt.errorbar(mag, tmean, yerr=tstd, fmt='o', label='Average true lag')
plt.errorbar(mag+0.1, sgmean, yerr=[sglo, sghi], fmt='x', label='Skewed Gaussian')
plt.errorbar(mag+0.2, llhmean, yerr=[llhlo, llhhi], fmt='x', label='Maximum likelihood')
plt.errorbar(mag+0.3, med1, yerr=[med1lo, med1hi], fmt='x', label='Stacked median')
plt.errorbar(mag+0.4, med2, yerr=[med2lo, med2hi], fmt='x', label='Bootstrap median')
plt.plot(m, tau, label=r'$\tau$ = C$\cdot$10$^{-0.22M}$')
plt.plot(absmag, lag, label='R-L relationship')
# Customise figure
plt.xlabel('Magnitude', fontsize='large')
plt.ylabel('Time lag [days]', fontsize='large')
plt.legend(loc="center right", bbox_to_anchor=(1.02, 1.13), ncol=2)
plt.axis([np.min(absmag), np.max(absmag), np.min(lag), np.max(lag)])
ax.xaxis.set_major_locator(tck.MultipleLocator(1))
ax.xaxis.set_minor_locator(tck.MultipleLocator(0.5))
ax.yaxis.set_major_locator(tck.MultipleLocator(50))
ax.yaxis.set_minor_locator(tck.MultipleLocator(25))
ax.set_aspect(ratio)
plt.subplots_adjust(wspace=0, hspace=0, bottom=0.015) # Ensure neat output
plt.savefig('/Users/natsomme/Documents/TheoryAndObservations.png', bbox_inches='tight', figsize=(10,10), dpi=500)
plt.figure()
plt.subplot(211)
plt.plot(absmag, gradlag)
plt.xlabel('Magnitude', fontsize='large')
plt.ylabel('Gradient of time lag', fontsize='large')
plt.axis([np.min(absmag), np.max(absmag), np.min(gradlag), np.max(gradlag)])
plt.subplot(212)
plt.plot(absmag, graddiv)
plt.xlabel('Magnitude', fontsize='large')
plt.ylabel('Gradient of time lag by lag', fontsize='large')
plt.axis([np.min(absmag), np.max(absmag), np.min(graddiv), np.max(graddiv)])
def plotOzDESAGN(QSOTemplate, plotIndividuals=False, binends=None, name=None):
AGN = createOzDESAGN(QSOTemplate)
rFilter = AGN.getFilter('rFilter')
iFilter = AGN.getFilter('iFilter')
# Extract data from file
catalogue = '/Users/natsomme/Dropbox/Uni/Thesis/Run007_MasterInput.cat'
data = np.loadtxt(catalogue)
mag = data[:, 3]
z = data[:, 4]
# Extract physically reasonable AGN
mag = mag[z>0]
z = z[z>0]
n = np.size(z)
Mag = np.zeros(n)
# Calculate absMag using QSO-functions
for i in range(n):
Lbol = AGN.getQSObol(mag[i], z[i], rFilter)
Mag[i] = AGN.getQSOabsMag(Lbol, 0.0, iFilter)
# End for-loop
# Plot redshift and magnitude histograms if desired
if (plotIndividuals == True):
# Create bins
zBins = np.logspace(-3, np.log10(0.69), 20)
zBins = np.concatenate(( zBins, np.logspace(np.log10(0.74), np.log10(1.58), 4) ))
zBins = np.concatenate(( zBins, np.logspace(np.log10(1.94), np.log10(4.0), 5)))
MagBins = np.linspace(np.min(Mag), np.max(Mag), 50)
plt.figure(dpi=500)
plt.hist(z, bins=zBins, facecolor='g')
plt.xlabel('Redshift z')
plt.ylabel('Number of objects')
plt.figure(dpi=500)
plt.hist(Mag, bins=MagBins, facecolor='g')
plt.xlabel('Absolute magnitude')
plt.ylabel('Number of objects')
# End if-statement
# Create a figure with the distributions of the OzDES AGN sample
plt.figure(figsize=(9,15), dpi=500)
# Annotate z-values corresponding to emission lines
plt.plot([0.69, 0.69], [-40, -10], color = '0.4')
plt.plot([0.74, 0.74], [-40, -10], color = '0.4')
plt.plot([1.58, 1.58], [-40, -10], color = '0.4')
plt.plot([1.94, 1.94], [-40, -10], color = '0.4')
if (binends != None):
# Fill out optimised bins
for i in range(np.size(binends)):
if (i < np.size(binends)-1):
if (np.remainder(i, 2) == 0):
shade = 0.2
else:
shade = 0.4
# End if-statement
plt.fill_between([0.69, 1.94], [binends[i], binends[i]], y2=[binends[i+1], binends[i+1]],
color='deeppink', alpha=shade)
# End if-statement
plt.plot([0.69, 1.94], [binends[i], binends[i]], color = '0.3', alpha=0.2)
# End for-loop
# End if-statement
# Plot the AGN in the OzDES sample
plt.plot(z, Mag, 'o', color='mediumvioletred')
# Figure properties
plt.axis([0.1, 4, -27.7, -18.9])
plt.xlabel('Redshift', fontsize='large')
plt.ylabel('Absolute magnitude', fontsize='large')
ax = plt.axes()
ax.xaxis.set_major_locator(MultipleLocator(0.5))
ax.xaxis.set_minor_locator(MultipleLocator(0.125))
ax.yaxis.set_major_locator(MultipleLocator(1))
ax.yaxis.set_minor_locator(MultipleLocator(0.25))
ax.set_aspect(3.9/(27.7-18.9))
if (name==None):
plt.savefig('/Users/natsomme/Documents/OzDES_AGN.png', dpi=500, bbox_inches='tight')
else:
plt.savefig('/Users/natsomme/Documents/OzDES_AGN'+name+'.png', dpi=500, bbox_inches='tight')
fName = 'fakedata.dat' section = 'oneline' emLine = 'Hb' binends = np.array([-26.15, -25.75, -25.35, -24.95, -24.55, -24.15, -23.75, -23.35, -22.85, -22.35])#, -21.55]) pm = 1.3 limits = np.array([0., 0.74, -22.90-pm, -22.90])#MagBins[30], MagBins[31]]) number = findAGNnumber(limits, QSOtemplate) #number = 1 print ' ' print 'Limits:', limits print 'Number of AGN within limits:', number # Initialisation AGN = createOzDESAGN(QSOtemplate) stack = Stacking() stats = Stats() conv = Convergence() plotOzDESAGN(QSOtemplate, plotIndividuals=False, binends=binends, name='MgII') # Create AGN AGN.generateAGN(limits, True, number, resDir, fName) output = AGN.generateLightCurveParameters(section, resDir+fName) AGN.printLightCurveParametersToFile(output, resDir+params) lag = printLagValueInParameterBin(resDir+params, limits, emLine=emLine) # Do the Javelin magic runJavelinOnFakeData(resDir+params, limits, number, resDir, string, emLine=emLine)
