def plot_syvsz(z_phot, z_true, binning, s, s_label, sp_label, x_label, y_ref, sig="nan"):
z_bin = tool.c_binning(binning)
if x_label == "z(true)":
z = z_true
if x_label == "z(phot)":
z = z_phot
if delta_z_str == "$\Delta z$":
delta_z = z_phot - z_true
if delta_z_str == "$\Delta z/(1+z)$":
delta_z = (z_phot - z_true) / (1 + z_true)
# sp: estimator percentage (median, sigma68, ...)
# s: estimator (mean, sigma, ...)
if s == "bias":
sp_y, err_sp_y, s_y, err_s_y = tool.biasvsz(z, delta_z, binning)
np.savetxt(
"/Users/polstein/Desktop/" + s + "_" + x_label + ".txt", np.array([z_bin, sp_y, err_sp_y]).T, fmt="%5.5f"
)
if s == "sigma":
sp_y, err_sp_y, s_y, err_s_y = tool.sigmavsz(z, delta_z, binning)
plt.ylim(ymax=yrange_sigma * sigma_ref)
np.savetxt(
"/Users/polstein/Desktop/" + s + "_" + x_label + ".txt", np.array([z_bin, sp_y, err_sp_y]).T, fmt="%5.5f"
)
if s == str(sig) + "$\sigma$ outliers fraction":
sp_y, err_sp_y_low, err_sp_y_high, s_y, err_s_y_low, err_s_y_high = tool.outliersvsz(z, delta_z, binning, sig)
err_sp_y = [err_sp_y_low, err_sp_y_high]
err_s_y = [err_s_y_low, err_s_y_high]
np.savetxt(
"/Users/polstein/Desktop/" + s + "_" + x_label + ".txt",
np.array([z_bin, s_y, err_s_y_low, err_s_y_high]).T,
fmt="%5.5f",
)
if sig == 2:
plt.ylim(ymax=yrange_out2)
if sig == 3:
plt.ylim(ymax=yrange_out3)
plt.errorbar(z_bin, s_y, err_s_y, color="black", label=s_label)
plt.errorbar(z_bin, sp_y, err_sp_y, color="blue", label=sp_label)
plt.axhline(y=y_ref, linewidth=1, color="red")
if s != "bias":
plt.ylim(ymin=0)
# plt.ylim(ymax = 0.25)
plt.xlim(xmin=z_min, xmax=z_max)
plt.xlabel(x_label)
plt.ylabel(s + " " + delta_z_str)
plt.legend(loc="best")
return sp_y, err_sp_y
def plot_syvsz(z_phot, z_true, binning, s ,s_label, sp_label, x_label , y_ref, sig = "nan"):
z_bin = tool.c_binning(binning)
if(x_label == "z(true)"): z = z_true
if(x_label == "z(phot)"): z = z_phot
if(delta_z_str == "$\Delta z$"): delta_z = z_phot - z_true
if(delta_z_str == "$\Delta z/(1+z)$"): delta_z = (z_phot - z_true) / (1 + z_true)
#sp: estimator percentage (median, sigma68, ...)
#s: estimator (mean, sigma, ...)
if(s == "bias"):
sp_y, err_sp_y, s_y , err_s_y = tool.biasvsz(z, delta_z, binning)
np.savetxt('/Users/pmarti/Desktop/' + s + '_' + x_label + '.txt', np.array([z_bin, sp_y, err_sp_y]).T, fmt = '%5.5f')
if(s == "sigma"):
sp_y, err_sp_y, s_y , err_s_y = tool.sigmavsz(z, delta_z, binning)
plt.ylim(ymax = yrange_sigma * sigma_ref)
np.savetxt('/Users/pmarti/Desktop/' + s + '_' + x_label + '.txt', np.array([z_bin, sp_y, err_sp_y]).T, fmt = '%5.5f')
if(s == str(sig) + "$\sigma$ outliers fraction"):
sp_y, err_sp_y_low, err_sp_y_high, s_y , err_s_y_low, err_s_y_high = tool.outliersvsz(z, delta_z, binning, sig)
err_sp_y = [err_sp_y_low, err_sp_y_high]
err_s_y = [err_s_y_low, err_s_y_high]
np.savetxt('/Users/pmarti/Desktop/' + s + '_' + x_label + '.txt', np.array([z_bin, s_y, err_s_y_low, err_s_y_high]).T, fmt = '%5.5f')
if(sig == 2): plt.ylim(ymax = yrange_out2)
if(sig == 3): plt.ylim(ymax = yrange_out3)
plt.errorbar(z_bin, s_y, err_s_y, color = 'black', label = s_label)
plt.errorbar(z_bin, sp_y, err_sp_y, color = 'blue', label = sp_label)
plt.axhline( y = y_ref, linewidth = 1, color = 'red')
if(s != "bias"): plt.ylim(ymin = 0)
#plt.ylim(ymax = 0.25)
plt.xlim(xmin = z_min, xmax = z_max)
plt.xlabel(x_label)
plt.ylabel(s + " " + delta_z_str)
plt.legend(loc = 'best')
return sp_y, err_sp_y
def plot_dzvsodds(cat_before, cat, binning, Dz_range, z_min, z_max, qual_para, od_bin):
if(delta_z_str == "$\Delta z$"): delta_z = cat_before['z_phot'] - cat_before['z_true']
if(delta_z_str == "$\Delta z/(1+z)$"): delta_z = (cat_before['z_phot'] - cat_before['z_true']) / (1 + cat_before['z_true'])
od_min = cat_before['odds'].min()
od_max = cat_before['odds'].max()
#if(od_max > 1.0): od_max = 1.
odds_binning = np.arange(od_min, od_max, od_bin)
odds_bin = tool.c_binning(odds_binning)
s_y, err_s_y = tool.sigmavsodds(cat_before['odds'], delta_z, odds_bin = odds_binning)
mult = 1
plt.figure(figsize=(13, 7))
plt.subplot(121)
plt.plot(cat_before['odds'], delta_z, 'o', markersize = 1)
plt.errorbar(odds_bin, mult*s_y, mult*err_s_y, color = 'green', label = "cumulative RMSx" + str(mult))
plt.axhline( 0, linewidth = 1, color = 'red')
plt.axvline( odds_cut, linewidth = 1, color = 'blue', label = qual_para + " cut = " + str(odds_cut))
plt.xlim(xmin = 0, xmax = 1)
if (qual_para == 'odds'): plt.xlim(xmin = 0, xmax = 1)
if (qual_para == 'PDZbest'): plt.xlim(xmin = 0, xmax = 100)
plt.ylim(ymin = -Dz_range, ymax = Dz_range)
plt.xlabel(qual_para)
plt.ylabel(delta_z_str)
z_bin = tool.c_binning(binning)
n_true, N_true = tool.hist_oddscut(cat_before['z_true'], cat['z_true'], binning)
n_phot, N_phot = tool.hist_oddscut(cat_before['z_phot'], cat['z_phot'], binning)
#c:completeness
c_true = np.array([])
c_true_high = np.array([])
c_true_low = np.array([])
c_phot = np.array([])
c_phot_high = np.array([])
c_phot_low = np.array([])
for i in range(len(binning)-1):
if N_true[i] > 1: c, c_low, c_high = tool.Completeness(0, n_true[i], N_true[i])
else:
c = nan
c_high = nan
c_low = nan
c_true = np.append(c_true, c)
c_true_high = np.append(c_true_high, c_high)
c_true_low = np.append(c_true_low, c_low)
if N_phot[i] > 1: c, c_low, c_high = tool.Completeness(0, n_phot[i], N_phot[i])
else:
c = nan
c_high = nan
c_low = nan
c_phot = np.append(c_phot, c)
c_phot_high = np.append(c_phot_high, c_high)
c_phot_low = np.append(c_phot_low, c_low)
err_c_true = [c_true_low, c_true_high]
err_c_phot = [c_phot_low, c_phot_high]
plt.subplot(122)
plt.errorbar(z_bin, c_true, err_c_true, color = 'blue', label = "z(true)")
plt.errorbar(z_bin, c_phot, err_c_phot, color = 'red', label = "z(phot)")
plt.ylabel("Completeness")
plt.xlabel("z")
plt.xlim(xmin = z_min, xmax = z_max)
plt.ylim(ymin = 0, ymax = 1.)
plt.legend(loc = 'best')
plt.savefig(plot_folder + "odds_cut.png", bbox_inches='tight')
plt.close()
def plot_dzvserrz(cat_before, cat, binning, Dz_range, z_min, z_max):
if(delta_z_str == "$\Delta z$"): delta_z = cat_before['z_phot'] - cat_before['z_true']
if(delta_z_str == "$\Delta z/(1+z)$"): delta_z = (cat_before['z_phot'] - cat_before['z_true']) / (1 + cat_before['z_true'])
errz_bin = tool.c_binning(errz_binning)
s_y, err_s_y = tool.sigmavserrz(cat_before['err_z_phot'], delta_z)
mult = 1
plt.figure(figsize=(13, 7))
plt.subplot(121)
plt.plot(cat_before['err_z_phot'], delta_z, 'o', markersize = 1)
plt.errorbar(errz_bin, mult*s_y, mult*err_s_y, color = 'green', label = "cumulative RMSx" + str(mult))
plt.axhline( 0, linewidth = 1, color = 'red')
plt.axvline( errz_cut, linewidth = 1, color = 'blue', label = "ERR z(phot) cut = " + str(errz_cut))
plt.xlim(xmin = 0, xmax = yrange_sigma * sigma_ref)
plt.ylim(ymin = -Dz_range, ymax = Dz_range)
plt.xlabel("ERR z(phot)")
plt.ylabel(delta_z_str)
plt.legend(loc = 'best')
z_bin = tool.c_binning(binning)
n_true, N_true = tool.hist_oddscut(cat_before['z_true'], cat['z_true'], binning)
n_phot, N_phot = tool.hist_oddscut(cat_before['z_phot'], cat['z_phot'], binning)
#c:completeness
c_true = np.array([])
c_true_high = np.array([])
c_true_low = np.array([])
c_phot = np.array([])
c_phot_high = np.array([])
c_phot_low = np.array([])
for i in range(len(binning)-1):
if N_true[i] > 1: c, c_low, c_high = tool.Completeness(0, n_true[i], N_true[i])
else:
c = nan
c_high = nan
c_low = nan
c_true = np.append(c_true, c)
c_true_high = np.append(c_true_high, c_high)
c_true_low = np.append(c_true_low, c_low)
if N_phot[i] > 1: c, c_low, c_high = tool.Completeness(0, n_phot[i], N_phot[i])
else:
c = nan
c_high = nan
c_low = nan
c_phot = np.append(c_phot, c)
c_phot_high = np.append(c_phot_high, c_high)
c_phot_low = np.append(c_phot_low, c_low)
err_c_true = [c_true_low, c_true_high]
err_c_phot = [c_phot_low, c_phot_high]
plt.subplot(122)
plt.errorbar(z_bin, c_true, err_c_true, color = 'blue', label = "z(true)")
plt.errorbar(z_bin, c_phot, err_c_phot, color = 'red', label = "z(phot)")
plt.ylabel("Completeness")
plt.xlabel("z")
plt.xlim(xmin = z_min, xmax = z_max)
plt.ylim(ymin = 0, ymax = 1.)
plt.legend(loc = 'best')
plt.savefig(plot_folder + "errz_cut.png")
plt.close()
#print "Num. gal. after z(photo)<%2.2f cut: %d %1.1f%%" % (z_cut, n , 100 * float(n)/float(N) )
#Seeting dictionary before odds cut......................
cat_before = pt.dict(cat_col)
#odds cut...............................................
cat_col, n, N = pt.odds_cut(cat_col, odds_cut)
#print "Num. gal. after odds> %2.2f cut: %d %1.1f%%" % (odds_cut, n , 100 * float(n)/float(N) )
#Seeting dictionary.....................................
cat = pt.dict(cat_col)
#Defining binning.......................................
binning = pt.set_binning(cat)
z_bin = pt.c_binning(binning)
sp_y, err_sp_y = pl.plot_syvsz(cat['z_phot'], cat['z_true'], binning, "sigma", "$\sigma$", "$\sigma_{68}$", "z(true)" , sigma_ref)
sig68 = np.append(sig68, sp_y)
err_sig68 = np.append(err_sig68, err_sp_y)
sig68 = np.reshape(sig68, (shape, -1))
err_sig68 = np.reshape(err_sig68, (shape, -1))
plt.close()
for i in range(len(sig68)): plt.errorbar(z_bin, sig68[i], err_sig68[i], label = files[i])
#for i in range(len(sig68)): plt.plot(z_bin, sig68[i], label = files[i])
plt.rcParams['legend.fontsize'] = 'small'
plt.axhline( y = sigma_ref, linewidth = 1, color = 'red')
#Gencat_nonobs............................................
cat_file = "des_auto_bpz.bpz"
#Main.....................................................
cat = np.loadtxt(cat_file, usecols = (1,5,9), unpack = True)
zp = cat[0]
zs = cat[2]
od = cat[1]
dz = zs - zp
print zp, zs, od
binning = np.arange(0,1.01,0.05)
c_od = pt.c_binning(binning)
n_bin = len(binning) - 1
sig68 = np.zeros(n_bin)
std = np.zeros(n_bin)
for i in range(n_bin):
mask = (od > binning[i]) & (od < 1.0)
x = np.compress(mask, dz)
sig68[i] = pt.Sigma68(x)
std[i] = np.std(x)
plt.plot(od, dz, 'o', color = 'black', markersize = 2)
plt.plot(c_od, sig68, 'g', linewidth = 2, label = '$\sigma_{68}$')
plt.plot(c_od, std, 'b', linewidth = 2, label = 'RMS')
plt.axhline(y = 0, linestyle = 'dashed', linewidth = 1, color = 'red')
