def pdf_func(data_arr, bins_arr,): N_pix, edg_f = binned(data_arr, data_arr, statistic = 'count', bins = bins_arr)[:2] pdf_pix = (N_pix / np.sum(N_pix) ) / (edg_f[1] - edg_f[0]) pdf_err = ( np.sqrt(N_pix) / np.sum(N_pix) ) / (edg_f[1] - edg_f[0]) f_cen = 0.5 * ( edg_f[1:] + edg_f[:-1]) id_zero = N_pix < 1. pdf_arr = pdf_pix[ id_zero == False] err_arr = pdf_err[ id_zero == False] pdf_x = f_cen[ id_zero == False] return pdf_arr, err_arr, pdf_x
samp_flux = samp_flux[id_nn == False] / pixel**2
bin_wide_samp = 1e-2
bin_edgs_samp = np.arange(
samp_flux.min(),
samp_flux.max(),
bin_wide_samp,
)
sam_tmp_mean_f.append(np.mean(samp_flux))
sam_tmp_medi_f.append(np.median(samp_flux))
sam_tmp_err.append(np.std(samp_flux))
# pdf of sample img flux
N_pix, edg_f = binned(samp_flux,
samp_flux,
statistic='count',
bins=bin_edgs_samp)[:2]
pdf_pix = (N_pix / np.sum(N_pix)) / (edg_f[1] - edg_f[0])
pdf_err = (np.sqrt(N_pix) / np.sum(N_pix)) / (edg_f[1] - edg_f[0])
f_cen = 0.5 * (edg_f[1:] + edg_f[:-1])
mean_f = np.mean(samp_flux)
medi_f = np.median(samp_flux)
try:
idu = N_pix != 0.
popt, pcov = optimize.curve_fit(
gau_func,
f_cen[idu],
bin_pix_cont = bin_pix_cont[id_nan == False] #bin_pix_cont = bin_pix_cont.astype( np.int32 ) lis_f = [] for dd in range( len(bin_flux) ): rep_arr = np.ones( bin_pix_cont[dd], dtype = np.float32) * bin_flux[dd] lis_f.append( rep_arr) F_arr = np.hstack(lis_f) / pixel**2 mean_f = np.mean( F_arr ) medi_f = np.median( F_arr ) bin_wide = 5e-4 bin_edgs = np.arange( F_arr.min(), F_arr.max(), bin_wide) N_pix, edg_f = binned(F_arr, F_arr, statistic = 'count', bins = bin_edgs)[:2] pdf_pix = (N_pix / np.sum(N_pix) ) / (edg_f[1] - edg_f[0]) pdf_err = ( np.sqrt(N_pix) / np.sum(N_pix) ) / (edg_f[1] - edg_f[0]) f_cen = 0.5 * ( edg_f[1:] + edg_f[:-1]) abs_min_f = np.abs( F_arr.min() ) try: idu = N_pix != 0. ''' popt, pcov = optimize.curve_fit(gau_func, f_cen[idu], pdf_pix[idu], p0 = [mean_f, np.std(F_arr)], sigma = pdf_err[idu],) e_mu, e_chi = popt[0], popt[1] fit_line = gau_func(f_cen, e_mu, e_chi) ''' eta_0, lamda_0 = 3, -0.5
lc0 = np.array([np.min(ll) for ll in tot_flux])
lc1 = np.array([np.max(ll) for ll in tot_flux])
bins_2 = np.linspace(lc0.min(), lc1.max(), 60)
chi_test = []
nu_on_chi = []
for kk in range(len(k_r)):
## parameter fit
tt_data = mean_percen[kk]
bin_di = bins_0
mu = np.mean(tt_data)
sqr_var = np.std(tt_data)
pix_n, edgs = binned(tt_data, tt_data, statistic='count', bins=bin_di)[:2]
pdf_pix = (pix_n / np.sum(pix_n)) / (edgs[1] - edgs[0])
pdf_err = (np.sqrt(pix_n) / np.sum(pix_n)) / (edgs[1] - edgs[0])
x_cen = 0.5 * (edgs[1:] + edgs[:-1])
idu = pix_n != 0.
use_obs = pix_n[idu]
use_err = np.sqrt(use_obs)
use_x = x_cen[idu]
popt, pcov = curve_fit(
gau_func,
use_x,
pdf_pix[idu],
p0=[mu, sqr_var],
sigma=pdf_err[idu],
)
img_1 = data_1[0].data
idnn = np.isnan(img_1)
sub_flux_1 = img_1[idnn == False]
pix_flux_1sigm.append(sub_flux_0)
pix_flux_2sigm.append(sub_flux_1)
## stack test
stack_1sigm.append(img_0)
stack_2sigm.append(img_1)
eta_0 = len(sub_flux_0) / (1489 * 2048)
eta_1 = len(sub_flux_1) / (1489 * 2048)
pix_remain_1sigm.append(eta_0)
pix_remain_2sigm.append(eta_1)
pock_n, edgs = binned(sub_flux_0, sub_flux_0, statistic='count',
bins=60)[:2]
id_zero = pock_n == 0.
pock_n = pock_n[id_zero == False]
sub_pdf0 = (pock_n / np.sum(pock_n)) / (edgs[1] - edgs[0])
sub_x0 = 0.5 * (edgs[1:] + edgs[:-1])
sub_x0 = sub_x0[id_zero == False]
popt, pcov = curve_fit(gau_func,
sub_x0,
sub_pdf0,
p0=[np.mean(sub_flux_0),
np.std(sub_flux_0)])
e_mu0, e_sigm0 = popt[0], popt[1]
fit_l0 = gau_func(
sub_x0,
e_mu0,
e_sigm0,
def hist_analysis_func(
cat_file,
img_file,
mask_img_file,
band,
star_cat,
gal_cat,
out_imgs,
pix_scale,
N_step,
):
"""
cat_file : cluster catalog (.csv file, including path information)
img_file : imgs have no mask
mask_img_file : imgs have been masked
band : band (ie. 'g', 'r', 'i',), str type
star_cat : star catalog. (.txt or .csv file, including path information)
gal_cat : source catalog in image region (builded by Source Extractor, .cat file, including path information)
out_imgs : ouput of figs,(.png or .pdf files, including path information)
pix_scale : pixel size of CCD, in unit of 'arcsec'
N_step : grid size for img division.
"""
### cat and select
dat = pds.read_csv(cat_file)
ra, dec, z = np.array(dat.ra), np.array(dat.dec), np.array(dat.z)
clus_x, clus_y = np.array(dat.bcg_x), np.array(dat.bcg_y)
Da = Test_model.angular_diameter_distance(z).value
N_samp = len(z)
pixel = pix_scale * 1.
Angu_l = (1 / Da) * rad2asec
R_pix = Angu_l / pixel
cen_sigm, cen_mu = [], []
img_mu, img_sigm = [], []
### overview on img structure
for kk in range(N_samp):
ra_g, dec_g, z_g = ra[kk], dec[kk], z[kk]
# original img
data = fits.open(img_file % (band, ra_g, dec_g, z_g), )
img = data[0].data
head = data[0].header
wcs_lis = awc.WCS(head)
xn, yn = clus_x[kk], clus_y[kk]
# mask imgs
res_data = fits.open(mask_img_file % (band, ra_g, dec_g, z_g), )
remain_img = res_data[0].data
# grid img (for selecting flare, saturated region...)
block_m, block_pix, block_Var, block_S0, x_edgs, y_edgs = cc_grid_img(
remain_img, N_step, N_step)
idzo = block_pix < 1.
pix_eta = block_pix / block_S0
idnn = np.isnan(pix_eta)
pix_eta[idnn] = 0.
idnul = pix_eta < 5e-2
block_m[idnul] = 0.
img_mu.append(np.nanmean(block_m[idnul == False]))
img_sigm.append(np.nanstd(block_m[idnul == False]))
idnn = np.isnan(remain_img)
bin_flux = remain_img[idnn == False]
bin_di = np.linspace(bin_flux.min(), bin_flux.max(), 51) / pixel**2
pix_n, edgs = binned(bin_flux / pixel**2,
bin_flux / pixel**2,
statistic='count',
bins=bin_di)[:2]
pdf_pix = (pix_n / np.sum(pix_n)) / (edgs[1] - edgs[0])
pdf_err = (np.sqrt(pix_n) / np.sum(pix_n)) / (edgs[1] - edgs[0])
x_cen = 0.5 * (edgs[1:] + edgs[:-1])
idu = pix_n != 0.
use_obs = pix_n[idu]
use_err = np.sqrt(use_obs)
use_x = x_cen[idu]
popt, pcov = curve_fit(
gau_func,
use_x,
pdf_pix[idu],
p0=[np.mean(bin_flux / pixel**2),
np.std(bin_flux / pixel**2)],
sigma=pdf_err[idu],
)
e_mu, e_chi = popt[0], popt[1]
fit_line = gau_func(x_cen, e_mu, e_chi)
### applied the mask region
star_file = star_cat % (z_g, ra_g, dec_g)
galax_file = gal_cat % (ra_g, dec_g, z_g)
tot_Numb, tot_cx, tot_cy, tot_a, tot_b, tot_theta = get_cat(
star_file,
galax_file,
pixel,
wcs_lis,
)
sc_x, sc_y = tot_cx / (img.shape[1] / block_m.shape[1]), tot_cy / (
img.shape[0] / block_m.shape[0])
sc_a, sc_b = tot_a * (block_m.shape[1] / img.shape[1]), tot_b * (
block_m.shape[0] / img.shape[0])
sc_x, sc_y = sc_x - 0.5, sc_y - 0.5
Rpp = R_pix[kk]
fig = plt.figure(figsize=(13.12, 9.84))
ax0 = fig.add_axes([0.05, 0.55, 0.40, 0.45])
ax1 = fig.add_axes([0.55, 0.55, 0.40, 0.45])
ax2 = fig.add_axes([0.05, 0.05, 0.40, 0.45])
ax3 = fig.add_axes([0.55, 0.05, 0.40, 0.40])
ax0.set_title('img ra%.3f dec%.3f z%.3f' % (ra_g, dec_g, z_g), )
tf = ax0.imshow(img / pixel**2,
cmap='Greys',
origin='lower',
vmin=1e-4,
vmax=1e0,
norm=mpl.colors.LogNorm())
clust = Circle(
xy=(xn, yn),
radius=Rpp,
fill=False,
ec='b',
ls='-',
linewidth=1,
alpha=0.50,
)
ax0.add_patch(clust)
cb = plt.colorbar(
tf,
ax=ax0,
fraction=0.035,
pad=0.01,
label='SB [nanomaggies / $arcsec^2$]',
)
ax1.set_title('after masking')
tg = ax1.imshow(
remain_img / pixel**2,
origin='lower',
cmap='seismic',
vmin=-4e-1,
vmax=4e-1,
)
cb = plt.colorbar(
tg,
ax=ax1,
fraction=0.035,
pad=0.01,
label='SB [nanomaggies / $arcsec^2$]',
)
cb.formatter.set_powerlimits((0, 0))
### image patch_mean case
ax2.set_title('2D hist of sub-patch mean value')
th = ax2.imshow(
block_m / pixel**2,
origin='lower',
cmap='seismic',
vmin=-4e-2,
vmax=4e-2,
)
cb = plt.colorbar(
th,
ax=ax2,
fraction=0.034,
pad=0.01,
label='SB [nanomaggies / $arcsec^2$]',
)
cb.formatter.set_powerlimits((0, 0))
for mm in range(tot_Numb):
ellips = Ellipse(
xy=(sc_x[mm], sc_y[mm]),
width=sc_a[mm],
height=sc_b[mm],
angle=tot_theta[mm],
fill=True,
fc='w',
ec='w',
ls='-',
linewidth=0.75,
)
ax2.add_patch(ellips)
for mm in range(block_m.shape[1]):
for nn in range(block_m.shape[0]):
ax2.text(mm,
nn,
s='%.3f' % pix_eta[nn, mm],
ha='center',
va='center',
color='g',
fontsize=8,
alpha=0.5)
ax2.set_xlabel('effective pixel ratio shown in green text')
ax3.set_title('pixel SB PDF [after masking]')
ax3.hist(
bin_flux / pixel**2,
bins=bin_di,
density=True,
color='b',
alpha=0.5,
)
ax3.plot(
x_cen,
fit_line,
color='r',
alpha=0.5,
label='Gaussian \n $\\mu=%.4f$ \n $\\sigma=%.4f$' % (e_mu, e_chi),
)
ax3.axvline(
x=0,
ls='-',
color='k',
alpha=0.5,
)
ax3.axvline(x=e_mu - e_chi,
ls='--',
color='k',
alpha=0.5,
label='1 $\\sigma$')
ax3.axvline(
x=e_mu + e_chi,
ls='--',
color='k',
alpha=0.5,
)
ax3.legend(
loc=1,
frameon=False,
)
ax3.set_xlabel('pixel SB [nanomaggies / $arcsec^2$]')
ax3.set_ylabel('PDF')
plt.savefig(out_imgs % (band, ra_g, dec_g, z_g), dpi=300)
plt.close()
raise
return
def binned_img_flux():
N_bin = 30
bin_side = np.linspace(-1e-1, 1e-1, N_bin + 1)
for kk in range(rank, rank + 1):
with h5py.File(load + 'random_cat/rand_%s_band_catalog.h5' % (band[kk]), 'r') as f:
tmp_array = np.array(f['a'])
ra, dec, z, rich = np.array(tmp_array[0]), np.array(tmp_array[1]), np.array(tmp_array[2]), np.array(tmp_array[3])
zN = len(z)
bin_flux_mean = np.zeros((zN, N_bin), dtype = np.float)
bin_flux_median = np.zeros((zN, N_bin), dtype = np.float)
bin_num = np.zeros((zN, N_bin), dtype = np.float)
for jj in range(zN):
ra_g, dec_g, z_g = ra[jj], dec[jj], z[jj]
data_A = fits.open(load + 'random_cat/edge_cut_img/rand_pont_Edg_cut-%s-ra%.3f-dec%.3f-redshift%.3f.fits' % (band[kk], ra_g, dec_g, z_g) )
img_A = data_A[0].data
id_nan = np.isnan(img_A)
flux_array = img_A[id_nan == False]
value_mean = binned(flux_array, flux_array, statistic='mean', bins = bin_side)[0]
value_median = binned(flux_array, flux_array, statistic='median', bins = bin_side)[0]
value_number = binned(flux_array, flux_array, statistic='count', bins = bin_side)[0]
bin_flux_mean[jj,:] = value_mean
bin_flux_median[jj,:] = value_median
bin_num[jj,:] = value_number / np.nansum(value_number)
flux_mean = np.nanmean(bin_flux_mean, axis = 0) / pixel**2
flux_median = np.nanmean(bin_flux_median, axis = 0) / pixel**2
num_mean = np.nanmean(bin_num, axis = 0)
dmp_array = np.array([flux_mean, num_mean])
with h5py.File(load + 'random_cat/stack/%s_band_random_field_pix_SB_mean_pdf.h5' % band[kk], 'w') as f:
f['a'] = np.array(dmp_array)
with h5py.File(load + 'random_cat/stack/%s_band_random_field_pix_SB_mean_pdf.h5' % band[kk], ) as f:
for ll in range(len(dmp_array)):
f['a'][ll,:] = dmp_array[ll,:]
dmp_array = np.array([flux_median, num_mean])
with h5py.File(load + 'random_cat/stack/%s_band_random_field_pix_SB_median_pdf.h5' % band[kk], 'w') as f:
f['a'] = np.array(dmp_array)
with h5py.File(load + 'random_cat/stack/%s_band_random_field_pix_SB_median_pdf.h5' % band[kk], ) as f:
for ll in range(len(dmp_array)):
f['a'][ll,:] = dmp_array[ll,:]
with h5py.File(load + 'random_cat/stack/%s_band_random_field_pix_SB_mean_pdf.h5' % band[kk], 'r') as f:
dmp_array = np.array(f['a'])
flux_mean, num_mean = dmp_array[0], dmp_array[1]
dx = flux_mean[1:] - flux_mean[:-1]
dx = np.r_[dx[0], dx]
M_flux_mean = np.sum(flux_mean * num_mean * dx)
with h5py.File(load + 'random_cat/stack/%s_band_random_field_pix_SB_median_pdf.h5' % band[kk], 'r') as f:
dmp_array = np.array(f['a'])
flux_median, num_mean = dmp_array[0], dmp_array[1]
dx = flux_median[1:] - flux_median[:-1]
dx = np.r_[dx[0], dx]
M_flux_median = np.sum(flux_median * num_mean * dx)
fig = plt.figure()
ax = plt.subplot(111)
ax.set_title('%s band random filed pixel SB pdf' % band[kk])
ax.plot(flux_mean, num_mean, c = 'r', ls = '-', label = 'Mean pixel SB')
ax.axvline(x = M_flux_mean, color = 'r', linestyle = '-', alpha = 0.5)
ax.plot(flux_median, num_mean, c = 'g', ls = '--', label = 'Median pixel SB')
ax.axvline(x = M_flux_median, color = 'g', linestyle = '--', alpha = 0.5)
ax.set_xlabel('$ pixel \; SB \; [nanomaggies / arcsec^2]$')
ax.set_ylabel('pdf')
ax.legend(loc = 1, frameon = False)
ax.tick_params(axis = 'both', which = 'both', direction = 'in')
subax = fig.add_axes([0.2, 0.35, 0.25, 0.4])
subax.plot(flux_mean, num_mean, c = 'r', ls = '-', label = 'Mean pixel SB')
subax.axvline(x = M_flux_mean, color = 'r', linestyle = '-', alpha = 0.5)
subax.set_xlim(-5e-4, 5e-4)
xtick = subax.get_xticks()
re_xtick = xtick * 1e4
subax.set_xticks(xtick)
subax.set_xticklabels(['%.1f' % ll for ll in re_xtick])
subax.set_xlabel('1e4 * pixel SB')
subax.tick_params(axis = 'both', which = 'both', direction = 'in')
plt.savefig(load + 'random_cat/stack/%s_band_pix_SB_pdf.png' % band[kk], dpi = 300)
plt.close()
return