# TODO: reproduce the results of the other dust model also
# load BOSS wavelengths, prep for interpolation
boss_lambdas = np.load('/Users/blakechellew/Documents/DustProject/alphas_and_stds/wavelength_boss.npy') # angstroms
total_int_f = interp1d(lambdas, total_int, kind='cubic')
if dust_model == 'zd':
bd12_factor = 0.52
elif dust_model == 'wd':
bd12_factor = 0.49
plot1 = lambdas * bc03_f(lambdas)
plot2 = total_int / nu_I_nu
plot3 = bd12_factor * boss_lambdas * total_int_f(boss_lambdas) / nu_I_nu * bc03_f(boss_lambdas)
# bin the main plot
lambdas_boss_bin, plot3_bin, stds = generate_binned_alphas([plot3], [np.ones(plot3.shape)], boss_lambdas)
plot3_bin = plot3_bin[0]
plot1_mean = np.mean(plot1)
plot2_mean = np.mean(plot2)
plot3_mean = np.mean(plot3)
print("plot1", plot1)
print("plot2", plot2)
print("plot3", plot3)
print("plot 1 mean:", plot1_mean)
print("plot 2 mean:", plot2_mean)
print("plot 3 mean:", plot3_mean)
plt.plot(lambdas, plot1 * plot3_mean / plot1_mean, label='bc03 * wav')
plt.plot(lambdas, plot2 * plot3_mean / plot2_mean, label='just the integral')
plt.plot(lambdas_boss_bin, plot3_bin, step='pre', label='predicted alphas')
plt.legend()
# scaling factors
boss_fluxfactor = 1.38
sdss_fluxfactor = 1.38
# apply correction factor
correction_factors = [
np.load('../alphas_and_stds/correction_factor_boss_iris_smooth.npy'),
np.load('../alphas_and_stds/correction_factor_boss_iris_north_smooth.npy'),
np.load('../alphas_and_stds/correction_factor_boss_iris_south_smooth.npy')
]
correction_factor_sdss = np.load(
'../alphas_and_stds/correction_factor_sdss_iris_smooth.npy')
lambdas_bin, binned_corrections, _ = generate_binned_alphas(
correction_factors,
3 * [np.ones(len(correction_factors[0]))],
wavelength,
boss=True,
bin_width=bin_width)
_, binned_sdss_correction, _ = generate_binned_alphas(
[correction_factor_sdss], [np.ones(len(correction_factor_sdss))],
wavelength_sdss,
boss=False,
bin_width=bin_width)
binned_sdss_correction = binned_sdss_correction[0]
alphas_norad = [
a / boss_fluxfactor * corr
for a, corr in zip(alphas_norad, correction_factors)
]
# bin
alphas_norad_bin_wav, alphas_norad_bin, _ = generate_binned_alphas(
if wd01_model:
bd12_alphas = np.loadtxt('/Users/blakechellew/Documents/DustProject/alphas_and_stds/bd12_fig3_green_052621.csv',
delimiter=",")
else:
bd12_alphas = np.loadtxt('/Users/blakechellew/Documents/DustProject/alphas_and_stds/bd12_fig3_blue_052621.csv',
delimiter=",")
# load plot from brandt code
brandt_path = '/Users/blakechellew/Documents/DustProject/BrandtFiles/brandt_radiative/integrals/'
if wd01_model:
brandt_alphas = np.load(brandt_path + 'alphas_wd.npy')
else:
brandt_alphas = np.load(brandt_path + 'alphas_zd.npy')
# bin some alphas
lambdas_bin, alphas_bin, _ = generate_binned_alphas([alphas], [np.ones(alphas.shape)], wavelength, bin_offset=0)
alphas_bin = alphas_bin[0]
alphas_norad = [np.load('../alphas_and_stds/alphas_boss_iris_2d_012720.npy'),
np.load('../alphas_and_stds/alphas_north011720.npy'),
np.load('../alphas_and_stds/alphas_south011720.npy')]
# apply correction factor
correction_factors = [np.load('../alphas_and_stds/correction_factor_boss_iris_smooth.npy'),
np.load('../alphas_and_stds/correction_factor_boss_iris_north_smooth.npy'),
np.load('../alphas_and_stds/correction_factor_boss_iris_south_smooth.npy')]
_, binned_corrections, _ = generate_binned_alphas(correction_factors, 3 * [np.ones(len(correction_factors[0]))], wavelength, boss=boss)
alphas_norad = [a / boss_fluxfactor * corr for a, corr in zip(alphas_norad, correction_factors)]
# bin
alphas_norad_bin_wav, alphas_norad_bin, _ = generate_binned_alphas(alphas_norad, [np.ones(alphas_norad[0].shape) for i in range(len(alphas_norad))],
wavelength, bin_offset=0)
np.load(alpha_direc_boot + 'bootstrap_alpha_stds_boss_iris_2d_' + loadkey +
'_10.npy'),
np.load(alpha_direc_boot + 'bootstrap_alpha_stds_boss_iris_2d_' + loadkey +
'_15.npy'),
np.load(alpha_direc_boot + 'bootstrap_alpha_stds_boss_iris_2d_' + loadkey +
'_20.npy'),
np.load(alpha_direc_boot + 'bootstrap_alpha_stds_boss_iris_2d_' + loadkey +
'_25.npy'),
np.load(alpha_direc_boot + 'bootstrap_alpha_stds_boss_iris_2d_' + loadkey +
'_30.npy')
]
print("loaded threshold alphas")
_, bootstrap_binned_alphas_thresh_1d, _ = generate_binned_alphas(
bootstrap_alphas_thresh_1d,
bootstrap_alpha_stds_thresh_1d,
wavelength_boss,
boss=True)
_, bootstrap_binned_alphas_thresh_2d, _ = generate_binned_alphas(
bootstrap_alphas_thresh_2d,
bootstrap_alpha_stds_thresh_2d,
wavelength_boss,
boss=True)
print("done binning")
# bootstrap percentiles
bootstrap_binned_lower_thresh_1d = [
np.percentile(b, 16, axis=0) for b in bootstrap_binned_alphas_thresh_1d
] # 68 percent confidence interval
bootstrap_binned_upper_thresh_1d = [
np.percentile(b, 84, axis=0) for b in bootstrap_binned_alphas_thresh_1d
else:
hdulist_direc = '/Users/blakechellew/Documents/DustProject/BrandtFiles/'
hdulist_sdsswav = fits.open('/Users/blakechellew/Documents/DustProject/BrandtFiles/SDSS_allskyspec.fits')
survey_lambdas = np.array(hdulist_sdsswav[1].data)
total_int_f = interp1d(lambdas, total_int, kind='cubic')
if dust_model == 'zd':
bd12_factor = 0.52
elif dust_model == 'wd':
bd12_factor = 0.49
plot1 = lambdas * bc03_f(lambdas)
plot2 = total_int / nu_I_nu
plot3 = bd12_factor * survey_lambdas * total_int_f(survey_lambdas) / nu_I_nu * bc03_f(survey_lambdas)
# bin the main plot
lambdas_bin, plot3_bin, stds = generate_binned_alphas([plot3], [np.ones(plot3.shape)], survey_lambdas)
plot3_bin = plot3_bin[0]
# load bd12 plot for comparison
if dust_model == 'wd':
bd12_plot = np.loadtxt('/Users/blakechellew/Documents/DustProject/alphas_and_stds/bd12_fig3_green_052621.csv',
delimiter=",")
else:
bd12_plot = np.loadtxt('/Users/blakechellew/Documents/DustProject/alphas_and_stds/bd12_fig3_blue_052621.csv',
delimiter=",")
plot1_mean = np.mean(plot1)
plot2_mean = np.mean(plot2)
plot3_mean = np.mean(plot3)
print("plot1", plot1)
print("plot2", plot2)
idx_array_north = [
378, 1271, 2498, 211, 212, 905, 938, 1265, 1509, 2141, 2495, 2500, 2502
]
idx_array_south = [
22, 23, 389, 419, 440, 536, 667, 1065, 1081, 1106, 1684, 1690, 1739, 1786,
2229, 2380, 2381, 2383, 2388, 74, 1078, 1104, 1110, 1393, 1696, 1990, 1997,
2009, 2011, 2382, 2416
]
# override to check just one
# idx_array_north = [2141]
# idx_array_south = []
for bad_idx in idx_array_north:
wav_binned, alphas_binned_1, _ = generate_binned_alphas(
[north[bad_idx]], [np.ones(len(boss_wavelength))], boss_wavelength)
wav_binned, alphas_binned_2, _ = generate_binned_alphas(
[north[1000]], [np.ones(len(boss_wavelength))], boss_wavelength)
alphas_binned_1 = alphas_binned_1[0]
alphas_binned_2 = alphas_binned_2[0]
plt.plot(wav_binned, alphas_binned_1, 'k', label='good')
plt.plot(wav_binned, alphas_binned_2, 'r', label='bad')
plt.legend()
plt.ylim(0, 1)
plt.title("North idx " + str(bad_idx))
plt.show()
for bad_idx in idx_array_south:
wav_binned, alphas_binned_3, _ = generate_binned_alphas(
[south[bad_idx]], [np.ones(len(boss_wavelength))], boss_wavelength)
wav_binned, alphas_binned_4, _ = generate_binned_alphas(
paths_bc03,
showPlots=False) # TEST: was using alphas_bootstrap[i]
print(coeffs_array)
# combine the coefficients:
avg_coeffs = np.mean(coeffs_array, axis=0)
coeff_stds = np.std(coeffs_array, axis=0)
print("coeffs with stds:")
print(avg_coeffs)
print(coeff_stds)
# bin the bootstrap stuff
# I think the std input here can be ones, but maybe there is something better?
print("best fit array:", best_fit_array.shape)
lambdas_bin, best_fit_bin, _ = generate_binned_alphas(
[best_fit_array], [np.ones(best_fit_array.shape)],
best_fit_wavelengths,
bin_offset=0)
best_fit_bin = best_fit_bin[0]
print("best fit bin:", best_fit_bin.shape)
# bin the boss stuff
lambdas_bin_boss, boss_bin, _ = generate_binned_alphas([alphas_boss],
[alpha_stds_boss],
wavelength,
bin_offset=0)
boss_bin = boss_bin[0]
# percentiles
bootstrap_binned_lower = np.nanpercentile(
best_fit_bin, 16, axis=0) # 68 percent confidence interval
bootstrap_binned_upper = np.nanpercentile(best_fit_bin, 84, axis=0)
# TODO: need to run with the original alphas. for now take the mean.
plt.plot(lambdas_bin,