def getdata():
data = fetch_sdss_specgals()
m_max = 17.7
# redshift and magnitude cuts
data = data[data['z'] > 0.08]
data = data[data['z'] < 0.12]
data = data[data['petroMag_r'] < m_max]
# RA/DEC cuts
RAmin, RAmax = 140, 220
DECmin, DECmax = 5, 45
data = data[data['ra'] < RAmax]
data = data[data['ra'] > RAmin]
data = data[data['dec'] < DECmax]
data = data[data['dec'] > DECmin]
ur = data['modelMag_u'] - data['modelMag_r']
flag_red = (ur > 2.22)
flag_blue = ~flag_red
data_red = data[flag_red]
data_blue = data[flag_blue]
return [data, data_red, data_blue]
from sklearn.ensemble import GradientBoostingRegressor from astroML.datasets import fetch_sdss_specgals from astroML.decorators import pickle_results #---------------------------------------------------------------------- # This function adjusts matplotlib settings for a uniform feel in the textbook. # Note that with usetex=True, fonts are rendered with LaTeX. This may # result in an error if LaTeX is not installed on your system. In that case, # you can set usetex to False. from astroML.plotting import setup_text_plots setup_text_plots(fontsize=8, usetex=True) #------------------------------------------------------------ # Fetch and prepare the data data = fetch_sdss_specgals() # put magnitudes in a matrix mag = np.vstack([data['modelMag_%s' % f] for f in 'ugriz']).T z = data['z'] # train on ~60,000 points mag_train = mag[::10] z_train = z[::10] # test on ~6,000 distinct points mag_test = mag[1::100] z_test = z[1::100] #------------------------------------------------------------
""" SDSS Spectroscopic Galaxy Sample -------------------------------- This figure shows photometric colors of the SDSS spectroscopic galaxy sample. """ # Author: Jake VanderPlas <*****@*****.**> # License: BSD # The figure is an example from astroML: see http://astroML.github.com import numpy as np from matplotlib import pyplot as plt from astroML.datasets import fetch_sdss_specgals data = fetch_sdss_specgals() #------------------------------------------------------------ # plot the RA/DEC in an area-preserving projection RA = data['ra'] DEC = data['dec'] # convert coordinates to degrees RA -= 180 RA *= np.pi / 180 DEC *= np.pi / 180 ax = plt.axes(projection='mollweide') ax = plt.axes() ax.grid()
def main():
''' Your goal is to recalculate the luminosity function for the red and
blue galaxies presented in the lower right of Figure 4.10 using
reproducible methods for histogram binning. This activity divides the
script for making these plots into more modular format. The above functions
are only suggestions for how to divide the script, feel free to modularize
the script in a different fashion.
Figure 4.10 shows how the probability density of galaxies as function of
redshift and absolute magnitude varies for two different galaxy populations
selected by color. See the left panels of Figure 4.10 showing the
probability density function p(M, z) for the two populations as a function
of absolute magnitude M and redshift z. p(M, z) = Phi(M) * rho(z) where Phi
is the normalized luminosity function, and rho is the sold angle averaged
number density of galaxies. The normalization of rho in the figure is
confusing, the top-right plot shows that the number density of redder
galaxies, u-r > 2.22, is more dense than blue galaxies at smaller
redshifts. The luminosity function plotted in the bottom right show that
there are fewer bright blue galaxies, u-r < 2.22, than bright red galaxies,
u-r > 2.22.
Read through the script
http://www.astroml.org/book_figures/chapter4/fig_lyndenbell_gals.html
to understand the steps the authors take to calculate the luminosity and
volume number density functions for red and blue galaxies from the SDSS
archive.
*** Truncate the sample to 1/10 the size as commented in line 48 and 49 ***
*** Else computation time exorbitant ***
The author create redshift and absolute magnitude bins to compute the
luminosity function. Find where these bins are initialized, and change the
binning method to either Freedman-Diaconis or Scott. Use the most
appropriate binning strategy for the data.
Plot the results on the same scale and with the same limits as the code in
the example script.
Does the histogram change how you interpret the results of this data? Are
these binning methods sensitive to outliers?
Notes
-----
-> The Freedman-Diaconis and Scott binning methods can imported as
functions in the following way:
from astroML.density_estimation import scotts_bin_width as scott_bin
from astroML.density_estimation import freedman_bin_width as free_bin
'''
from astroML.datasets import fetch_sdss_specgals
data = fetch_sdss_specgals()
z_min = 0.08
z_max = 0.12
m_max = 17.7
data_red, data_blue = cut_data(data, z_min, z_max, m_max)
data_red = data_red[::10]
data_blue = data_blue[::10]
data_samples = (data_red, data_blue)
mu_z, z_mu = derive_dist_mod_tables((0.01, 1.5), 100)
# Initialize data lists to contain blue and red data
Mbins_list = []
zbins_list = []
dist_M_list = []
err_M_list = []
dist_z_list = []
err_z_list = []
archive_files = ['lumfunc_red.npz', 'lumfunc_blue.npz']
# calculate luminosity function for blue and red galaxies
for i in xrange(2):
m = data_samples[i]['petroMag_r']
z = data_samples[i]['z']
M = m - mu_z(z)
zbins, dist_z, err_z, Mbins, dist_M, err_M = \
compute_luminosity_function(z, m, M, m_max, archive_files[i],
Nbootstraps=20, bin_type='freedman', z_mu=z_mu)
Mbins_list.append(Mbins)
zbins_list.append(zbins)
dist_M_list.append(dist_M)
err_M_list.append(err_M)
dist_z_list.append(dist_z)
err_z_list.append(err_z)
# Plot the volume number density and the luminosity function
titles = (r'u-r > 2.22', r'u-r < 2.22')
markers = ['o', '^']
plot_num_density(zbins_list,
dist_z_list,
err_z_list,
titles=titles,
markers=markers)
plot_luminosity_function(Mbins_list,
dist_M_list,
err_M_list,
titles=titles,
markers=markers)
import numpy as np import scipy as sp import scipy.interpolate as spi import matplotlib.pyplot as plt from astroML.datasets import fetch_sdss_specgals from astroML.datasets import fetch_sdss_spectrum from sklearn.manifold import Isomap d_galaxies= fetch_sdss_specgals() Ngals = 1000 k = 20 n = 2 d_galaxies = d_galaxies[0:Ngals] #print d_galaxies #print d_galaxies.shape mjd_galaxies = d_galaxies['mjd'] plate_galaxies = d_galaxies['plate'] fiberID_galaxies = d_galaxies['fiberID'] h_alpha_flux_galaxies = d_galaxies['h_alpha_flux'] extinction_r_galaxies = d_galaxies['extinction_r'] velDisp_galaxies = d_galaxies['velDisp']
def main():
''' Your goal is to recalculate the luminosity function for the red and
blue galaxies presented in the lower right of Figure 4.10 using
reproducible methods for histogram binning. This activity divides the
script for making these plots into more modular format. The above functions
are only suggestions for how to divide the script, feel free to modularize
the script in a different fashion.
Figure 4.10 shows how the probability density of galaxies as function of
redshift and absolute magnitude varies for two different galaxy populations
selected by color. See the left panels of Figure 4.10 showing the
probability density function p(M, z) for the two populations as a function
of absolute magnitude M and redshift z. p(M, z) = Phi(M) * rho(z) where Phi
is the normalized luminosity function, and rho is the sold angle averaged
number density of galaxies. The normalization of rho in the figure is
confusing, the top-right plot shows that the number density of redder
galaxies, u-r > 2.22, is more dense than blue galaxies at smaller
redshifts. The luminosity function plotted in the bottom right show that
there are fewer bright blue galaxies, u-r < 2.22, than bright red galaxies,
u-r > 2.22.
Read through the script
http://www.astroml.org/book_figures/chapter4/fig_lyndenbell_gals.html
to understand the steps the authors take to calculate the luminosity and
volume number density functions for red and blue galaxies from the SDSS
archive.
*** Truncate the sample to 1/10 the size as commented in line 48 and 49 ***
*** Else computation time exorbitant ***
The author create redshift and absolute magnitude bins to compute the
luminosity function. Find where these bins are initialized, and change the
binning method to either Freedman-Diaconis or Scott. Use the most
appropriate binning strategy for the data.
Plot the results on the same scale and with the same limits as the code in
the example script.
Does the histogram change how you interpret the results of this data? Are
these binning methods sensitive to outliers?
Notes
-----
-> The Freedman-Diaconis and Scott binning methods can imported as
functions in the following way:
from astroML.density_estimation import scotts_bin_width as scott_bin
from astroML.density_estimation import freedman_bin_width as free_bin
'''
from astroML.datasets import fetch_sdss_specgals
data = fetch_sdss_specgals()
z_min = 0.08
z_max = 0.12
m_max = 17.7
data_red, data_blue = cut_data(data, z_min, z_max, m_max)
data_red = data_red[::10]
data_blue = data_blue[::10]
data_samples = (data_red, data_blue)
mu_z, z_mu = derive_dist_mod_tables((0.01, 1.5), 100)
# Initialize data lists to contain blue and red data
Mbins_list = []
zbins_list = []
dist_M_list = []
err_M_list = []
dist_z_list = []
err_z_list = []
archive_files = ['lumfunc_red.npz', 'lumfunc_blue.npz']
# calculate luminosity function for blue and red galaxies
for i in xrange(2):
m = data_samples[i]['petroMag_r']
z = data_samples[i]['z']
M = m - mu_z(z)
zbins, dist_z, err_z, Mbins, dist_M, err_M = \
compute_luminosity_function(z, m, M, m_max, archive_files[i],
Nbootstraps=20, bin_type='freedman', z_mu=z_mu)
Mbins_list.append(Mbins)
zbins_list.append(zbins)
dist_M_list.append(dist_M)
err_M_list.append(err_M)
dist_z_list.append(dist_z)
err_z_list.append(err_z)
# Plot the volume number density and the luminosity function
titles = (r'u-r > 2.22', r'u-r < 2.22')
markers = ['o', '^']
plot_num_density(zbins_list, dist_z_list, err_z_list, titles=titles,
markers=markers)
plot_luminosity_function(Mbins_list, dist_M_list, err_M_list,
titles=titles, markers=markers)