def test_tduty_tdelay_dt_grid(run):
'''
'''
file_name = ''.join([UT.fig_dir(), 'tduty_tdelay_dt_grid.', run, '.p'])
grid = pickle.load(open(file_name, 'rb'))
fig = plt.figure(figsize=(10,10))
sub = fig.add_subplot(111, projection='3d')
#sub.view_init(45, 60)
tduty = grid[0,:]
tdelay = grid[1,:]
dt_abias = grid[2,:]
sigMstar = grid[3,:]
l2 = grid[4,:]
print 'L2 = ', l2.min(), l2.max()
scat = sub.scatter(dt_abias, tdelay, sigMstar, c=tduty)#, facecolors=cm.viridis(tduty))
sub.set_xlabel('$\Delta t_{abias}$ Gyr', fontsize=15)
sub.set_ylabel('$t_{delay}$ Gyr', fontsize=15)
sub.set_zlabel('$\sigma_{log\,M_*}$ Gyr', fontsize=15)
fig.colorbar(scat, shrink=0.5, aspect=10, cmap='hot', label='$t_{duty}$ Gyr')
for angle in range(0, 360):
sub.view_init(10, angle)
fig_name = ''.join([UT.fig_dir(), 'tduty_tdelay_dt_grid.', run, '.', str(angle), '.png'])
fig.savefig(fig_name)
return None
def lnL_pca_kde(mock, ell=0, rebin=None, krange=None):
''' ***TESTED: expectedly, more discrepant for low number of
mock catalogs. For Nseries monopole with 1000 mocks, no
significant discrepancy in the likelihood distribution
***
Test whether or not the Gaussian KDE approximation of pdfs
is sufficiently accurate by comparing the likelihood estimated
from NG.lnL_pca vs NG.lnL_pca_gauss. If they are highly
discrepant, then KDE estimate of the pdfs are not very accurate.
'''
Pk = NG.dataX(mock, ell=ell, rebin=rebin, krange=krange)
pca_gauss = NG.lnL_pca_gauss(Pk, Pk)
pca_kde = NG.lnL_pca(Pk, Pk)
prettyplot()
fig = plt.figure()
sub = fig.add_subplot(111)
nbin = 32
sub.hist(pca_gauss, bins=nbin, range=[-2.2*Pk.shape[1], -0.5*Pk.shape[1]],
normed=True, alpha=0.75, label='Gaussian $\mathcal{L^\mathtt{pseudo}}$')
sub.hist(pca_kde, bins=nbin, range=[-2.2*Pk.shape[1], -0.8*Pk.shape[1]],
normed=True, alpha=0.75, label='$\mathcal{L^\mathtt{pseudo}}$ KDE estimate')
sub.set_xlabel('log $\mathcal{L}$', fontsize=25)
sub.set_xlim([-2.2*Pk.shape[1], -0.5*Pk.shape[1]])
sub.legend(loc='upper left', prop={'size': 20})
if rebin is None: # save fig
f = ''.join([UT.fig_dir(), 'tests/test.lnL_kde_test.', mock, '.ell', str(ell), '.png'])
else:
f = ''.join([UT.fig_dir(), 'tests/test.lnL_kde_test.', mock, '.ell', str(ell), '.rebin', str(rebin), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def lnL_sys(mock, ell=0, rebin=None, sys='fc'):
''' Compare the pseudo gaussian L with no systematics, ICA L estimation with
no systematics, and ICA L estimation with fiber collisions.
'''
# Likelihood without systematics
Pk_nosys = NG.dataX(mock, ell=ell, rebin=rebin, sys=None)
gauss = NG.lnL_pca_gauss(Pk_nosys, Pk_nosys)
ica_nosys = NG.lnL_ica(Pk_nosys, Pk_nosys)
# Likelihood with specified systematics
Pk_sys = NG.dataX(mock, ell=ell, rebin=rebin, sys=sys)
ica_sys = NG.lnL_ica(Pk_sys, Pk_sys)
prettyplot()
fig = plt.figure()
sub = fig.add_subplot(111)
nbin = 32
sub.hist(gauss, bins=nbin, range=[-2.2*Pk_nosys.shape[0], -0.8*Pk_nosys.shape[0]],
normed=True, alpha=0.75, label='Gaussian $\mathcal{L^\mathtt{pseudo}}$; no sys.')
sub.hist(ica_nosys, bins=nbin, range=[-2.2*Pk_nosys.shape[0], -0.8*Pk_nosys.shape[0]],
normed=True, alpha=0.75, label='ICA; no sys.')
sub.hist(ica_sys, bins=nbin, range=[-2.2*Pk_nosys.shape[0], -0.8*Pk_nosys.shape[0]],
normed=True, alpha=0.75, label='ICA; w/ sys.')
sub.set_xlabel('log $\mathcal{L}$', fontsize=25)
sub.set_xlim([-2.2*Pk_nosys.shape[0], -0.5*Pk_nosys.shape[0]])
sub.legend(loc='upper left', prop={'size': 20})
if rebin is None: # save fig
f = ''.join([UT.fig_dir(), 'tests/test.lnL_sys.', mock, '.ell', str(ell), '.png'])
else:
f = ''.join([UT.fig_dir(), 'tests/test.lnL_sys.', mock, '.ell', str(ell), '.rebin', str(rebin), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def whiten(mock, ell=0, rebin=None, krange=None, method='choletsky'):
''' ***TESTED: Choletsky decomposition fails for full binned Nseries
P(k) because the precision matrix estimate is not positive definite***
test the data whitening.
'''
Pk = NG.dataX(mock, ell=ell, rebin=rebin, krange=krange)
X, _ = NG.meansub(Pk)
X_w, W = NG.whiten(X, method=method) # whitened data
prettyplot()
fig = plt.figure(figsize=(15,7))
sub = fig.add_subplot(121)
for i in range(X.shape[1]):
sub.plot(range(X_w.shape[0]), X_w[:,i])
sub.set_xlim([0, X.shape[0]])
sub.set_xlabel('$\mathtt{k}$ bins', fontsize=25)
sub.set_ylim([-7., 7.])
sub.set_ylabel('$\mathtt{W^{T} (P^i_'+str(ell)+'- \overline{P_'+str(ell)+'})}$', fontsize=25)
C_Xw = np.cov(X_w.T)
sub = fig.add_subplot(122)
im = sub.imshow(C_Xw, interpolation='none')
fig.colorbar(im, ax=sub)
if rebin is None:
f = ''.join([UT.fig_dir(), 'tests/test.whiten.', method, '.', mock, '.ell', str(ell), '.png'])
else:
f = ''.join([UT.fig_dir(), 'tests/test.whiten.', method, '.', mock, '.ell', str(ell), '.rebin', str(rebin), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def ica(mock, ell=0, rebin=None):
''' *** TESTED ***
Test that the ICA works!
'''
Pk = NG.dataX(mock, ell=ell, rebin=rebin)
X, _ = NG.meansub(Pk)
X_w, W = NG.whiten(X) # whitened data
X_ica, W = NG.Ica(X_w)
# compare covariance?
C_X = np.cov(X.T)
C_Xica = np.cov(X_ica.T)
prettyplot()
fig = plt.figure(figsize=(20, 8))
sub = fig.add_subplot(121)
im = sub.imshow(np.log10(C_X), interpolation='none')
sub.set_title('log(Cov.) of Data')
fig.colorbar(im, ax=sub)
sub = fig.add_subplot(122)
im = sub.imshow(C_Xica, interpolation='none')
fig.colorbar(im, ax=sub)
sub.set_title('Cov. of ICA transformed Data')
# save fig
if rebin is None:
f = ''.join([UT.fig_dir(), 'tests/test.ICAcov.', mock, '.ell', str(ell), '.png'])
else:
f = ''.join([UT.fig_dir(), 'tests/test.ICAcov.', mock, '.ell', str(ell), '.rebin', str(rebin), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def test_SumData():
''' Make sure abcee.SumData returns something sensible with some hardcoded values
Takes roughly 0.7 seconds
Notes
-----
* SumData returns sensible SMFs when compared to Li & White (2009) scaled by
1-f_sat from Wetzel et al.(2013).
'''
t0 = time.time()
output = abcee.SumData(['smf'], info=True, nsnap0=15, sigma_smhm=0.2)
m_arr, phi_arr, err_arr = Obvs.MF_data()
fcen = np.array([1. - Obvs.f_sat(mm, 0.05) for mm in m_arr])
fig = plt.figure()
sub = fig.add_subplot(111)
sub.plot(output[0][0], output[0][1])
sub.errorbar(m_arr, fcen * phi_arr, yerr=err_arr*np.sqrt(1./fcen))
sub.set_xlim([9., 12.])
sub.set_xlabel('$log\;M_*$', fontsize=25)
sub.set_ylim([1e-6, 10**-1.75])
sub.set_yscale('log')
sub.set_ylabel('$\Phi$', fontsize=25)
fig.savefig(''.join([UT.fig_dir(), 'tests/test.SumData.smf.png']), bbox_inches='tight')
return None
def lnL_pca_gauss(mock, ell=0, krange=None, NorS='ngc'):
''' ***TESTED***
Test that lnL_pca_gauss is consistent with chi-squared calculated directly
from the mocks with the covariance matrix.
'''
# Calculate the lnL_pca_gauss
Pk = NG.X_pk(mock, ell=ell, krange=krange, NorS=NorS)
dpk , mu_pk = NG.meansub(Pk)
pca_gauss = NG.lnL_pca_gauss(dpk, Pk)
# calculate chi-squared
C_X = np.cov(Pk.T)
chi2 = np.zeros(Pk.shape[0])
for i in range(Pk.shape[0]):
chi2[i] = -0.5 * np.sum(np.dot(dpk[i,:], np.linalg.solve(C_X, dpk[i,:].T)))
offset = chi2 - pca_gauss
print offset
fig = plt.figure()
sub = fig.add_subplot(111)
nbin = 32
sub.hist(pca_gauss, bins=nbin, normed=True, alpha=0.75, label='PCA Gaussian')
sub.hist(chi2, bins=nbin, normed=True, alpha=0.75, label=r'$\chi^2$')
sub.set_xlabel('log $\mathcal{L}$', fontsize=25)
sub.set_xlim([-2.2*Pk.shape[1], 0.])
sub.legend(loc='upper left', prop={'size': 20})
f = ''.join([UT.fig_dir(), 'tests/test.lnL_pca_gauss_test.', mock, '.ell', str(ell), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def gmf(Mr=20, Nmock=500):
'''
Plot Group Multiplicty Function of fake observations
'''
prettyplot()
pretty_colors = prettycolors()
# import fake obs GMF
gmf, sig_gmf = Data.data_gmf(Mr=Mr, Nmock=Nmock)
# group richness bins
gmf_bin = Data.data_gmf_bins()
fig = plt.figure(1)
sub = fig.add_subplot(111)
sub.errorbar(
0.5*(gmf_bin[:-1]+gmf_bin[1:]), gmf, yerr=sig_gmf,
fmt="ok", capsize=1.0
)
sub.set_xlim([1, 60])
sub.set_yscale('log')
sub.set_ylabel(r"Group Multiplicity Function (h$^{3}$ Mpc$^{-3}$)", fontsize=20)
sub.set_xlabel(r"$\mathtt{Group\;\;Richness}$", fontsize=20)
# save to file
fig_file = ''.join([util.fig_dir(),
'gmf.Mr', str(Mr), '.Nmock', str(Nmock), '.png'])
fig.savefig(fig_file, bbox_inches='tight')
return None
def test_SFMS_highz(run, T, nsnap=15, lit='lee', nsnap0=15, downsampled='14'):
''' Compare the best-fit SFMS parameters from ABC to literature at high z
'''
if lit == 'lee':
lit_label = 'Lee et al. (2015)'
# median ABC theta
abcout = abcee.readABC(run, T)
theta_med = [UT.median(abcout['theta'][:, i], weights=abcout['w'][:]) for i in range(len(abcout['theta'][0]))]
subcat_sim = abcee.model(run, theta_med, nsnap0=nsnap0, downsampled=downsampled)
m_arr = np.arange(8., 12.1, 0.1)
sfr_abc = Obvs.SSFR_SFMS(m_arr, UT.z_nsnap(nsnap), theta_SFMS=subcat_sim['theta_sfms']) + m_arr
sfr_obv = Obvs.SSFR_SFMS_obvs(m_arr, UT.z_nsnap(nsnap), lit=lit) + m_arr
fig = plt.figure()
sub = fig.add_subplot(111)
sub.fill_between(m_arr, sfr_abc-0.3, sfr_abc+0.3, color='b', alpha=0.25, linewidth=0, edgecolor=None,
label=r'ABC $\theta_{median}$')
sub.plot(m_arr, sfr_obv+0.3, ls='--', c='k', label=lit_label)
sub.plot(m_arr, sfr_obv-0.3, ls='--', c='k')
sub.set_xlim([8., 12.])
sub.set_xlabel('$\mathtt{log\;M_*}$', fontsize=25)
sub.set_ylim([-4., 3.])
sub.set_ylabel('$\mathtt{log\;SFR}$', fontsize=25)
sub.legend(loc='best')
fig.savefig(UT.fig_dir()+'SFMS.z'+str(round(UT.z_nsnap(nsnap),2))+'.'+run+'.'+lit+'.png', bbox_inches='tight')
plt.close()
return None
def SFMS_z1():
''' Comparison of different SFMS observations at z~1.
'''
# Lee et al. (2015)
logSFR_lee = lambda mm: 1.53 - np.log10(1 + (10**mm/10**(10.10))**-1.26)
# Noeske et al. (2007) 0.85 < z< 1.10 (by eye)
logSFR_noeske = lambda mm: (1.580 - 1.064)/(11.229 - 10.029)*(mm - 10.0285) + 1.0637
# Moustakas et al. (2013) 0.8 < z < 1. (by eye)
logSFR_primus = lambda mm: (1.3320 - 1.296)/(10.49 - 9.555) * (mm-9.555) + 1.297
# prior range
logSFR_prior_min = lambda mm: (0.6 * (mm - 10.5) + 0.9 * (1.-0.0502) - 0.11)
logSFR_prior_max = lambda mm: (0.6 * (mm - 10.5) + 2. * (1.-0.0502) - 0.11)
fig = plt.figure(1)
sub = fig.add_subplot(111)
marr = np.linspace(9., 12., 100)
sub.plot(marr, logSFR_lee(marr), label='Lee et al.(2015)')
sub.plot(marr, logSFR_noeske(marr), label='Noeske et al.(2007)')
sub.plot(marr, logSFR_primus(marr), label='Moustakas et al.(2013)')
sub.fill_between(marr, logSFR_prior_min(marr), logSFR_prior_max(marr), label='Prior', alpha=0.5)
sub.legend(loc='lower right')
sub.set_xlim([9.5, 11.5])
fig.savefig(UT.fig_dir()+'test.z_1.sfms.png')
plt.close()
return None
def plotCMS_SMF_MS(cenque='default', evol_dict=None):
'''Plot the stellar mass function of the integrated SFR main sequence population
and compare it to the theoretic stellar mass function of the main sequence
population
'''
# import evolved galaxy population
MSpop = CMS.EvolvedGalPop(cenque=cenque, evol_dict=evol_dict)
print MSpop.File()
MSpop.Read()
MSonly = np.where(MSpop.t_quench == 999.) # remove the quenching galaxies
smf = Obvs.getSMF(MSpop.mass[MSonly], weights=MSpop.weight_down[MSonly])
prettyplot()
pretty_colors = prettycolors()
# analytic SMF
theory_smf = Obvs.getSMF(MSpop.M_sham[MSonly], weights=MSpop.weight_down[MSonly])
fig = plt.figure()
sub = fig.add_subplot(111)
sub.plot(theory_smf[0], theory_smf[1], lw=3, c='k', label='Theory')
sub.plot(smf[0], smf[1], lw=3, c=pretty_colors[3], ls='--', label='MS Only')
sub.set_ylim([10**-5, 10**-1])
sub.set_xlim([6., 12.0])
sub.set_yscale('log')
sub.set_xlabel(r'Mass $\mathtt{M_*}$', fontsize=25)
sub.set_ylabel(r'Stellar Mass Function $\mathtt{\Phi}$', fontsize=25)
sub.legend(loc='upper right', frameon=False)
fig_file = ''.join([UT.fig_dir(), 'SMF_MS.CMS', MSpop._Spec_str(), '.png'])
fig.savefig(fig_file, bbox_inches='tight')
plt.close()
return None
def test_Svd():
'''
'''
k_arr, pk_lhd = Dat.X_lhd(0.0,
1,
4,
1,
obvs='pk',
ell=0,
Nmesh=360,
rsd=True,
krange=[0.01, 0.5],
karr=True,
HODrange='sinha2017prior_narrow',
samples=40,
method='mdu',
silent=True)
svd = Comp.Svd()
_ = svd.fit(pk_lhd)
pcs = svd.transform(pk_lhd)
assert np.allclose(svd.inv_transform(pcs), pk_lhd)
fig = plt.figure()
sub = fig.add_subplot(111)
sub.plot(range(len(svd.exp_var_ratio) + 1),
np.concatenate([np.array([0]),
np.cumsum(svd.exp_var_ratio)]))
sub.set_ylim([5e-1, 1.])
sub.set_yscale('log')
f = ''.join([UT.fig_dir(), 'tests/plk_svd.exp_var.png'])
fig.savefig(f, bbox_inches='tight')
return None
def test_PC(n_comp):
'''
'''
k_arr, pk_lhd = Dat.X_lhd(0.0,
1,
4,
1,
obvs='pk',
ell=0,
Nmesh=360,
rsd=True,
krange=[0.01, 0.5],
karr=True,
HODrange='sinha2017prior_narrow',
samples=40,
method='mdu',
silent=True)
svd = Comp.Svd(n_comp=n_comp)
_ = svd.fit(pk_lhd)
pcs = svd.transform(pk_lhd)
fig = plt.figure(figsize=(4 * n_comp, 4))
for i in range(n_comp):
sub = fig.add_subplot(1, n_comp, i + 1)
sub.hist(pcs[:, i], normed=True)
sub.set_xlabel('PC ' + str(i + 1), fontsize=20)
f = ''.join([UT.fig_dir(), 'tests/plk_svd.pc.', str(n_comp), 'comp.png'])
fig.savefig(f, bbox_inches='tight')
return None
def xi(Mr=20, Nmock=500):
'''
Plot xi(r) of the fake observations
'''
prettyplot()
pretty_colors = prettycolors()
xir, cii = Data.data_xi(Mr=Mr, Nmock=Nmock)
rbin = Data.data_xi_bins(Mr=Mr)
fig = plt.figure(1)
sub = fig.add_subplot(111)
sub.plot(rbin, rbin*xir, c='k', lw=1)
sub.errorbar(rbin, rbin*xir, yerr = rbin*cii**0.5 , fmt="ok", ms=1, capsize=2, alpha=1.)
sub.set_xlim([0.1, 15])
sub.set_ylim([1, 10])
sub.set_yscale("log")
sub.set_xscale("log")
sub.set_xlabel(r'$\mathtt{r}\; (\mathtt{Mpc})$', fontsize=25)
sub.set_ylabel(r'$\mathtt{r} \xi_{\rm gg}$', fontsize=25)
fig_file = ''.join([util.fig_dir(),
'xi.Mr', str(Mr), '.Nmock', str(Nmock), '.png'])
fig.savefig(fig_file, bbox_inches='tight')
plt.close()
return None
def lnL(mock, ell=0, rebin=None, krange=None):
''' Test the ICA likelihood estimation and pseudo gaussian likelihood
'''
Pk = NG.dataX(mock, ell=ell, rebin=rebin, krange=krange)
ica = NG.lnL_ica(Pk, Pk)
gauss = NG.lnL_pca_gauss(Pk, Pk)
prettyplot()
fig = plt.figure()
sub = fig.add_subplot(111)
nbin = 32
sub.hist(gauss, bins=nbin, range=[-2.2*Pk.shape[1], -0.8*Pk.shape[1]],
normed=True, alpha=0.75, label='Gaussian $\mathcal{L^\mathtt{pseudo}}$')
sub.hist(ica, bins=nbin, range=[-2.2*Pk.shape[1], -0.8*Pk.shape[1]],
normed=True, alpha=0.75, label='ICA')
sub.set_xlabel('log $\mathcal{L}$', fontsize=25)
sub.set_xlim([-2.2*Pk.shape[1], -0.5*Pk.shape[1]])
sub.legend(loc='upper left', prop={'size': 20})
str_rebin = ''
if rebin is not None:
str_rebin = '.rebin'+str(rebin)
str_krange = ''
if krange is not None:
str_krange = '.kmin'+str(krange[0])+'.kmax'+str(krange[1])
f = ''.join([UT.fig_dir(), 'tests/test.lnL.', mock, '.ell', str(ell),
str_rebin, str_krange, '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def Plk_fiducial(nreal):
''' ***TESTED***
Test that the methods Data.X_fid and Data.Fiducial_Obvs
are returning sensible results for obvs = 'plk'
'''
karr, Xfid = Data.X_fid(nreal, 4, obvs='plk', karr=True)
muX = np.sum(Xfid, axis=0)/float(Xfid.shape[0])
sigX = np.sqrt(np.diag(np.cov(Xfid.T)))
fig = plt.figure()
sub = fig.add_subplot(111)
sub.errorbar(karr, muX, yerr=sigX, fmt='.k')
for i in range(Xfid.shape[0]):
sub.plot(karr, Xfid[i,:], c='k', lw=0.1)
sub.set_xlim([1e-2, 0.5])
sub.set_xlabel('$k$', fontsize=20)
sub.set_xscale('log')
sub.set_ylim([2e3, 2e5])
sub.set_ylabel('$P(k)$', fontsize=20)
sub.set_yscale('log')
f = ''.join([UT.fig_dir(), 'tests/test_data.Plk_fiducial.png'])
fig.savefig(f, bbox_inches='tight')
return None
def TrueObservables(Mr=21, b_normal=0.25):
''' Plot xi and gmf for the data
'''
# xi data
xi_data = Data.data_xi(Mr=Mr, b_normal=b_normal)
cov_data = Data.data_cov(Mr=Mr, b_normal=b_normal, inference='mcmc')
data_xi_cov = cov_data[1:16, 1:16]
xi_r_bin = Data.data_xi_bin(Mr=Mr)
# gmf data
data_gmf = Data.data_gmf(Mr=Mr, b_normal=b_normal)
cov_data = Data.data_cov(Mr=Mr, b_normal=b_normal, inference='mcmc')
data_gmf_cov = cov_data[16:, 16:]
r_binedge = Data.data_gmf_bins()
gmf_r_bin = 0.5 * (r_binedge[:-1] + r_binedge[1:])
prettyplot()
pretty_colors = prettycolors()
fig = plt.figure(figsize=(12,6))
sub_xi = fig.add_subplot(121)
sub_gmf = fig.add_subplot(122)
#sub_xi.errorbar(xi_r_bin, xi_data, yerr = np.sqrt(np.diag(data_xi_cov)), fmt="o", color='k',
# markersize=0, lw=0, capsize=3, elinewidth=1.5)
#sub_xi.scatter(xi_r_bin, xi_data, c='k', s=10, lw=0)
sub_xi.fill_between(xi_r_bin,
xi_data-np.sqrt(np.diag(data_xi_cov)), xi_data+np.sqrt(np.diag(data_xi_cov)),
color=pretty_colors[1])
sub_xi.set_yscale('log')
sub_xi.set_xscale('log')
sub_xi.set_xlim(0.1, 20)
sub_xi.set_xlabel(r'$\mathtt{r}\; [\mathtt{Mpc}/h]$', fontsize=25)
sub_xi.set_ylabel(r'$\xi(r)$', fontsize=25)
#sub_gmf.errorbar(gmf_r_bin, data_gmf, yerr=np.sqrt(np.diag(data_gmf_cov)),
# fmt="o", color='k',
# markersize=0, lw=0, capsize=4, elinewidth=2)
#sub_gmf.scatter(gmf_r_bin, data_gmf, s=15, lw=0, c='k', label='Mock Observation')
sub_gmf.fill_between(gmf_r_bin,
data_gmf-np.sqrt(np.diag(data_gmf_cov)), data_gmf+np.sqrt(np.diag(data_gmf_cov)),
color=pretty_colors[1])
sub_gmf.set_xlim(1, 20)
sub_gmf.set_xlabel(r'$\mathtt{N}$ (Group Richness)', fontsize=25)
sub_gmf.yaxis.tick_right()
sub_gmf.yaxis.set_ticks_position('both')
sub_gmf.yaxis.set_label_position('right')
sub_gmf.set_ylim([10**-7, 2.0*10**-4])
sub_gmf.set_yscale('log')
sub_gmf.set_ylabel(r'$\zeta(\mathtt{N})$', fontsize=25)
fig.subplots_adjust(hspace=0.05)
fig_name = ''.join([ut.fig_dir(),
'paper.data_observables',
'.pdf'])
fig.savefig(fig_name, bbox_inches='tight', dpi=150)
plt.close()
return None
def TrueObservables(Mr=21, b_normal=0.25):
''' Plot xi and gmf for the data
'''
# xi data
xi_data = Data.data_xi(Mr=Mr, b_normal=b_normal)
cov_data = Data.data_cov(Mr=Mr, b_normal=b_normal, inference='mcmc')
data_xi_cov = cov_data[1:16, 1:16]
xi_r_bin = Data.data_xi_bin(Mr=Mr)
# gmf data
data_gmf = Data.data_gmf(Mr=Mr, b_normal=b_normal)
cov_data = Data.data_cov(Mr=Mr, b_normal=b_normal, inference='mcmc')
data_gmf_cov = cov_data[16:, 16:]
r_binedge = Data.data_gmf_bins()
gmf_r_bin = 0.5 * (r_binedge[:-1] + r_binedge[1:])
prettyplot()
pretty_colors = prettycolors()
fig = plt.figure(figsize=(12, 6))
sub_xi = fig.add_subplot(121)
sub_gmf = fig.add_subplot(122)
#sub_xi.errorbar(xi_r_bin, xi_data, yerr = np.sqrt(np.diag(data_xi_cov)), fmt="o", color='k',
# markersize=0, lw=0, capsize=3, elinewidth=1.5)
#sub_xi.scatter(xi_r_bin, xi_data, c='k', s=10, lw=0)
sub_xi.fill_between(xi_r_bin,
xi_data - np.sqrt(np.diag(data_xi_cov)),
xi_data + np.sqrt(np.diag(data_xi_cov)),
color=pretty_colors[1])
sub_xi.set_yscale('log')
sub_xi.set_xscale('log')
sub_xi.set_xlim(0.1, 20)
sub_xi.set_xlabel(r'$\mathtt{r}\; [\mathtt{Mpc}/h]$', fontsize=25)
sub_xi.set_ylabel(r'$\xi(r)$', fontsize=25)
#sub_gmf.errorbar(gmf_r_bin, data_gmf, yerr=np.sqrt(np.diag(data_gmf_cov)),
# fmt="o", color='k',
# markersize=0, lw=0, capsize=4, elinewidth=2)
#sub_gmf.scatter(gmf_r_bin, data_gmf, s=15, lw=0, c='k', label='Mock Observation')
sub_gmf.fill_between(gmf_r_bin,
data_gmf - np.sqrt(np.diag(data_gmf_cov)),
data_gmf + np.sqrt(np.diag(data_gmf_cov)),
color=pretty_colors[1])
sub_gmf.set_xlim(1, 20)
sub_gmf.set_xlabel(r'$\mathtt{N}$ (Group Richness)', fontsize=25)
sub_gmf.yaxis.tick_right()
sub_gmf.yaxis.set_ticks_position('both')
sub_gmf.yaxis.set_label_position('right')
sub_gmf.set_ylim([10**-7, 2.0 * 10**-4])
sub_gmf.set_yscale('log')
sub_gmf.set_ylabel(r'$\zeta(\mathtt{N})$', fontsize=25)
fig.subplots_adjust(hspace=0.05)
fig_name = ''.join([ut.fig_dir(), 'paper.data_observables', '.pdf'])
fig.savefig(fig_name, bbox_inches='tight', dpi=150)
plt.close()
return None
def plotCMS_SFH_AsBias(cenque='default', evol_dict=None):
''' Plot the star formation history of star forming main sequence galaxies
and their corresponding halo mass acretion history
'''
# import evolved galaxy population
egp = CMS.EvolvedGalPop(cenque=cenque, evol_dict=evol_dict)
egp.Read()
MSonly = np.where(egp.t_quench == 999.) # only keep main-sequence galaxies
MSonly_rand = np.random.choice(MSonly[0], size=100)
z_acc, t_acc = UT.zt_table()
t_cosmic = t_acc[1:16]
dt = 0.1
t_arr = np.arange(t_cosmic.min(), t_cosmic.max()+dt, dt)
prettyplot()
pretty_colors = prettycolors()
fig = plt.figure()
sub1 = fig.add_subplot(211)
sub2 = fig.add_subplot(212)
dsfrt = np.zeros((len(t_arr), len(MSonly_rand)))
for i_t, tt in enumerate(t_arr):
dsfr = SFR.dSFR_MS(tt, egp.sfh_dict)
for ii, i_gal in enumerate(MSonly_rand):
dsfrt[i_t,ii] = dsfr[i_gal]
i_col = 0
for ii, i_gal in enumerate(MSonly_rand):
tlim = np.where(t_arr > egp.tsnap_genesis[i_gal])
tcoslim = np.where(t_cosmic >= egp.tsnap_genesis[i_gal])
halo_hist = (egp.Mhalo_hist[i_gal])[tcoslim][::-1]
halo_exists = np.where(halo_hist != -999.)
if np.power(10, halo_hist[halo_exists]-egp.halo_mass[i_gal]).min() < 0.7:
if i_col < 5:
sub1.plot(t_arr[tlim], dsfrt[:,ii][tlim], lw=3, c=pretty_colors[i_col])
sub2.plot(t_cosmic[tcoslim][::-1][halo_exists],
np.power(10, halo_hist[halo_exists]-egp.halo_mass[i_gal]),
lw=3, c=pretty_colors[i_col])
i_col += 1
sub1.set_xlim([5.0, 13.5])
sub1.set_xticklabels([])
sub1.set_ylim([-1., 1.])
sub1.set_ylabel(r'$\mathtt{SFR(t) - <SFR_{MS}(t)>}$', fontsize=25)
sub2.set_xlim([5.0, 13.5])
sub2.set_xlabel(r'$\mathtt{t_{cosmic}}$', fontsize=25)
sub2.set_ylim([0.0, 1.1])
sub2.set_ylabel(r'$\mathtt{M_h(t)}/\mathtt{M_h(t_f)}$', fontsize=25)
fig_file = ''.join([UT.fig_dir(), 'SFH_AsBias.CMS', egp._Spec_str(), '.png'])
fig.savefig(fig_file, bbox_inches='tight')
plt.close()
return None
def plotCMS_SFMS(cenque='default', evol_dict=None):
''' Plot the SFR-M* relation of the SFMS and compare to expectations
'''
# import evolved galaxy population
cms = CMS.CentralMS(cenque=cenque)
cms._Read_CenQue()
MSpop = CMS.EvolvedGalPop(cenque=cenque, evol_dict=evol_dict)
MSpop.Read()
MSonly = np.where(MSpop.t_quench == 999.)
delta_m = 0.5
mbins = np.arange(MSpop.mass[MSonly].min(), MSpop.mass[MSonly].max(), delta_m)
sfr_percent = np.zeros((len(mbins)-1, 4))
for im, mbin in enumerate(mbins[:-1]):
within = np.where(
(MSpop.mass[MSonly] > mbin) &
(MSpop.mass[MSonly] <= mbin + delta_m)
)
#sfr_percent[im,:] = np.percentile(MSpop.sfr[MSonly[0][within]], [2.5, 16, 84, 97.5])
sfr_percent[im,:] = UT.weighted_quantile(MSpop.sfr[MSonly[0][within]],
[0.025, 0.16, 0.84, 0.975], weights=MSpop.weight_down[MSonly[0][within]])
prettyplot()
pretty_colors = prettycolors()
fig = plt.figure()
sub = fig.add_subplot(111)
# parameterized observed SFMS
obvs_mu_sfr = SFR.AverageLogSFR_sfms(0.5*(mbins[:-1]+mbins[1:]), 0.05, sfms_dict=cms.sfms_dict)
obvs_sig_sfr = SFR.ScatterLogSFR_sfms(0.5*(mbins[:-1]+mbins[1:]), 0.05, sfms_dict=cms.sfms_dict)
sub.fill_between(0.5*(mbins[:-1] + mbins[1:]),
obvs_mu_sfr - 2.*obvs_sig_sfr, obvs_mu_sfr + 2.*obvs_sig_sfr,
color=pretty_colors[1], alpha=0.3, edgecolor=None, lw=0)
sub.fill_between(0.5*(mbins[:-1] + mbins[1:]),
obvs_mu_sfr - obvs_sig_sfr, obvs_mu_sfr + obvs_sig_sfr,
color=pretty_colors[1], alpha=0.5, edgecolor=None, lw=0, label='Observations')
# simulation
sub.fill_between(0.5*(mbins[:-1] + mbins[1:]), sfr_percent[:,0], sfr_percent[:,-1],
color=pretty_colors[3], alpha=0.3, edgecolor=None, lw=0)
sub.fill_between(0.5*(mbins[:-1] + mbins[1:]), sfr_percent[:,1], sfr_percent[:,-2],
color=pretty_colors[3], alpha=0.5, edgecolor=None, lw=0, label='Simulated')
# x,y axis
sub.set_xlim([9.0, 12.0])
sub.set_xlabel(r'$\mathtt{M_*}\; [\mathtt{M}_\odot]$', fontsize=25)
sub.set_ylim([-2., 1.])
sub.set_ylabel(r'$\mathtt{SFR}\; [\mathtt{M}_\odot/\mathtt{yr}]$', fontsize=25)
sub.legend(loc='lower right', numpoints=1, markerscale=1.)
fig_file = ''.join([UT.fig_dir(), 'SFMS.CMS', MSpop._Spec_str(), '.png'])
plt.savefig(fig_file, bbox_inches='tight')
plt.close()
return None
def whiten_recon(mock, ell=0, rebin=None, krange=None, method='choletsky'):
''' ***TESTED: The whitening matrices reconstruct the P(k)s***
Test whether P(k) can be reconstructed using the whitening matrix
'''
Pk, k = NG.dataX(mock, ell=ell, rebin=rebin, krange=krange, k_arr=True)
X, mu_X = NG.meansub(Pk)
X_w, W = NG.whiten(X, method=method) # whitened data
prettyplot()
fig = plt.figure(figsize=(15,7))
sub = fig.add_subplot(121)
for i in range(X.shape[0]):
sub.plot(k, Pk[i,:])
if krange is None:
sub.set_xlim([1e-3, 0.5])
else:
sub.set_xlim(krange)
sub.set_xscale('log')
sub.set_xlabel('$\mathtt{k}$', fontsize=25)
sub.set_yscale('log')
sub.set_ylim([2e3, 2.5e5])
np.random.seed(7)
sub = fig.add_subplot(122)
for i in range(X.shape[0]):
X_noise = np.random.normal(size=X_w.shape[1])
X_rec = np.linalg.solve(W.T, X_noise.T)
sub.plot(k, X_rec.T + mu_X)
if krange is None:
sub.set_xlim([1e-3, 0.5])
else:
sub.set_xlim(krange)
sub.set_xscale('log')
sub.set_xlabel('$\mathtt{k}$', fontsize=25)
sub.set_yscale('log')
sub.set_ylim([2e3, 2.5e5])
if rebin is None:
f = ''.join([UT.fig_dir(), 'tests/test.whiten_recon.', method, '.', mock, '.ell', str(ell), '.png'])
else:
f = ''.join([UT.fig_dir(), 'tests/test.whiten_recon.', method, '.', mock, '.ell', str(ell), '.rebin', str(rebin), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def Plk_halo(mneut=0.0, nzbin=4, zspace=False):
''' **TESTED --- Nov 7, 2017 **
Test the Plk_halo
'''
p0ks, p2ks, p4ks = [], [], []
for ireal in range(1, 101):
# read all 100 realizations
plk_i = Obvs.Plk_halo(mneut, ireal, nzbin, zspace=zspace)
if ireal == 1: k = plk_i['k']
p0ks.append(plk_i['p0k'])
p2ks.append(plk_i['p2k'])
p4ks.append(plk_i['p4k'])
fig = plt.figure()
sub = fig.add_subplot(111)
for p0k, p2k, p4k in zip(p0ks, p2ks, p4ks):
sub.plot(k, k * p0k, c='k', lw=0.1)
sub.plot(k, k * p2k, c='b', lw=0.1)
sub.plot(k, k * p4k, c='r', lw=0.1)
# plot the average
sub.plot(k,
k * np.average(np.array(p0ks), axis=0),
c='k',
lw=2,
ls='--',
label='$\ell=0$')
sub.plot(k,
k * np.average(np.array(p2ks), axis=0),
c='b',
lw=2,
ls='--',
label='$\ell=2$')
sub.plot(k,
k * np.average(np.array(p4ks), axis=0),
c='r',
lw=2,
ls='--',
label='$\ell=4$')
sub.set_xlim([0.01, 0.15])
sub.set_xlabel('k', fontsize=20)
sub.set_ylim([-2000., 2500.])
sub.set_ylabel('$k P(k)$', fontsize=20)
sub.legend(loc='lower right', prop={'size': 15})
if zspace: str_space = 'z'
else: str_space = 'r'
fig.savefig(''.join([
UT.fig_dir(), 'tests/plk_halo.',
str(mneut), 'eV.nzbin',
str(nzbin), '.', str_space, 'space.png'
]),
bbox_inches='tight')
return None
def plotCMS_SMHMR_MS(cenque='default', evol_dict=None):
''' Plot the Stellar Mass to Halo Mass Relation of the evolved Gal population
'''
# import evolved galaxy population
egp = CMS.EvolvedGalPop(cenque=cenque, evol_dict=evol_dict)
egp.Read()
# Calculate the SMHMR for the EvolvedGalPop
m_halo_bin = np.arange(10., 15.25, 0.25) # M_halo bins
m_halo_mid = 0.5 * (m_halo_bin[:-1] + m_halo_bin[1:]) # M_halo bin mid
smhmr = np.zeros((len(m_halo_bin)-1, 5))
for im, m_mid in enumerate(m_halo_mid):
inbin = np.where(
(egp.halo_mass > m_halo_bin[im]) &
(egp.halo_mass <= m_halo_bin[im+1]))
smhmr[im,:] = np.percentile(egp.mass[inbin], [2.5, 16, 50, 84, 97.5])
#smhmr[im,:] = UT.weighted_quantile(egp.mass[inbin], [0.025, 0.16, 0.50, 0.84, 0.975],
# weights=egp.weight_down[inbin])
# "observations" with observed scatter of 0.2 dex
obvs_smhmr = np.zeros((len(m_halo_bin)-1, 2))
obvs_smhmr[:,0] = smhmr[:,2] - 0.2
obvs_smhmr[:,1] = smhmr[:,2] + 0.2
prettyplot()
pretty_colors = prettycolors()
fig = plt.figure()
sub = fig.add_subplot(111)
sub.fill_between(m_halo_mid, smhmr[:,0], smhmr[:,-1],
color=pretty_colors[1], alpha=0.25, edgecolor=None, lw=0)
sub.fill_between(m_halo_mid, smhmr[:,1], smhmr[:,-2],
color=pretty_colors[1], alpha=0.5, edgecolor=None, lw=0,
label=r'$\mathtt{Simulated}$')
sub.plot(m_halo_mid, obvs_smhmr[:,0],
color='k', ls='--', lw=3)
sub.plot(m_halo_mid, obvs_smhmr[:,1],
color='k', ls='--', lw=3,
label=r"$\sigma = \mathtt{0.2}$")
sub.set_xlim([10., 15.0])
sub.set_ylim([9., 12.0])
sub.set_xlabel(r'$\mathtt{log}\;\mathtt{M_{halo}}\;\;[\mathtt{M_\odot}]$', fontsize=25)
sub.set_ylabel(r'$\mathtt{log}\;\mathtt{M_{*}}\;\;[\mathtt{M_\odot}]$', fontsize=25)
sub.legend(loc='upper right')
fig_file = ''.join([UT.fig_dir(), 'SMHMR_MS.CMS', egp._Spec_str(), '.png'])
fig.savefig(fig_file, bbox_inches='tight')
plt.close()
return None
def SFMS_Mfid_z():
''' Comparison of different SFMS observations from z~0.05 to 1 at fiducial
mass log M = 10.5. We include the compilation fit from Speagle et al. (2014).
These comparisons are mainly intended to determine whether the prior for
the logSFR_MS parameters are sensible.
'''
# z = 0
# fit from SDSS group catalog
logSFR_sdss = lambda mm: (0.5757 * (mm - 10.5) - 0.13868)
# Lee et al. (2015)
logSFR_lee = lambda mm: 1.53 - np.log10(1 + (10**mm/10**(10.10))**-1.26)
# Noeske et al. (2007) 0.85 < z< 1.10 (by eye)
logSFR_noeske = lambda mm: (1.580 - 1.064)/(11.229 - 10.029)*(mm - 10.0285) + 1.0637
# Moustakas et al. (2013) 0.8 < z < 1. (by eye)
logSFR_primus = lambda mm: (1.3320 - 1.296)/(10.49 - 9.555) * (mm-9.555) + 1.297
# Speagle's SFMS bestfit
logSFR_speagle = lambda mm, tt: (0.84 - 0.026 * tt) * mm - (6.51 - 0.11 * tt)
logSFR_min = lambda mm, zz: logSFR_sdss(mm) + (zz - 0.05)
logSFR_max = lambda mm, zz: logSFR_sdss(mm) + 2. * (zz - 0.05)
zarr = np.linspace(0., 1., 5)
tarr = UT.t_from_z(zarr)
fig = plt.figure(1)
sub = fig.add_subplot(111)
sub.scatter([0.05], [logSFR_sdss(10.5)], label='SDSS group catalog')
sub.scatter([0.95], [logSFR_lee(10.5)], label='Lee et al.(2015)')
sub.scatter([1.00], [logSFR_noeske(10.5)], label='Noeske et al.(2007)')
sub.scatter([0.97], [logSFR_primus(10.5)], label='Moustakas et al.(2013)')
sub.plot(zarr, [logSFR_speagle(10.5, tt) for tt in tarr],
c='k', lw=2, label='Speagle')
sub.plot(zarr,
[logSFR_speagle(10.5, tt) - \
(logSFR_speagle(10.5, UT.t_from_z(0.05)) - logSFR_sdss(10.5))
for tt in tarr],
c='k', lw=2, ls='--', label='Speagle offset')
sub.fill_between(zarr,
[logSFR_min(10.5, zz) for zz in zarr],
[logSFR_max(10.5, zz) for zz in zarr], label='Prior', alpha=0.5)
sub.legend(loc='lower right')
sub.set_xlim([0., 1.])
sub.set_xlabel('Redshift ($z$)', fontsize=20)
sub.set_ylim([-2, 2])
sub.set_ylabel('log SFR ($M_{*,fid} = 10^{10.5} M_\odot$)', fontsize=20)
fig.savefig(UT.fig_dir()+'test.sfms.Mfid.z.png', bbox_inches='tight')
plt.close()
return None
def test_dMh_dMstar(run, theta, sigma_smhm=0.2, nsnap0=15, downsampled='14', flag=None):
'''
'''
subcat_sim = abcee.model(run, theta,
nsnap0=nsnap0, sigma_smhm=sigma_smhm, downsampled=downsampled)
isSF = np.where((subcat_sim['gclass'] == 'sf') &
(subcat_sim['weights'] > 0.) &
(subcat_sim['nsnap_start'] == nsnap0))[0] # only SF galaxies
assert subcat_sim['m.star'][isSF].min() > 0.
#print subcat_sim['m.star'][isSF].min(), subcat_sim['m.star'][isSF].max()
#print subcat_sim['m.star0'][isSF].min(), subcat_sim['m.star0'][isSF].max()
dMh = np.log10(10**subcat_sim['halo.m'][isSF] - 10**subcat_sim['halo.m0'][isSF])
dMstar = np.log10(10**subcat_sim['m.star'][isSF] - 10**subcat_sim['m.star0'][isSF])
fig = plt.figure(figsize=(26,6))
sub = fig.add_subplot(141)
scat = sub.scatter(subcat_sim['halo.m0'][isSF], subcat_sim['halo.m'][isSF],
lw=0, s=5, c='k', cmap='hot')
sub.set_xlim([9.75, 14.5])
sub.set_xlabel('$\mathtt{log\;M_h(z_0 \sim 1)}$', fontsize=20)
sub.set_ylim([9.75, 14.5])
sub.set_ylabel('$\mathtt{log\;M_h(z_f \sim 0)}$', fontsize=20)
sub = fig.add_subplot(142)
scat = sub.scatter(subcat_sim['m.star0'][isSF], subcat_sim['m.star'][isSF],
lw=0, s=5, c='k', cmap='hot')
sub.set_xlim([7., 12.])
sub.set_xlabel('$\mathtt{log\;M_*(z_0 \sim 1)}$', fontsize=20)
sub.set_ylim([7., 12.])
sub.set_ylabel('$\mathtt{log\;M_*(z_f \sim 0)}$', fontsize=20)
sub = fig.add_subplot(143)
scat = sub.scatter(dMh, dMstar, lw=0, s=5, c=subcat_sim['halo.m'][isSF], cmap='hot',
vmin=10., vmax=14.5)
fig.colorbar(scat, ax=sub)
sub.set_xlim([8., 14.5])
sub.set_xlabel('$\mathtt{log(\; \Delta M_h\;)}$', fontsize=20)
sub.set_ylim([9, 11.25])
sub.set_ylabel('$\mathtt{log(\; \Delta M_*\;)}$', fontsize=20)
plt.show()
raise ValueError
if flag is None:
flag_str = ''
else:
flag_str = '.'+flag
fig.savefig(''.join([UT.fig_dir(), 'dMh_dMstar.', run, '.sig_smhm', str(sigma_smhm),
flag_str, '.png']), bbox_inches='tight')
plt.close()
return None
def test_tdelay_dt_grid(run, tduty):
file_name = ''.join([UT.fig_dir(), 'tduty_tdelay_dt_grid.', run, '.p'])
grid = pickle.load(open(file_name, 'rb'))
tdutys = grid[0,:]
tdelay = grid[1,:]
dt_abias = grid[2,:]
sigMstar = grid[3,:]
l2 = grid[4,:]
duty = np.where(tdutys == tduty)
fig = plt.figure()
sub = fig.add_subplot(111)
scat = sub.scatter(dt_abias[duty], tdelay[duty], c=sigMstar[duty], lw=0, s=100)
plt.colorbar(scat)
sub.set_xlabel('$\Delta t$ Gyr', fontsize=25)
sub.set_ylabel('$t_{delay}$ Gyr', fontsize=25)
fig_name = ''.join([UT.fig_dir(), 'tdelay_dt_grid.', run, '.tduty', str(tduty), '.png'])
fig.savefig(fig_name, bbox_inches='tight')
return None
def thetaLHD():
''' ***TESTED***
Test thetaLHD using the HOD parameter priors
'''
# from Sinha et al (2017) prior
theta_lbl = [
'log $M_\mathrm{min}$', '$\sigma_{\mathrm{log}\,M}$', 'log $M_0$',
'log $M_1$', r'$\alpha$'
]
HOD_range_min = [11., 0.001, 6., 12., 0.001]
HOD_range_max = [12.2, 1., 14., 14., 2.]
samples = 17
m_list = ['maximin', 'centermaximin', 'mdu', 'nohl']
for meth in m_list:
lhcube = lhd.thetaLHD([HOD_range_min, HOD_range_max],
samples=samples,
method=meth)
fig = plt.figure(figsize=(9, 9))
for i in range(lhcube.shape[1]):
for j in range(lhcube.shape[1]):
if i < j:
sub = fig.add_subplot(lhcube.shape[1], lhcube.shape[1],
lhcube.shape[1] * j + i + 1)
sub.scatter(lhcube[:, i], lhcube[:, j])
sub.set_xlim([HOD_range_min[i], HOD_range_max[i]])
if j == lhcube.shape[1] - 1:
sub.set_xlabel(theta_lbl[i], fontsize=20)
else:
sub.set_xticks([])
sub.set_ylim([HOD_range_min[j], HOD_range_max[j]])
if i == 0: sub.set_ylabel(theta_lbl[j], fontsize=20)
else: sub.set_yticks([])
elif i == j:
sub = fig.add_subplot(lhcube.shape[1], lhcube.shape[1],
lhcube.shape[1] * j + i + 1)
sub.hist(lhcube[:, i],
range=[HOD_range_min[i], HOD_range_max[i]],
normed=True)
sub.set_xlim([HOD_range_min[i], HOD_range_max[i]])
if i != 0: sub.set_yticks([])
else: sub.set_ylabel(theta_lbl[j], fontsize=20)
if i != lhcube.shape[1] - 1: sub.set_xticks([])
else: sub.set_xlabel(theta_lbl[i], fontsize=20)
f = ''.join([
UT.fig_dir(), 'tests/test.thetaLHD.HOD.', meth, '.',
str(samples), '.samples.png'
])
fig.savefig(f, bbox_inches='tight')
return None
def SFMS_comparison():
''' Comparison of different SFMS observations at z~0.05 and 1. We
include the compilation fit from Speagle et al. (2014).
'''
# z = 0
# fit from SDSS group catalog
logSFR_sdss = lambda mm: (0.5757 * (mm - 10.5) - 0.13868)
# Lee et al. (2015)
logSFR_lee = lambda mm: 1.53 - np.log10(1 + (10**mm/10**(10.10))**-1.26)
# Noeske et al. (2007) 0.85 < z< 1.10 (by eye)
logSFR_noeske = lambda mm: (1.580 - 1.064)/(11.229 - 10.029)*(mm - 10.0285) + 1.0637
# Moustakas et al. (2013) 0.8 < z < 1. (by eye)
logSFR_primus = lambda mm: (1.3320 - 1.296)/(10.49 - 9.555) * (mm-9.555) + 1.297
# Speagle's SFMS bestfit
logSFR_speagle = lambda mm, tt: (0.84 - 0.026 * tt) * mm - (6.51 - 0.11 * tt)
# priors
logSFR_prior = lambda mm, zz, p1, p2: logSFR_sdss(mm) + (zz - 0.05) * (p1 * (mm - 10.5) + p2)
marr = np.linspace(9., 12., 100)
fig = plt.figure(1)
sub = fig.add_subplot(121) # z = 0.05 subplot
sub.plot(marr, logSFR_sdss(marr), label='SDSS group catalog')
sub.plot(marr, logSFR_speagle(marr, 13.5), label='Speagle $z \sim 0.05$')
sub.legend(loc='lower right')
sub.set_xlim([9.5, 11.5])
sub.set_ylim([-2, 3])
sub = fig.add_subplot(122) # z = 1 subplot
sub.plot(marr, logSFR_lee(marr), label='Lee et al.(2015)')
sub.plot(marr, logSFR_noeske(marr), label='Noeske et al.(2007)')
sub.plot(marr, logSFR_primus(marr), label='Moustakas et al.(2013)')
sub.plot(marr, logSFR_speagle(marr, 5.7), label='Speagle $z \sim 1$')
logSFRmin, logSFRmax = [], []
parr = [[-0.5, 1], [-0.5, 2.], [0.5, 1.], [0.5, 2.]]
for mm in marr:
logSFRmin.append(
np.min([logSFR_prior(mm, 1., pp[0], pp[1]) for pp in parr]))
logSFRmax.append(
np.max([logSFR_prior(mm, 1., pp[0], pp[1]) for pp in parr]))
sub.fill_between(marr, logSFRmin, logSFRmax, label='Prior', alpha=0.5)
sub.legend(loc='lower right')
sub.set_xlim([9.5, 11.5])
sub.set_ylim([-2, 3])
fig.savefig(UT.fig_dir()+'test.sfms.comparison.png')
plt.close()
return None
def Plk_halo_mneut_ratio(nzbin=4, zspace=False):
''' Plot the ratio of P_l^mneut(k)/P_l^0.0eV
for different neutrino masses
'''
mneuts = [0.0, 0.06, 0.10, 0.15, 0.6] # eV
p0ks_mneut, p2ks_mneut, p4ks_mneut = [], [], []
for mneut in mneuts:
p0ks, p2ks, p4ks = [], [], []
for ireal in range(1, 101):
# read all 100 realizations
plk_i = Obvs.Plk_halo(mneut, ireal, nzbin, zspace=zspace)
if ireal == 1: k = plk_i['k']
p0ks.append(plk_i['p0k'])
p2ks.append(plk_i['p2k'])
p4ks.append(plk_i['p4k'])
# plot the average
p0ks_mneut.append(np.average(np.array(p0ks), axis=0))
p2ks_mneut.append(np.average(np.array(p2ks), axis=0))
p4ks_mneut.append(np.average(np.array(p4ks), axis=0))
plks_mneut = [p0ks_mneut, p2ks_mneut, p4ks_mneut]
fig = plt.figure(figsize=(15, 5))
for i, ell in enumerate([0, 2, 4]):
sub = fig.add_subplot(1, 3, i + 1)
for ii in range(len(mneuts)):
sub.plot(k,
plks_mneut[i][ii] / plks_mneut[i][0],
lw=2,
label=r'$\sum m_\nu = $ ' + str(mneuts[ii]) + 'eV')
if i == 0:
sub.legend(loc='lower right', prop={'size': 12})
else:
sub.set_yticks([])
sub.set_xscale('log')
sub.set_xlim([0.01, 0.5])
sub.set_xlabel('k', fontsize=20)
sub.set_ylim([0.9, 1.15])
sub.set_ylabel('$P_{' + str(ell) + '}(k)/P_{' + str(ell) +
'}^{0.0\mathrm{eV}}(k)$',
fontsize=20)
if zspace: str_space = 'z'
else: str_space = 'r'
fig.savefig(''.join([
UT.fig_dir(), 'tests/plk_halo.mneuts_ratio.nzbin',
str(nzbin), '.', str_space, 'space.png'
]),
bbox_inches='tight')
return None
def p_Xw_i_outlier(mock, ell=0, rebin=None, krange=None, method='choletsky'):
''' Examine the pdf of X_w^i components that deviate significantly from
N(0,1)
'''
Pk = NG.dataX(mock, ell=ell, rebin=rebin, krange=krange)
X, _ = NG.meansub(Pk)
X_w, W = NG.whiten(X, method=method) # whitened data
# calculate the chi-squared values of each p(X_w^i)
x = np.arange(-5., 5.1, 0.1)
chi2 = np.zeros(X_w.shape[1])
for i_bin in range(X_w.shape[1]):
kern = gkde(X_w[:,i_bin]) # gaussian KDE kernel using "rule of thumb" scott's rule.
chi2[i_bin] = np.sum((UT.gauss(x, 1., 0.) - kern.evaluate(x))**2)/np.float(len(x))
# plot the most discrepant components.
prettyplot()
fig = plt.figure()
sub = fig.add_subplot(111)
i_sort = np.argsort(chi2)
print 'outlier bins = ', i_sort[-5:]
for i_bin in i_sort[-10:]:
kern = gkde(X_w[:,i_bin]) # gaussian KDE kernel using "rule of thumb" scott's rule.
sub.plot(x, kern.evaluate(x))
sub.plot(x, UT.gauss(x, 1., 0.), c='k', lw=3, label='$\mathcal{N}(0,1)$')
sub.set_xlim([-2.5, 2.5])
sub.set_xlabel('$\mathtt{X^{i}_{W}}$', fontsize=25)
sub.set_ylim([0., 0.6])
sub.set_ylabel('$\mathtt{P(X^{i}_{W})}$', fontsize=25)
sub.legend(loc='upper right')
if rebin is None:
f = ''.join([UT.fig_dir(), 'tests/test.p_Xw_i_outlier.', method, '.', mock, '.ell', str(ell), '.png'])
else:
f = ''.join([UT.fig_dir(), 'tests/test.p_Xw_i_outlier.', method, '.', mock, '.ell', str(ell), '.rebin', str(rebin), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def dataX(mock, ell=0, rebin=None, krange=None):
''' ***TESTED***
Test the data X calculation
'''
Pk = NG.dataX(mock, ell=ell, rebin=rebin, krange=krange)
X, _ = NG.meansub(Pk)
prettyplot()
fig = plt.figure()
sub = fig.add_subplot(111)
for i in range(X.shape[0]):
sub.plot(range(X.shape[1]), X[i,:])
sub.set_xlim([0, X.shape[1]])
sub.set_xlabel('$\mathtt{k}$ bins', fontsize=25)
sub.set_ylim([-1e5, 1e5])
sub.set_ylabel('$\mathtt{P^i_'+str(ell)+'(k) - \overline{P_'+str(ell)+'(k)}}$', fontsize=25)
if rebin is not None:
f = ''.join([UT.fig_dir(), 'tests/test.dataX.', mock, '.ell', str(ell), '.rebin', str(rebin), '.png'])
else:
f = ''.join([UT.fig_dir(), 'tests/test.dataX.', mock, '.ell', str(ell), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def plot_thetas(theta,
w,
t,
truths=None,
plot_range=None,
theta_filename=None,
output_dir=None):
''' Corner plots of input theta values
'''
if output_dir is None:
fig_dir = util.fig_dir()
else:
fig_dir = output_dir
for ww in ['weighted', 'unweighted']:
if ww == 'weighted':
weights = w.flatten()
elif ww == 'unweighted':
weights = None
# weighted theta
fig = corner.corner(theta,
weights=weights,
truths=truths,
truth_color='#ee6a50',
labels=[
r'$\mathtt{\log\;M_{0}}$',
r'$\mathtt{\log\;\sigma_{\logM}}$',
r'$\mathtt{\log\;M_{min}}$',
r'$\mathtt{\alpha}$', r'$\mathtt{\log\;M_{1}}$'
],
label_kwargs={'fontsize': 25},
range=plot_range,
quantiles=[0.16, 0.5, 0.84],
show_titles=True,
title_args={"fontsize": 12},
plot_datapoints=True,
fill_contours=True,
levels=[0.68, 0.95],
color='b',
bins=20,
smooth=1.0)
fig_file = ''.join([theta_filename.split('.dat')[0], '.', ww, '.png'])
plt.savefig(fig_file)
plt.close()
return None
def IntegTest():
''' Test the convergence of integration methods by trying both euler and
RK4 integration for different time steps and then comparing the resulting SMF
Conclusions
-----------
* The integration is more or less converged after tstep <= 0.5 dt_min especially for
RK4. So for calculations use tstep = 0.5 dt_min.
'''
prettyplot()
pretty_colors = prettycolors()
fig = plt.figure(figsize=(15,6))
for ii, integ in enumerate(['euler', 'rk4']):
sub = fig.add_subplot(1,2,ii+1)
for i_t, tstep in enumerate([0.001, 0.01, 0.1]):
evol_dict = {
'sfh': {'name': 'random_step', 'dt_min': 0.1, 'dt_max': 0.5, 'sigma': 0.3,
'assembly_bias': 'acc_hist', 'halo_prop': 'frac', 'sigma_bias': 0.3},
'mass': {'type': integ, 'f_retain': 0.6, 't_step': tstep}
}
EGP = CMS.EvolvedGalPop(cenque='default', evol_dict=evol_dict)
if not os.path.isfile(EGP.File()):
EGP.Write()
EGP.Read()
MSonly = np.where(EGP.t_quench == 999.) # remove the quenching galaxies
smf = Obvs.getSMF(EGP.mass[MSonly], weights=EGP.weight_down[MSonly])
sub.plot(smf[0], smf[1], lw=3,
c=pretty_colors[i_t], ls='--',
label=''.join([integ, ';', r'$\mathtt{t_{step} =', str(tstep),'}$']))
# analytic SMF for comparison
theory_smf = Obvs.getSMF(EGP.M_sham[MSonly], weights=EGP.weight_down[MSonly])
sub.plot(theory_smf[0], theory_smf[1], lw=2, c='k', label='Theory')
sub.set_ylim([10**-5, 10**-1])
sub.set_xlim([8., 12.0])
sub.set_yscale('log')
sub.set_xlabel(r'Mass $\mathtt{M_*}$', fontsize=25)
if ii == 0:
sub.set_ylabel(r'Stellar Mass Function $\mathtt{\Phi}$', fontsize=25)
sub.legend(loc='upper right', frameon=False)
fig_file = ''.join([UT.fig_dir(), 'IntegTest.png'])
fig.savefig(fig_file, bbox_inches='tight')
plt.close()
return None
def Galaxies():
''' Test that the Galaxies function correctly populats halo catalogs with an HOD
'''
# HOD parameters (Zheng+2007 Mr < -21)
p_hod = {
'logM0': 11.92,
'sigma_logM': 0.39,
'logMmin': 12.79,
'alpha': 1.15,
'logM1': 13.94
}
# import Neutrino halo with 0.0eV realization 1 at z=0
halos = Dat.NeutHalos(0.0, 1, 4)
halos['RSDPosition'] = FM.RSD(halos)
# measure the P_l(k) of the galaxy catalog
plk_halo = FM.Observables(halos, observable='plk', rsd=True, Nmesh=256)
# now run HOD on the halo catalog
gals = FM.Galaxies(halos, p_hod, seed=10)
gals['RSDPosition'] = FM.RSD(gals)
# measure the P_l(k) of the galaxy catalog
plk = FM.Observables(gals, observable='plk', rsd=True, Nmesh=256)
fig = plt.figure()
sub = fig.add_subplot(121)
for ell in [0, 2, 4]:
sub.plot(plk['k'], plk['k'] * plk['p' + str(ell) + 'k'])
sub.plot(plk_halo['k'],
plk_halo['k'] * plk_halo['p' + str(ell) + 'k'],
ls='--')
sub.set_xlabel('$k$', fontsize=20)
sub.set_xlim([0.01, 0.5])
sub.set_xscale('log')
sub.set_ylabel('$k P_\ell(k)$', fontsize=20)
sub = fig.add_subplot(122)
sub.plot(plk['k'], plk['p0k'])
sub.plot(plk_halo['k'], plk_halo['p0k'], ls='--')
sub.set_xlabel('$k$', fontsize=20)
sub.set_xlim([0.01, 0.5])
sub.set_xscale('log')
sub.set_ylabel('$P_\ell(k)$', fontsize=20)
sub.set_yscale('log')
fig.savefig(''.join([UT.fig_dir(), 'tests/hodgalaxy_pk.png']),
bbox_inches='tight')
return None
def LHD():
''' *** TESTED ***
Testing the different methods of LHD
'''
dim = 5
samples = 17
m_list = ['maximin', 'centermaximin', 'mdu', 'nohl']
for meth in m_list:
lhcube = lhd.LHD(dim, samples=samples, method=meth, niter=1000)
fig = plt.figure(figsize=(9, 9))
for i in range(lhcube.shape[1]):
for j in range(lhcube.shape[1]):
if i < j:
sub = fig.add_subplot(lhcube.shape[1], lhcube.shape[1],
lhcube.shape[1] * j + i + 1)
sub.scatter(lhcube[:, i], lhcube[:, j])
sub.set_xlim([0., 1.])
if j == lhcube.shape[1] - 1:
sub.set_xticks([0., 0.5, 1.])
sub.set_xlabel(r'$\theta_' + str(i) + '$', fontsize=20)
else:
sub.set_xticks([])
sub.set_ylim([0., 1.])
if i == 0:
sub.set_yticks([0., 0.5, 1.])
sub.set_ylabel(r'$\theta_' + str(j) + '$', fontsize=20)
else:
sub.set_yticks([])
elif i == j:
sub = fig.add_subplot(lhcube.shape[1], lhcube.shape[1],
lhcube.shape[1] * j + i + 1)
sub.hist(lhcube[:, i], range=[0., 1], normed=True)
sub.set_xlim([0., 1.])
if i != 0:
sub.set_yticks([])
if i != lhcube.shape[1] - 1:
sub.set_xticks([])
f = ''.join([
UT.fig_dir(), 'tests/test.LHD.', meth, '.',
str(dim), 'dim.',
str(samples), '.samples.png'
])
fig.savefig(f, bbox_inches='tight')
return None
def tdelay_dt_mcmc(run, theta, Niter=20, Nwalkers=10, Ndim=2, sigma_smhm=0.2, nsnap0=15, downsampled='14', flag=None, continue_chain=False):
'''
'''
if Ndim == 2:
tdelay_range = [0., 3.]#np.arange(0., 3., 0.5)
dt_range = [0.1, 4.]
# new chain
chain_file = ''.join([UT.fig_dir(), run, '.tdelay_dt_mcmc.chain.dat'])
if os.path.isfile(chain_file) and continue_chain:
print 'Continuing previous MCMC chain!'
sample = np.loadtxt(chain_file)
Niter = Niter - (np.float(len(sample))/np.float(Nwalkers)) # Number of chains left to finish
if Niter <= 0:
raise ValueError
print Niter, ' iterations left to finish'
else:
f = open(chain_file, 'w')
f.close()
# Initializing Walkers
pos0 = [np.array([np.random.uniform(tdelay_range[0], tdelay_range[1]), np.random.uniform(dt_range[0], dt_range[1])]) for i in range(Nwalkers)]
pool = MPIPool()
if not pool.is_master():
pool.wait()
sys.exit(0)
# Initializing the emcee sampler
kwargs = {
'theta': theta,
'sigma_smhm': 0.2,
'nsnap0': 15,
'downsampled': '14',
}
sampler = emcee.EnsembleSampler(Nwalkers, Ndim, sigM, pool=pool, kwargs=kwargs)
for result in sampler.sample(pos0, iterations=Niter, storechain=False):
position = result[0]
#print position
f = open(chain_file, 'a')
for k in range(position.shape[0]):
output_str = '\t'.join(position[k].astype('str')) + '\n'
f.write(output_str)
f.close()
pool.close()
return None
def plot_thetas(theta, w , t, truths=None, plot_range=None,
theta_filename=None, output_dir=None):
''' Corner plots of input theta values
'''
if output_dir is None:
fig_dir = util.fig_dir()
else:
fig_dir = output_dir
for ww in [ 'weighted', 'unweighted']:
if ww == 'weighted':
weights = w.flatten()
elif ww == 'unweighted':
weights = None
# weighted theta
fig = corner.corner(
theta,
weights=weights,
truths=truths,
truth_color='#ee6a50',
labels=[
r'$\mathtt{\log\;M_{0}}$',
r'$\mathtt{\log\;\sigma_{\logM}}$',
r'$\mathtt{\log\;M_{min}}$',
r'$\mathtt{\alpha}$',
r'$\mathtt{\log\;M_{1}}$'
],
label_kwargs={'fontsize': 25},
range=plot_range,
quantiles=[0.16,0.5,0.84],
show_titles=True,
title_args={"fontsize": 12},
plot_datapoints=True,
fill_contours=True,
levels=[0.68, 0.95],
color='b',
bins=20,
smooth=1.0)
fig_file = ''.join([theta_filename.split('.dat')[0], '.', ww, '.png'])
plt.savefig(fig_file)
plt.close()
return None
def tduty_tdelay_dt_grid(run, theta, sigma_smhm=0.2, nsnap0=15, downsampled='14', flag=None):
''' plot sigma_M* on a grid of t_delay and dt
'''
delt = 0.5
tdelays = np.arange(0., 3., delt)
dts = np.arange(0., 4.5, delt)
dts[0] += 0.1 # dt = 0 not allowed
tdutys = np.array([0.5, 1., 2., 3., 5., 10.])
subcat_dat = abcee.Data(nsnap0=nsnap0, sigma_smhm=sigma_smhm) # 'data'
sumdata = abcee.SumData(['smf'], nsnap0=nsnap0, sigma_smhm=sigma_smhm)
smhmr = Obvs.Smhmr()
grid = np.zeros((5, len(tdutys)*len(tdelays)*len(dts)))
ii = 0
for i_duty, tduty in enumerate(tdutys):
for i_t, tdelay in enumerate(tdelays):
for i_d, dt in enumerate(dts):
theta_i = np.concatenate([theta, np.array([tduty, tdelay, dt])])
#try:
subcat_sim = abcee.model(run, theta_i,
nsnap0=nsnap0, sigma_smhm=sigma_smhm, downsampled=downsampled)
sumsim = abcee.SumSim(['smf'], subcat_sim)
isSF = np.where(subcat_sim['gclass'] == 'sf') # only SF galaxies
# calculate sigma_M* at M_h = 12
try:
m_mid, mu_mhalo, sig_mhalo, cnts = smhmr.Calculate(subcat_sim['halo.m'][isSF], subcat_sim['m.star'][isSF],
dmhalo=0.2, weights=subcat_sim['weights'][isSF])
grid[3,ii] = sig_mhalo[np.argmin(np.abs(m_mid-12.))]
# rho
grid[4, ii] = abcee.L2_logSMF(sumsim, sumdata)
except ValueError:
grid[3,ii] = -999.
grid[4, ii] = 999.
print 'tduty = ', tduty, 'tdelay = ', tdelay, ', dt = ', dt, ', sigma = ', grid[3, ii]
grid[0, ii] = tduty
grid[1, ii] = tdelay
grid[2, ii] = dt
ii += 1
# save sig_Mstar values to file
file_name = ''.join([UT.fig_dir(), 'tduty_tdelay_dt_grid.', run, '.p'])
pickle.dump(grid, open(file_name, 'wb'))
return None
def p_Xw_i_MISE(mock, ell=0, rebin=None, krange=None, method='choletsky', b=0.1):
''' Examine the pdf of X_w^i components that deviate significantly from
N(0,1) based on MISE
'''
Pk = NG.dataX(mock, ell=ell, rebin=rebin, krange=krange)
X, _ = NG.meansub(Pk)
X_w, W = NG.whiten(X, method=method) # whitened data
# calculate the chi-squared values of each p(X_w^i)
x = np.arange(-5., 5.1, 0.1)
mise = np.zeros(X_w.shape[1])
for i_bin in range(X_w.shape[1]):
mise[i_bin] = NG.MISE(X_w[:,i_bin], b=b)
# plot the most discrepant components.
prettyplot()
fig = plt.figure()
sub = fig.add_subplot(111)
i_sort = np.argsort(mise)
print 'outlier bins = ', i_sort[-5:]
print 'mise = ', mise[i_sort[-5:]]
nbin = int(10./b)
for i_bin in i_sort[-10:]:
hb_Xi, Xi_edges = np.histogram(X_w[:,i_bin], bins=nbin, range=[-5., 5.], normed=True)
p_X_w_arr = UT.bar_plot(Xi_edges, hb_Xi)
sub.plot(p_X_w_arr[0], p_X_w_arr[1])
sub.plot(x, UT.gauss(x, 1., 0.), c='k', lw=3, label='$\mathcal{N}(0,1)$')
sub.set_xlim([-2.5, 2.5])
sub.set_xlabel('$\mathtt{X^{i}_{W}}$', fontsize=25)
sub.set_ylim([0., 0.6])
sub.set_ylabel('$\mathtt{P(X^{i}_{W})}$', fontsize=25)
sub.legend(loc='upper right')
str_rebin = ''
if rebin is not None:
str_rebin = '.rebin'+str(rebin)
f = ''.join([UT.fig_dir(), 'tests/test.p_Xw_i_outlier.', method, '.', mock, '.ell', str(ell),
str_rebin, '.b', str(b), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def PlotCovariance(obvs, Mr=21, b_normal=0.25, inference='mcmc'):
''' Plot the covariance matrix for a specified obvs
'''
# import the covariance matrix
covar = Data.data_cov(Mr=Mr, b_normal=b_normal, inference=inference)
if obvs == 'xi':
obvs_cov = covar[1:16 , 1:16]
r_bin = Data.xi_binedges()
elif obvs == 'gmf':
obvs_cov = covar[17:, 17:]
binedges = Data.data_gmf_bins()
r_bin = 0.5 * (binedges[:-1] + binedges[1:])
n_bin = int(np.sqrt(obvs_cov.size))
# calculate the reduced covariance for plotting
red_covar = np.zeros([n_bin, n_bin])
for ii in range(n_bin):
for jj in range(n_bin):
red_covar[ii][jj] = obvs_cov[ii][jj]/np.sqrt(obvs_cov[ii][ii] * obvs_cov[jj][jj])
prettyplot()
fig = plt.figure()
sub = fig.add_subplot(111)
cont = sub.pcolormesh(r_bin, r_bin, red_covar, cmap=plt.cm.afmhot_r)
plt.colorbar(cont)
sub.set_xlim([r_bin[0], r_bin[-1]])
sub.set_ylim([r_bin[0], r_bin[-1]])
sub.set_xscale('log')
sub.set_yscale('log')
sub.set_xlabel(r'$\mathtt{r}\;[\mathtt{Mpc/h}$]', fontsize=25)
sub.set_ylabel(r'$\mathtt{r}\;[\mathtt{Mpc/h}$]', fontsize=25)
fig_file = ''.join([util.fig_dir(),
obvs.upper(), 'covariance',
'.Mr', str(Mr),
'.bnorm', str(round(b_normal,2)),
'.', inference, '_inf.png'])
fig.savefig(fig_file, bbox_inches='tight')
plt.close()
return None
def HODemulator_readHODLHD(HODrange='sinha2017prior_narrow',
method='mdu',
samples=17):
''' ***TESTED Nov 13, 2017***
Test HODemulator.read_HODLHD
'''
emu = Emu.HODemulator()
lhcube = emu.read_HODLHD(HODrange=HODrange, method=method, samples=samples)
fig = plt.figure(figsize=(9, 9))
for i in range(lhcube.shape[1]):
for j in range(lhcube.shape[1]):
if i < j:
sub = fig.add_subplot(lhcube.shape[1], lhcube.shape[1],
lhcube.shape[1] * j + i + 1)
sub.scatter(lhcube[:, i], lhcube[:, j])
sub.set_xlim([emu.HODrange_min[i], emu.HODrange_max[i]])
if j == lhcube.shape[1] - 1:
sub.set_xlabel(emu.HOD_labels[i], fontsize=20)
else:
sub.set_xticks([])
sub.set_ylim([emu.HODrange_min[j], emu.HODrange_max[j]])
if i == 0:
sub.set_ylabel(emu.HOD_labels[j], fontsize=20)
else:
sub.set_yticks([])
elif i == j:
sub = fig.add_subplot(lhcube.shape[1], lhcube.shape[1],
lhcube.shape[1] * j + i + 1)
sub.hist(lhcube[:, i],
range=[emu.HODrange_min[i], emu.HODrange_max[i]],
normed=True)
sub.set_xlim([emu.HODrange_min[i], emu.HODrange_max[i]])
if i != 0:
sub.set_yticks([])
if i != lhcube.shape[1] - 1:
sub.set_xticks([])
f = ''.join([
UT.fig_dir(), 'tests/test.HODemulator.LHD.', method, '.',
str(samples), '.samples.png'
])
fig.savefig(f, bbox_inches='tight')
return None
def plotCMS_SMF(cenque='default', evol_dict=None):
''' Plot the stellar mass function of the integrated SFR population
and compare it to the theoretic stellar mass function
'''
# calculate SMF
cq = CMS.CentralQuenched(cenque=cenque) # quenched population
cq._Read_CenQue()
# import evolved galaxy population
MSpop = CMS.EvolvedGalPop(cenque=cenque, evol_dict=evol_dict)
MSpop.Read()
smf = Obvs.getSMF(
np.array(list(MSpop.mass)+list(cq.mass)),
weights=np.array(list(MSpop.weight_down) + list(cq.weight_down))
)
prettyplot()
pretty_colors = prettycolors()
# analytic SMF
theory_smf = Obvs.getSMF(
np.array(list(MSpop.M_sham)+list(cq.M_sham)),
weights=np.array(list(MSpop.weight_down) + list(cq.weight_down))
)
#Obvs.analyticSMF(0.1, source='li-march')
fig = plt.figure()
sub = fig.add_subplot(111)
sub.plot(theory_smf[0], theory_smf[1], lw=3, c='k', label='Theory')
sub.plot(smf[0], smf[1], lw=3, c=pretty_colors[3], ls='--', label='Simul.')
sub.set_ylim([10**-5, 10**-1])
sub.set_xlim([6., 12.0])
sub.set_yscale('log')
sub.set_xlabel(r'Mass $\mathtt{M_*}$', fontsize=25)
sub.set_ylabel(r'Stellar Mass Function $\mathtt{\Phi}$', fontsize=25)
sub.legend(loc='upper right', frameon=False)
fig_file = ''.join([UT.fig_dir(), 'SMF.CMS', MSpop._Spec_str(), '.png'])
fig.savefig(fig_file, bbox_inches='tight')
plt.close()
return None
def test_tduty_tdelay_dt_grid_dtaxis(run, tdutyy, tdelayy):
''' Check the dependence of sigma_logM* to dt_abias. It's weird that
dt_abias has such a significant effect...
'''
file_name = ''.join([UT.fig_dir(), 'tduty_tdelay_dt_grid.', run, '.p'])
grid = pickle.load(open(file_name, 'rb'))
tduty = grid[0,:]
tdelay = grid[1,:]
dt_abias = grid[2,:]
sigMstar = grid[3,:]
l2 = grid[4,:]
lim = np.where((tdelay == tdelayy) & (tduty == tdutyy))
for ii in range(len(lim[0])):
print dt_abias[lim[0][ii]], sigMstar[lim[0][ii]]
#abcee.qaplotABC(run, 10, sigma_smhm=0.2, theta=np.array([1.35, 0.6, grid[0,i_min], grid[1,i_min], grid[2,i_min]]))
return None
def test_tduty_tdelay_dt_grid_best(run):
'''
'''
file_name = ''.join([UT.fig_dir(), 'tduty_tdelay_dt_grid.', run, '.p'])
grid = pickle.load(open(file_name, 'rb'))
tduty = grid[0,:]
tdelay = grid[1,:]
dt_abias = grid[2,:]
sigMstar = grid[3,:]
l2 = grid[4,:]
lim = np.where(tdelay > 1.)
i_min = np.argmin(sigMstar[lim])
i_min = lim[0][i_min]
print grid[:,i_min]
abcee.qaplotABC(run, 10, sigma_smhm=0.2, theta=np.array([1.35, 0.6, grid[0,i_min], grid[1,i_min], grid[2,i_min]]))
return None
def plotCMS_SFH(cenque='default', evol_dict=None):
''' Plot the star formation history of star forming main sequence galaxies
'''
# import evolved galaxy population
egp = CMS.EvolvedGalPop(cenque=cenque, evol_dict=evol_dict)
egp.Read()
MSonly = np.where(egp.t_quench == 999.) # only keep main-sequence galaxies
MSonly_rand = np.random.choice(MSonly[0], size=10)
z_acc, t_acc = UT.zt_table()
t_cosmic = t_acc[1:16]
dt = 0.1
t_arr = np.arange(t_cosmic.min(), t_cosmic.max()+dt, dt)
prettyplot()
pretty_colors = prettycolors()
fig = plt.figure()
sub = fig.add_subplot(111)
dsfrt = np.zeros((len(t_arr), len(MSonly_rand)))
for i_t, tt in enumerate(t_arr):
dsfr = SFR.dSFR_MS(tt, egp.sfh_dict)
for ii, i_gal in enumerate(MSonly_rand):
dsfrt[i_t,ii] = dsfr[i_gal]
for ii, i_gal in enumerate(MSonly_rand):
tlim = np.where(t_arr > egp.tsnap_genesis[i_gal])
sub.plot(t_arr[tlim], dsfrt[:,ii][tlim], lw=3, c=pretty_colors[ii])
sub.set_xlim([5.0, 13.5])
sub.set_xlabel(r'$\mathtt{t_{cosmic}}$', fontsize=25)
sub.set_ylim([-1., 1.])
sub.set_ylabel(r'$\mathtt{SFR(t) - <SFR_{MS}(t)>}$', fontsize=25)
fig_file = ''.join([UT.fig_dir(), 'SFH.CMS', egp._Spec_str(), '.png'])
fig.savefig(fig_file, bbox_inches='tight')
plt.close()
def X_gmf_all():
''' ***TESTED -- Nov 8, 2017***
Test to make sure that NG.X_gmf_all returns correct values
'''
X, nbins = NG.X_gmf_all(n_arr=True)
nmid = 0.5*(nbins[1:] + nbins[:-1])
assert X.shape[1] == len(nmid)
assert X.shape[0] == 20000
fig = plt.figure(figsize=(5,5))
sub = fig.add_subplot(111)
for i in np.random.choice(range(X.shape[0]), 1000, replace=False):
sub.plot(nmid, X[i,:], c='k', lw=0.01)
sub.plot(nmid, np.average(X, axis=0), c='r', lw=2, ls='--')
# x-axis
sub.set_xlim([0., 180.])
sub.set_xlabel('$N$', fontsize=20)
# y-axis
sub.set_yscale('log')
sub.set_ylabel(r'$\zeta(N)$', fontsize=20)
fig.savefig(''.join([UT.fig_dir(), 'tests/X_gmf_all.png']), bbox_inches='tight')
plt.close()
return None
def Plk_LHD(nreal):
''' Test that Data.X_lhd returns sensible P_l(k)
for the LHD
'''
karr, Xlhd = Data.X_lhd(0.0, nreal, 4, 1, obvs='plk', karr=True,
HODrange='sinha2017prior_narrow', samples=17, method='mdu')
lhcube = LHD.HOD_LHD(HODrange='sinha2017prior_narrow', samples=17, method='mdu')
fig = plt.figure(figsize=(12,4))
sub = fig.add_subplot(131)
sub1 = fig.add_subplot(132)
sub2 = fig.add_subplot(133)
for i in range(10):#Xlhd.shape[0]):
sub.plot(karr, Xlhd[i,:], c='C'+str(i % 10))
sub1.scatter([lhcube[i,0]], [lhcube[i,1]], c='C'+str(i % 10))
sub2.scatter([lhcube[i,3]], [lhcube[i,4]], c='C'+str(i % 10))
sub.set_xlim([1e-2, 0.5])
sub.set_xlabel('$k$', fontsize=20)
sub.set_xscale('log')
sub.set_ylim([2e3, 2e5])
sub.set_ylabel('$P(k)$', fontsize=20)
sub.set_yscale('log')
sub1.set_xlim([11., 12.2])
sub1.set_xlabel('log $M_\mathrm{min}$', fontsize=20)
sub1.set_ylim([0.001, 1.])
sub1.set_ylabel('$\sigma_{\mathrm{log}\,M}$', fontsize=20)
sub2.set_xlim([12., 14])
sub2.set_xlabel('log $M_1$', fontsize=20)
sub2.set_ylim([0.5, 1.5])
sub2.set_ylabel(r'$\alpha$', fontsize=20)
f = ''.join([UT.fig_dir(), 'tests/test_data.Plk_lhd.png'])
fig.savefig(f, bbox_inches='tight')
return None
def plot_data(Mr=21, output_dir=None):
'''
plot summary statistics of the data
'''
if output_dir is None:
fig_dir = util.fig_dir()
else:
fig_dir = output_dir
cov = Data.data_cov()
xi_err = np.sqrt(np.diag(cov)[1:16])
gmf_err = np.sqrt(np.diag(cov)[16:])
fig, axes = plt.subplots(1, 2, figsize=(10, 5))
fig.subplots_adjust(wspace=0.4, hspace=0.4)
ax = axes[0]
x = Data.xi_binedges()
y = Data.data_xi()
ax.errorbar(0.5 * (x[:-1] + x[1:]), y, yerr=xi_err, fmt=".k", capsize=0)
ax.set_yscale('log')
ax.set_xscale('log')
ax.set_xlim(0.2, 22)
ax.set_xlabel(r'$r[\mathrm{Mpc}/h]$')
ax.set_ylabel(r'$\xi(r)$')
ax = axes[1]
x = Data.gmf_bins()
y = Data.data_gmf()
ax.errorbar(0.5 * (x[:-1] + x[1:]), y, yerr=gmf_err, fmt=".k", capsize=0)
ax.set_xlim(1, 20)
ax.set_yscale('log')
ax.set_xlabel(r'Group Richness $N$')
ax.set_ylabel(r'$\zeta(N)$')
fig_file = ''.join([fig_dir, "data", '_Mr', str(Mr), '.pdf'])
plt.savefig(fig_file)
plt.close()
def plot_data(Mr = 21 , output_dir = None):
'''
plot summary statistics of the data
'''
if output_dir is None:
fig_dir = util.fig_dir()
else:
fig_dir = output_dir
cov = Data.data_cov()
xi_err = np.sqrt(np.diag(cov)[1:16])
gmf_err = np.sqrt(np.diag(cov)[16:])
fig , axes = plt.subplots(1, 2, figsize=(10, 5))
fig.subplots_adjust(wspace=0.4, hspace=0.4)
ax = axes[0]
x = Data.xi_binedges()
y = Data.data_xi()
ax.errorbar(0.5*(x[:-1]+x[1:]), y, yerr=xi_err, fmt=".k", capsize=0)
ax.set_yscale('log')
ax.set_xscale('log')
ax.set_xlim(0.2, 22)
ax.set_xlabel(r'$r[\mathrm{Mpc}/h]$')
ax.set_ylabel(r'$\xi(r)$')
ax = axes[1]
x = Data.gmf_bins()
y = Data.data_gmf()
ax.errorbar(0.5*(x[:-1]+x[1:]), y, yerr=gmf_err, fmt=".k", capsize=0)
ax.set_xlim(1, 20)
ax.set_yscale('log')
ax.set_xlabel(r'Group Richness $N$')
ax.set_ylabel(r'$\zeta(N)$')
fig_file = ''.join([fig_dir, "data", '_Mr', str(Mr),'.pdf'])
plt.savefig(fig_file)
plt.close()
def HODemulator_readNeutObvs(obvs='p0k',
mneut=0.0,
nreal=1,
nzbin=4,
seed_hod=1,
Nmesh=360,
rsd=True,
HODrange='sinha2017prior_narrow',
method='mdu',
samples=17):
''' ***TESTED Nov 13, 2017***
Test HODemulator.read_HODLHD
'''
emu = Emu.HODemulator()
lhcube = emu.read_HODLHD(HODrange=HODrange, method=method, samples=samples)
plks = emu.read_NeutObvs(obvs,
mneut,
nreal,
nzbin,
seed_hod,
Nmesh=Nmesh,
rsd=rsd)
fig = plt.figure()
sub = fig.add_subplot(111)
for i in range(plks.shape[0]):
sub.plot(emu.k, plks[i, :])
# x-axis
sub.set_xscale('log')
sub.set_xlim([0.01, 0.5])
sub.set_xlabel('k', fontsize=25)
# y-axis
sub.set_yscale('log')
sub.set_ylabel('$P(k)$', fontsize=25)
f = ''.join([UT.fig_dir(), 'tests/test.HODemulator.', obvs, '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def Subvolume_Analytic(N_sub, ratio=False):
''' Test the 2PCF estimates from MultiDark subvolume versus the
analytic 2PCF for the entire MultiDark volume
Parameters
----------
N_sub : (int)
Number of subvolumes to sample
'''
prettyplot()
pretty_colors = prettycolors()
pickle_file = ''.join([
'/export/bbq2/hahn/ccppabc/dump/',
'xi_subvolume_test',
'.Nsub', str(N_sub),
'.p'])
fig = plt.figure(1)
sub = fig.add_subplot(111)
xi_bin = xi_binedges()
if os.path.isfile(pickle_file):
data_dump = pickle.load(open(pickle_file, 'rb'))
full_xi = data_dump['full_xi']
else:
# Entire MultiDark Volume (Analytic xi)
model = PrebuiltHodModelFactory('zheng07', threshold=-21)
halocat = CachedHaloCatalog(simname = 'multidark', redshift = 0, halo_finder = 'rockstar')
model.populate_mock(halocat)
pos = three_dim_pos_bundle(model.mock.galaxy_table, 'x', 'y', 'z')
# while the estimator claims to be Landy-Szalay, I highly suspect it
# actually uses Landy-Szalay since DR pairs cannot be calculated from
# analytic randoms
full_xi = tpcf(pos, xi_bin, period=model.mock.Lbox, max_sample_size=int(2e5), estimator='Landy-Szalay', num_threads=1)
data_dump = {}
data_dump['full_xi'] = full_xi
if not ratio:
sub.plot(0.5*(xi_bin[:-1]+xi_bin[1:]), full_xi, lw=2, ls='-', c='k', label=r'Analytic $\xi$ Entire Volume')
if not os.path.isfile(pickle_file):
# MultiDark SubVolume (precomputed RR pairs)
sub_model = PrebuiltHodModelFactory('zheng07', threshold=-21)
sub_model.new_haloprop_func_dict = {'sim_subvol': util.mk_id_column}
sub_halocat = CachedHaloCatalog(simname = 'multidark', redshift = 0, halo_finder = 'rockstar')
RR = data_RR()
randoms = data_random()
NR = len(randoms)
for method in ['Landy-Szalay', 'Natural']:
if method == 'Landy-Szalay':
iii = 3
elif method == 'Natural':
iii = 5
if not os.path.isfile(pickle_file):
sub_xis_list = []
sub_xis = np.zeros(len(full_xi))
for ii in range(1,N_sub+1):
# randomly sample one of the subvolumes
rint = ii #np.random.randint(1, 125)
simsubvol = lambda x: util.mask_func(x, rint)
sub_model.populate_mock(sub_halocat, masking_function=simsubvol, enforce_PBC=False)
pos = three_dim_pos_bundle(sub_model.mock.galaxy_table, 'x', 'y', 'z')
xi, yi , zi = util.random_shifter(rint)
temp_randoms = randoms.copy()
temp_randoms[:,0] += xi
temp_randoms[:,1] += yi
temp_randoms[:,2] += zi
rmax = xi_bin.max()
approx_cell1_size = [rmax , rmax , rmax]
approx_cellran_size = [rmax , rmax , rmax]
sub_xi = tpcf(
pos, xi_bin, pos,
randoms=temp_randoms,
period = None,
max_sample_size=int(1e5),
estimator=method,
approx_cell1_size = approx_cell1_size,
approx_cellran_size = approx_cellran_size,
RR_precomputed=RR,
NR_precomputed=NR)
label = None
if ii == N_sub - 1:
label = 'Subvolumes'
#if not ratio:
# sub.plot(0.5*(xi_bin[:-1]+xi_bin[1:]), sub_xi, lw=0.5, ls='--', c=pretty_colors[iii])
sub_xis += sub_xi
sub_xis_list.append(sub_xi)
sub_xi_avg = sub_xis/np.float(N_sub)
data_dump[method] = {}
data_dump[method]['sub_xi_avg'] = sub_xi_avg
data_dump[method]['sub_xis_list'] = sub_xis_list
else:
sub_xis_list = data_dump[method]['sub_xis_list']
sub_xi_avg = data_dump[method]['sub_xi_avg']
if not ratio:
sub.plot(0.5*(xi_bin[:-1]+xi_bin[1:]), sub_xi_avg,
lw=2, ls='--', c=pretty_colors[iii], label='Subvolume '+method)
else:
sub.plot(0.5*(xi_bin[:-1]+xi_bin[1:]), sub_xi_avg/full_xi,
lw=2, ls='--', c=pretty_colors[iii], label='Subvolume '+method)
if not os.path.isfile(pickle_file):
pickle.dump(data_dump, open(pickle_file, 'wb'))
sub.set_xlim([0.1, 50.])
sub.set_xlabel('r', fontsize=30)
sub.set_xscale('log')
if not ratio:
sub.set_ylabel(r"$\xi \mathtt{(r)}$", fontsize=25)
sub.set_yscale('log')
else:
sub.set_ylabel(r"$\overline{\xi^\mathtt{sub}}/\xi^\mathtt{all}$", fontsize=25)
sub.legend(loc='lower left')
if ratio:
fig_file = ''.join([util.fig_dir(), 'test_xi_subvolume_analytic.Nsub', str(N_sub), '.ratio.png'])
else:
fig_file = ''.join([util.fig_dir(), 'test_xi_subvolume_analytic.Nsub', str(N_sub), '.png'])
fig.savefig(fig_file, bbox_inches='tight', dpi=100)
plt.close()
return None
def plot_occupations_total(obs = "gmf", clotter = True):
npts = 2e3
mass = np.logspace(10, 14, npts)
pretty_color = prettycolors()
if clotter == False:
print "running compute_occupation_prediction first"
else:
ylabel = r'$\langle N_{\rm g}(M_{h})\rangle$'
xlabel = r'$M_{h}\; [h^{-1} \; M_{\odot}]$'
totdec_18 = np.loadtxt("results/ntot_"+obs+"_dec_18.0.dat")
totdec_19 = np.loadtxt("results/ntot_"+obs+"_dec_19.0.dat")
totdec_20 = np.loadtxt("results/ntot_"+obs+"_dec_20.0.dat")
tothod_18 = np.loadtxt("results/ntot_"+obs+"_hod_18.0.dat")
tothod_19 = np.loadtxt("results/ntot_"+obs+"_hod_19.0.dat")
tothod_20 = np.loadtxt("results/ntot_"+obs+"_hod_20.0.dat")
fig = plt.figure(1, figsize=(15,5))
gs = gridspec.GridSpec(1, 3)# height_ratios=[2.5, 1], width_ratios=[1,1])
pretty_colors=prettycolors()
##### MR18 DECvsHOD ######
ax = plt.subplot(gs[0])
a_dec, b_dec, c_dec = np.percentile(totdec_18, [16, 50, 84], axis=0)
a_hod, b_hod, c_hod = np.percentile(tothod_18, [16, 50, 84], axis=0)
ax.plot(mass, b_dec, color=pretty_colors[1], linewidth=2.)
ax.plot(mass, b_hod, color=pretty_colors[7], linewidth=2.)
ax.fill_between(mass, a_dec, c_dec, color=pretty_colors[1], alpha=0.3, edgecolor=pretty_colors[1] )
ax.fill_between(mass, a_hod, c_hod, color=pretty_colors[7], alpha=0.3, edgecolor=pretty_colors[7])
blue_line = mlines.Line2D([], [], ls = '-', c = pretty_color[1], linewidth=2, label = '$Heaviside$ $AB$')
red_line = mlines.Line2D([], [], ls = '-', c = pretty_color[7], linewidth=2, label = '$Standard$ $HOD$')
import matplotlib as mpl
#mpl.rc('font',family='serif')
plt.legend(handles=[blue_line, red_line], frameon=False, loc= 2 , fontsize=15)
ax.set_ylabel(ylabel, fontsize=27)
ax.set_xlabel(xlabel, fontsize=27)
ax.set_yscale('log')
ax.set_xscale('log')
ax.set_xlim([10**10, 8. * 10**13])
ax.set_ylim([0.1,100])
ax.text(4 * 10**12. , 50, r'$M_{r}<-18$', fontsize=18)
##### Mr19 DECvsHOD ######
ax = plt.subplot(gs[1])
a_dec, b_dec, c_dec = np.percentile(totdec_19, [16, 50, 84], axis=0)
a_hod, b_hod, c_hod = np.percentile(tothod_19, [16, 50, 84], axis=0)
ax.plot(mass, b_dec, color=pretty_colors[1], linewidth=2.)
ax.plot(mass, b_hod, color=pretty_colors[7], linewidth=2.)
ax.fill_between(mass, a_dec, c_dec, color=pretty_colors[1], alpha=0.3, edgecolor=pretty_colors[1] )
ax.fill_between(mass, a_hod, c_hod, color=pretty_colors[7], alpha=0.3, edgecolor=pretty_colors[7])
blue_line = mlines.Line2D([], [], ls = '-', c = pretty_color[1], linewidth=2, label = '$Heaviside$ $AB$ ')
red_line = mlines.Line2D([], [], ls = '-', c = pretty_color[7], linewidth=2, label = '$Standard$ $HOD$')
import matplotlib as mpl
#mpl.rc('font',family='serif')
#plt.legend(handles=[blue_line, red_line], frameon=False, loc= 2 , fontsize=15)
ax.set_xlabel(xlabel, fontsize=27)
ax.set_yscale('log')
ax.set_xscale('log')
ax.set_xlim([10**10, 8. * 10**13])
ax.set_ylim([0.1,100])
ax.text(4 * 10**12. , 50, r'$M_{r}<-19$', fontsize=18)
ax.set_yticklabels([])
##### Mr20 DECvsHOD #######
ax = plt.subplot(gs[2])
a_dec, b_dec, c_dec = np.percentile(totdec_20, [16, 50, 84], axis=0)
a_hod, b_hod, c_hod = np.percentile(tothod_20, [16, 50, 84], axis=0)
ax.plot(mass, b_dec, color=pretty_colors[1], linewidth=2.)
ax.plot(mass, b_hod, color=pretty_colors[7], linewidth=2.)
ax.fill_between(mass, a_dec, c_dec, color=pretty_colors[1], alpha=0.3, edgecolor=pretty_colors[1] )
ax.fill_between(mass, a_hod, c_hod, color=pretty_colors[7], alpha=0.3, edgecolor=pretty_colors[7])
blue_line = mlines.Line2D([], [], ls = '-', c = pretty_color[1], linewidth=2, label = '$Heaviside$ $AB$')
red_line = mlines.Line2D([], [], ls = '-', c = pretty_color[7], linewidth=2, label = '$Standard$ $HOD$')
import matplotlib as mpl
#mpl.rc('font',family='serif')
#plt.legend(handles=[blue_line, red_line], frameon=False, loc= 2 , fontsize=15)
ax.set_xlabel(xlabel, fontsize=27)
ax.set_yscale('log')
ax.set_xscale('log')
ax.set_xlim([10**10, 8. * 10**13])
ax.set_ylim([0.1,100])
ax.text(4 * 10**12. , 50, r'$M_{r}<-20$', fontsize=18)
ax.set_yticklabels([])
fig.subplots_adjust(wspace=0.0, hspace=0.0)
fig_name = ''.join([ut.fig_dir(),
'paper',
'.abvshod.',
obs,
'.pdf'])
fig.savefig(fig_name, bbox_inches='tight')
plt.close()
return None
def logSFR_initiate(SHsnaps, indices, theta_sfh=None, theta_sfms=None, testing=False):
''' initiate log SFR function for Evolver.Evolve() method
'''
nsnap0 = SHsnaps['metadata']['nsnap0']
mu_sfr0 = SFR_sfms(SHsnaps['m.star0'][indices], UT.z_nsnap(SHsnaps['nsnap_start'][indices]), theta_sfms)
if theta_sfh['name'] == 'constant_offset':
# constant d_logSFR
F_sfr = _logSFR_dSFR
sfr_kwargs = {'dSFR': SHsnaps['sfr0'][indices] - mu_sfr0, # offset
'theta_sfms': theta_sfms}
elif theta_sfh['name'] == 'corr_constant_offset':
# constant d_logSFR (assigned based on halo accretion rate)
#dSFR0 = SHsnaps['sfr0'][indices] - mu_sfr
# now rank order it based on halo accretion rate
dMhalo = SHsnaps['halo.m'][indices] - SHsnaps['halo.m0'][indices]
m_kind = SHsnaps[theta_sfh['m.kind']+'0'][indices] # how do we bin the SFRs?
dm_kind = theta_sfh['dm.kind'] # bins to rank order
# scatter from noise -- subtract intrinsic assembly bias scatter from sig_SFMS
sig_noise = np.sqrt(0.3**2 - theta_sfh['sig_abias']**2)
# slow and inefficient
dsfr = np.repeat(-999., len(dMhalo))
for nn in np.arange(1, 21):
started = np.where(SHsnaps['nsnap_start'][indices] == nn)[0]
m_kind_bin = np.arange(m_kind[started].min(), m_kind[started].max()+dm_kind, dm_kind)
ibin = np.digitize(m_kind[started], m_kind_bin)
for i in np.unique(ibin):
inbin = np.where(ibin == i)
isort = np.argsort(dMhalo[started[inbin]])
dsfr[started[inbin[0][isort]]] = \
np.sort(np.random.randn(len(inbin[0]))) * theta_sfh['sig_abias']
assert dsfr.min() != -999.
if sig_noise > 0.:
dsfr += sig_noise * np.random.randn(len(dMhalo))
# correct initial SFR to match long-term assembly bias
SHsnaps['sfr0'][indices] = mu_sfr0 + dsfr
F_sfr = _logSFR_dSFR
sfr_kwargs = {'dSFR': dsfr, 'theta_sfms': theta_sfms}
elif theta_sfh['name'] == 'random_step':
# completely random
# amplitude is sampled randomly from a Gaussian with sig_logSFR = 0.3
# time steps are sampled randomly from a unifrom distribution [dt_min, dt_max]
if 'dt_min' not in theta_sfh:
raise ValueError
if 'dt_max' not in theta_sfh:
raise ValueError
# Random step function duty cycle
del_t_max = UT.t_nsnap(1) - UT.t_nsnap(nsnap0) #'nsnap_start'][indices].max())
# the range of the steps
tshift_min = theta_sfh['dt_min']
tshift_max = theta_sfh['dt_max']
# get the times when the amplitude changes
n_col = int(np.ceil(del_t_max/tshift_min))+1 # number of columns
n_gal = len(indices)
tshift = np.zeros((n_gal, n_col))
tshift[:,1:] = np.random.uniform(tshift_min, tshift_max, size=(n_gal, n_col-1))
tsteps = np.cumsum(tshift , axis=1) + np.tile(UT.t_nsnap(SHsnaps['nsnap_start'][indices]), (n_col, 1)).T
del tshift
# make sure everything evolves properly until the end
assert tsteps[range(n_gal), n_col-1].min() > UT.t_nsnap(1)
dlogSFR_amp = np.random.randn(n_gal, n_col) * theta_sfh['sigma']
dlogSFR_amp[:,0] = SHsnaps['sfr0'][indices] - mu_sfr0
F_sfr = _logSFR_dSFR_tsteps
sfr_kwargs = {'dlogSFR_amp': dlogSFR_amp, 'tsteps': tsteps,'theta_sfms': theta_sfms}
elif theta_sfh['name'] == 'random_step_fluct':
# completely random amplitude that is sampled from a Gaussian with sig_logSFR
# EXCEPT each adjacent timesteps have alternating sign amplitudes
# time steps have width specified tduty
dt_tot= UT.t_nsnap(1) - UT.t_nsnap(nsnap0) # total cosmic time of evolution
tduty = theta_sfh['tduty'] # dutycycle time
# get the times when the amplitude changes
n_col = int(np.ceil(dt_tot / tduty))+2 # number of columns
n_gal = len(indices) # number of galaxies
tshift = np.tile(tduty, (n_gal, n_col))
tshift[:,0] = 0.
tshift[:,1] = np.random.uniform(0., tduty, n_gal)
tsteps = np.cumsum(tshift , axis=1) + np.tile(UT.t_nsnap(SHsnaps['nsnap_start'][indices]), (n_col, 1)).T
del tshift
# make sure everything evolves properly until the end
assert tsteps[range(n_gal), n_col-1].min() > UT.t_nsnap(1)
# dlogSFR absolute amplitue
dlogSFR_amp = np.abs(np.random.randn(n_gal, n_col)) * theta_sfh['sigma']
dlogSFR_amp[:,0] = SHsnaps['sfr0'][indices] - mu_sfr0 # dlogSFR at nsnap_start
# now make every other time step has fluctuating signs!
pos = dlogSFR_amp[:,0] >= 0.
neg = ~pos
fluct = np.ones(n_col-1)
fluct[2 * np.arange(int(np.ceil(float(n_col-1)/2.)))] *= -1.
dlogSFR_amp[pos,1:] *= fluct
dlogSFR_amp[neg,1:] *= -1. * fluct
# testing the distribution of delta SFR
if testing:
for i in range(n_col):
print('std of dlogSFR amp = %f' % np.std(dlogSFR_amp))
plt.hist(dlogSFR_amp[:,i], range=(-1., 1.),
linewidth=2, histtype='step', density=True, label='Step '+str(i))
plt.legend()
plt.xlabel(r'$\Delta\log\,\mathrm{SFR}$', fontsize=20)
plt.xlim([-1., 1.])
plt.savefig(''.join([UT.fig_dir(), 'random_step_fluct_test.png']), bbox_inches='tight')
F_sfr = _logSFR_dSFR_tsteps
sfr_kwargs = {'dlogSFR_amp': dlogSFR_amp, 'tsteps': tsteps,'theta_sfms': theta_sfms}
elif theta_sfh['name'] == 'random_step_abias_dt':
# SFH where the amplitude w.r.t. the SFS is correlated with halo mass growth over
# tdyn(t). The amplitude is sampled at every time step specified by `tduty`
dt_tot= UT.t_nsnap(1) - UT.t_nsnap(nsnap0) # total cosmic time of evolution
tduty = theta_sfh['tduty'] # dutycycle time
sigma_int = np.sqrt(theta_sfh['sigma_tot']**2 - theta_sfh['sigma_corr']**2) # intrinsic sigma
# get the times when the amplitude changes
n_col = int(np.ceil(dt_tot / tduty))+2 # number of columns
n_gal = len(indices) # number of galaxies
tshift = np.tile(tduty, (n_gal, n_col))
tshift[:,0] = 0.
tshift[:,1] = np.random.uniform(0., tduty, n_gal)
tsteps = np.cumsum(tshift , axis=1) + \
np.tile(UT.t_nsnap(SHsnaps['nsnap_start'][indices]), (n_col, 1)).T
del tshift
# M_h of the galaxies throughout the snapshots
Mh_snaps = np.zeros((n_gal, nsnap0+9), dtype=np.float32)
Mh_snaps[:,0] = SHsnaps['halo.m'][indices]
Mm_snaps = np.zeros((n_gal, nsnap0+9), dtype=np.float32)
Mm_snaps[:,0] = SHsnaps['m.max'][indices]
for isnap in range(2, nsnap0+10):
Mh_snaps[:,isnap-1] = SHsnaps['halo.m.snap'+str(isnap)][indices]
Mm_snaps[:,isnap-1] = SHsnaps['m.max.snap'+str(isnap)][indices]
t_snaps = UT.t_nsnap(range(1, nsnap0+10))
# now we need to use these M_h to calculate the halo growth rates
# at every time step t_i over the range t_i - dt_dMh to t_i. This means
# we need to get M_h at both t_i and t_i-dt_dMh...
f_dMh = np.zeros(tsteps.shape, dtype=np.float32) # growth rate of halos over t_i and t_i - dt_Mh
#Mh_ts = np.zeros(tsteps.shape, dtype=np.float32)
Mm_ts = np.zeros(tsteps.shape, dtype=np.float32)
iii, iiii = 0, 0
for i_g in range(n_gal):
in_sim = (Mm_snaps[i_g,:] > 0.)
t_snaps_i = t_snaps[in_sim]
Mh_snaps_i = np.power(10., Mh_snaps[i_g, in_sim])
Mm_snaps_i = np.power(10., Mm_snaps[i_g, in_sim])
t_i = tsteps[i_g,:] # t_i
t_imdt = t_i - UT.tdyn_t(t_i) # t_i - tdyn
#t_imdt = tsteps[i_g,:] - theta_sfh['dt_dMh'] # t_i - dt_dMh
Mh_ti = np.interp(t_i, t_snaps_i[::-1], Mh_snaps_i[::-1])
Mh_timdt = np.interp(t_imdt, t_snaps_i[::-1], Mh_snaps_i[::-1])
f_dMh[i_g,:] = Mh_ti / Mh_timdt #1. - Mh_timdt /
#Mh_ts[i_g,:] = np.log10(Mh_ti)
Mm_ti = np.interp(t_i, t_snaps_i[::-1], Mm_snaps_i[::-1])
Mm_ts[i_g,:] = np.log10(Mm_ti)
if testing:
if (SHsnaps['nsnap_start'][indices][i_g] == 15) and (iii < 10):
fig = plt.figure(figsize=(8,4))
sub = fig.add_subplot(121)
sub.plot(t_snaps_i, Mh_snaps_i/Mh_snaps_i[0], c='k', ls='--')
sub.fill_between([t_imdt[0], t_i[0]], [0., 0.], [1.2, 1.2], color='C0')
sub.fill_between([t_imdt[-2], t_i[-2]], [0., 0.], [1.2, 1.2], color='C1')
sub.vlines(UT.t_nsnap(15), 0., 1.2)
sub.vlines(UT.t_nsnap(1), 0., 1.2)
sub.set_xlim([0., 14.])
sub.set_ylim([0., 1.2])
sub = fig.add_subplot(122)
sub.plot(t_i, f_dMh[i_g,:], c='k', ls='--')
sub.scatter([t_i[0]], [f_dMh[i_g,0]], c='C0')
sub.scatter([t_i[-2]], [f_dMh[i_g,-2]], c='C1')
sub.vlines(UT.t_nsnap(15), -0.2, 5.)
sub.vlines(UT.t_nsnap(1), -0.2, 5.)
sub.set_xlim([0., 14.])
sub.set_ylim([0., 5.])
fig.savefig(''.join([UT.fig_dir(), 'abiastest', str(iii), '.png']), bbox_inches='tight')
plt.close()
iii += 1
if in_sim[15]:
if Mh_snaps_i[0] < Mh_snaps_i[14]:
if iiii > 10: continue
fig = plt.figure(figsize=(8,4))
sub = fig.add_subplot(121)
sub.plot(t_snaps_i, Mh_snaps_i/Mh_snaps_i[0], c='k', ls='--')
sub.fill_between([t_imdt[0], t_i[0]], [0., 0.], [1.2, 1.2], color='C0')
sub.fill_between([t_imdt[-2], t_i[-2]], [0., 0.], [1.2, 1.2], color='C1')
sub.vlines(UT.t_nsnap(15), 0., 1.2)
sub.vlines(UT.t_nsnap(1), 0., 1.2)
sub.set_xlim([0., 14.])
sub.set_ylim([0., 1.2])
sub = fig.add_subplot(122)
sub.plot(t_i, f_dMh[i_g,:], c='k', ls='--')
sub.scatter([t_i[0]], [f_dMh[i_g,0]], c='C0')
sub.scatter([t_i[-2]], [f_dMh[i_g,-2]], c='C1')
sub.vlines(UT.t_nsnap(15), -0.2, 1.)
sub.vlines(UT.t_nsnap(1), -0.2, 1.)
sub.set_xlim([0., 14.])
sub.set_ylim([0., 5.])
sub.text(0.05, 0.05, r'$\log\,M_h='+str(round(np.log10(Mh_snaps_i[0]),2))+'$',
fontsize=15, ha='left', va='bottom', transform=sub.transAxes)
fig.savefig(''.join([UT.fig_dir(), 'weird_abiastest', str(iiii), '.png']), bbox_inches='tight')
plt.close()
iiii += 1
# calculate the d(log SFR) amplitude at t_steps
# at each t_step, for a given halo mass bin of 0.2 dex,
# rank order the SFRs and the halo growth rate
dlogSFR_amp = np.zeros(f_dMh.shape, dtype=np.float32)
dlogMh = 0.1
for ii in range(f_dMh.shape[1]):
f_dMh_i = f_dMh[:,ii]
Mh_ti = Mm_ts[:,ii]
mh_bins = np.arange(Mh_ti.min(), Mh_ti.max(), dlogMh)
#mh_bins = np.linspace(Mh_ti.min(), Mh_ti.max(), 100)
if testing:
fig = plt.figure(1, figsize=(4*np.int(np.ceil(float(len(mh_bins))/3.)+1.), 8))
for i_m in range(len(mh_bins)-1):
inbin = ((Mh_ti >= mh_bins[i_m]) & (Mh_ti < mh_bins[i_m]+dlogMh))
#inbin = ((Mh_ti >= mh_bins[i_m]) & (Mh_ti < mh_bins[i_m+1]))
n_bin = np.sum(inbin)
isort = np.argsort(-1. * f_dMh_i[inbin])
irank = np.zeros(n_bin)
irank[isort] = (np.arange(n_bin) + 0.5)/np.float(n_bin)
# i_rank = 1/2 (1 - erf(x/sqrt(2)))
# x = (SFR - avg_SFR)/sigma_SFR
# dSFR = sigma_SFR * sqrt(2) * erfinv(1 - 2 i_rank)
dlogSFR_amp[inbin, ii] = \
theta_sfh['sigma_corr'] * 1.41421356 * erfinv(1. - 2. * irank)
if testing:
sub = fig.add_subplot(3, np.int(np.ceil(np.float(len(mh_bins))/3.)+1), i_m+1)
sub.scatter(f_dMh_i[inbin], 0.3 * np.random.randn(n_bin), c='k', s=2)
sub.scatter(f_dMh_i[inbin], dlogSFR_amp[inbin, ii] + sigma_int * np.random.randn(n_bin), c='r', s=2, lw=0)
sub.set_xlim([-0.5, 1.])
sub.set_ylim([-1., 1.])
if testing:
plt.savefig(''.join([UT.fig_dir(), 'random_step_abias_dt_test', str(ii), '.png']), bbox_inches='tight')
plt.close()
fig = plt.figure(figsize=(8,4))
sub = fig.add_subplot(121)
DFM.hist2d(Mh_ti, f_dMh_i, levels=[0.68, 0.95], range=[[10., 14.],[0., 10.]], color='k',
plot_datapoints=False, fill_contours=False, plot_density=True, ax=sub)
sub.set_xlim([10., 14.])
sub = fig.add_subplot(122)
DFM.hist2d(Mh_ti, dlogSFR_amp[:,ii], levels=[0.68, 0.95], range=[[10., 14.],[-1., 1.]], color='k',
plot_datapoints=False, fill_contours=False, plot_density=True, ax=sub)
sub.set_xlim([10., 14.])
fig.savefig(''.join([UT.fig_dir(), 'abias_fdMh', str(ii), '.png']), bbox_inches='tight')
plt.close()
del f_dMh
del Mm_ts
# add in intrinsic scatter
dlogSFR_int = np.random.randn(n_gal, n_col) * sigma_int
dlogSFR_amp += dlogSFR_int
SHsnaps['sfr0'][indices] = mu_sfr0 + dlogSFR_amp[:,0] # change SFR_0 so that it's assembly biased as well
F_sfr = _logSFR_dSFR_tsteps
sfr_kwargs = {'dlogSFR_amp': dlogSFR_amp, 'tsteps': tsteps,'theta_sfms': theta_sfms}
else:
raise NotImplementedError
return F_sfr, sfr_kwargs
def plot_occupations_decs(obs = "gmf", galtype = "ncen", clotter = True):
npts = 2e3
mass = np.logspace(10, 14, npts)
pretty_color = prettycolors()
if clotter == False:
print "running compute_occupation_prediction first"
else:
if galtype == "ncen":
ylabel = r'$\langle N_{\rm cen}(M_{h})\rangle$'
xlabel = r'$M_{h}\; [h^{-1} \; M_{\odot}]$'
if galtype == "nsat":
ylabel = r'$\langle N_{\rm sat}(M_{h})\rangle$'
xlabel = r'$M_{h}\; [h^{-1} \; M_{\odot}]$'
old_18 = np.loadtxt("results/"+galtype+"_old_"+obs+"_dec_18.0.dat")
young_18 = np.loadtxt("results/"+galtype+"_young_"+obs+"_dec_18.0.dat")
old_19 = np.loadtxt("results/"+galtype+"_old_"+obs+"_dec_19.0.dat")
young_19 = np.loadtxt("results/"+galtype+"_young_"+obs+"_dec_19.0.dat")
old_20 = np.loadtxt("results/"+galtype+"_old_"+obs+"_dec_20.0.dat")
young_20 = np.loadtxt("results/"+galtype+"_young_"+obs+"_dec_20.0.dat")
fig = plt.figure(1, figsize=(15,5))
gs = gridspec.GridSpec(1, 3)# height_ratios=[2.5, 1], width_ratios=[1,1])
pretty_colors=prettycolors()
##### MR18 DEC ######
ax = plt.subplot(gs[0])
a_old, b_old, c_old = np.percentile(old_18, [16, 50, 84], axis=0)
a_young, b_young, c_young = np.percentile(young_18, [16, 50, 84], axis=0)
ax.plot(mass, b_young, color=pretty_colors[1], linewidth=2.)
ax.plot(mass, b_old, color=pretty_colors[4], linewidth=2.)
ax.fill_between(mass, a_young, c_young, color=pretty_colors[1], alpha=0.5, edgecolor=pretty_colors[1] )
ax.fill_between(mass, a_old, c_old, color=pretty_colors[4], alpha=0.5, edgecolor=pretty_colors[4])
blue_line = mlines.Line2D([], [], ls = '-', c = pretty_color[1], linewidth=2, label = '$type-1$ $halos$')
red_line = mlines.Line2D([], [], ls = '-', c = pretty_color[4], linewidth=2, label = '$type-2$ $halos$')
import matplotlib as mpl
#mpl.rc('font',family='serif')
plt.legend(handles=[blue_line, red_line], frameon=False, loc= 2 , fontsize=15)
ax.set_ylabel(ylabel, fontsize=27)
ax.set_xlabel(xlabel, fontsize=27)
ax.set_yscale('log')
ax.set_xscale('log')
ax.set_xlim([10**10, 8. * 10**13])
if galtype == "ncen":
ax.set_ylim([0.1, 2])
ax.text(4 * 10**12. , 1.3, r'$M_{r}<-18$', fontsize=18)
if galtype == "nsat":
ax.set_ylim([0.1,100])
ax.text(4 * 10**12. , 50, r'$M_{r}<-18$', fontsize=18)
##### Mr19 DEC ######
ax = plt.subplot(gs[1])
a_old, b_old, c_old = np.percentile(old_19, [16, 50, 84], axis=0)
a_young, b_young, c_young = np.percentile(young_19, [16, 50, 84], axis=0)
ax.plot(mass, b_young, color=pretty_colors[1], linewidth=2.)
ax.plot(mass, b_old, color=pretty_colors[4], linewidth=2.)
ax.fill_between(mass, a_young, c_young, color=pretty_colors[1], alpha=0.5, edgecolor=pretty_colors[1])
ax.fill_between(mass, a_old, c_old, color=pretty_colors[4], alpha=0.5, edgecolor=pretty_colors[4])
blue_line = mlines.Line2D([], [], ls = '-', c = pretty_color[3], linewidth=2, label = 'high-concentration halos')
red_line = mlines.Line2D([], [], ls = '-', c = pretty_color[4], linewidth=2, label = 'low-concentration halos')
ax.set_yscale('log')
ax.set_xscale('log')
ax.set_yticklabels([])
ax.set_xlabel(xlabel, fontsize=27)
ax.set_xlim([10**10, 8.*10**13])
if galtype == "ncen":
ax.set_ylim([0.1, 2])
ax.text(4 * 10**12. , 1.3, r'$M_{r}<-19$', fontsize=18)
if galtype == "nsat":
ax.set_ylim([0.1,100])
ax.text(4 * 10**12. , 50, r'$M_{r}<-19$', fontsize=18)
##### Mr20 DEC #######
ax = plt.subplot(gs[2])
a_old, b_old, c_old = np.percentile(old_20, [16, 50, 84], axis=0)
a_young, b_young, c_young = np.percentile(young_20, [16, 50, 84], axis=0)
ax.plot(mass, b_young, color=pretty_colors[1], linewidth=2.)
ax.plot(mass, b_old, color=pretty_colors[4], linewidth=2.)
ax.fill_between(mass, a_young, c_young, color=pretty_colors[1], alpha=0.5, edgecolor=pretty_colors[1])
ax.fill_between(mass, a_old, c_old, color=pretty_colors[4], alpha=0.5, edgecolor=pretty_colors[4])
blue_line = mlines.Line2D([], [], ls = '-', c = pretty_color[3], linewidth=2, label = 'high-concentration halos')
red_line = mlines.Line2D([], [], ls = '-', c = pretty_color[4], linewidth=2, label = 'low-concentration halos')
ax.set_yscale('log')
ax.set_xscale('log')
ax.set_yticklabels([])
ax.set_xlabel(xlabel, fontsize=27)
ax.set_xlim([10**10, 8. * 10**13])
if galtype == "ncen":
ax.set_ylim([0.1, 2])
ax.text(4 * 10**12. , 1.3, r'$M_{r}<-20$', fontsize=18)
if galtype == "nsat":
ax.set_ylim([0.1,100])
ax.text(4 * 10**12. , 50, r'$M_{r}<-20$', fontsize=18)
fig.subplots_adjust(wspace=0.0, hspace=0.0)
fig_name = ''.join([ut.fig_dir(),
'paper',
'.abhod.',
galtype,
obs,
'.pdf'])
fig.savefig(fig_name, bbox_inches='tight')
plt.close()
return None
def div_K(div_func='kl'):
''' compare the KL or Renyi divergence for the following with their using different K values
- D( gauss(C_X) || gauss(C_X) )
- D( mock X || gauss(C_X))
- D( mock X || p(X) KDE)
- D( mock X || p(X) GMM)
- D( mock X || PI p(X^i_ICA) KDE)
- D( mock X || PI p(X^i_ICA) GMM)
'''
lbls = [r'$D( P(k) \parallel \mathcal{N}({\bf C}))$',
r'$D( P(k) \parallel p_\mathrm{KDE}(P(k)))$',
r'$D( P(k) \parallel p_\mathrm{GMM}(P(k)))$',
r'$D( P(k) \parallel \prod_i p_\mathrm{KDE}(P(k)_i^\mathrm{ICA}))$',
r'$D( P(k) \parallel \prod_i p_\mathrm{GMM}(P(k)_i^\mathrm{ICA}))$']
fig = plt.figure(figsize=(20,4))
for i_obv, obv in enumerate(['pk.ngc', 'gmf']):
if obv == 'pk.ngc':
Nref = 2000
if div_func == 'kl': hranges = [[-0.5, 0.5], [-0.5, 7.], [-0.5, 0.5], [-0.5, 0.5], [-0.5, 0.5]]##7.]
else: hranges = [[-0.5, 0.5] for i in range(5)]
Ks = [5, 10, 15]
elif obv == 'gmf':
Nref = 10000
hranges = [[-0.1, 0.4], [-0.1, 0.4], [-0.1, 0.4], [-0.1, 0.4], [-0.1, 0.4]]##7.]
Ks = [10]
for K in Ks:
fs = ['pX_gauss.K'+str(K), 'pX_scottKDE.K'+str(K), 'pX_GMM.K'+str(K)+'.ncomp30',
'pXi_ICA_scottKDE.K'+str(K), 'pXi_ICA_GMM.K'+str(K)+'.ncomp30']
divs, divs_ref = [], []
for f in fs:
f_div = ''.join([UT.dat_dir(), 'diverg.', obv, '.', f, '.Nref', str(Nref), '.',
div_func, '.dat'])
try:
div = np.loadtxt(f_div)
except IOError:
print f_div
continue
divs.append(div)
nbins = 50
bkgd = fig.add_subplot(2,1,i_obv+1, frameon=False)
for i_div, div, lbl in zip(range(len(fs)), divs, lbls):
sub = fig.add_subplot(2,5,len(fs)*i_obv+i_div+1)
y_max = 0.
hh = np.histogram(div, normed=True, range=hranges[i_div], bins=nbins)
bp = UT.bar_plot(*hh)
sub.fill_between(bp[0], np.zeros(len(bp[0])), bp[1], edgecolor='none')
y_max = max(y_max, bp[1].max())
sub.set_xlim(hranges[i_div])
sub.set_ylim([0., y_max*1.4])
if i_obv == 0:
sub.set_title(lbl)
if div_func == 'kl':
bkgd.set_xlabel(r'KL divergence', fontsize=20, labelpad=20)
elif div_func == 'renyi0.5':
bkgd.set_xlabel(r'R\'enyi-$\alpha$ divergence', fontsize=20, labelpad=20)
bkgd.set_xticklabels([])
bkgd.set_yticklabels([])
bkgd.tick_params(labelcolor='none', top='off', bottom='off', left='off', right='off')
fig.subplots_adjust(wspace=.15, hspace=0.3)
f_fig = ''.join([UT.fig_dir(), 'tests/Ktest_kNNdiverg.', div_func, '.png'])
fig.savefig(f_fig, bbox_inches='tight')
return None
def p_Xw_i(mock, ell=0, rebin=None, krange=None, ica=False, pca=False):
''' Test the probability distribution function of each X_w^i
component -- p(X_w^i). First compare the histograms of p(X_w^i)
with N(0,1). Then compare the gaussian KDE of p(X_w^i).
'''
Pk = NG.dataX(mock, ell=ell, rebin=rebin, krange=krange)
X, _ = NG.meansub(Pk)
str_w = 'W'
if ica and pca:
raise ValueError
if ica: # ICA components
# ICA components do not need to be Gaussian.
# in fact the whole point of the ICA transform
# is to capture the non-Gaussianity...
X_white, _ = NG.whiten(X) # whitened data
X_w, _ = NG.Ica(X_white)
str_w = 'ICA'
if pca: # PCA components
X_w, _ = NG.whiten(X, method='pca') # whitened data
str_w = 'PCA'
if not ica and not pca: # just whitened
X_w, W = NG.whiten(X) # whitened data
# p(X_w^i) histograms
fig = plt.figure(figsize=(15,7))
sub = fig.add_subplot(121)
for i_bin in range(X_w.shape[1]):
p_X_w, edges = np.histogram(X_w[:,i_bin], normed=True)
p_X_w_arr = UT.bar_plot(edges, p_X_w)
sub.plot(p_X_w_arr[0], p_X_w_arr[1])
x = np.arange(-5., 5.1, 0.1)
sub.plot(x, UT.gauss(x, 1., 0.), c='k', lw=3, label='$\mathcal{N}(0,1)$')
sub.set_xlim([-2.5, 2.5])
sub.set_xlabel('$\mathtt{X_{'+str_w+'}}$', fontsize=25)
sub.set_ylim([0., 0.6])
sub.set_ylabel('$\mathtt{P(X_{'+str_w+'})}$', fontsize=25)
sub.legend(loc='upper right')
# p(X_w^i) gaussian KDE fits
pdfs = NG.p_Xw_i(X_w, range(X_w.shape[1]), x=x)
sub = fig.add_subplot(122)
for i_bin in range(X_w.shape[1]):
sub.plot(x, pdfs[i_bin])
sub.plot(x, UT.gauss(x, 1., 0.), c='k', lw=3, label='$\mathcal{N}(0,1)$')
sub.set_xlim([-2.5, 2.5])
sub.set_xlabel('$\mathtt{X_{W}}$', fontsize=25)
sub.set_ylim([0., 0.6])
sub.set_ylabel('$\mathtt{P(X_{W})}$', fontsize=25)
sub.legend(loc='upper right')
str_ica, str_pca = '', ''
if ica:
str_ica = '.ICA'
if pca:
str_pca = '.PCA'
if rebin is None:
f = ''.join([UT.fig_dir(), 'tests/test.p_Xw_i', str_pca, str_ica, '.', mock, '.ell', str(ell), '.png'])
else:
f = ''.join([UT.fig_dir(), 'tests/test.p_Xw_i', str_pca, str_ica, '.', mock, '.ell', str(ell), '.rebin', str(rebin), '.png'])
fig.savefig(f, bbox_inches='tight')
return None
def Subvolume_FullvolumeCut(N_sub, ratio=False):
''' Test the 2PCF estimates from MultiDark subvolume versus the
2PCF for the entire MultiDark volume WITHOUT periodic boundary conditions
and actual pair counts, CUT into subvolumes of the same size *AFTER*
populate mock
Parameters
----------
N_sub : (int)
Number of subvolumes to sample
'''
prettyplot()
pretty_colors = prettycolors()
pickle_file = ''.join([
'/export/bbq2/hahn/ccppabc/dump/',
'xi_subvolume_fullvolume_cut_test',
'.Nsub', str(N_sub),
'.p'])
fig = plt.figure(1)
sub = fig.add_subplot(111)
xi_bin = xi_binedges()
# Entire MultiDark Volume (No Periodic Boundary Conditions)
model = PrebuiltHodModelFactory('zheng07', threshold=-21)
halocat = CachedHaloCatalog(simname = 'multidark', redshift = 0, halo_finder = 'rockstar')
sub_RR = data_RR(box='md_sub')
sub_randoms = data_random(box='md_sub')
sub_NR = len(sub_randoms)
rmax = xi_bin.max()
full_approx_cell1_size = [rmax , rmax , rmax]
full_approx_cellran_size = [rmax , rmax , rmax]
model.populate_mock(halocat, enforce_PBC=False)
subvol_id = util.mk_id_column(table=model.mock.galaxy_table)
full_pos = three_dim_pos_bundle(model.mock.galaxy_table, 'x', 'y', 'z')
# Full Volume
if os.path.isfile(pickle_file):
data_dump = pickle.load(open(pickle_file, 'rb'))
full_xi = data_dump['full_xi']
else:
model = PrebuiltHodModelFactory('zheng07', threshold=-21)
halocat = CachedHaloCatalog(simname = 'multidark', redshift = 0, halo_finder = 'rockstar')
full_randoms = data_random(box='md_all')
full_RR = data_RR(box='md_all')
full_NR = len(full_randoms)
rmax = xi_bin.max()
full_approx_cell1_size = [rmax , rmax , rmax]
full_approx_cellran_size = [rmax , rmax , rmax]
model.populate_mock(halocat, enforce_PBC=False)
full_pos = three_dim_pos_bundle(model.mock.galaxy_table, 'x', 'y', 'z')
full_xi = tpcf(
full_pos, xi_bin,
randoms=full_randoms, period=None,
do_auto=True,
do_cross=False,
num_threads=5,
max_sample_size=int(full_pos.shape[0]), estimator='Natural',
approx_cell1_size=full_approx_cell1_size,
approx_cellran_size=full_approx_cellran_size,
RR_precomputed = full_RR,
NR_precomputed = full_NR)
data_dump = {}
data_dump['full_xi'] = full_xi
if not ratio:
sub.plot(0.5*(xi_bin[:-1]+xi_bin[1:]), full_xi,
lw=2, ls='-', c='k', label=r'Full Volume')
if os.path.isfile(pickle_file):
fullcut_xi_list = data_dump['fullcut_xi']['fullcut_xi_list']
fullcut_xi_avg = data_dump['fullcut_xi']['fullcut_xi_avg']
else:
data_dump['fullcut_xi'] = {}
fullcut_xi_list = []
fullcut_xi_tot = np.zeros(len(xi_bin)-1)
for id in np.unique(subvol_id)[:N_sub]:
print 'Subvolume ', id
in_cut = np.where(subvol_id == id)
fullcut_pos = full_pos[in_cut]
fullcut_xi = tpcf(
fullcut_pos, xi_bin,
randoms=sub_randoms, period=None,
do_auto=True,
do_cross=False,
num_threads=5,
max_sample_size=int(fullcut_pos.shape[0]), estimator='Natural',
approx_cell1_size=full_approx_cell1_size,
approx_cellran_size=full_approx_cellran_size,
RR_precomputed=sub_RR,
NR_precomputed=sub_NR)
fullcut_xi_list.append(fullcut_xi)
fullcut_xi_tot += fullcut_xi
fullcut_xi_avg = fullcut_xi_tot / np.float(N_sub)
data_dump['fullcut_xi']['fullcut_xi_list']= fullcut_xi_list
data_dump['fullcut_xi']['fullcut_xi_avg']= fullcut_xi_avg
if not ratio:
sub.plot(0.5*(xi_bin[:-1]+xi_bin[1:]), fullcut_xi_avg,
lw=2, ls='-', c='k', label=r'Full Volume Cut Average')
else:
sub.plot(0.5*(xi_bin[:-1]+xi_bin[1:]), fullcut_xi_avg/full_xi,
lw=2, ls='-', c='k', label=r'Full Volume Cut Average')
if not os.path.isfile(pickle_file):
# MultiDark SubVolume (precomputed RR pairs)
sub_model = PrebuiltHodModelFactory('zheng07', threshold=-21)
sub_model.new_haloprop_func_dict = {'sim_subvol': util.mk_id_column}
sub_halocat = CachedHaloCatalog(simname = 'multidark', redshift = 0, halo_finder = 'rockstar')
sub_RR = data_RR(box='md_sub')
sub_randoms = data_random(box='md_sub')
sub_NR = len(sub_randoms)
sub_xis_list = []
sub_xis = np.zeros(len(full_xi))
for ii in range(1,N_sub):
print 'Subvolume ', ii
# randomly sample one of the subvolumes
rint = ii #np.random.randint(1, 125)
simsubvol = lambda x: util.mask_func(x, rint)
sub_model.populate_mock(sub_halocat, masking_function=simsubvol, enforce_PBC=False)
pos = three_dim_pos_bundle(sub_model.mock.galaxy_table, 'x', 'y', 'z')
xi, yi , zi = util.random_shifter(rint)
temp_randoms = sub_randoms.copy()
temp_randoms[:,0] += xi
temp_randoms[:,1] += yi
temp_randoms[:,2] += zi
rmax = xi_bin.max()
sub_approx_cell1_size = [rmax , rmax , rmax]
sub_approx_cellran_size = [rmax , rmax , rmax]
sub_xi = tpcf(
pos, xi_bin,
randoms=temp_randoms,
period = None,
do_auto=True,
do_cross=False,
num_threads=5,
max_sample_size=int(pos.shape[0]),
estimator='Natural',
approx_cell1_size = sub_approx_cell1_size,
approx_cellran_size = sub_approx_cellran_size,
RR_precomputed=sub_RR,
NR_precomputed=sub_NR)
label = None
if ii == N_sub - 1:
label = 'Subvolumes'
sub_xis += sub_xi
sub_xis_list.append(sub_xi)
sub_xi_avg = sub_xis/np.float(N_sub)
data_dump['Natural'] = {}
data_dump['Natural']['sub_xi_avg'] = sub_xi_avg
data_dump['Natural']['sub_xis_list'] = sub_xis_list
else:
sub_xis_list = data_dump['Natural']['sub_xis_list']
sub_xi_avg = data_dump['Natural']['sub_xi_avg']
if not os.path.isfile(pickle_file):
pickle.dump(data_dump, open(pickle_file, 'wb'))
if not ratio:
sub.plot(0.5*(xi_bin[:-1]+xi_bin[1:]), sub_xi_avg,
lw=2, ls='--', c=pretty_colors[3], label='Subvolume')
else:
sub.plot(0.5*(xi_bin[:-1]+xi_bin[1:]), sub_xi_avg/full_xi,
lw=2, ls='--', c=pretty_colors[3], label='Subvolume')
sub.set_xlim([0.1, 50.])
sub.set_xlabel('r', fontsize=30)
sub.set_xscale('log')
if not ratio:
sub.set_ylabel(r"$\xi \mathtt{(r)}$", fontsize=25)
sub.set_yscale('log')
else:
sub.set_ylabel(r"$\overline{\xi^\mathtt{sub}}/\xi^\mathtt{all}$", fontsize=25)
sub.legend(loc='lower left')
if ratio:
fig_file = ''.join([util.fig_dir(), 'test_xi_subvolume_fullvolume_cut.Nsub', str(N_sub), '.ratio.png'])
else:
fig_file = ''.join([util.fig_dir(), 'test_xi_subvolume_fullvolume_cut.Nsub', str(N_sub), '.png'])
fig.savefig(fig_file, bbox_inches='tight', dpi=100)
plt.close()
return None
