def main():
# cluster parameters
redshift = 0.7019
NH_1022pcm2 = 0.0137
Z_solar = 0.3
# for calculating distances, etc.
cosmology = mb.Cosmology(redshift)
# annuli object contains edges of annuli
annuli = mb.Annuli(getEdges(), cosmology)
# load each band, chopping outer radius
bands = []
for bandE in bandEs:
bands.append(loadBand(bandE))
# Data object represents annuli and bands
data = mb.Data(bands, annuli)
# flat metallicity profile
Z_cmpt = mb.CmptFlat('Z', annuli, defval=Z_solar)
# this is the modified beta model density described in Sanders+17 (used in McDonald+12)
# (single mode means only one beta model, as described in Vikhlinin+06)
ne_cmpt = mb.CmptVikhDensity('ne', annuli, mode='single')
# nfw mass model
nfw = mb.CmptMassNFW(annuli)
# hydrostatic model combining mass, density and metallicity
model = mb.ModelHydro(
annuli, nfw, ne_cmpt, Z_cmpt, NH_1022pcm2=NH_1022pcm2)
# get default parameters
pars = model.defPars()
# add parameter which allows variation of background with a
# Gaussian prior with sigma=0.1
pars['backscale'] = mb.ParamGaussian(1., prior_mu=1, prior_sigma=0.1)
# freeze metallicity at fixed value
pars['Z'].frozen = True
# change range of NFW concentration and r200
# (val is initial value, minval and maxval give allowed range)
pars['nfw_logconc'].val = N.log10(4)
pars['nfw_logconc'].minval = N.log10(1)
pars['nfw_logconc'].maxval = N.log10(10)
pars['nfw_r200_logMpc'].minval = -2
pars['nfw_r200_logMpc'].maxval = 1
# override default outer pressure
pars['Pout_logergpcm3'].minval = -16
pars['Pout_logergpcm3'].maxval = -8
# stop radii going beyond edge of data
pars['ne_logrc_1'].maxval = annuli.edges_logkpc[-2]
pars['ne_logr_s'].maxval = annuli.edges_logkpc[-2]
# some ranges of parameters to allow for the density model
pars['ne_gamma'].val = 3.
pars['ne_gamma'].frozen = True
pars['ne_logrc_1'].val = 2.
pars['ne_alpha'].val = 0.1
pars['ne_alpha'].maxval = 4
pars['ne_alpha'].minval = 0.
pars['ne_epsilon'].maxval = 10
# do fitting of data with model
fit = mb.Fit(pars, model, data)
# refreshThawed is required if frozen is changed after Fit is
# constructed before doFitting (it's not required here)
fit.refreshThawed()
fit.doFitting()
# construct MCMC object and do burn in
mcmc = mb.MCMC(fit, walkers=nwalkers, processes=nthreads)
mcmc.burnIn(nburn)
# save best fit
with open('%s_fit.pickle' % name, 'wb') as f:
pickle.dump(fit, f, -1)
# run mcmc proper
chainfilename = '%s_chain.hdf5' % name
mcmc.run(nlength)
mcmc.save(chainfilename)
# construct a set of physical median profiles from the chain and
# save
profs = mb.replayChainPhys(
chainfilename, model, pars, thin=10, confint=68.269)
mb.savePhysProfilesHDF5('%s_medians.hdf5' % name, profs)
mb.savePhysProfilesText('%s_medians.txt' % name, profs)
from joxsz_plots import traceplot, triangle, best_fit_prof, fitwithmod, comp_rad_profs, plot_rad_profs, comp_mass_prof, mass_plot, frac_gas_prof, frac_gas_plot from types import MethodType import emcee as mc from multiprocessing import Pool ################# ## Global and local variables mystep = 2. # constant sampling step in arcsec for SZ analysis (values higher than (1/3)*FWHM of the SZ beam are not recommended) m_e = 0.5109989 * 1e3 # electron rest mass (keV/c^2) sigma_T = 6.6524587158 * 1e-25 # Thomson cross section (cm^2) R_b = 5000. # Radial cluster extent (kpc), serves as upper bound for Compton y parameter integration # Cluster cosmology redshift = 0.888 cosmology = mb.Cosmology(redshift) cosmology.H0 = 67.32 # Hubble's constant (km/s/Mpc) cosmology.WM = 0.3158 # matter density cosmology.WV = 0.6842 # vacuum density # name for outputs name = 'joxsz' plotdir = './' # directory for the plots savedir = './' # directory for saved files # Uncertainty level ci = 95 # MCMC parameters nburn = 2000 # number of burn-in iterations nlength = 5000 # number of chain iterations (after burn-in)
def main():
indir = '.'
cosmology = mbproj2.Cosmology(0.1028)
# define annuli using 1st two columns in this file
annuli = mbproj2.loadAnnuli(
os.path.join(indir, 'sb_profile_1200_2500.dat.rebin'), cosmology)
# load count profiles for each band
bands = []
for emin, emax in ((0.5, 1.2), (1.2, 2.5), (2.5, 6.0)):
band = mbproj2.loadBand(
os.path.join(indir, 'sb_profile_%i_%i.dat.rebin' % (
emin*1000, emax*1000)),
emin, emax,
os.path.join(indir, 'ann_0_total_rmf.fits'),
os.path.join(indir, 'ann_0_total_arf.fits'))
band.backrates = N.loadtxt(
os.path.join(indir, 'sb_bgcomb_%i_%i.dat.rebin.norm' % (
emin*1000, emax*1000)))[:,2]
bands.append(band)
data = mbproj2.Data(bands, annuli)
# this is the mass component to use
nfw = mbproj2.CmptMassNFW(annuli)
# density component
ne_cmpt = mbproj2.CmptBinned(
'ne', annuli, log=True, defval=-2, minval=-6., maxval=1.)
# metallicity component
Z_cmpt = mbproj2.CmptFlat(
'Z', annuli, defval=N.log10(0.3), log=True, minval=-2., maxval=1.)
# combine components to create model
model = mbproj2.ModelHydro(
annuli, nfw, ne_cmpt, Z_cmpt, NH_1022pcm2=0.378)
# get default parameters
pars = model.defPars()
# example change in parameter
pars['Z'].val = N.log10(0.4)
pars['Z'].frozen = True
# estimate densities from surface brightness to avoid fits
mbproj2.estimateDensityProfile(model, data, pars)
# fit parameters to data
fit = mbproj2.Fit(pars, model, data)
fit.doFitting()
# now start MCMC using 4 processes
mcmc = mbproj2.MCMC(fit, walkers=200, processes=4)
mcmc.burnIn(1000)
mcmc.run(1000)
mcmc.save('chain.hdf5')
# compute physical profiles and save as medians
profs = mbproj2.replayChainPhys('chain.hdf5', model, pars)
mbproj2.savePhysProfilesHDF5('medians.hdf5', profs)