figure(1)
clf()
errorbar(xdata, ydata, xerr=xerr, yerr=yerr, marker='.', linestyle='none')
MC = pymc_linear_fit_withoutliers(xdata, ydata, data1err=xerr, data2err=yerr, return_MC=True)
MC.sample(50000,burn=10000,verbose=1)
mmean = MC.stats()['slope']['mean']
bmean = MC.stats()['intercept']['mean']
plot(linspace(0,300),linspace(0,300)*mmean+bmean,color='k',linewidth=2)
for m,b in zip(MC.trace('slope')[-100:],MC.trace('intercept')[-100:]):
plot(linspace(0,300),linspace(0,300)*m+b, color='k', alpha=0.05)
scatter(xdata[MC.badvals.value.astype('bool')],
ydata[MC.badvals.value.astype('bool')], color='r')
from agpy import pymc_plotting
import pymc
figure(2)
clf()
pymc_plotting.hist2d(MC,'slope','intercept',fignum=2,bins=50)
figure(3)
clf()
pymc.Matplot.plot(MC.slope,new=False)
figure(4)
clf()
pymc.Matplot.plot(MC.intercept,new=False)
lm = agpy.blackbody.fit_blackbody_lmfit(wavelength, flux, err=err,
blackbody_function=agpy.blackbody.modified_blackbody_wavelength, logscale=16,
guesses=(tguess,bguess,nguess),
wavelength_units='microns')
print lm.params
"""
Incomplete testing so far. Looks like luminosity comes out very wrong. Other
parameters seem to be recovered well enough.
"""
wavelength = np.array([20,70,160,250,350,500,850,1100])
frequency = 3e14 / wavelength
flux = agpy.blackbody.modified_blackbody(frequency, 15, beta=1.75,
normalize=False, logN=22, logscale=16)
err = 0.1 * flux
flux += np.random.randn(len(wavelength)) * err
tguess, bguess, nguess = 20.,2.,21.5
mc = agpy.blackbody.fit_blackbody_montecarlo(frequency, flux, err=err,
blackbody_function=agpy.blackbody.modified_blackbody,
logscale=16,
temperature_guess=tguess,
beta_guess=bguess,
scale_guess=nguess,
#guesses=(tguess,bguess,nguess),
scale_keyword='logscale',
return_MC=True)
from agpy import pymc_plotting
pymc_plotting.hist2d(mc, mc.temperature, mc.beta)
from measure_tau import savefig, savepath
#import astropy.io.fits as pyfits
from agpy import pymc_plotting
import pylab as pl
import numpy as np
pl.rc('font',size=24)
for fignum,abundance in enumerate((-8.5,-9.0)):
execfile('load_parameter_traces.py')
fig = pl.figure(1+fignum)
pl.clf()
ax1 = pymc_plotting.hist2d(mc_lognormal_freemach_traces, 'tauoneone_mu','sigma', bins=50,
clear=True, fignum=1, varslice=(2.5e5,None,None), colorbar=False,
axis=pl.subplot(231))
ax2 = pymc_plotting.hist2d(mc_lognormal_freemach_traces, 'tautwotwo_mu','sigma', bins=50,
clear=True, fignum=1, varslice=(2.5e5,None,None), colorbar=False,
axis=pl.subplot(232))
ax3 = pymc_plotting.hist2d(mc_lognormal_freemach_traces, 'tau_ratio','sigma', bins=50,
clear=True, fignum=1, varslice=(2.5e5,None,None), colorbar=False,
axis=pl.subplot(233))
ax4 = pymc_plotting.hist2d(mc_hopkins_freemach_traces, 'tauoneone_mu','sigma', bins=50,
clear=True, fignum=1, varslice=(2.5e5,None,None), colorbar=False,
axis=pl.subplot(234))
ax5 = pymc_plotting.hist2d(mc_hopkins_freemach_traces, 'tautwotwo_mu','sigma', bins=50,
clear=True, fignum=1, varslice=(2.5e5,None,None), colorbar=False,
axis=pl.subplot(235))
ax6 = pymc_plotting.hist2d(mc_hopkins_freemach_traces, 'tau_ratio','sigma', bins=50,
from measure_tau import trace_data_path, abundance, savefig, savepath
import astropy.io.fits as pyfits
from agpy import pymc_plotting
#import pylab as pl
lognormal_freemach_statstable = pyfits.getdata(trace_data_path+'lognormal_freemach_statstable_abundance%s.fits' % abundance)
mc_lognormal_freemach_traces = pyfits.getdata(trace_data_path+"mc_lognormal_freemach_traces.fits")
ax = pymc_plotting.hist2d(mc_lognormal_freemach_traces, 'tau_ratio','mach', bins=50,
clear=True, fignum=1, varslice=(None,None,None), colorbar=True)
ax.hlines([3.68],*ax.get_xlim(),linestyles=['--'], colors=['k'], label="H$_2$CO Lower-limit")
ax.hlines([6.58],*ax.get_xlim(),linestyles=[':'], colors=['k'], label="CO Lower-limit")
ax.legend(loc='best')
ax.set_ylabel(r"$\mathcal{M}_{3D}$")
ax.set_xlabel('$\\tau_{1-1}/\\tau_{2-2}$',fontsize=24)
savefig(savepath+"mach_vs_tauratio_lognormal_mcmc_contours.png")
from measure_tau import trace_data_path, savefig, savepath
import astropy.io.fits as pyfits
from agpy import pymc_plotting
import pylab as pl
pl.rc('font',size=30)
for abundance in (-8.5,-9.0):
for withfree in ('with','free'):
logtraces = pyfits.getdata(trace_data_path+"mc_lognormal_%smach_traces%s.fits" % (withfree,abundance))
ax = pymc_plotting.hist2d(logtraces, 'b','mach', bins=50,
clear=True, fignum=1, varslice=(2.5e5,None,None), colorbar=True)
ax.hlines([3.68],*ax.get_xlim(),linestyles=['--'], colors=['k'], label="H$_2$CO")
ax.hlines([6.58],*ax.get_xlim(),linestyles=[':'], colors=['k'], label="CO")
ax.legend(loc='best')
ax.set_ylabel(r"$\mathcal{M}_{3D}$")
ax.set_xlabel('$b$',fontsize=36)
savefig(savepath+"mach_vs_b_lognormal_mcmc_contours_%smach_abundance%s.png" % (withfree,abundance))
hoptraces = pyfits.getdata(trace_data_path+"mc_hopkins_%smach_traces_abundance%s.fits" % (withfree,abundance))
if 'mach' in hoptraces.names:
mach = 'mach'
else:
"""
Incomplete testing so far. Looks like luminosity comes out very wrong. Other
parameters seem to be recovered well enough.
"""
wavelength = np.array([20, 70, 160, 250, 350, 500, 850, 1100])
frequency = 3e14 / wavelength
flux = agpy.blackbody.modified_blackbody(frequency,
15,
beta=1.75,
normalize=False,
logN=22,
logscale=16)
err = 0.1 * flux
flux += np.random.randn(len(wavelength)) * err
tguess, bguess, nguess = 20., 2., 21.5
mc = agpy.blackbody.fit_blackbody_montecarlo(
frequency,
flux,
err=err,
blackbody_function=agpy.blackbody.modified_blackbody,
logscale=16,
temperature_guess=tguess,
beta_guess=bguess,
scale_guess=nguess,
#guesses=(tguess,bguess,nguess),
scale_keyword='logscale',
return_MC=True)
from agpy import pymc_plotting
pymc_plotting.hist2d(mc, mc.temperature, mc.beta)
