def test_GTC_truthLineStyles():
truthLineStyles = ['-', '-']
truths = ((4, .5, None, .1), (None, None, .3, 1))
pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
truths=truths,
truthLineStyles=truthLineStyles,
smoothingKernel=0)
def test_GTC_smoothingKernel():
if not HAS_SCIPY:
warnings.warn("Can't test smoothing without scipy. Skipping test.", UserWarning)
raise nose.SkipTest
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
smoothingKernel = 2)
def test_GTC_smoothingKernel():
if not HAS_SCIPY:
warnings.warn("Can't test smoothing without scipy. Skipping test.",
UserWarning)
raise nose.SkipTest
pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2], smoothingKernel=2)
def test_GTC_truthLabels_withTex():
truths = ((4, .5, None, .1), (None, None, .3, 1))
truthLabelsWithTex = ('the truth $f_0$', 'alternative truth $\\lambda$')
pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
truths=truths,
truthLabels=truthLabelsWithTex,
smoothingKernel=0)
def test_GTC_truthLabels_noTex():
truths = ((4, .5, None, .1), (None, None, .3, 1))
truthLabelsNoTex = ('the truth', 'alternative truth')
pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
truths=truths,
truthLabels=truthLabelsNoTex,
smoothingKernel=0)
def test_GTC_paramNames_withTex():
namesWithTex = [
'param name', '$B_\mathrm{\lambda}$', '$Q^a$', '$\\lambda$'
]
pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
paramNames=namesWithTex,
smoothingKernel=0)
def test_GTC_truthLineStyles():
truthLineStyles = ['-', '-']
truths = ((4, .5, None, .1),
(None, None, .3, 1))
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
truths = truths,
truthLineStyles = truthLineStyles,
smoothingKernel = 0)
def test_GTC_truthLabels_withTex():
truths = ((4, .5, None, .1),
(None, None, .3, 1))
truthLabelsWithTex = ('the truth $f_0$', 'alternative truth $\\lambda$')
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
truths = truths,
truthLabels = truthLabelsWithTex,
smoothingKernel = 0)
def test_GTC_truthLabels_noTex():
truths = ((4, .5, None, .1),
(None, None, .3, 1))
truthLabelsNoTex = ('the truth', 'alternative truth')
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
truths = truths,
truthLabels = truthLabelsNoTex,
smoothingKernel = 0)
def test_GTC_priors():
if not HAS_SCIPY:
warnings.warn("Can't test priors without scipy installed. Skipping test.", UserWarning)
raise nose.SkipTest
priors = (None, (2, 1), (.5, 2), ())
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
priors = priors,
smoothingKernel = 0)
def test_GTC_priors():
if not HAS_SCIPY:
warnings.warn(
"Can't test priors without scipy installed. Skipping test.",
UserWarning)
raise nose.SkipTest
priors = (None, (2, 1), (.5, 2), ())
pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
priors=priors,
smoothingKernel=0)
def test_GTC_pandas():
namesNoTex = ['param name', 'B_labmda', 'C', 'lambda']
if HAS_PANDAS:
samples1_pd = pd.DataFrame(SAMPLES_1, columns=namesNoTex)
samples2_pd = pd.DataFrame(SAMPLES_2, columns=namesNoTex)
else:
warnings.warn("Can't test pandas auto-name without pandas. Skipping test.", UserWarning)
raise nose.SkipTest
pygtc.plotGTC(chains=[samples1_pd,samples2_pd],
smoothingKernel = 0)
def test_GTC_pandas():
namesNoTex = ['param name', 'B_labmda', 'C', 'lambda']
if HAS_PANDAS:
samples1_pd = pd.DataFrame(SAMPLES_1, columns=namesNoTex)
samples2_pd = pd.DataFrame(SAMPLES_2, columns=namesNoTex)
else:
warnings.warn(
"Can't test pandas auto-name without pandas. Skipping test.",
UserWarning)
raise nose.SkipTest
pygtc.plotGTC(chains=[samples1_pd, samples2_pd], smoothingKernel=0)
def SamplePosterior(filebasename, showplots):
#print(np.exp(gp.kernel.vector))
#obsll= lambda *args: -lnlike(*args)
#results=op.minimize(obsll,[fixx1v,fixx2v,fixx3v,fixx4v])
#print(results.x)
#exit()
sampler = emcee.EnsembleSampler(nwalkers, ndim, obslnprob)
#p0 = [results.x + 1e-4 * np.random.randn(ndim)
# for i in range(nwalkers)]
p0 = [[fixx1v, fixx2v, fixx3v, fixx4v] + 1e-4 * np.random.randn(ndim)
for i in range(nwalkers)]
#print p0,obslnprob(p0[0],y)
print("Running burn-in")
p0, _, _ = sampler.run_mcmc(p0, nburnin)
print("Running production chain")
sampler.run_mcmc(p0, nproduction)
samples = sampler.flatchain
g4.create_dataset('posterior samples', data=samples)
labels = [r"$x_1$", r"$x_2$", r"$x_3$", r"$x_4$"]
#fig = triangle.corner(samples[:, 2:], labels=labels)
#print np.array(samples)
#fig = triangle.corner(np.array(samples), bins=100, labels=labels)
nburn = nburnin * nwalkers
fig = pygtc.plotGTC(chains=[samples[nburn:, :]])
fig.savefig("%s_Triangle.png" % (filebasename), dpi=150)
if showplots:
fig.show()
def test_GTC_priors():
if not HAS_SCIPY:
pytest.skip("Can't test priors without scipy installed.")
priors = (None, (2, 1), (.5, 2), ())
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
priors=priors,
smoothingKernel=0)
def test_GTC_mathTextFontSet():
namesWithTex = [
'param name', '$B_\\mathrm{\\lambda}$', '$Q^a$', '$\\lambda$'
]
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
paramNames=namesWithTex,
mathTextFontSet=None,
smoothingKernel=0)
def test_GTC_doOnly1dPlot():
input_chains = [
np.array([SAMPLES_1[:, 0]]).T,
np.array([SAMPLES_2[:, 0]]).T
]
return pygtc.plotGTC(chains=input_chains,
doOnly1dPlot=True,
smoothingKernel=0)
def plot_gtc(self, resObj, chains):
#Plot the triangle plot, and overlay the best fit values with dashed black lines (default)
#You can see that the least-squares fitter did a very nice job of getting the values right
#You can also see that there is some strange non-gaussian parameter space that the MCMC
#analysis maps out! This is kind of wierd, but not too worrisome. It is probably suggestive
#that more care is needed in choosing good options for the MCMC engine.
figGTC = pygtc.plotGTC(chains, truths=[resObj.lmfit_vals])
plt.show()
def test_GTC_pandas():
namesNoTex = ['param name', 'B_labmda', 'C', 'lambda']
if HAS_PANDAS:
samples1_pd = pd.DataFrame(SAMPLES_1, columns=namesNoTex)
samples2_pd = pd.DataFrame(SAMPLES_2, columns=namesNoTex)
else:
pytest.skip("Can't test pandas auto-name without pandas.")
return pygtc.plotGTC(chains=[samples1_pd, samples2_pd], smoothingKernel=0)
def test_GTC_figureSize():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
figureSize = 'APJ_page',
smoothingKernel = 0)
def test_GTC_do1dPlots():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
do1dPlots=False,
smoothingKernel=0)
def test_GTC_tickShifts():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
tickShifts=(0.2, 0.2),
smoothingKernel=0)
def test_GTC_panelSpacing():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
panelSpacing='loose',
smoothingKernel=0)
def test_GTC_plotDensity():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
plotDensity=True,
smoothingKernel=0)
def test_GTC_nBins():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
nBins=20,
smoothingKernel=0)
def test_GTC_nContourLevels():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
nContourLevels=3,
smoothingKernel=0)
def test_GTC_sigmaContourLevels():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
sigmaContourLevels = True,
smoothingKernel = 0)
def test_GTC_colorsOrder():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
colorsOrder = ['purples', 'yellows'],
smoothingKernel = 0)
def test_GTC_nBins():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
nBins = 20,
smoothingKernel = 0)
def test_GTC_tickShifts():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
tickShifts = (0.2, 0.2),
smoothingKernel = 0)
def test_GTC_labelRotation():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
labelRotation = (False, False),
smoothingKernel = 0)
def test_GTC_panelSpacing():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
panelSpacing = 'loose',
smoothingKernel = 0)
def test_GTC_chainLabels_withTex():
chainLabelsWithTex = ['data1 $\\lambda$', 'data 2']
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
chainLabels=chainLabelsWithTex,
smoothingKernel=0)
def test_GTC_nContourLevels():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
nContourLevels = 3,
smoothingKernel = 0)
def test_GTC_chainLabels_noTex():
chainLabelsNoTex = ['data1', 'data 2']
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
chainLabels = chainLabelsNoTex,
smoothingKernel = 0)
def test_GTC_paramNames_noTex():
namesNoTex = ['param name', 'B_labmda', 'C', 'lambda']
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
paramNames = namesNoTex,
smoothingKernel = 0)
def test_GTC_sigmaContourLevels():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
sigmaContourLevels=True,
smoothingKernel=0)
def test_GTC_paramNames_withTex():
namesWithTex = ['param name', '$B_\mathrm{\lambda}$', '$Q^a$', '$\\lambda$']
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
paramNames = namesWithTex,
smoothingKernel = 0)
def test_GTC_smoothingKernel():
if not HAS_SCIPY:
pytest.skip("Can't test smoothing without scipy.")
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2], smoothingKernel=2)
def test_GTC_chainLabels_withTex():
chainLabelsWithTex = ['data1 $\lambda$', 'data 2']
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
chainLabels = chainLabelsWithTex,
smoothingKernel = 0)
def test_GTC_figureSize():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
figureSize='APJ_page',
smoothingKernel=0)
####### CORNER PLOT ESTIMATES #######################################
import corner
fig = corner.corner(samples_plot, labels=[param1[0], param2[0]],
truths=[param1[1], param2[1]])
fig.savefig('corner_'+fileOut+'.png')
import pygtc
fig = pygtc.plotGTC(samples_plot, paramNames=[param1[0], param2[0]], truths=[ param1[1],
param2[1]] ,
figureSize='MNRAS_page' )
fig.savefig('pygtc_'+fileOut+'.png')
####### FINAL PARAMETER ESTIMATES #######################################
#
# plt.figure(1432)
#
# xl = np.array([0, 10])
# for m, b, lnf in samples[np.random.randint(len(samples), size=100)]:
# plt.plot(xl, m*xl+b, color="k", alpha=0.1)
# plt.plot(xl, m_true*xl+b_true, color="r", lw=2, alpha=0.8)
# plt.errorbar(x, y, yerr=yerr, fmt=".k", alpha=0.1)
def test_GTC_labelRotation():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
labelRotation=(False, False),
smoothingKernel=0)
plt.tight_layout()
if bayesian_model == 'multinest':
plt.savefig('multinest_best_fit_{}.png'.format(time_stamp))
if bayesian_model == 'emcee':
plt.savefig('emcee_best_fit_{}walkers_{}steps_{}.png'.format(nwalkers, nsteps, time_stamp))
res_flatchain_df = DataFrame(result.samples, columns=fit_info.fit_param_names)
if 'Rs' in fit_info.fit_param_names: res_flatchain_df['Rs'] = res_flatchain_df['Rs'] / R_sun
if 'Rp' in fit_info.fit_param_names: res_flatchain_df['Rp'] = res_flatchain_df['Rp'] / R_jup
if 'Mp' in fit_info.fit_param_names: res_flatchain_df['Mp'] = res_flatchain_df['Mp'] / M_jup
plt.rcParams['figure.dpi'] = 150
plt.rcParams['figure.figsize'] = (10,10)
fit_param_names = ['Rs [Rsun]', 'Mp [Mjup]', 'Rp [Rjup]', 'T [K]', 'log(haze)', 'logZ [FeH]', 'log(P_c) [mbar]', 'err_mod']
pygtc_out = plotGTC(res_flatchain_df.values,
weights=result.weights,
nContourLevels=3,
paramNames=fit_param_names,
# range=[0.99] * result.flatchain.shape[1],
# chainLabels=['Not Cheap', 'Cheap'],
customLabelFont={'size':10},
customTickFont={'size':7},
# customLegendFont={'size':30},
figureSize=12);
plt.savefig('multinest_gtc_{}.png'.format(time_stamp))
def test_GTC_colorsOrder():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
colorsOrder=['purples', 'yellows'],
smoothingKernel=0)
def test_GTC_chainLabels_noTex():
chainLabelsNoTex = ['data1', 'data 2']
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
chainLabels=chainLabelsNoTex,
smoothingKernel=0)
def test_GTC_do1dPlots():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
do1dPlots = False,
smoothingKernel = 0)
def test_GTC_plotDensity():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
plotDensity = True,
smoothingKernel = 0)
def test_GTC_bare():
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2], smoothingKernel=0)
def test_GTC_doOnly1dPlot():
input_chains = [np.array([SAMPLES_1[:,0]]).T, np.array([SAMPLES_2[:,0]]).T]
pygtc.plotGTC(chains=input_chains,
doOnly1dPlot = True,
smoothingKernel = 0)
def test_GTC_paramNames_noTex():
namesNoTex = ['param name', 'B_labmda', 'C', 'lambda']
return pygtc.plotGTC(chains=[SAMPLES_1, SAMPLES_2],
paramNames=namesNoTex,
smoothingKernel=0)
def test_GTC_bare():
pygtc.plotGTC(chains=[SAMPLES_1,SAMPLES_2],
smoothingKernel = 0)
