def posterior_predictions_plot(plottype,
callback,
analyzer,
transformations,
nsamples=None):
"""
Internal Routine used by posterior_predictions_unconvolved, posterior_predictions_convolved
"""
posterior = analyzer.get_equal_weighted_posterior()
# for plotting, we don't need so many points, and especially the
# points that barely made it into the analysis are not that interesting.
# so pick a random subset of at least nsamples points
if nsamples is not None and len(posterior) > nsamples:
if hasattr(numpy.random, 'choice'):
chosen = numpy.random.choice(numpy.arange(len(posterior)),
replace=False,
size=nsamples)
else:
chosen = list(
set(numpy.random.randint(0, len(posterior),
size=10 * nsamples)))[:nsamples]
posterior = posterior[chosen, :]
assert len(posterior) == nsamples
prefix = analyzer.outputfiles_basename
tmpfilename = '%s-wdatatmp.qdp' % prefix
olddevice = Plot.device
Plot.device = '/null'
modelnames = set([t['model'].name for t in transformations])
while len(Plot.commands) > 0:
Plot.delCommand(1)
Plot.addCommand('wdata "%s"' % tmpfilename.replace('.qdp', ''))
oldchatter = Xset.chatter, Xset.logChatter
Xset.chatter, Xset.logChatter = 0, 0
# plot models
widgets = [progressbar.Percentage()]
if hasattr(progressbar, 'Counter'):
widgets += [progressbar.Counter('%5d')]
widgets += [progressbar.Bar(), progressbar.ETA()]
pbar = progressbar.ProgressBar(widgets=widgets,
maxval=len(posterior)).start()
for k, row in enumerate(posterior):
set_parameters(values=row[:-1], transformations=transformations)
if os.path.exists(tmpfilename):
os.remove(tmpfilename)
xspec.Plot(plottype)
content = numpy.genfromtxt(tmpfilename, skip_header=3)
os.remove(tmpfilename)
callback(content)
pbar.update(k)
pbar.finish()
Xset.chatter, Xset.logChatter = oldchatter
xspec.Plot.device = olddevice
def posterior_predictions_plot(plottype, callback, analyzer, transformations,
nsamples = None):
"""
Internal Routine used by posterior_predictions_unconvolved, posterior_predictions_convolved
"""
posterior = analyzer.get_equal_weighted_posterior()
# for plotting, we don't need so many points, and especially the
# points that barely made it into the analysis are not that interesting.
# so pick a random subset of at least nsamples points
if nsamples is not None and len(posterior) > nsamples:
if hasattr(numpy.random, 'choice'):
chosen = numpy.random.choice(
numpy.arange(len(posterior)),
replace=False, size=nsamples)
else:
chosen = list(set(numpy.random.randint(0, len(posterior),
size=10*nsamples)))[:nsamples]
posterior = posterior[chosen,:]
assert len(posterior) == nsamples
prefix = analyzer.outputfiles_basename
tmpfilename = '%s-wdatatmp.qdp' % prefix
olddevice = Plot.device
Plot.device = '/null'
modelnames = set([t['model'].name for t in transformations])
while len(Plot.commands) > 0:
Plot.delCommand(1)
Plot.addCommand('wdata "%s"' % tmpfilename.replace('.qdp', ''))
oldchatter = Xset.chatter, Xset.logChatter
Xset.chatter, Xset.logChatter = 0, 0
# plot models
widgets = [progressbar.Percentage()]
if hasattr(progressbar, 'Counter'):
widgets += [progressbar.Counter('%5d')]
widgets += [progressbar.Bar(), progressbar.ETA()]
pbar = progressbar.ProgressBar(widgets=widgets, maxval=len(posterior)).start()
for k, row in enumerate(posterior):
set_parameters(values=row[:-1], transformations=transformations)
if os.path.exists(tmpfilename):
os.remove(tmpfilename)
xspec.Plot(plottype)
content = numpy.genfromtxt(tmpfilename, skip_header=3)
os.remove(tmpfilename)
callback(content)
pbar.update(k)
pbar.finish()
Xset.chatter, Xset.logChatter = oldchatter
xspec.Plot.device = olddevice
def qq(analyzer, markers = 5, annotate = True):
"""
Create a quantile-quantile plot for model discovery (deviations in data from model).
The current data and model is used, so call *set_best_fit(analyzer, transformations)*
before, to get the qq plot at the best fit.
* markers: list of energies/channels (whichever the current plotting xaxis unit)
or number of equally spaced markers between minimum+maximum.
* annotate: add information to the plot
"""
olddevice = Plot.device
prefix = analyzer.outputfiles_basename
Plot.device = '/null'
tmpfilename = '%s-wdatatmp.qdp' % prefix
while len(Plot.commands) > 0:
Plot.delCommand(1)
Plot.addCommand('wdata "%s"' % tmpfilename.replace('.qdp', ''))
if os.path.exists(tmpfilename):
os.remove(tmpfilename)
Plot("counts")
content = numpy.genfromtxt(tmpfilename, skip_header=3)
#os.remove(tmpfilename)
bins, width, data, dataerror, model = content[:,0:5].transpose()
if not hasattr(markers, '__len__'):
nmarkers = int(markers)
decimals = int(-numpy.log10(bins[-1] - bins[0] + 1e-4))
markers = numpy.linspace(bins[0], bins[-1], nmarkers+2)[1:-1]
markers = set(numpy.round(markers, decimals=decimals))
# make qq plot, with 1:1 line
qq_plot(bins=bins, data=data * width * 2, model=model * width * 2,
markers = markers, annotate = annotate, unit=Plot.xAxis)
Plot.device = olddevice
def posterior_predictions_plot(self, plottype, callback, nsamples=None):
"""
Internal Routine used by posterior_predictions_unconvolved, posterior_predictions_convolved
"""
posterior = self.posterior
# for plotting, we don't need so many points, and especially the
# points that barely made it into the analysis are not that interesting.
# so pick a random subset of at least nsamples points
if nsamples is not None and len(posterior) > nsamples:
if hasattr(numpy.random, 'choice'):
chosen = numpy.random.choice(numpy.arange(len(posterior)),
replace=False,
size=nsamples)
else:
chosen = list(
set(
numpy.random.randint(0,
len(posterior),
size=10 * nsamples)))[:nsamples]
posterior = posterior[chosen, :]
assert len(posterior) == nsamples
prefix = self.outputfiles_basename
tmpfilename = os.path.join(
os.path.dirname(prefix),
os.path.basename(prefix).replace('.', '_') + '-wdatatmp.qdp')
if os.path.exists(tmpfilename):
os.remove(tmpfilename)
with XSilence():
olddevice = Plot.device
Plot.device = '/null'
#modelnames = set([t['model'].name for t in transformations])
while len(Plot.commands) > 0:
Plot.delCommand(1)
Plot.addCommand('wdata "%s"' % tmpfilename.replace('.qdp', ''))
# plot models
for k, row in enumerate(tqdm(posterior, disable=None)):
set_parameters(values=row,
transformations=self.transformations)
if os.path.exists(tmpfilename):
os.remove(tmpfilename)
xspec.Plot(plottype)
content = numpy.genfromtxt(tmpfilename, skip_header=3)
os.remove(tmpfilename)
callback(content)
xspec.Plot.device = olddevice
while len(Plot.commands) > 0:
Plot.delCommand(1)
if os.path.exists(tmpfilename):
os.remove(tmpfilename)
def posterior_predictions_plot(plottype, callback, analyzer, transformations,
nsamples = None):
"""
Internal Routine used by posterior_predictions_unconvolved, posterior_predictions_convolved
"""
posterior = analyzer.get_equal_weighted_posterior()
# for plotting, we don't need so many points, and especially the
# points that barely made it into the analysis are not that interesting.
# so pick a random subset of at least nsamples points
if nsamples is not None and len(posterior) > nsamples:
if hasattr(numpy.random, 'choice'):
chosen = numpy.random.choice(
numpy.arange(len(posterior)),
replace=False, size=nsamples)
else:
chosen = list(set(numpy.random.randint(0, len(posterior),
size=10*nsamples)))[:nsamples]
posterior = posterior[chosen,:]
assert len(posterior) == nsamples
prefix = analyzer.outputfiles_basename
tmpfilename = os.path.join(os.path.dirname(prefix), os.path.basename(prefix).replace('.','_') + '-wdatatmp.qdp')
olddevice = Plot.device
Plot.device = '/null'
#modelnames = set([t['model'].name for t in transformations])
while len(Plot.commands) > 0:
Plot.delCommand(1)
Plot.addCommand('wdata "%s"' % tmpfilename.replace('.qdp', ''))
oldchatter = Xset.chatter, Xset.logChatter
Xset.chatter, Xset.logChatter = 0, 0
# plot models
for k, row in enumerate(tqdm(posterior, disable=None)):
set_parameters(values=row[:-1], transformations=transformations)
if os.path.exists(tmpfilename):
os.remove(tmpfilename)
xspec.Plot(plottype)
content = numpy.genfromtxt(tmpfilename, skip_header=3)
os.remove(tmpfilename)
callback(content)
Xset.chatter, Xset.logChatter = oldchatter
xspec.Plot.device = olddevice
while len(Plot.commands) > 0:
Plot.delCommand(1)