def test_proj_bubble(self):
def cmp_proj_bubble(mylocals, convarerr, proj, conf):
mek1 = mylocals['mek1']
assert covarerr[0] == approx(0, rel=1e-4)
assert covarerr[1] == approx(8.74608e-07, rel=1e-3)
assert mek1.kt.val == approx(17.8849, rel=1e-2)
assert mek1.norm.val == approx(4.15418e-06, rel=1e-2)
# Proj -- Upper bound of kT can't be found
#
assert proj.parmins[0] == approx(-12.048069, rel=0.01)
assert proj.parmins[1] == approx(-9.510913e-07, rel=0.01)
assert proj.parmaxes[1] == approx(2.403640e-06, rel=0.01)
assert proj.parmaxes[0] is None
assert conf.parmins[0] == approx(-12.1073, rel=0.01)
assert conf.parmaxes[0] == approx(62.0585, rel=0.01)
assert conf.parmins[1] == approx(-9.5568e-07, rel=0.01)
assert conf.parmaxes[1] == approx(2.39937e-06, rel=0.01)
self.run_thread('proj_bubble')
fit_results = ui.get_fit_results()
covarerr = sqrt(fit_results.extra_output['covar'].diagonal())
proj = ui.get_proj_results()
conf = ui.get_conf_results()
cmp_proj_bubble(self.locals, covarerr, proj, conf)
self.run_thread('proj_bubble_ncpus')
fit_results = ui.get_fit_results()
covarerr = sqrt(fit_results.extra_output['covar'].diagonal())
proj = ui.get_proj_results()
conf = ui.get_conf_results()
cmp_proj_bubble(self.locals, covarerr, proj, conf)
def fit(self, conf = False):
ui.ignore(None, None)
ui.notice(self.start, self.stop)
self.set_source()
ui.fit(1)
if conf:
ui.conf()
res = ui.get_fit_results()
for line in (self.H2lines + self.nonH2lines):
sourcename = line['source'].split('.')[1]
print sourcename
for p in ['pos', 'fwhm', 'ampl']:
n = '{0}.{1}'.format(sourcename, p)
_place_val(line, p, ui.get_par(n).val)
self.const = ui.get_par('c1.c0').val
self.redchi2 = res.rstat
if conf:
res = ui.get_conf_results()
for line in (self.H2lines + self.nonH2lines):
sourcename = line['source'].split('.')[1]
for p in ['pos', 'fwhm', 'ampl']:
n = '{0}.{1}'.format(sourcename, p)
parmin, parmax = _parminmax(res, n)
line[p+'_max'] = parmax
line[p+'_min'] = parmin
# deal with error on const
parmin, parmax = _parminmax(res, 'c1.c0')
self.const_min = parmin
self.const_max = parmax
def save_conf(filename, energy_flux=None, absorbpars=[]):
'''
Parameters
----------
filename : string
energy_flux : string or integer
Identifier of Sherpa model for which the energy flux will be calculated.
absorbpars : sherpa parameter instances
These parameters are set to 0 before the flux is calculated.
'''
confres = ui.get_conf_results()
fitres = ui.get_fit_results()
if energy_flux is not None:
oldval = [getattr(a, 'val') for a in absorbpars]
for a in absorbpars:
setattr(a, 'val', 0.)
energy_flux = ui.calc_energy_flux(0.3, 9., energy_flux)
for a, v in zip(absorbpars, oldval):
setattr(a, 'val', v)
with open(join(articlepath, filename), 'w') as f:
json.dump(
{
'name': confres.parnames,
'val': confres.parvals,
'up': confres.parmaxes,
'down': confres.parmins,
'redchi2': fitres.rstat,
'dof': fitres.dof,
'flux': energy_flux
}, f)
def test_proj_bubble(parallel, clean_astro_ui, fix_xspec, run_thread):
# How sensitive are the results to the change from bcmc to vern
# made in XSPEC 12.10.1? It looks like the mekal best-fit
# temperature can jump from ~17.9 to 18.6, so require bcmc
# in this test (handled by fix_xspec).
#
# Note that the error on kT is large, so we can expect that
# changes to the system could change these results. In particular,
# the covariance errors on kt are < 1 but from other error
# analysis they are > 10 or even unbound, so it is likely that the
# covariance error results can change significantly.
#
# The fit results change in XSPEC 12.10.0 since the mekal model
# was changed (FORTRAN to C++). A 1% difference is used for the
# parameter ranges from covar and proj (matches the tolerance for
# the fit results). Note that this tolerance has been relaced to
# 10% for the kT errors, as there is a significant change seen
# with different XSPEC versions for the covariance results.
#
# fit_results is unused
def cmp_thread(fit_results, mek1, covarerr):
assert covarerr[0] == approx(0, rel=1e-4)
assert covarerr[1] == approx(8.74608e-07, rel=1e-3)
assert mek1.kt.val == approx(17.8849, rel=1e-2)
assert mek1.norm.val == approx(4.15418e-06, rel=1e-2)
check_thread(run_thread, 'proj_bubble', parallel, cmp_thread, ['mek1'])
# Covar
#
# TODO: should this check that parmaxes is -1 * parmins instead?
covar = ui.get_covar_results()
assert covar.parmins[0] == approx(-0.653884, rel=0.1)
assert covar.parmins[1] == approx(-8.94436e-07, rel=0.01)
assert covar.parmaxes[0] == approx(0.653884, rel=0.1)
assert covar.parmaxes[1] == approx(8.94436e-07, rel=0.01)
# Proj -- Upper bound of kT can't be found
#
proj = ui.get_proj_results()
assert proj.parmins[0] == approx(-12.048069, rel=0.01)
assert proj.parmins[1] == approx(-9.510913e-07, rel=0.01)
assert proj.parmaxes[0] is None
assert proj.parmaxes[1] == approx(2.403640e-06, rel=0.01)
# Conf
#
conf = ui.get_conf_results()
assert conf.parmins[0] == approx(-12.1073, rel=0.01)
assert conf.parmins[1] == approx(-9.5568e-07, rel=0.01)
assert conf.parmaxes[0] is None
assert conf.parmaxes[1] == approx(2.39937e-06, rel=0.01)
def get_gamma_param(params, fitstats, fixgamma):
if fixgamma:
params["PhoIndex"] = fixgamma
else:
if fitstats.rstat <= 3:
dsmod.conf(po1.PhoIndex)
confstats = shp.get_conf_results()
params["PhoIndex"] = confstats.parvals[0]
params["PhoIndex_ErrMin"] = confstats.parmins[0]
params["PhoIndex_ErrMax"] = confstats.parmaxes[0]
else:
params["PhoIndex"] = fitstats.parvals[1]
return params
def fit_multiplets(multipletlist, id = None, outpath = None, plot = False, delta_lam = .2):
#
n_lines = np.sum([len(mult['wave']) for mult in multipletlist])
result = np.zeros(n_lines, dtype = {'names': ['multname', 'linename', 'wave', 'flux', 'errup', 'errdown', 'photons', 'photonflux'], 'formats': ['S30', 'S30', 'f4', 'f4', 'f4', 'f4', 'f4', 'f4']})
currentline = 0
#
for mult in multipletlist:
if outpath is not None:
outfile = os.path.join(outpath, filter(lambda x: x.isalnum(), mult['name']))
else:
outfile = None
fit_lines(mult, id, delta_lam = delta_lam, plot = plot, outfile = outfile)
#
ui.conf(id)
conf_res = ui.get_conf_results()
source = ui.get_source(id)
#set all ampl to 0 and only for 1 line to real value
set_all_val('c0', 0., id)
for lname, lfili, lwave in zip(mult['linename'], mult['fililiname'], mult['wave']):
print 'Fitting line '+str(currentline+1)+'/'+str(n_lines)
par = ui.get_model_component(lfili)
indconf_res = ( np.array(conf_res.parnames) == lfili+'.ampl').nonzero()[0]
set_all_val('ampl', 0., id)
par.ampl.val = conf_res.parvals[indconf_res]
counts = ui.calc_model_sum(None, None, id)
#
print(lname, counts)
photonflux = ui.calc_photon_flux(None, None, id)
# determine scaling between ampl and flux
par.ampl.val = 1
amp2flux = ui.calc_energy_flux(None, None, id)
par.ampl.val = conf_res.parvals[indconf_res]
#
val = conf_res.parvals[indconf_res] * amp2flux
errdown = conf_res.parmins[indconf_res] * amp2flux if conf_res.parmins[indconf_res] else np.nan
errup = conf_res.parmaxes[indconf_res] * amp2flux if conf_res.parmaxes[indconf_res] else np.nan
#
result[currentline] = (mult['name'], lname, lwave, val, errup, errdown, counts, photonflux)
#
currentline +=1
return result
def get_nh_param(params, fitstats, fixgamma):
nH = fitstats.parvals[0]
if fitstats.rstat <= 3.0:
dsmod.conf(abs1.nH)
confstats = shp.get_conf_results()
nHmin = confstats.parmins[0]
nHmax = confstats.parmaxes[0]
# Estimate nH upper-limit if nH=0, or nH-nHmin<0, or nHmin = nan
if (nH == 0) or (nHmin is None) or (nH - nHmin <= 0):
params["nH"] = _nH_uplimit(fixgamma)
else:
params["nH"] = nH
params["nH_ErrMin"] = nHmin
params["nH_ErrMax"] = nHmax
else:
params["nH"] = nH
return params
def conf(self):
"""
Run conf on each of the model parameters using the "onion-peeling" method:
- First conf the outside shell model using the outer annulus spectrum
- Freeze the model parameters for the outside shell
- get confidences for the next inward shell / annulus and freeze those parameters
- Repeat until all datasets have been conf()-ed and all shell-by-shell error parameters determined.
- Return model parameters to original thawed/frozen status
- WARNING: This ignores the correlations between parameters
:rtype: None
"""
thawed = [] # Parameter objects that are not already frozen
conf_results = []
this_conf_result = []
for annulus in reversed(range(self.nshell)):
dataids = [x['id'] for x in self.datasets if x['annulus'] == annulus]
print 'Getting shell-by-shell confidence for dataset ', dataids
SherpaUI.conf(*dataids)
this_conf_result = SherpaUI.get_conf_results()
conf_results.insert(0, this_conf_result)
for model_comp in self.model_comps:
name = model_comp['name']
if model_comp['shell'] == annulus:
# Remember parameters that are currently thawed
for par in [SherpaUI.get_par('%s.%s'%(name, x))
for x in SherpaUI.get_model_pars(name)]:
if not par.frozen:
thawed.append(par)
print 'Freezing', model_comp['name']
SherpaUI.freeze(model_comp['name'])
# Unfreeze parameters
for par in thawed:
print 'Thawing', par.fullname
par.thaw()
return conf_results
def write_all(filename='results.json'):
"""Dump source, fit results and conf results to a JSON file.
http://www.astropython.org/snippet/2010/7/Save-sherpa-fit-and-conf-results-to-a-JSON-file
"""
import sherpa.astro.ui as sau
out = dict()
if 0:
src = sau.get_source()
src_par_attrs = ('name', 'frozen', 'modelname', 'units', 'val',
'fullname')
out['src'] = dict(name=src.name,
pars=[
dict((attr, getattr(par, attr))
for attr in src_par_attrs)
for par in src.pars
])
try:
fit_attrs = ('methodname', 'statname', 'succeeded', 'statval',
'numpoints', 'dof', 'rstat', 'qval', 'nfev', 'message',
'parnames', 'parvals')
fit = sau.get_fit_results()
out['fit'] = dict((attr, getattr(fit, attr)) for attr in fit_attrs)
except Exception as err:
print(err)
try:
conf_attrs = ('datasets', 'methodname', 'fitname', 'statname', 'sigma',
'percent', 'parnames', 'parvals', 'parmins', 'parmaxes',
'nfits')
conf = sau.get_conf_results()
out['conf'] = dict((attr, getattr(conf, attr)) for attr in conf_attrs)
except Exception as err:
print(err)
try:
covar_attrs = ('datasets', 'methodname', 'fitname', 'statname',
'sigma', 'percent', 'parnames', 'parvals', 'parmins',
'parmaxes', 'nfits')
covar = sau.get_covar_results()
out['covar'] = dict(
(attr, getattr(covar, attr)) for attr in covar_attrs)
except Exception as err:
print(err)
if 0:
out['pars'] = []
for par in src.pars:
fullname = par.fullname
if any(fullname == x['name'] for x in out['pars']):
continue # Parameter was already processed
outpar = dict(name=fullname, kind=par.name)
# None implies no calculated confidence interval for Measurement
parmin = None
parmax = None
try:
if fullname in conf.parnames: # Confidence limits available from conf
i = conf.parnames.index(fullname)
parval = conf.parvals[i]
parmin = conf.parmins[i]
parmax = conf.parmaxes[i]
if parmin is None:
parmin = -float(
'inf'
) # None from conf means infinity, so set accordingly
if parmax is None:
parmax = float('inf')
elif fullname in fit.parnames: # Conf failed or par is uninteresting and wasn't sent to conf
i = fit.parnames.index(fullname)
parval = fit.parvals[i]
else: # No fit or conf value (maybe frozen)
parval = par.val
except Exception as err:
print(err)
out['pars'].append(outpar)
if filename is None:
return out
else:
json.dump(out, open(filename, 'w'), sort_keys=True, indent=4)
def print_conf():
res = sau.get_conf_results()
print("Confidence limits on fitted parameters:")
for index, parn in enumerate(res.parnames):
print(parn, "\t\t{0:.3e}".format(res.parvals[index]), "\t{0:.3e}".format(res.parmins[index]),
"\t{0:.3e}".format(res.parmaxes[index]))
def write_all(filename='results.json'):
"""Dump source, fit results and conf results to a JSON file.
http://www.astropython.org/snippet/2010/7/Save-sherpa-fit-and-conf-results-to-a-JSON-file
"""
import sherpa.astro.ui as sau
out = dict()
if 0:
src = sau.get_source()
src_par_attrs = ('name', 'frozen', 'modelname', 'units', 'val', 'fullname')
out['src'] = dict(name=src.name,
pars=[dict((attr, getattr(par, attr)) for attr in src_par_attrs)
for par in src.pars])
try:
fit_attrs = ('methodname', 'statname', 'succeeded', 'statval', 'numpoints', 'dof',
'rstat', 'qval', 'nfev', 'message', 'parnames', 'parvals')
fit = sau.get_fit_results()
out['fit'] = dict((attr, getattr(fit, attr)) for attr in fit_attrs)
except Exception as err:
print(err)
try:
conf_attrs = ('datasets', 'methodname', 'fitname', 'statname', 'sigma', 'percent',
'parnames', 'parvals', 'parmins', 'parmaxes', 'nfits')
conf = sau.get_conf_results()
out['conf'] = dict((attr, getattr(conf, attr)) for attr in conf_attrs)
except Exception as err:
print(err)
try:
covar_attrs = ('datasets', 'methodname', 'fitname', 'statname', 'sigma', 'percent',
'parnames', 'parvals', 'parmins', 'parmaxes', 'nfits')
covar = sau.get_covar_results()
out['covar'] = dict((attr, getattr(covar, attr)) for attr in covar_attrs)
except Exception as err:
print(err)
if 0:
out['pars'] = []
for par in src.pars:
fullname = par.fullname
if any(fullname == x['name'] for x in out['pars']):
continue # Parameter was already processed
outpar = dict(name=fullname, kind=par.name)
# None implies no calculated confidence interval for Measurement
parmin = None
parmax = None
try:
if fullname in conf.parnames: # Confidence limits available from conf
i = conf.parnames.index(fullname)
parval = conf.parvals[i]
parmin = conf.parmins[i]
parmax = conf.parmaxes[i]
if parmin is None:
parmin = -float('inf') # None from conf means infinity, so set accordingly
if parmax is None:
parmax = float('inf')
elif fullname in fit.parnames: # Conf failed or par is uninteresting and wasn't sent to conf
i = fit.parnames.index(fullname)
parval = fit.parvals[i]
else: # No fit or conf value (maybe frozen)
parval = par.val
except Exception as err:
print(err)
out['pars'].append(outpar)
if filename is None:
return out
else:
json.dump(out, open(filename, 'w'), sort_keys=True, indent=4)
