def setup_class(self):
self.model = Polynomial1D(2)
self.x = np.arange(0, 10, 0.1)
self.params = (2, 0.5, 3)
# a simple polynomial
self.y = self.params[0]
self.y += self.params[1] * self.x
self.y += self.params[2] * self.x**2
self.y += np.random.uniform(-0.1, 0.1, self.x.size)
sfit = SherpaFitter(statistic="cash", estmethod='covariance')
sfit(self.model, self.x, self.y)
self.sampler = sfit.get_sampler()
def setup_class(self):
self.model = Polynomial1D(2)
self.x = np.arange(0, 10, 0.1)
self.params = (2, 0.5, 3)
# a simple polynomial
self.y = self.params[0]
self.y += self.params[1] * self.x
self.y += self.params[2] * self.x ** 2
self.y += np.random.uniform(-0.1, 0.1, self.x.size)
sfit = SherpaFitter(statistic="cash", estmethod='covariance')
sfit(self.model, self.x, self.y)
self.sampler = sfit.get_sampler()
def test_bkg_doesnt_explode(self):
"""
Check this goes through the motions
"""
m = Polynomial1D(2)
x = np.arange(0, 10, 0.1)
y = 2 + 0.5 * x + 3 * x**2
bkg = x
sfit = SherpaFitter(statistic="cash", estmethod='covariance')
sfit(m, x, y, bkg=bkg)
def fit_knot_unified(hdu, j1, j2, u0, lineid='nii'):
NS, NV = hdu.data.shape
w = WCS(hdu.header)
vels, _ = w.all_pix2world(np.arange(NV), [0]*NV, 0)
vels /= 1000.0
# Ensure we don't go out of bounds
j1 = max(j1, 0)
j2 = min(j2, NS)
print('Slit pixels {}:{} out of {}'.format(j1, j2, NS))
knotspec = hdu.data[j1:j2, :].sum(axis=0)
# make sure all pixels are positive, since that helps the fitting/plotting
knotspec -= knotspec.min()
# Levenberg-Marquardt for easy jobs
lmfitter = SherpaFitter(statistic='chi2',
optimizer='levmar',
estmethod='confidence')
# Simulated annealing for trickier jobs
safitter = SherpaFitter(statistic='chi2',
optimizer='neldermead',
estmethod='covariance')
# The idea is that this strategy should work for all knots
# Estimate error from the BG: < -120 or > +100
bgmask = np.abs(vels + 10.0) >= 110.0
bgerr = np.std(knotspec[bgmask]) * np.ones_like(vels)
# Define core as [-10, 50], or 20 +/- 30
coremask = np.abs(vels - 20.0) < 30.0
# Fit to the BG with constant plus Lorentz
try:
vmean = np.average(vels[coremask], weights=knotspec[coremask])
except ZeroDivisionError:
vmean = 15.0
bgmodel = lmfitter(_init_bgmodel(vmean),
vels[bgmask], knotspec[bgmask],
err=bgerr[bgmask])
# Now freeze the BG model and add it to the initial core model
#bgmodel['Lorentz'].fixed['amplitude'] = True
#bgmodel['Constant'].fixed['amplitude'] = True
# Increase the data err in the bright part of the line to mimic Poisson noise
# Even though we don't know what the normalization is really, we will guess ...
spec_err = bgerr + POISSON_SCALE*np.sqrt(knotspec)
## Now for the exciting bit, fit everything at once
##
knotmask = np.abs(vels - u0) <= KNOT_WIDTH
# For low-velocity knots, we need to exclude positive velocities
# from the mask, since they will have large residual errors from
# the core subtraction
knotmask = knotmask & (vels < 0.0)
# Start off with the frozen BG model
fullmodel = bgmodel.copy()
core_components = list(fullmodel.submodel_names)
# Add in a model for the core
DV_INIT = [-15.0, -5.0, 5.0, 10.0, 30.0]
NCORE = len(DV_INIT)
BASE_WIDTH = 10.0 if lineid == 'ha' else 5.0
W_INIT = [BASE_WIDTH]*4 + [1.5*BASE_WIDTH]
for i in range(NCORE):
v0 = vmean + DV_INIT[i]
w0 = W_INIT[i]
component = 'G{}'.format(i)
fullmodel += Gaussian1D(
3.0, v0, w0,
bounds={'amplitude': [0, None],
'mean': [v0 - 10, v0 + 10],
'stddev': [w0, 1.5*w0]},
name=component)
core_components.append(component)
# Now, add in components for the knot to extract
knotmodel_init = Gaussian1D(
0.01, u0, BASE_WIDTH,
# Allow +/- 10 km/s leeway around nominal knot velocity
bounds={'amplitude': [0, None],
'mean': [u0 - 10, u0 + 10],
'stddev': [BASE_WIDTH, 25.0]},
name='Knot')
fullmodel += knotmodel_init
knot_components = ['Knot']
other_components = []
# Depending on the knot velocity, we may need other components to
# take up the slack too
if u0 <= -75.0 or u0 >= -50.0:
# Add in a generic fast knot
fullmodel += Gaussian1D(
0.01, -60.0, BASE_WIDTH,
bounds={'amplitude': [0, None],
'mean': [-70.0, -50.0],
'stddev': [BASE_WIDTH, 25.0]},
name='Fast other')
other_components.append('Fast other')
if u0 <= -50.0:
# Add in a generic slow knot
fullmodel += Gaussian1D(
0.01, -30.0, BASE_WIDTH,
bounds={'amplitude': [0, None],
'mean': [-40.0, -10.0],
'stddev': [BASE_WIDTH, 25.0]},
name='Slow other')
other_components.append('Slow other')
if u0 >= -75.0:
# Add in a very fast component
fullmodel += Gaussian1D(
0.001, -90.0, BASE_WIDTH,
bounds={'amplitude': [0, None],
'mean': [-110.0, -75.0],
'stddev': [BASE_WIDTH, 25.0]},
name='Ultra-fast other')
other_components.append('Ultra-fast other')
if u0 <= 30.0:
# Add in a red-shifted component just in case
fullmodel += Gaussian1D(
0.01, 40.0, BASE_WIDTH,
bounds={'amplitude': [0, None],
'mean': [30.0, 200.0],
'stddev': [BASE_WIDTH, 25.0]},
name='Red other')
other_components.append('Red other')
# Moment of truth: fit models to data
fullmodel = safitter(fullmodel, vels, knotspec, err=spec_err)
full_fit_info = safitter.fit_info
# Isolate the core+other model components
coremodel = fullmodel[core_components[0]]
for component in core_components[1:] + other_components:
coremodel += fullmodel[component]
# Subtract the core model from the data
residspec = knotspec - coremodel(vels)
# Now re-fit the knot model to the residual
# Calculate running std of residual spectrum
NWIN = 11
running_mean = generic_filter(residspec, np.mean, size=(NWIN,))
running_std = generic_filter(residspec, np.std, size=(NWIN,))
# Increase error estimate for data points where this is larger
# than spec_err, but only for velocities that are not in knotmask
residerr = bgerr
# residerr = spec_err
mask = (~knotmask) & (running_std > bgerr)
residerr[mask] = running_std[mask]
# The reason for this is so that poor modelling of the core is
# accounted for in the errors. Otherwise the reduced chi2 of the
# knot model will be too high
# Make an extended mask for fitting the knot, omitting the
# redshifted half of the spectrum since it is irrelevant and we
# don't want it to affect tha chi2 or the confidance intervals
bmask = vels < 50.0
knotmodel = lmfitter(knotmodel_init,
vels[bmask], residspec[bmask],
err=residerr[bmask])
# Calculate the final residuals, which should be flat
final_residual = residspec - knotmodel(vels)
# Look at stddev of the final residuals and use them to rescale
# the residual errors. Then re-fit the knot with this better
# estimate of the errors. But only if rescaling would reduce the
# data error estimate.
residerr_rescale = final_residual[bmask].std() / residerr[bmask].mean()
if residerr_rescale < 1.0:
print('Rescaling data errors by', residerr_rescale)
residerr *= residerr_rescale
knotmodel = lmfitter(knotmodel,
vels[bmask], residspec[bmask],
err=residerr[bmask])
else:
residerr_rescale = 1.0
knot_fit_info = lmfitter.fit_info
lmfitter._fitter.estmethod.config['max_rstat'] = MAX_RSTAT
if knot_fit_info.rstat < MAX_RSTAT:
knot_fit_errors = lmfitter.est_errors(sigma=3)
else:
knot_fit_errors = None
return {
'nominal knot velocity': u0,
'velocities': vels,
'full profile': knotspec,
'error profile': residerr,
'core fit model': coremodel,
'core fit profile': coremodel(vels),
'core fit components': {k: coremodel[k](vels) for k in coremodel.submodel_names},
'core fit info': full_fit_info,
'core-subtracted profile': residspec,
'knot fit model': knotmodel,
'knot fit profile': knotmodel(vels),
'knot fit info': knot_fit_info,
'knot fit errors': knot_fit_errors,
'error rescale factor': residerr_rescale,
'knot j range': (j1, j2),
}
def fit_knot(hdu, j1, j2, u0):
NS, NV = hdu.data.shape
w = WCS(hdu.header)
vels, _ = w.all_pix2world(np.arange(NV), [0]*NV, 0)
vels /= 1000.0
# Ensure we don't go out of bounds
j1 = max(j1, 0)
j2 = min(j2, NS)
print('Slit pixels {}:{} out of {}'.format(j1, j2, NS))
knotspec = hdu.data[j1:j2, :].sum(axis=0)
# make sure all pixels are positive, since that helps the fitting/plotting
knotspec -= knotspec.min()
# Levenberg-Marquardt for easy jobs
lmfitter = SherpaFitter(statistic='chi2',
optimizer='levmar',
estmethod='confidence')
# Simulated annealing for trickier jobs
safitter = SherpaFitter(statistic='chi2',
optimizer='neldermead',
estmethod='covariance')
# First do the strategy for typical knots (u0 = [-30, -80])
# Estimate error from the BG: < -120 or > +100
bgmask = np.abs(vels + 10.0) >= 110.0
bgerr = np.std(knotspec[bgmask]) * np.ones_like(vels)
# Fit to the BG with constant plus Lorentz
try:
vmean = np.average(vels, weights=knotspec)
except ZeroDivisionError:
vmean = 15.0
bgmodel = lmfitter(_init_bgmodel(vmean),
vels[bgmask], knotspec[bgmask],
err=bgerr[bgmask])
# Now freeze the BG model and add it to the initial core model
bgmodel['Lorentz'].fixed['amplitude'] = True
bgmodel['Constant'].fixed['amplitude'] = True
# Increase the data err in the bright part of the line to mimic Poisson noise
# Even though we don't know what the normalization is really, we will guess ...
spec_err = bgerr + POISSON_SCALE*np.sqrt(knotspec)
# Fit to the line core
knotmask = np.abs(vels - u0) <= KNOT_WIDTH
coremodel = safitter(_init_coremodel() + bgmodel,
vels[~knotmask], knotspec[~knotmask],
err=spec_err[~knotmask])
core_fit_info = safitter.fit_info
# Residual should contain just knot
residspec = knotspec - coremodel(vels)
# Calculate running std of residual spectrum
NWIN = 11
running_mean = generic_filter(residspec, np.mean, size=(NWIN,))
running_std = generic_filter(residspec, np.std, size=(NWIN,))
# Increase error estimate for data points where this is larger
# than spec_err, but only for velocities that are not in knotmask
residerr = bgerr
# residerr = spec_err
mask = (~knotmask) & (running_std > bgerr)
residerr[mask] = running_std[mask]
# The reason for this is so that poor modelling of the core is
# accounted for in the errors. Otherwise the reduced chi2 of the
# knot model will be too high
# Make an extended mask for fitting the knot, omitting the
# redshifted half of the spectrum since it is irrelevant and we
# don't want it to affect tha chi2 or the confidance intervals
bmask = vels < 50.0
# Fit single Gaussian to knot
amplitude_init = residspec[knotmask].max()
if amplitude_init < 0.0:
# ... pure desperation here
amplitude_init = residspec[bmask].max()
knotmodel = lmfitter(_init_knotmodel(amplitude_init, u0),
vels[bmask], residspec[bmask],
err=residerr[bmask])
# Calculate the final residuals, which should be flat
final_residual = residspec - knotmodel(vels)
# Look at stddev of the final residuals and use them to rescale
# the residual errors. Then re-fit the knot with this better
# estimate of the errors. But only if rescaling would reduce the
# data error estimate.
residerr_rescale = final_residual[bmask].std() / residerr[bmask].mean()
if residerr_rescale < 1.0:
print('Rescaling data errors by', residerr_rescale)
residerr *= residerr_rescale
knotmodel = lmfitter(knotmodel,
vels[bmask], residspec[bmask],
err=residerr[bmask])
else:
residerr_rescale = 1.0
knot_fit_info = lmfitter.fit_info
lmfitter._fitter.estmethod.config['max_rstat'] = MAX_RSTAT
if knot_fit_info.rstat < MAX_RSTAT:
knot_fit_errors = lmfitter.est_errors(sigma=3)
else:
knot_fit_errors = None
return {
'nominal knot velocity': u0,
'velocities': vels,
'full profile': knotspec,
'error profile': residerr,
'core fit model': coremodel,
'core fit profile': coremodel(vels),
'core fit components': {k: coremodel[k](vels) for k in coremodel.submodel_names},
'core fit info': core_fit_info,
'core-subtracted profile': residspec,
'knot fit model': knotmodel,
'knot fit profile': knotmodel(vels),
'knot fit info': knot_fit_info,
'knot fit errors': knot_fit_errors,
'error rescale factor': residerr_rescale,
}
def fit_knot_unified(hdu, j1, j2, u0, lineid='nii'):
NS, NV = hdu.data.shape
w = WCS(hdu.header)
vels, _ = w.all_pix2world(np.arange(NV), [0] * NV, 0)
vels /= 1000.0
# Ensure we don't go out of bounds
j1 = max(j1, 0)
j2 = min(j2, NS)
print('Slit pixels {}:{} out of {}'.format(j1, j2, NS))
knotspec = hdu.data[j1:j2, :].sum(axis=0)
# make sure all pixels are positive, since that helps the fitting/plotting
knotspec -= knotspec.min()
# Levenberg-Marquardt for easy jobs
lmfitter = SherpaFitter(statistic='chi2',
optimizer='levmar',
estmethod='confidence')
# Simulated annealing for trickier jobs
safitter = SherpaFitter(statistic='chi2',
optimizer='neldermead',
estmethod='covariance')
# The idea is that this strategy should work for all knots
# Estimate error from the BG: < -120 or > +100
bgmask = np.abs(vels + 10.0) >= 110.0
bgerr = np.std(knotspec[bgmask]) * np.ones_like(vels)
# Define core as [-10, 50], or 20 +/- 30
coremask = np.abs(vels - 20.0) < 30.0
# Fit to the BG with constant plus Lorentz
try:
vmean = np.average(vels[coremask], weights=knotspec[coremask])
except ZeroDivisionError:
vmean = 15.0
bgmodel = lmfitter(_init_bgmodel(vmean),
vels[bgmask],
knotspec[bgmask],
err=bgerr[bgmask])
# Now freeze the BG model and add it to the initial core model
#bgmodel['Lorentz'].fixed['amplitude'] = True
#bgmodel['Constant'].fixed['amplitude'] = True
# Increase the data err in the bright part of the line to mimic Poisson noise
# Even though we don't know what the normalization is really, we will guess ...
spec_err = bgerr + POISSON_SCALE * np.sqrt(knotspec)
## Now for the exciting bit, fit everything at once
##
knotmask = np.abs(vels - u0) <= KNOT_WIDTH
# For low-velocity knots, we need to exclude positive velocities
# from the mask, since they will have large residual errors from
# the core subtraction
knotmask = knotmask & (vels < 0.0)
# Start off with the frozen BG model
fullmodel = bgmodel.copy()
core_components = list(fullmodel.submodel_names)
# Add in a model for the core
DV_INIT = [-15.0, -5.0, 5.0, 10.0, 30.0]
NCORE = len(DV_INIT)
BASE_WIDTH = 10.0 if lineid == 'ha' else 5.0
W_INIT = [BASE_WIDTH] * 4 + [1.5 * BASE_WIDTH]
for i in range(NCORE):
v0 = vmean + DV_INIT[i]
w0 = W_INIT[i]
component = 'G{}'.format(i)
fullmodel += Gaussian1D(3.0,
v0,
w0,
bounds={
'amplitude': [0, None],
'mean': [v0 - 10, v0 + 10],
'stddev': [w0, 1.5 * w0]
},
name=component)
core_components.append(component)
# Now, add in components for the knot to extract
knotmodel_init = Gaussian1D(
0.01,
u0,
BASE_WIDTH,
# Allow +/- 10 km/s leeway around nominal knot velocity
bounds={
'amplitude': [0, None],
'mean': [u0 - 10, u0 + 10],
'stddev': [BASE_WIDTH, 25.0]
},
name='Knot')
fullmodel += knotmodel_init
knot_components = ['Knot']
other_components = []
# Depending on the knot velocity, we may need other components to
# take up the slack too
if u0 <= -75.0 or u0 >= -50.0:
# Add in a generic fast knot
fullmodel += Gaussian1D(0.01,
-60.0,
BASE_WIDTH,
bounds={
'amplitude': [0, None],
'mean': [-70.0, -50.0],
'stddev': [BASE_WIDTH, 25.0]
},
name='Fast other')
other_components.append('Fast other')
if u0 <= -50.0:
# Add in a generic slow knot
fullmodel += Gaussian1D(0.01,
-30.0,
BASE_WIDTH,
bounds={
'amplitude': [0, None],
'mean': [-40.0, -10.0],
'stddev': [BASE_WIDTH, 25.0]
},
name='Slow other')
other_components.append('Slow other')
if u0 >= -75.0:
# Add in a very fast component
fullmodel += Gaussian1D(0.001,
-90.0,
BASE_WIDTH,
bounds={
'amplitude': [0, None],
'mean': [-110.0, -75.0],
'stddev': [BASE_WIDTH, 25.0]
},
name='Ultra-fast other')
other_components.append('Ultra-fast other')
if u0 <= 30.0:
# Add in a red-shifted component just in case
fullmodel += Gaussian1D(0.01,
40.0,
BASE_WIDTH,
bounds={
'amplitude': [0, None],
'mean': [30.0, 200.0],
'stddev': [BASE_WIDTH, 25.0]
},
name='Red other')
other_components.append('Red other')
# Moment of truth: fit models to data
fullmodel = safitter(fullmodel, vels, knotspec, err=spec_err)
full_fit_info = safitter.fit_info
# Isolate the core+other model components
coremodel = fullmodel[core_components[0]]
for component in core_components[1:] + other_components:
coremodel += fullmodel[component]
# Subtract the core model from the data
residspec = knotspec - coremodel(vels)
# Now re-fit the knot model to the residual
# Calculate running std of residual spectrum
NWIN = 11
running_mean = generic_filter(residspec, np.mean, size=(NWIN, ))
running_std = generic_filter(residspec, np.std, size=(NWIN, ))
# Increase error estimate for data points where this is larger
# than spec_err, but only for velocities that are not in knotmask
residerr = bgerr
# residerr = spec_err
mask = (~knotmask) & (running_std > bgerr)
residerr[mask] = running_std[mask]
# The reason for this is so that poor modelling of the core is
# accounted for in the errors. Otherwise the reduced chi2 of the
# knot model will be too high
# Make an extended mask for fitting the knot, omitting the
# redshifted half of the spectrum since it is irrelevant and we
# don't want it to affect tha chi2 or the confidance intervals
bmask = vels < 50.0
knotmodel = lmfitter(knotmodel_init,
vels[bmask],
residspec[bmask],
err=residerr[bmask])
# Calculate the final residuals, which should be flat
final_residual = residspec - knotmodel(vels)
# Look at stddev of the final residuals and use them to rescale
# the residual errors. Then re-fit the knot with this better
# estimate of the errors. But only if rescaling would reduce the
# data error estimate.
residerr_rescale = final_residual[bmask].std() / residerr[bmask].mean()
if residerr_rescale < 1.0:
print('Rescaling data errors by', residerr_rescale)
residerr *= residerr_rescale
knotmodel = lmfitter(knotmodel,
vels[bmask],
residspec[bmask],
err=residerr[bmask])
else:
residerr_rescale = 1.0
knot_fit_info = lmfitter.fit_info
lmfitter._fitter.estmethod.config['max_rstat'] = MAX_RSTAT
if knot_fit_info.rstat < MAX_RSTAT:
knot_fit_errors = lmfitter.est_errors(sigma=3)
else:
knot_fit_errors = None
return {
'nominal knot velocity': u0,
'velocities': vels,
'full profile': knotspec,
'error profile': residerr,
'core fit model': coremodel,
'core fit profile': coremodel(vels),
'core fit components':
{k: coremodel[k](vels)
for k in coremodel.submodel_names},
'core fit info': full_fit_info,
'core-subtracted profile': residspec,
'knot fit model': knotmodel,
'knot fit profile': knotmodel(vels),
'knot fit info': knot_fit_info,
'knot fit errors': knot_fit_errors,
'error rescale factor': residerr_rescale,
'knot j range': (j1, j2),
}
def fit_knot(hdu, j1, j2, u0):
NS, NV = hdu.data.shape
w = WCS(hdu.header)
vels, _ = w.all_pix2world(np.arange(NV), [0] * NV, 0)
vels /= 1000.0
# Ensure we don't go out of bounds
j1 = max(j1, 0)
j2 = min(j2, NS)
print('Slit pixels {}:{} out of {}'.format(j1, j2, NS))
knotspec = hdu.data[j1:j2, :].sum(axis=0)
# make sure all pixels are positive, since that helps the fitting/plotting
knotspec -= knotspec.min()
# Levenberg-Marquardt for easy jobs
lmfitter = SherpaFitter(statistic='chi2',
optimizer='levmar',
estmethod='confidence')
# Simulated annealing for trickier jobs
safitter = SherpaFitter(statistic='chi2',
optimizer='neldermead',
estmethod='covariance')
# First do the strategy for typical knots (u0 = [-30, -80])
# Estimate error from the BG: < -120 or > +100
bgmask = np.abs(vels + 10.0) >= 110.0
bgerr = np.std(knotspec[bgmask]) * np.ones_like(vels)
# Fit to the BG with constant plus Lorentz
try:
vmean = np.average(vels, weights=knotspec)
except ZeroDivisionError:
vmean = 15.0
bgmodel = lmfitter(_init_bgmodel(vmean),
vels[bgmask],
knotspec[bgmask],
err=bgerr[bgmask])
# Now freeze the BG model and add it to the initial core model
bgmodel['Lorentz'].fixed['amplitude'] = True
bgmodel['Constant'].fixed['amplitude'] = True
# Increase the data err in the bright part of the line to mimic Poisson noise
# Even though we don't know what the normalization is really, we will guess ...
spec_err = bgerr + POISSON_SCALE * np.sqrt(knotspec)
# Fit to the line core
knotmask = np.abs(vels - u0) <= KNOT_WIDTH
coremodel = safitter(_init_coremodel() + bgmodel,
vels[~knotmask],
knotspec[~knotmask],
err=spec_err[~knotmask])
core_fit_info = safitter.fit_info
# Residual should contain just knot
residspec = knotspec - coremodel(vels)
# Calculate running std of residual spectrum
NWIN = 11
running_mean = generic_filter(residspec, np.mean, size=(NWIN, ))
running_std = generic_filter(residspec, np.std, size=(NWIN, ))
# Increase error estimate for data points where this is larger
# than spec_err, but only for velocities that are not in knotmask
residerr = bgerr
# residerr = spec_err
mask = (~knotmask) & (running_std > bgerr)
residerr[mask] = running_std[mask]
# The reason for this is so that poor modelling of the core is
# accounted for in the errors. Otherwise the reduced chi2 of the
# knot model will be too high
# Make an extended mask for fitting the knot, omitting the
# redshifted half of the spectrum since it is irrelevant and we
# don't want it to affect tha chi2 or the confidance intervals
bmask = vels < 50.0
# Fit single Gaussian to knot
amplitude_init = residspec[knotmask].max()
if amplitude_init < 0.0:
# ... pure desperation here
amplitude_init = residspec[bmask].max()
knotmodel = lmfitter(_init_knotmodel(amplitude_init, u0),
vels[bmask],
residspec[bmask],
err=residerr[bmask])
# Calculate the final residuals, which should be flat
final_residual = residspec - knotmodel(vels)
# Look at stddev of the final residuals and use them to rescale
# the residual errors. Then re-fit the knot with this better
# estimate of the errors. But only if rescaling would reduce the
# data error estimate.
residerr_rescale = final_residual[bmask].std() / residerr[bmask].mean()
if residerr_rescale < 1.0:
print('Rescaling data errors by', residerr_rescale)
residerr *= residerr_rescale
knotmodel = lmfitter(knotmodel,
vels[bmask],
residspec[bmask],
err=residerr[bmask])
else:
residerr_rescale = 1.0
knot_fit_info = lmfitter.fit_info
lmfitter._fitter.estmethod.config['max_rstat'] = MAX_RSTAT
if knot_fit_info.rstat < MAX_RSTAT:
knot_fit_errors = lmfitter.est_errors(sigma=3)
else:
knot_fit_errors = None
return {
'nominal knot velocity': u0,
'velocities': vels,
'full profile': knotspec,
'error profile': residerr,
'core fit model': coremodel,
'core fit profile': coremodel(vels),
'core fit components':
{k: coremodel[k](vels)
for k in coremodel.submodel_names},
'core fit info': core_fit_info,
'core-subtracted profile': residspec,
'knot fit model': knotmodel,
'knot fit profile': knotmodel(vels),
'knot fit info': knot_fit_info,
'knot fit errors': knot_fit_errors,
'error rescale factor': residerr_rescale,
}
def setup_class(self):
# make data and models to use later!
err = 0.1
self.x1 = np.arange(1, 10, .1)
self.dx1 = np.ones(self.x1.shape) * (.1 / 2.)
self.x2 = np.arange(1, 10, .05)
self.dx2 = np.ones(self.x2.shape) * (.05 / 2.)
self.model1d = Gaussian1D(mean=5, amplitude=10, stddev=0.8)
self.model1d_2 = Gaussian1D(mean=4, amplitude=5, stddev=0.2)
self.tmodel1d = self.model1d.copy()
self.tmodel1d_2 = self.model1d_2.copy()
self.y1 = self.model1d(self.x1)
self.y1 += err * np.random.uniform(-1., 1., size=self.y1.size)
self.dy1 = err * np.random.uniform(0.5, 1., size=self.y1.size)
self.y2 = self.model1d_2(self.x2)
self.y2 += err * np.random.uniform(-1., 1., size=self.y2.size)
self.dy2 = err * np.random.uniform(0.5, 1., size=self.y2.size)
self.model1d.mean = 4
self.model1d.amplitude = 6
self.model1d.stddev = 0.5
self.model1d_2.mean = 5
self.model1d_2.amplitude = 10
self.model1d_2.stddev = 0.3
self.xx2, self.xx1 = np.mgrid[2:12:1, 2:12:1]
self.shape = self.xx2.shape
self.xx1 = self.xx1.flatten()
self.xx2 = self.xx2.flatten()
self.model2d = Gaussian2D(amplitude=10,
x_mean=5,
y_mean=6,
x_stddev=0.8,
y_stddev=1.1)
self.model2d.theta.fixed = True
self.yy = self.model2d(self.xx1, self.xx2)
self.dxx1 = err * np.random.uniform(0.5, 1., size=self.xx1.size)
self.dxx2 = err * np.random.uniform(0.5, 1., size=self.xx2.size)
self.dyy = err * np.random.uniform(0.5, 1., size=self.yy.size)
self.tmodel2d = self.model2d.copy()
self.model2d.amplitude = 5
self.model2d.x_mean = 6
self.model2d.y_mean = 5
self.model2d.x_stddev = 0.2
self.model2d.y_stddev = 0.7
# to stop stddev going negitive and getting div by zero error
self.model1d.stddev.min = 1e-99
self.model1d_2.stddev.min = 1e-99
self.model2d.x_stddev.min = 1e-99
self.model2d.y_stddev.min = 1e-99
#Lets define some tophats
self.rsp1 = np.zeros_like(self.x1)
self.rsp1[(self.x1 > 4) & (self.x1 < 6)] = 1
self.rsp2 = np.zeros_like(self.x2)
self.rsp2[(self.x2 > 4) & (self.x2 < 6)] = 1
self.rsp2d = np.zeros_like(self.xx1)
self.rsp2d[(self.xx1 > 4) & (self.xx1 < 6) & (self.xx2 > 4) &
(self.xx2 < 6)] = 1
self.rsp2d.flatten()
self.fitter = SherpaFitter(statistic="Chi2")
def fit_lines_sherpa(spec_file,
z_init=0.,
file_out=None,
do_plot=True,
monte_carlo=False):
"""Fit an HII region spectrum using Sherpa package. """
# from astropy.modeling.fitting import SherpaFitter
from saba import SherpaFitter
import matplotlib.pyplot as plt
from linetools.spectra.xspectrum1d import XSpectrum1D
# Redshift scale:
scale_factor = (1. + z_init)
# Read in the spectrum. **ASSUME VACUUM WAVELENGTHS?**
mods_spec = XSpectrum1D.from_file(spec_file)
# Set up a convenient wavelength, flux, error arrays
wave = mods_spec.wavelength.value
flux = mods_spec.flux.value
err = mods_spec.sig.value
###### ------ FOR TESTING!! ------
### To test this, let's constrain ourselves to only the wavelengths between ~Hbeta, OIII
# g = np.where((wave >= 4000) & (wave <= 5400.))
# wave = wave[g]
# flux = flux[g]
# err = err[g]
# Load the data for the lines to be fit. Starts with MANGA line list, modified for MODS.
line_data = get_linelist()
# Exclude lines outside of the wavelength coverage.
keep_lines = np.where(
(line_data['lambda'] <= np.max(wave) / scale_factor)
& (line_data['lambda'] >= np.min(wave) / scale_factor))[0]
keep_line_index = line_data['indx'][keep_lines]
# For now...debugging. jch
keep_lines = np.array(len(line_data))
keep_line_index = line_data['indx'][keep_lines]
##### MODEL DEFINITIONS
# Define initial parameters
amplitude_init = 0.1 * np.max(mods_spec.flux)
stddev_init = 1.5
amplitude_bounds, stddev_bounds, velocity_range = _define_bounds()
# Calculate the redshift delta
z_bounds_scale = (velocity_range / c.c.to('km/s').value) * scale_factor
z_bounds = (z_init - z_bounds_scale, z_init + z_bounds_scale)
# Define a joint model as the sums of Gaussians for each line
# Gaussian for first line:
j = 0
wave0 = line_data['lambda'][j]
line_center = wave0 * scale_factor
model_name = np.str(line_data['name'][j])
# Here we use a custom Gaussian class to fix redshifts together
joint_model = GaussianEmission(amplitude=amplitude_init,
redshift=z_init,
stddev=stddev_init,
wave0=wave0,
name=model_name)
# The rest wavelength is not a free parameter:
joint_model.wave0.fixed = True
# Loop through the remaining lines:
for j in np.arange(1, np.size(line_data)):
wave0 = line_data['lambda'][j]
line_center = wave0 * scale_factor
model_name = np.str(line_data['name'][j])
joint_model += GaussianEmission(amplitude=amplitude_init,
redshift=z_init,
stddev=stddev_init,
wave0=wave0,
name=model_name)
# Extract the model names:
model_names = joint_model.submodel_names
# Now we have to loop through the same models, applying the bounds:
for mdlnms in model_names:
joint_model[mdlnms].bounds['amplitude'] = amplitude_bounds
joint_model[mdlnms].bounds['redshift'] = z_bounds
joint_model[mdlnms].bounds['stddev'] = stddev_bounds
# The rest wavelength is not a free parameter:
joint_model[mdlnms].wave0.fixed = True
# TODO Get tied parameters to work.
# Tie some parameters together, checking that reference lines
# are actually covered by the spectrum:
for k in np.arange(0, np.size(line_data)):
mdlnm = model_names[k]
if (line_data['mode'][k] == 't33') & (np.in1d(33, keep_line_index)):
joint_model[mdlnm].stddev.tied = _tie_sigma_4862
joint_model[mdlnm].redshift.tied = _tie_redshift_4862
elif (line_data['mode'][k] == 't35') & (np.in1d(35, keep_line_index)):
joint_model[mdlnm].stddev.tied = _tie_sigma_5008
joint_model[mdlnm].redshift.tied = _tie_redshift_5008
elif (line_data['mode'][k] == 't45') & (np.in1d(45, keep_line_index)):
joint_model[mdlnm].stddev.tied = _tie_sigma_6585
joint_model[mdlnm].redshift.tied = _tie_redshift_6585
elif (line_data['mode'][k] == 't46') & (np.in1d(46, keep_line_index)):
joint_model[mdlnm].stddev.tied = _tie_sigma_6718
joint_model[mdlnm].redshift.tied = _tie_redshift_6718
# 3727/3729 lines:
if mdlnm == '[OII]3727':
import IPython
IPython.embed()
joint_model[mdlnm].stddev.tied = _tie_sigma_3729
joint_model[mdlnm].redshift.tied = _tie_redshift_3729
# Tie amplitudes of doublets
if (line_data['line'][k] == 'd35') & (np.in1d(35, keep_line_index)):
joint_model[mdlnm].amplitude.tied = _tie_ampl_5008 # 4959/5008
if (line_data['line'][k] == 'd45') & (np.in1d(45, keep_line_index)):
joint_model[mdlnm].amplitude.tied = _tie_ampl_6585 # 6549/6585
##### FITTING
# Sherpa model fitting from SABA package
sfit = SherpaFitter(statistic='chi2',
optimizer='levmar',
estmethod='confidence')
sfit_lm = SherpaFitter(statistic='chi2',
optimizer='neldermead',
estmethod='confidence')
sfit_mc = SherpaFitter(statistic='chi2',
optimizer='moncar',
estmethod='confidence')
# Do the fit
sfitted_model = sfit(joint_model, wave, flux, err=err)
# Refine with different optimizer
temp_model = sfitted_model.copy()
sfitted_model = sfit_lm(temp_model, wave, flux, err=err)
if monte_carlo:
# If requested, do a second fit with the very slow Monte Carlo approach
sfitted_model = sfit_mc(sfitted_model.copy(), wave, flux, err=err)
# Create the fitted flux array
sfitted_flux = sfitted_model(wave)
# TODO Get error estimates from Sherpa
# Work out the errors...
#sfit.est_config['maxiters']=200
#sfitted_err = sfit.est_errors(sigma=3)
# Plot the results
if do_plot:
plt.clf()
plt.plot(wave, flux, drawstyle='steps-mid', linewidth=2)
plt.plot(wave, sfitted_flux, color='orange', linewidth=2)
##### Create integrated fluxes and errors
# The integration range is over +/-stddev * int_delta_factor
int_delta_factor = _define_integration_delta()
output_construct = 0
for j in np.arange(np.size(line_data)):
# Calculate integrated fluxes, errors;
# -- First test that the lines are in the range covered by data
if np.in1d(j, keep_lines):
mean_lambda = line_data[j]['lambda'] * (1. +
sfitted_model[j].redshift)
# deal with blended O II 3727/3729 doublet
if line_data[j]['name'] == '[OII]3727':
# Calculate the integrated fluxes and errors
iflux, ierr = integrate_line_flux(wave,
flux,
err,
mean_lambda,
sfitted_model[j].stddev *
int_delta_factor,
line3727=True)
elif line_data[j]['name'] == '[OII]3729':
# pdb.set_trace()
# For 3729, use the flux derived for 3726
iflux = output_table['int_flux'][j - 1]
#iflux = 0.
# For 3729, use its own error. This is appropriate for the fitted errors of both lines
crap, ierr = integrate_line_flux(
wave, flux, err, mean_lambda,
sfitted_model[j].stddev * int_delta_factor)
else:
# Calculate the integrated fluxes and errors
iflux, ierr = integrate_line_flux(
wave, flux, err, mean_lambda,
sfitted_model[j].stddev * int_delta_factor)
redshift_out = (sfitted_model[j].redshift)[0]
sfitted_flux_out = np.sqrt(
2. *
np.pi) * sfitted_model[j].amplitude * sfitted_model[j].stddev
if output_construct == 0:
# Define and construct the initial table to hold the results
output_col_names, output_format, output_dtype = _define_output_table(
)
output_data = [[line_data[j]['name']], [line_data[j]['ion']],
[line_data[j]['lambda']],
[line_data[j]['indx']], [line_data[j]['mode']],
[mean_lambda],
[sfitted_model[j].amplitude.value],
[sfitted_model[j].stddev.value], [redshift_out],
[sfitted_flux_out], [iflux], [ierr],
[iflux / ierr]]
output_table = Table(output_data,
names=output_col_names,
dtype=output_dtype)
output_construct = 1
else:
output_table.add_row([
line_data[j]['name'], line_data[j]['ion'],
line_data[j]['lambda'], line_data[j]['indx'],
line_data[j]['mode'], mean_lambda,
sfitted_model[j].amplitude.value,
sfitted_model[j].stddev.value, redshift_out,
sfitted_flux_out, iflux, ierr, iflux / ierr
])
# Set the output format of the results table:
colnames = output_table.colnames
for j in np.arange(np.size(colnames)):
output_table[colnames[j]].format = output_format[j]
# Set up the spectral table:
spec_table = Table([wave, flux, err, sfitted_flux],
names=['wave', 'flux', 'err', 'spec_fit'])
# Write summary FITS files
if file_out is None:
file_base = spec_file.strip('.fits')
else:
file_base = file_out
table_file = file_base + '.HIIFitTable.fits'
fit_file = file_base + '.HIIFitSpec.fits'
output_table.write(table_file, overwrite=True)
spec_table.write(fit_file, overwrite=True)
return output_table
