def _fit_1D(initial_model, spectrum, run_fitter):
"""
Fits an astropy CompoundModel to a Spectrum1D instance.
Parameters
----------
spectrum : :class:`specutils.spectrum.Spectrum1D`
The spectrum to be fitted.
initial_model : :class: `astropy.modeling.CompoundModel`
Initial guess for the model to be fitted.
run_fitter : bool
When False (the default), the function composes the compound
model and returns it without fitting.
Returns
-------
:class: `astropy.modeling.CompoundModel`
The model resulting from the fit.
:class:`specutils.spectrum.Spectrum1D`
The realization of the fitted model as a spectrum.
"""
if run_fitter:
output_model = fit_lines(spectrum, initial_model)
output_values = output_model(spectrum.spectral_axis)
else:
# Return without fitting.
output_model = initial_model
output_values = initial_model(spectrum.spectral_axis)
# Build return spectrum
output_spectrum = Spectrum1D(spectral_axis=spectrum.spectral_axis,
flux=output_values)
return output_model, output_spectrum
def __call__(self):
results = {'x': [], 'y': [], 'fitted_model': [], 'fitted_values': []}
for parameters in self.param_set:
x = parameters[0]
y = parameters[1]
# Calling the Spectrum1D constructor for every spaxel
# turned out to be less expensive than expected. Experiments
# show that the cost amounts to a couple percent additional
# running time in comparison with a version that uses a 3D
# spectrum as input. Besides, letting an externally-created
# spectrum reference into the callable somehow prevents it
# to execute. This behavior was seen also with other functions
# passed to the callable.
flux = self.cube[x, y, :] # transposed!
sp = Spectrum1D(spectral_axis=self.wave, flux=flux)
import sys, traceback
try:
fitted_model = fit_lines(sp, self.model)
fitted_values = fitted_model(self.wave)
results['x'].append(x)
results['y'].append(y)
results['fitted_model'].append(fitted_model.unitless_model)
results['fitted_values'].append(fitted_values.value)
except:
pass
# print("Exception on [{},{}]:".format(x, y), sys.exc_info()[0])
# print(results)
return results
def _on_fit_clicked(self, model_plot_data_item):
fit_mod = fit_lines(self.hub.data_item.spectrum, result)
flux = fit_mod(self.hub.data_item.spectrum.spectral_axis)
new_spec = Spectrum1D(
flux=flux, spectral_axis=self.hub.data_item.spectrum.spectral_axis)
# self.hub.model.add_data(new_spec, "Fitted Model Spectrum")
# Update the stored plot data item object for this model editor model
# self._model_editor_model.plot_data_item.data_item.set_data(new_spec)
# Fitted quantity models do not preserve the names of the sub models
# which are used to relate the fitted sub models back to the displayed
# models in the model editor. Go through and hope that their order is
# preserved.
if result.n_submodels() > 1:
for i, x in enumerate(result):
fit_mod.unitless_model._submodels[i].name = x.name
sub_mods = [x for x in fit_mod.unitless_model]
else:
fit_mod.unitless_model.name = result.name
sub_mods = [fit_mod.unitless_model]
disp_mods = {item.text(): item for item in model_editor_model.items}
for i, sub_mod in enumerate(sub_mods):
# Get the base astropy model object
model_item = disp_mods.get(sub_mod.name)
# For each of the children `StandardItem`s, parse out their
# individual stored values
for cidx in range(model_item.rowCount()):
param_name = model_item.child(cidx, 0).data()
if result.n_submodels() > 1:
parameter = getattr(fit_mod,
"{0}_{1}".format(param_name, i))
else:
parameter = getattr(fit_mod, param_name)
model_item.child(cidx,
1).setText("{:.4g}".format(parameter.value))
model_item.child(cidx, 1).setData(parameter.value,
Qt.UserRole + 1)
model_item.child(cidx, 3).setData(parameter.fixed,
Qt.UserRole + 1)
for i in range(0, 3):
self.model_tree_view.resizeColumnToContents(i)
def run(self):
"""
Implicitly called when the thread is started. Performs the operation.
"""
self.status.emit("Fitting model...", 0)
fit_mod = fit_lines(self.spectrum, self.model, fitter=self.fitter,
window=self.window, **self.fitter_kwargs)
if not self.fitter.fit_info.get('message', ""):
self.status.emit("Fit completed successfully!", 5000)
else:
self.status.emit("Fit completed, but with warnings.", 5000)
self.result.emit(fit_mod)
def __call__(self, parameters):
x = parameters[0]
y = parameters[1]
# Calling the Spectrum1D constructor for every spaxel
# turned out to be less expensive than expected. Experiments
# show that the cost amounts to a couple percent additional
# running time in comparison with a version that uses a 3D
# spectrum as input. Besides, letting an externally-created
# spectrum reference into the callable somehow prevents it
# to execute. This behavior was seen also with other functions
# passed to the callable.
flux = self.cube[x, y, :] # transposed!
sp = Spectrum1D(spectral_axis=self.wave, flux=flux)
fitted_model = fit_lines(sp, self.model)
fitted_values = fitted_model(self.wave)
return (x, y, fitted_model, fitted_values)
def get_toi837_li_equivalent_width():
spectrum_path = '../data/spectra/TOI-837_FEROS.fits'
plotpath = '../results/TOI_837/feros_spectrum_get_li_equivalent_width.png'
#
# fit out the continuum to get the continuum normalized flux, over the
# window of 6670 angstrom to 6713 angstrom.
#
xlim = [6670, 6713]
hdul = fits.open(spectrum_path)
d = hdul[0].data
wav = d[0, 0]
flx = d[3, 0]
if isinstance(xlim, list):
xmin = xlim[0]
xmax = xlim[1]
sel = (wav > xmin) & (wav < xmax)
wav = wav[sel]
flx = flx[sel]
spec = Spectrum1D(spectral_axis=wav * u.AA,
flux=flx * u.dimensionless_unscaled)
if isinstance(xlim, list):
exclude_regions = []
if xmin < 6709.2 and xmax > 6710.2:
exclude_regions.append(SpectralRegion(6709.2 * u.AA,
6710.2 * u.AA))
if xmin < 6679 and xmax > 6681:
exclude_regions.append(SpectralRegion(6679 * u.AA, 6681 * u.AA))
cont_flx = (fit_generic_continuum(spec, exclude_regions=exclude_regions)(
spec.spectral_axis))
cont_norm_spec = spec / cont_flx
#
# to fit gaussians, look at 1-flux.
#
full_spec = Spectrum1D(spectral_axis=cont_norm_spec.wavelength,
flux=(1 - cont_norm_spec.flux))
#
# get the Li EW
#
region = SpectralRegion(6708.5 * u.AA, 6711.5 * u.AA)
li_equiv_width = equivalent_width(cont_norm_spec, regions=region)
li_centroid = centroid(full_spec, region)
#
# fit a gaussian too, and get ITS equiv width
# https://specutils.readthedocs.io/en/stable/fitting.html
#
g_init = models.Gaussian1D(amplitude=0.2 * u.dimensionless_unscaled,
mean=6709.7 * u.AA,
stddev=0.5 * u.AA)
g_fit = fit_lines(full_spec, g_init, window=(region.lower, region.upper))
y_fit = g_fit(full_spec.wavelength)
fitted_spec = Spectrum1D(spectral_axis=full_spec.wavelength,
flux=(1 - y_fit) * u.dimensionless_unscaled)
fitted_li_equiv_width = equivalent_width(fitted_spec, regions=region)
#
# print bestfit params
#
print(42 * '=')
print('got Li equiv width of {}'.format(li_equiv_width))
print('got fitted Li equiv width of {}'.format(fitted_li_equiv_width))
print('got Li centroid of {}'.format(li_centroid))
print('fit gaussian1d params are\n{}'.format(repr(g_fit)))
print(42 * '=')
#
# plot the results
#
f, axs = plt.subplots(nrows=4, ncols=1, figsize=(6, 8))
axs[0].plot(wav, flx, c='k', zorder=3)
axs[0].plot(wav, cont_flx, c='r', zorder=2)
axs[1].plot(cont_norm_spec.wavelength, cont_norm_spec.flux, c='k')
axs[2].plot(cont_norm_spec.wavelength, cont_norm_spec.flux, c='k')
axs[3].plot(full_spec.wavelength, full_spec.flux, c='k')
axs[3].plot(full_spec.wavelength, y_fit, c='g')
txt = ('gaussian1d\namplitude:{:.3f}\nmean:{:.3f}\nstd:{:.3f}\nEW:{:.3f}'.
format(g_fit.amplitude.value, g_fit.mean.value, g_fit.stddev.value,
fitted_li_equiv_width))
axs[3].text(0.95,
0.95,
txt,
ha='right',
va='top',
transform=axs[3].transAxes,
fontsize='xx-small')
axs[0].set_ylabel('flux')
axs[1].set_ylabel('contnorm flux')
axs[2].set_ylabel('contnorm flux [zoom]')
axs[3].set_ylabel('1 - (contnorm flux)')
if isinstance(xlim, list):
for ax in axs:
ax.set_xlim(xlim)
axs[2].set_xlim([6708.5, 6711.5])
axs[3].set_xlim([6708.5, 6711.5])
axs[-1].set_xlabel('wavelength [angstrom]')
for ax in axs:
format_ax(ax)
outpath = '../results/TOI_837/toi837_li_equivalent_width_routine.png'
savefig(f, outpath)
def _on_fit_clicked(self, eq_pop_up=True):
if eq_pop_up:
self._on_equation_edit_button_clicked()
# Grab the currently selected plot data item from the data list
plot_data_item = self.hub.plot_item
# If this item is not a model data item, bail
if not isinstance(plot_data_item.data_item, ModelDataItem):
return
data_item = self._get_selected_data_item()
if data_item is None:
return
spectral_region = self.hub.spectral_regions
# Compose the compound model from the model editor sub model tree view
model_editor_model = plot_data_item.data_item.model_editor_model
result = model_editor_model.evaluate()
if result is None:
QMessageBox.warning(
self, "Please add models to fit.",
"Models can be added by clicking the"
" green \"add\" button and selecting a"
" model from the drop-down menu")
return
# Load options
fitter = FITTERS[self.fitting_options["fitter"]]
output_formatter = "{:0.%sg}" % self.fitting_options['displayed_digits']
kwargs = {}
if fitter is fitting.LevMarLSQFitter:
kwargs['maxiter'] = self.fitting_options['max_iterations']
kwargs['acc'] = self.fitting_options['relative_error']
kwargs['epsilon'] = self.fitting_options['epsilon']
# Run the compound model through the specutils fitting routine. Ensure
# that the returned values are always in units of the current plot by
# passing in the spectrum with the spectral axis and flux
# converted to plot units.
spectrum = data_item.spectrum.with_spectral_unit(
plot_data_item.spectral_axis_unit)
spectrum = spectrum.new_flux_unit(plot_data_item.data_unit)
fit_mod = fit_lines(spectrum,
result,
fitter=fitter(),
window=spectral_region,
**kwargs)
if fit_mod is None:
return
# Fitted quantity models do not preserve the names of the sub models
# which are used to relate the fitted sub models back to the displayed
# models in the model editor. Go through and hope that their order is
# preserved.
# TODO: Uncomment for when specutils function is working with units
# if result.n_submodels() > 1:
# for i, x in enumerate(result):
# fit_mod.unitless_model._submodels[i].name = x.name
# sub_mods = [x for x in fit_mod.unitless_model]
# else:
# fit_mod.unitless_model.name = result.name
# sub_mods = [fit_mod.unitless_model]
if result.n_submodels() > 1:
sub_mods = [x for x in fit_mod._submodels]
for i, x in enumerate(result):
fit_mod._submodels[i].name = x.name
else:
fit_mod.name = result.name
sub_mods = [fit_mod]
# Get a list of the displayed name for each sub model in the tree view
disp_mods = {item.text(): item for item in model_editor_model.items}
for i, sub_mod in enumerate(sub_mods):
# Get the base astropy model object
model_item = disp_mods.get(sub_mod.name)
# For each of the children `StandardItem`s, parse out their
# individual stored values
for cidx in range(model_item.rowCount()):
param_name = model_item.child(cidx, 0).data()
if result.n_submodels() > 1:
parameter = getattr(fit_mod,
"{0}_{1}".format(param_name, i))
else:
parameter = getattr(fit_mod, param_name)
model_item.child(cidx, 1).setText(
output_formatter.format(parameter.value))
model_item.child(cidx, 1).setData(parameter.value,
Qt.UserRole + 1)
model_item.child(cidx, 3).setData(parameter.fixed,
Qt.UserRole + 1)
for i in range(0, 4):
self.model_tree_view.resizeColumnToContents(i)
# Update the displayed data on the plot
self._redraw_model()
def __init__(self,
spectrumIn,
wline,
band,
polyOrder=1,
widthFactor=4.,
verbose=0):
# =======================================================================
# Initial check in spectral_axis
# =======================================================================
if not (isinstance(spectrumIn.spectral_axis, u.Quantity)
and isinstance(spectrumIn.flux, u.Quantity)):
raise ValueError("Spectral axis must be a `Quantity` object.")
if not spectrumIn.spectral_axis.unit == u.um:
raise ValueError("Spectral axis is not in units of microns")
self.polyOrder = polyOrder
resv = self.get_spec_resolution(int(band[1]),
np.array(wline, dtype=float))
# =======================================================================
# Convert delta velocity to delta lambda in microns with astropy units equivalencies
# =======================================================================
# c_kms = const.c.to('km/s')
# fwhmum = (resv * u.kilometer/ u.s / c_kms) * wline
rest_wline = wline * u.um
fwhmum_q = (resv * u.km / u.s).to(
u.um, equivalencies=u.doppler_optical(rest_wline)) - rest_wline
fwhmum = fwhmum_q.value
fwhm_to_sigma = 1. / (8 * np.log(2))**0.5
lineWidthEstimate = fwhmum * fwhm_to_sigma
wmin, wmax = wline[0] - (widthFactor *
fwhmum), wline[-1] + (widthFactor * fwhmum)
if verbose:
print("wline, wmin, wmax, resv, fwhmum, lineWidthEstimate: ",
wline, wmin, wmax, resv, fwhmum, lineWidthEstimate)
self.wmin = wmin
self.wmax = wmax
self.spectrumIn = spectrumIn
# =======================================================================
# mask non finite elements
# =======================================================================
spectrum = self.__finite(spectrumIn, verbose=verbose)
self.spectrum = spectrum
wunit = self.spectrum.spectral_axis.unit
region = SpectralRegion(self.wmin * wunit, self.wmax * wunit)
spectrum_region = extract_region(self.spectrum, region)
# =======================================================================
# Compute peak flux estimates for model parameters starting values in the region
# =======================================================================
peakFluxEstimate = []
for wsline in wline:
wave = spectrum_region.spectral_axis.value #just the ndarray and not Quantity
flux = spectrum_region.flux.value
wdist = np.abs(wsline - wave)
if verbose: print(wsline, min(wdist), max(wdist))
indexLine = np.where(wdist == min(wdist))[0][0]
if verbose: print("indexLine= {}".format(indexLine))
peakEstimate = np.mean(flux[indexLine - 1:indexLine + 1])
if verbose:
print('Estimates for peak init {}'.format(peakEstimate))
cont_sample = np.concatenate((flux[:5], flux[-5:]), axis=None)
continuumEstimate = np.median(
np.concatenate((flux[:5], flux[-5:]), axis=None))
peakFluxEstimate = np.append(peakFluxEstimate,
peakEstimate - continuumEstimate)
if verbose:
print('Estimates for peak & continuum {}, {}'.format(
peakFluxEstimate, continuumEstimate))
# =======================================================================
# Construct model compound (branching off lines+continuum or continuum)
# =======================================================================
try:
lineModel_init = models.Polynomial1D(self.polyOrder,
c0=continuumEstimate,
name='cont')
for xi in range(len(wline)):
lineModel_init += models.Gaussian1D(
amplitude=peakFluxEstimate[xi],
mean=wline[xi],
stddev=lineWidthEstimate[xi],
name='g{}'.format(xi + 1))
fitter = LevMarLSQFitter()
lineModel = fit_lines(self.spectrum,
lineModel_init,
fitter=fitter,
window=region)
fitResult = lineModel(self.spectrum.spectral_axis)
findLine = 1
self.flux = []
self.sigma = []
for idx in range(len(wline)):
#momentarily taking advantage of astropy units for conversion
line_amp = (lineModel.unitless_model[idx + 1].amplitude.value *
u.Jy).to(u.Watt / u.m**2 / u.Hz)
line_sig = (lineModel.unitless_model[idx + 1].stddev.value *
u.um).to(u.Hz, equivalencies=u.spectral())
self.flux = np.append(self.flux, (line_amp * line_sig *
np.sqrt(2. * np.pi)).value)
self.sigma = np.append(self.sigma, line_sig.value)
except:
if verbose: print('Exception')
lineModel_init = models.Polynomial1D(self.polyOrder,
c0=continuumEstimate,
name='cont')
fitter = LevMarLSQFitter()
lineModel = fit_lines(
self.spectrum, lineModel_init, fitter=fitter, window=region
) #the problem is narrow window where the contribution of the continuum sample is small
fitResult = lineModel(self.spectrum.spectral_axis)
findLine = 0
self.model = lineModel
self.fitResult = fitResult
self.findLine = findLine
self.fitter = fitter
# =======================================================================
# Preserve continuum Polynomial model
# =======================================================================
# there are two types of models, those that are based on
# `~astropy.modeling.models.PolynomialModel` and therefore
# require the ``degree`` parameter when instantiating the
# class , and "everything else" that does not require an
# "extra" parameter for class instantiation.
compound_model = lineModel.n_submodels > 1
if compound_model:
self.continuumModel = lineModel.unitless_model[0]
else:
self.continuumModel = lineModel.unitless_model
if findLine:
self.continuum = []
self.peak = []
self.centre = []
self.sigma = []
self.fwhm = []
self.chiSquared = (self.fitter.fit_info['fvec']**2).sum() / (
len(self.fitter.fit_info['fvec']) -
len(self.fitter.fit_info['param_cov'].data))
self.stddev = np.sqrt(
np.diag(fitter.fit_info['cov_x']
)) #standard deviations pertaining to all parameters.
params_idx = [
int(param.split('_', -1)[-1])
for param in self.model.param_names
]
self.fluxError = []
self.fluxErrorRelative = []
for idx in range(len(wline)):
self.continuum = np.append(
self.continuum,
self.continuumModel(
lineModel.unitless_model[idx + 1].mean.value))
self.peak = np.append(
self.peak,
lineModel.unitless_model[idx + 1].amplitude.value)
self.centre = np.append(
self.centre, lineModel.unitless_model[idx + 1].mean.value)
self.sigma = np.append(
self.sigma, lineModel.unitless_model[idx + 1].stddev.value)
self.fwhm = np.append(self.fwhm, self.sigma / fwhm_to_sigma)
line_amp = (lineModel.unitless_model[idx + 1].amplitude.value *
u.Jy).to(u.Watt / u.m**2 / u.Hz)
line_sig = (lineModel.unitless_model[idx + 1].stddev.value *
u.um).to(u.Hz, equivalencies=u.spectral())
param_idx = [
i for i, value in enumerate(params_idx)
if value == (idx + 1)
]
self.fluxErrorRelative = np.append(
self.fluxErrorRelative,
np.sqrt(
np.sum((self.stddev / self.model.parameters)[np.array(
[param_idx])][np.array([0, -1])]**2.)))
self.fluxError = self.fluxErrorRelative * self.flux
self.width = self.fwhm
else:
self.continuum = np.array(
[np.median(flux) for i in range(len(wline))])
self.flux = self.peak = self.sigma = self.fwhm = self.width = np.array(
[0. for i in range(len(wline))])
self.centre = wline
if verbose:
print('Line Not Detected. Continuum: {}'.format(
self.continuum))
self.line_spec = self.get_line_spec()
self.chiSquared = (self.fitter.fit_info['fvec']**2).sum() / (
len(self.fitter.fit_info['fvec']) -
len(self.fitter.fit_info['param_cov'].data))
return
def centroid_offsets(targ_bounds, data_wave, data_flux, sky_cents):
"""Returns amount by which extracted skylines are offset from model and the nearest wavelength value to each.
Parameters
----------
targ_bounds : tuple
List of tuples defining bounds of region around each skyline to examine
data_wave : tuple
Wavelength array
data_flux : tuple
Flux array
sky_cents : tuple
Skymodel centroids
Returns
-------
nearest_waves : tuple
Nearest wavelength value to centroid
offsets : tuple
Offset between data and skymodel
"""
regions = SpectralRegion(targ_bounds[0][0]*u.Angstrom,targ_bounds[0][-1]*u.Angstrom)
for i in range(1, len(targ_bounds)):
regions += SpectralRegion(targ_bounds[i][0]*u.Angstrom, targ_bounds[i][-1]*u.Angstrom)
#Normalize data
targ_norm_wave, targ_norm_flux, targ_norm_noise = continuum_normalize(np.min(data_wave), np.max(data_wave), data_flux,
data_wave, np.zeros(len(data_flux)))
#Find offsets
target = Spectrum1D(spectral_axis=targ_norm_wave*u.Angstrom, flux=targ_norm_flux*u.ct)
sub_spec = extract_region(target, regions)
offsets = np.zeros(len(sky_cents))
nearest_waves = np.zeros(len(sky_cents))
for i, sub in enumerate(sub_spec):
an_disp = sub.flux.max()
an_ampl = sub.flux.min()
an_mean = sub.spectral_axis[sub.flux.argmax()]
nearest_waves[i] = an_mean.value
an_stdv = np.sqrt(np.sum((sub.spectral_axis - an_mean)**2) / (len(sub.spectral_axis) - 1))
plt.figure()
plt.scatter(an_mean.value, an_disp.value, marker='o', color='#e41a1c', s=100, label='data')
plt.scatter(sky_cents[i], an_disp.value, marker='o', color='k', s=100, label='archive')
plt.vlines([an_mean.value - an_stdv.value, an_mean.value + an_stdv.value],
sub.flux.min().value, sub.flux.max().value,
color='#377eb8', ls='--', lw=2)
g_init = ( models.Const1D(an_disp) +
models.Gaussian1D(amplitude=(an_ampl - an_disp),
mean=an_mean, stddev=an_stdv) )
g_fit = fit_lines(sub, g_init)
line_fit = g_fit(sub.spectral_axis)
plt.plot(sub.spectral_axis, sub.flux, color='#e41a1c', lw=2)
plt.plot(sub.spectral_axis, line_fit, color='#377eb8', lw=2)
plt.axvline(an_mean.value, color='#e41a1c', ls='--', lw=2)
plt.legend()
offsets[i] = an_mean.value - sky_cents[i].value
return nearest_waves, offsets
def line_fit(spec,
spec_err,
wave_obj,
dwave=10. * u.AA,
dwave_cont=100. * u.AA,
sigmamax=14. * u.AA):
'''
Function to fit a 1D gaussian to a HETDEX spectrum from get_spec.py
Parameters
----------
spec
1D spectrum from a row in the table provided by get_spec.py.
Will assume unit of 10**-17*u.Unit('erg cm-2 s-1 AA-1') if no units
are provided.
spec_err
1D spectral uncertainty from table provided by get_spec.py.
Will assume unit of 10**-17*u.Unit('erg cm-2 s-1 AA-1') if no units
are provided.
wave_obj
wavelength you want to fit, an astropy quantity
dwave
spectral region above and below wave_obj to fit a line, an astropy quantity.
Default is 10.*u.AA
dwave_cont
spectral region to fit continuum. Default is +/- 100.*u.AA
sigmamax
Maximum linewidth (this is sigma/stdev of the gaussian fit) to allow
for a fit. Assumes unit of u.AA if not given
Returns
-------
'''
try:
spectrum = Spectrum1D(flux=spec,
spectral_axis=(2.0 * np.arange(1036) + 3470.) *
u.AA,
uncertainty=StdDevUncertainty(spec_err),
velocity_convention=None)
except ValueError:
spectrum = Spectrum1D(
flux=spec * 10**-17 * u.Unit('erg cm-2 s-1 AA-1'),
spectral_axis=(2.0 * np.arange(1036) + 3470.) * u.AA,
uncertainty=StdDevUncertainty(spec_err * 10**-17 *
u.Unit('erg cm-2 s-1 AA-1')),
velocity_convention=None)
# measure continuum over 2*dwave_cont wide window first:
cont_region = SpectralRegion((wave_obj - dwave_cont),
(wave_obj + dwave_cont))
cont_spectrum = extract_region(spectrum, cont_region)
cont = np.median(cont_spectrum.flux)
if np.isnan(cont):
#set continuum if its NaN
print('Continuum fit is NaN. Setting to 0.0')
cont = 0.0 * cont_spectrum.unit
# now get region to fit the continuum subtracted line
sub_region = SpectralRegion((wave_obj - dwave), (wave_obj + dwave))
sub_spectrum = extract_region(spectrum, sub_region)
try:
line_param = estimate_line_parameters(sub_spectrum - cont,
models.Gaussian1D())
except:
return None
if np.isnan(line_param.amplitude.value):
print('Line fit yields NaN result. Exiting.')
return None
try:
sigma = np.minimum(line_param.stddev, sigmamax)
except ValueError:
sigma = np.minimum(line_param.stddev, sigmamax * u.AA)
if np.isnan(sigma):
sigma = sigmamax
g_init = models.Gaussian1D(amplitude=line_param.amplitude,
mean=line_param.mean,
stddev=sigma)
# lineregion = SpectralRegion((wave_obj-2*sigma), (wave_obj+2*sigma))
# cont = fit_generic_continuum(sub_spectrum, exclude_regions=lineregion,
# model=models.Linear1D(slope=0))
#r1 = SpectralRegion((wave_obj-dwave), (wave_obj-2*sigma))
#r2 = SpectralRegion((wave_obj+2*sigma), (wave_obj+dwave))
#fitcontregion = r1 + r2
#fit_cont_spectrum = extract_region(sub_spectrum, fitcontregion)
#cont = np.mean(np.hstack([fit_cont_spectrum[0].flux, fit_cont_spectrum[1].flux]))
#contspec = cont(sub_spectrum.spectral_axis)
g_fit = fit_lines(sub_spectrum - cont, g_init)
x = np.arange(wave_obj.value - dwave.value, wave_obj.value + dwave.value,
0.5) * u.AA
y_fit = g_fit(x)
line_flux_model = np.sum(y_fit * 0.5 * u.AA)
chi2 = calc_chi2(sub_spectrum - cont, g_fit)
sn = np.sum(np.array(sub_spectrum.flux)) / np.sqrt(
np.sum(sub_spectrum.uncertainty.array**2))
line_flux_data = line_flux(sub_spectrum - cont).to(u.erg * u.cm**-2 *
u.s**-1)
line_flux_data_err = np.sqrt(np.sum(sub_spectrum.uncertainty.array**2))
#fitted_region = SpectralRegion((line_param.mean - 2*sigma),
# (line_param.mean + 2*sigma))
#fitted_spectrum = extract_region(spectrum, fitted_region)
#line_param = estimate_line_parameters(fitted_spectrum, models.Gaussian1D())
#sn = np.sum(np.array(fitted_spectrum.flux)) / np.sqrt(np.sum(
# fitted_spectrum.uncertainty.array**2))
#line_flux_data = line_flux(fitted_spectrum).to(u.erg * u.cm**-2 * u.s**-1)
#line_flux_data_err = np.sqrt(np.sum(fitted_spectrum.uncertainty.array**2))
return line_param, sn, chi2, sigma, line_flux_data, line_flux_model, line_flux_data_err, g_fit, cont
def gaussian_fit(self,spectrum,submin,submax,star_name):
#noise_region=SpectralRegion(2.26*u.um,2.3*u.um)
#spectrum=noise_region_uncertainty(spectrum,noise_region)
#lines = find_lines_threshold(spectrum, noise_factor=4)
#print(lines[lines['line_type'] == 'emission'])
#print(lines[lines['line_type'] == 'absorption'])
#amp=0.00075
#centre=2.1675
#sub region defined from function parameters
sub_region=SpectralRegion(submin*u.um,submax*u.um)
#spectrum extracted from subregion
sub_spectrum=extract_region(spectrum,sub_region)
#continuum fitted
g1_fit=fit_generic_continuum(spectrum)
y_continuum_fitted=g1_fit(sub_spectrum.spectral_axis)
#continuum fit plotted with subregion spectrum
plt.plot(sub_spectrum.spectral_axis,sub_spectrum.flux)
plt.plot(sub_spectrum.spectral_axis,y_continuum_fitted)
plt.title('Continuum Fitting')
plt.grid(True)
plt.show()
#continuum substracted to show intensity
sub_spectrum=sub_spectrum-y_continuum_fitted
#initial gaussian fitted to estimate parameters for more accurate fitting
estimates=estimate_line_parameters(sub_spectrum,models.Gaussian1D())
#new gauss_axis variable created for a smooth fit
gauss_axis=np.linspace(min(sub_spectrum.spectral_axis),max(sub_spectrum.spectral_axis),2000)
#gaussian fit from previous estimates produced
g_init=estimates
g_fit=fit_lines(sub_spectrum,g_init)
y_fit=g_fit(gauss_axis)
#fit plotted
plt.plot(sub_spectrum.spectral_axis,sub_spectrum.flux)
plt.plot(gauss_axis,y_fit)
#linestrength found
strength=(max(y_fit))
#f=open('brackett_strengths.txt','a')
#f.write(star_name + ' - '+ str(strength) + 'Jy')
#f.close()
plt.title('Single fit peak')
plt.grid(True)
plt.legend('Original Spectrum','Specutils Fit Result')
plt.show()
# define your array to store all of the arrays with points for a fit for each peak
all_peak_fits = []
# define your array to store the parameters of each fit
all_gaus_fits = []
for i in range(len(absorptions)):
# initialize your gaussian fit -- I approximated values for amplitude and stddev by looking a tthe plot
# then it will loop through each peak by its mean
gaus_init = models.Gaussian1D(amplitude=-100*u.Jy, mean=absorptions[i][0], stddev=20*u.Angstrom)
# fit the spectrum with the initialized gaussian looking only in a window of 100Å before and after each peak
# this will only fit one peak and not confuse the program to fit the entire spectrum
# gaus_fit returns the parameters of this fit
gaus_fit = fit_lines(spec_normalized, gaus_init, window =
SpectralRegion(absorptions[i][0]-100*u.Angstrom,absorptions[i][0]+100*u.Angstrom))
# peak fit applies those parameters to the wavelengths of the spectral axis
peak_fit = gaus_fit(spec_normalized.spectral_axis)
# append your findings
all_gaus_fits.append(gaus_fit)
all_peak_fits.append(peak_fit)
print('Absorption peak:',i,'\n','mean:',gaus_fit.mean[0],'Angstrom', '\n','FWHM',gaus_fit.fwhm)
plt.figure()
plt.plot(spec_normalized.spectral_axis, spec_normalized.flux)
# plot each gaussian fit