mangadap.proc.spectralfitting module

Implements a few base classes used during spectral fitting procedures.

Revision history

14 Apr 2016: Implementation begun by K. Westfall (KBW)
19 Apr 2016: (KBW) First version
26 Apr 2016: (KBW) Moved PPXFFit to a separate file (
03 Nov 2016: (KBW) Added USETPL column to stellar kinematics output table.
25 Oct 2017: (KBW) Added PLY columns to emission-line database
02 Feb 2018: (KBW) Added select_binned_spectra_to_fit().
15 Mar 2018: (KBW) Equivalenth width now measured at the velocity of each line if a redshift is not provided. See measure_equivalent_width().


Copyright © 2019, SDSS-IV/MaNGA Pipeline Group

class mangadap.proc.spectralfitting.CompositionFit(fit_method, bitmask, par=None)[source]

Bases: mangadap.proc.spectralfitting.SpectralFitting

Base class for fitting the spectral composition.

class mangadap.proc.spectralfitting.EmissionLineFit(fit_method, bitmask=None, par=None)[source]

Bases: mangadap.proc.spectralfitting.SpectralFitting

Base class for fitting emission lines.

static _per_emission_line_dtype(neml, nkin, mask_dtype)[source]

Construct the record array data type for the output fits extension.

static check_and_prep_input(wave, flux, ivar=None, mask=None, sres=None, continuum=None, redshift=None, dispersion=None, default_dispersion=100.0)[source]

Check the input used for emission-line measurements.

inverse variance is converted to 1-sigma error

mask must be a boolean array.

sres can be a single vector, but will be returned as an array with a size that matches flux.

output all converted to masked arrays with at least two dimensions

static check_emission_line_database(emldb, wave=None, check_par=True)[source]

Check the emission-line database. Modes are checked by mangadap.par.emissionlinedb.EmissionLinePar, and the indices are checked to be unique by mangadap.par.emissionlinedb.EmissionLineDB.

  • The type of the object must be mangadap.par.emissionlinedb.EmissionLineDB
  • The provided profile type of each line must be a defined class.
  • At least one line must have mode='f'
  • All tied lines must be tied to a line with a correctly specified index.
  • Warnings will be provided for any line with a centroid that falls outside of the provided wavelength range.
  • The database must provide at least one valid line.
  • TypeError – Raised if the provided object is not an instance of mangadap.par.emissionlinedb.EmissionLineDB.
  • ValueError – Raised if any line has a mode of x or if the database does not provide a valid definition for any templates.
  • NameError – Raised if a defined profile type is not known.
static get_spectra_to_fit(binned_spectra, pixelmask=None, select=None, error=False, original_spaxels=False)[source]

Get the spectra to fit during the emission-line fitting.

  • binned_spectra – (mangadap.proc.spatiallybinnedspectra.SpatiallBinnedSpectra): Object with the spectra to fit.
  • pixelmask – (mangadap.util.pixelmask.SpectralPixelMask, optional): Pixel mask to apply.
  • select (numpy.ndarray, optional) – Select specific spectra to return. Must have the correct shape; cf. original_spaxels.
  • error (bool, optional) – Return \(1\sigma\) errors instead of inverse variance.
  • original_spaxels (bool, optional) – Instead of the binned spectra, use the drpf attribute of the binned_spectra object to return the original spaxels, corrected for Galactic extinction.

    1. The wavelength vector of the spectra,
    1. a masked numpy array with the flux data,
  • (3) a masked numpy array with the error data (returned as either inverse variance or \(1\sigma\),
  • (4) and an array with the spectral resolution for each spectrum, based on the internal binned spectra parameters.

Return type:

Four objects are returned

static instrumental_dispersion(wave, sres, restwave, cz)[source]

Determine the instrumental dispersion for a set of rest wavelengths and velocities.

  • wave (numpy.ndarray) – Vector with the wavelengths of the spectrum.
  • sres (numpy.ndarray) – Vector with the spectral resolution as a function of wavelength.
  • restwave (float, numpy.ndarray) – Rest wavelengths for a set of measured lines.
  • cz (float, numpy.ndarray) – Redshifts (in km/s) of each or all lines.

The instrumental dispersions for each

provided line.

Return type:


static line_metrics(emission_lines, wave, flux, ferr, model_flux, model_eml_par, mask=None, model_mask=None, bitmask=None, window=15, fill_redshift=False)[source]

Calculate fit-quality metrics near each emission line.


  • Allow window to be defined in angstroms?
  • emission_lines (mangadap.par.emissionlinedb.EmissionLineDB) – Emission-line database use during the fit.
  • wave (numpy.ndarray) – Wavelength vector for object spectra. Shape is \((N_{\rm pix},)\).
  • flux (numpy.ndarray) – Object spectra that have been fit. Can be provided as a Shape is \((N_{\rm spec},N_{\rm pix})\).
  • ferr (numpy.ndarray) – \(1\sigma\) errors in the object spectra. Can be provided as a Shape is \((N_{\rm spec},N_{\rm pix})\).
  • model_flux (numpy.ndarray) – Best-fitting model spectra. Can be provided as a Shape is (\(N_{\rm spec},N_{\rm pix}\)).
  • model_eml_par (numpy.recarray) – A numpy record array with data type given by _per_emission_line_dtype(). Uses FLUX, KIN, and MASK; and assigns results to LINE_* columns.
  • mask (numpy.ndarray, optional) – A mask for the object spectra that have been fit. Added to mask attribute of flux if it is a
  • model_mask (numpy.ndarray, optional) – A boolean numpy array with the mask for the model spectra. Added to mask attribute of model_flux if it is a
  • bitmask (mangadap.util.bitmask.BitMask, optional) – The BitMask object used to interpret the MASK column in the model_eml_par object. If None, the MASK column is ignored.
  • window (int, optional) – The width of the window used to compute the metrics around each line in number of pixels.
  • fill_redshift (bool, optional) – Fill any masked velocity measurement to the masked median of the velocities for the unmasked lines in the same spectrum when constructing the redshifted bands. If False, the A/N measurement is masked.

Return the input model_eml_par after filling the LINE_PIXC, AMP, ANR, LINE_NSTAT, LINE_CHI2, LINE_RMS, and LINE_FRMS columns.

Return type:



ValueError – Raised if various checks of the input array sizes are incorrect.

static measure_equivalent_width(wave, flux, emission_lines, model_eml_par, mask=None, redshift=None, bitmask=None, checkdb=True)[source]

The flux array is expected to have size Nspec x Nwave.

Provided previous emission-line fits, this function adds the equivalent width measurements to the output database.

Errors currently do not include the errors in the continuum measurement; only the provided error in the flux.

Raises:ValueError – Raised if the length of the spectra, errors, or mask does not match the length of the wavelength array; raised if the wavelength, redshift, or dispersion arrays are not 1D vectors; and raised if the number of redshifts or dispersions is not a single value or the same as the number of input spectra.
static select_binned_spectra_to_fit(binned_spectra, minimum_snr=0.0, stellar_continuum=None, debug=False)[source]

Select binned spectra for which to fit emission lines.


This could be based on the moment assessment of the emission-line S/N instead; for now just based on continuum S/N.


Boolean vector with the spectra in the binned_spectra object to fit.

Return type:


static select_spaxels_to_fit(binned_spectra, bins_to_fit=None, debug=False)[source]

Select spaxels for which to fit emission lines.

static subtract_continuum(flux, continuum)[source]

Subtract the continuum. Does not check that shapes match. Returns the continuum subtracted flux and a boolean array setting where the continuum is not defined.

class mangadap.proc.spectralfitting.SpectralFitting(fit_type, bitmask=None, par=None)[source]

Bases: object

Base class for spectral fitting.

class mangadap.proc.spectralfitting.StellarKinematicsFit(fit_method, bitmask, par=None)[source]

Bases: mangadap.proc.spectralfitting.SpectralFitting

Base class for fitting stellar kinematics.

static _per_stellar_kinematics_dtype(ntpl, nadd, nmult, nkin, mask_dtype)[source]

Construct the record array data type for the output fits extension.