Emission-line Modeling

Analysis class: EmissionLineModel

Reference root: see method_path; $MANGA_SPECTRO_ANALYSIS/$MANGADRP_VER/$MANGADAP_VER/[METHOD]/[PLATE]/[IFUDESIGN]/ref for MaNGA

Reference file: see default_paths; manga-[PLATE]-[IFUDESIGN]-[RDXQA]-[BINNING]-[CONTINUUM]-[ELFIT].fits.gz for MaNGA

Optional Parameters: see Analysis Plans. The table below lists the parameters defined by EmissionLineModelDef

Key	Type	Options	Default	Description
key	str		EFITMPL11SSPDB	Keyword used to distinguish between different emission-line moment databases.
minimum_snr	int, float		0.0	Minimum S/N of spectrum to fit
mom_vel_name	str		Ha-6564	Name of the emission-line moments band used to set the initial velocity guess for each spaxel.
mom_disp_name	str			Name of the emission-line moments band used to set the initial velocity dispersion guess for each spaxel.
fitpar	ParSet, dict			Fitting function parameters
fitclass	Undefined			Class used to perform the fit.
fitfunc	Undefined			Function or method that performs the fit; must be callable.
overwrite	bool		False	If the output file already exists, redo all the calculations and overwrite it.

Important class dependencies:

• SpectralPixelMask: Used to mask spectral regions.

• EmissionLineDB: Generalized class that provides the detailed parameters for a set of emission lines to fit

• lineprofiles: Generalized class that provides the detailed profile shapes for each line.

• EmissionLineTemplates: Used to construct the emission-line templates to use with pPXF.

• EmissionLineFit: Provides functions common to both the moment and Gaussian-fit calculations.

• Sasuke: The equivalent of PPXFFit for the emission-line fits that is primarily a wrapper for pPXF. Critically, this uses the emission-line fitter contributed by Xihan Ji and Michele Cappellari, emline_fitter_with_ppxf().

• TemplateLibrary: If using Sasuke to perform the fit, this generalized interface provides the spectral-template library for use in modeling the stellar continuum.

• Elric: Provides the main fitting functions when just fitting Gaussians to continuum-subtracted spectra. BEWARE: This class has not been used or tested regularly since MPL-5.

Algorithm:

• Setup the fitting method:

  • Instantiate the SpectralPixelMask using the artifact_mask and waverange from config.

    • The BADSKY artifact mask is read and used to build an ArtifactDB instance that masks the typical residuals around the strong sky line at 5577 angstroms.

    • The waverange config parameter can be used to limit the fitted spectral range; will fit as much as possible if no range is provided.

  • If fit_method = elric, baseline_order sets the Legendre function used to set the baseline in each fitting window and window_buffer sets the +/- window in angstroms around each line to use during the fit.

  • If fit_method = sasuke:

    • etpl_line_sigma_mode and etpl_line_sigma_min determines the method used to set the emission-line template instrumental dispersion; the available options are set by etpl_line_sigma_options().

    • reject_boxcar sets the size of the boxcar (pixels) to use for the rejection iterations.

    • The templates used to fit the stellar continuum during the emission-line modeling can be different than those used during the stellar kinematics fit. Use continuum_templates and velscale_ratio to select the new templates and set their sampling. If continuum_templates is not given velscale_ratio is ignored and the templates are identical between the two modules. If the templates are switched, a new TemplateLibrary object is instantiated. When switching template libraries, the templates must have their resolution matched to the MaNGA data so that the corrected stellar kinematics from the existing StellarContinuumModel instance can be held fixed during the fitting.

    • bias, degree, mdegree are passed directly to pPXF (at the moment moments is ignored and always 2!)

    • internal_reddening = True forces use of the Calzetti (2000) attenuation law; will override any non-zero mdegree.

  • If deconstruct_bins = True, method will fit the emission lines on an individual spaxel basis instead of the binned spectra; currently this can only be used if fit_method = sasuke.

  • If mom_vel_name or mom_disp_name is defined, the DAP will use the corresponding moment measurements from the EmissionLineMoments object to set the initial guess for the velocity (default is the datacube metadata redshift) and/or velocity dispersion (default is 100 km/s) for the fit.

• If requested, call the “Elric” fitter:

  • WARNING: The Elric fitter has not been used in the DAP for some time. It should generally not be selected; if it is, one may need to spend some time debugging… For this reason, the method is not well documented here. Its main DAP wrapper fitting function is fit_SpatiallyBinnedSpectra() and its generalized fitting function is fit().

• Or, call “Sasuke” fitter:

  • The main DAP wrapper function is fit_SpatiallyBinnedSpectra() and does the following:

    • Get the binned spectra from the SpatiallyBinnedSpectra object

    • Either get the stellar templates from the TemplateLibrary object pointed to by the StellarContinuumModel object or, if new templates were selected, build the new TemplateLibrary instance (which must have its resolution matched to the MaNGA data).

    • Get the fitted stellar kinematics from the StellarContinuumModel object using matched_kinematics().

    • Determine which binned spectra have the minimum_snr from config, and have a good continuum model (cannot be flagged as NOVALUE or FITFAILED).

    • If deconstructing bins:

      • Get the individual spaxel spectra from the DataCube object

      • Apply the reddening defined in the SpatiallyBinnedSpectra object

      • Get the individual on-sky spaxel coordinates from the ReductionAssessment object and the unweighted on-sky binned-spectra coordinates from the SpatiallyBinnedSpectra object.

      • Run the generalized fitting function (see description below), providing the spectra to which the stellar-continuum results are “remapped” to for fitting the emission lines.

      • Measure the equivalent widths for the individual spaxels using measure_equivalent_width().

    • Otherwise:

      • Run the generalized fitting function (see description below), only providing the binned spectra.

      • Measure the equivalent widths for the binned spectra using measure_equivalent_width().

    • Construct the “emission-line baseline” as the difference between continuum+emission-line optimized fit from Sasuke and the stellar continuum fit for the stellar kinematics (from the StellarContinuumModel object)

  • The generalized fitter is mangadap.proc.sasuke.Sasuke.fit() and initially proceeds very similarly to mangadap.proc.ppxffit.PPXFFit.fit():

    • Check the input spectra to fit, guess kinematics, and remapping coordinates if provided

    • Check and set stellar templates and stellar kinematics if provided

    • Determine the spectral resolution to use for the emission-line templates; available options are set by etpl_line_sigma_options().

    • Construct and add emission-line templates using EmissionLineTemplates.

      • Parse the EmissionLineDB object: Determine which lines to fit and how to group lines into the same template (flux ratio fixed and same kinematics) kinematic components (same velocity and velocity dispersion), velocity groups, and sigma groups using _parse_emission_line_database().

      • Sample the desired spectral resolution at each input line center.

      • Convert the profile parameters into pixel coordinates.

      • Construct each template with the specified line profile using classes/methods in mangadap.util.lineprofiles. Lines with a fixed flux ratio are placed in the same template (this means they’ll also have tied velocities and velocity dispersions).

    • Parse the velocity and sigma groups into tied parameters to provide to pPXF.

    • Given the template and object spectral range, determine the maximum viable fitting range for pPXF using fitting_mask().

    • Run fit iterations using emline_fitter_with_ppxf().

      • If deconstructing the bins (for the Hybrid (HYB) binning scheme):

        • (a.) First fit the binned spectra (with a 3-sigma rejection iteration) forcing all the gas components into a single kinematic component (all velocities and velocity dispersions are tied).

        • (b.) Deconstruct the binned spectrum into its individual spaxels.

        • (c.) Use the result of the first fit to create a single, optimal stellar template and to set the starting kinematics for first fit to each spaxel.

        • (d.) Fit each spaxel (with a 3-sigma rejection iteration) with the optimized template and all gas components in a single kinematic component

        • (e.) Reset the starting guesses and refit each spaxel (without a rejection iteration) with the gas components in the appropriate velocity and sigma groups.

      • Otherwise:

        • Perform steps a, c, and e above, but just using the provided spectra (whether or not they’re bins or individual spaxels).

    • Parse the results of the fit iterations into the output using _save_results().