Skip to content

Stellar kinematics

Purpose

The stellarKinematics module measures the stellar kinematics of the observed spectra. nGIST is currently equipped with one routine that can readily be used by setting the configuration parameter KIN: METHOD to 'ppxf'. In particular, this module employs the pPXF routine of Cappellari and Emsellem (2004). More precisely:

  • Initial guesses: The used initial guess for the stellar velocity is 0 (as all spectra have been shifted to rest-frame wavelengths). The initial guess for the velocity dispersion is defined in the configuration parameter KIN: SIGMA. Alternatively, one can define initial guesses for all bins separately, i.e. by supplying the file *_kin-guess.fits in the output directory before the run. This file must contain a column V and SIGMA (as in the output file *_kin.fits), defining an initial guess for each spatial bin.

  • A spectral mask is applied, according to the masking file defined by KIN: SPEC_MASK. See Configuration for details.

  • The module calculates the stellar velocity, sigma, and higher-order velocity moments (depending on how many are specified, currently up to 6 have been tested). An iterative sigma-clipping process can be turned on to clean the spectra.

  • A set of stellar or SSP templates must be input. Currently, nGIST has been tested on the MILES stars and SSPs, eMILES, and SMILES libraries, the IndoUS library, the X-shooter stellar and SSP library, and Walcher+2009 templates. Only the MILES library is included in the default nGIST distribution, though any other library can be employed by writing a read-in module for it, following the example in your /ngist/ngistPipeline/prepareTemplates/miles.py directory.

Fitting procedure

This module uses a multi-step fitting approach to determine the best-fit parameters. nGIST first determines an optimal template set (OPT_TEMP, optional), measures and then applies (optional) a dust correction (DUST_CORR), rescales the noise, and cleans (DOCLEAN, optional) the spectrum of outliers (e.g. >3σ noise, sky lines, and emission lines).

To speed up the overall fitting process, the OPT_TEMP keyword can be used to pass a subset of templates to pPXF during the preparation steps (DUST_CORR, noise rescaling, and DOCLEAN). Note that in the final fit (used to determine the best-fit parameters), all templates are always provided. If OPT_TEMP is set to 'galaxy_single' or 'galaxy_set', pPXF is first run once on a single mean spectrum of the entire datacube to identify templates with non-zero weights. For 'galaxy_single', these templates are combined into a single ‘optimal’ template, whereas 'galaxy_set' passes all non-zero-weight templates to pPXF during the preparation steps. If OPT_TEMP is set to 'default', the optimal template set from the dust preparation step is used.

In the dust preparation step, nGIST determines a best-fit dust vector following the procedure described in Cappellari (2023). No additive or multiplicative polynomials are used in this fit. If DUST_CORR: True, all consecutive fits will use this dust vector within pPXF as a fixed component.

Next, a new estimate of the noise is derived by fitting the spectrum using pPXF with a constant noise vector. The input noise vector (either variance or constant) is rescaled using the bi-weight estimate (~standard deviation) of the fit residuals (data minus the best-fit model).

If DOCLEAN: True, we use the clip_outlier function to remove all 3σ outliers based on the best-fit from the noise estimation step.

In the final step, pPXF is run with all templates to determine the best-fit parameters.

Notes - The DEBUG keyword within the module can be used to fit specific bin(s), which can be useful when debugging. This option only works if PARALLEL is set to False in GENERAL. - The PLOT keyword within the module can be used to provide two figures for each spaxel. The first figure (step 1) shows the spectrum (black), the best-fit (red), and residuals (green) from the noise preparation step, whereas the second figure (step 3) shows the results of the final fit. Here, the vertical grey lines are the input mask, and the vertical pink lines show the spectral regions masked by DOCLEAN = True. Using PLOT = True will slow down the fit significantly and will create a large number of files if you are fitting many bins.

Config file input

KIN :
  METHOD : 'ppxf' # Name of the routine in stellarKinematics/ (without .py) to perform the tasks. Set 'False' to turn off module. Set 'ppxf' to use the standard GIST implementation, exploiting the pPXF routine of Cappellari & Emsellem (2004).
  SPEC_MASK : 'specMask_KIN' # File to define wavelength ranges to be masked during the stellar kinematics fit. The specified path is relative to the configDir path in defaultDir.
  LMIN : 4750 # Rest-frame wavelength range used for the stellar kinematics analysis [in Angst.]
  LMAX : 7100
  SIGMA : 100 # Initial guess of the velocity dispersion of the system [in km/s]
  MOM : 4 # Number of kinematic moments to be extracted
  ADEG : 12 # Degree of the additive Legendre polynomial. Set '-1' to not include any additive polynomials
  MDEG : 0 # Degree of the multiplicative Legendre polynomial. Set '0' to not include any multiplicative polynomials
  REDDENING : null # Initial guess on the stellar reddening E(B-V), in order to measure the stellar reddening. Note: This cannot be used together with multiplicative polynomials.
  MC_PPXF : 0 # Number of Monte-Carlo simulations to extract errors on the stellar kinematics. Formal errors are saved in any case.
  LSF_TEMP : 'lsf_MILES' # Path of the file specifying the line-spread-function of the spectral templates. The specified path is relative to the configDir path in defaultDir.
  TEMPLATE_SET : 'miles' # 
  LIBRARY : 'MILES/' # options are 'MILES/'
  NORM_TEMP : 'LIGHT' # Normalise the spectral template library to obtain light- or mass-weighted results [LIGHT / MASS]
  DOCLEAN : True # Keyword to turn on/off the sigma clipping. Set to 'False' for testing.
  OPT_TEMP : 'default' # Keyword to select Optimal Template method. Options are 'default', 'galaxy_single', 'galaxy_set'
  DUST_CORR : True # True/False, fixes the dust in the final pPXF fit 
  NOISE : 'variance' # Keyword to set noise to be used in pPXF run. Option 'constant' and 'variance'
  PLOT : False # produce extra output plots
  DEBUG_BIN : False # Optional Keyword - array of bins [1,2,3] - only works when parallel = False.

Output

Files

  • _kin.fits: Results of the extraction of stellar kinematics and their errors

    • Columns: V SIGMA H3 H4 etc Stellar kinematics. The number of columns here will correspond to value of the MOM keyword. For example, MOM: 4 will return V SIGMA H3 H4, while MOM:6 will return V SIGMA H3 H4 H5 H6 | ERR_* Errors on the stellar kinematics from MC-simulations | FORM_ERR_* Formal errors on the stellar kinematics | REDDENING Extinction calculated by PPXF

    • Rows: One line per bin.

  • _kin-bestfit.fits, Extension 1:

    • The best fit to the spectrum

    • Columns: BESTFIT The best fit to the spectrum

    • Rows: One fit per bin.

  • _kin-bestfit.fits, Extension 2:

    • Columns: LOGLAM The corresponding wavelength array

  • _kin-bestfit.fits, Extension 3:

    • Columns: GOODPIX The spectral pixels included in the fit

  • _kin-optimalTemplates.fits: Extension 1

    • Optimal template for the fit on each bin.

    • Columns: OPTIMAL_TEMPLATES Optimal templates

    • Rows: One template per bin.

  • _kin-optimalTemplates.fits: Extension 2

    • Columns: LOGLAM_TEMPLATE The corresponding wavelength array

  • _kin-SpectralMask.fits, Extension 1:

    • Columns: SPECTRAL_MASK The spectral pixels masked out in the fit

    • Rows: One mask per bin

Function returns

The module should end with return(None). If an uncaught exception occurs in the module, all following modules will be skipped for this galaxy. If you intend to manually skip all subsequent modules, simply raise an exception.

Examples

Comparison of the masking and subsequent pPXF best-fit for three Voronoi bins of the GIST (top) and nGIST (bottom) fits from Fraser-McKelvie et al., 2025. In each case the original spectrum of the bin is shown as a black line. The emission and sky line masks provided to pPXF are shown as the shaded grey regions. nGIST additionally performs further cleaning of the spectra, and these additional clipped regions are shown as shaded red regions. The best fit to the stellar continuum is shown in red, and the residuals in green. The derived stellar kinematic parameters for each spectrum are also listed.