4 Methods and validation of SED fitting

In Section 2 we have described how models predict the SEDs of galaxies. Yet the ultimate goal of these models is to allow the reverse process, that is to derive the properties of galaxies from their observed SEDs through fitting procedures. Therefore we describe in this section several of the main methods used to fit model or observed template SEDs to multiwavelength observations of galaxies. Exactly which of these procedures should be used is contingent on which physical parameters are required and with what precision. In turn, this, and the accuracy with which the models reproduce specific observations, determines what observational data are necessary.

The following sections deal almost exclusively with stellar emission. This reflects a real dearth of specialized codes for carrying out the fitting process in the dust emission range (besides simple χ2 minimization over a parameter grid, e.g.  Klaas et al. 2001). This state of affair is understandable, however, in light of the rapid development and large uncertainties still inherent to our modelling of the dust emission of galaxies.

 4.1 Parametrizing SED models
 4.2 Spectral indices
 4.3 Principal Component Analysis
 4.4 Spectral fitting by inversion
  4.4.1 Method
  4.4.2 Non-linear inversion codes
  4.4.3 Non-linear physics
  4.4.4 Validation
 4.5 Bayesian inference
  4.5.1 Method
  4.5.2 Libraries and priors
  4.5.3 Validation
 4.6 Method-independent caveats