3.1. Phosphoros overview

Phosphoros is a Python script packaging a number of C++ executables that implement a photometric redshift template fitting algorithm (see, e.g. []; []; [IlbertArnoutsMcCracken+06]). In short, multi-band photometric measurements from an input catalog of sources are compared to a grid of modeled photometry. For each source, the algorithm computes the likelihood that a model is representative of the observed photometry via a classical \(\chi^2\) calculation. Prior information can also be taken into account using Bayesian inference. The main outputs of Phosphoros are the best-fit model for input sources (in terms of redshift, spectral energy distribution, etc.), and the redshift Probability Density Function (PDF), which provides the probability of a source to be at a given redshift.

Phosphoros involves two main steps (discussed in detail in the Phosphoros paper []; see also the Methodology chapter for a shorter overview):

(Grid of Models) As a first step, Phosphoros generates a large set of models for which photometry are computed. Models are characterized by four parameters: redshift, restframe Spectral Energy Distribution (SED) template, intrinsic reddening curve and intrinsic color excess \(E_{(B-V)}\) value. The grid of models is created through the following operations (in this order):

  • Building a library of restframe SED templates (with the option to add emission lines to templates).

  • Correcting restframe SED templates for the effect of intrinsic dust attenuation.

  • Redshifting SED templates to the observed frame.

  • (optional) Applying intergalactic medium (IGM) redshift-dependent absorption.

  • Integrating modeled templates through a set of filter transmission curves.

Each model is characterized by a vector of computed photometric values, one value for each selected filter.

(Redshift Estimate) Phosphoros is now able to estimate the redshift of sources in the input catalog.

  • The likelihood \(\mathcal{L}\) of models is determined by computing the \(\chi^2\) between observed and modeled photometry.

  • (optional) Priors on input parameters are taken into account and the posterior distribution of a model is computed.

  • (optional) The SED template normalization (or scale factor), which is usually fixed to its best-likelihood value, can be treated as a free parameter in a fully Bayesian approach.

  • The output is a catalog containing the best-fit model for each input source and its redshift at the redshift PDF peak. Optionally, 1D PDFs for the model parameters can be saved, along with the multi-dimensional likelihood or posterior distribution.

Phosphoros includes also:

Additional functionalities, to be applied after generating the grid of models and before the redshift estimate:

  • (optional) Correction for the SED reddening due to Milky Way dust absorption.

  • (optional) Flux corrections due to variations in filter transmissions.

  • (optional) Zero-point correction, accounting for calibration issues and mismatches between galaxy colors and templates.

Tools to examine results, producing:

  • Statistical analysis of the redshift PDF of input sources.

  • Plots comparing reference and estimated redshifts.

Fig. 3.1 provides a sketch of main Phosphoros steps with their inputs and outputs, including optional functionalities.

../../_images/Flowchart_Phosphoros_v12.pdf

Fig. 3.1 Flowchart of the Phosphoros algorithm

Phosphoros is usable via two execution modes, i.e. the Command Line Interface (CLI) and the Graphical User Interface (GUI). The CLI allows to call pre-defined Phosphoros executables coupled with a list of parameters. A convinient way to use command lines is to create configuration files. On the other hand, the GUI allows for a more user-friendly interactive execution of Phosphoros. Users are encouraged to run Phosphoros with the GUI.