Decoding spectral energy distributions of dust-obscured starburst-AGN

We present BayeSED, a general purpose tool for doing Bayesian analysis of SEDs by using whatever pre-existing model SED libraries or their linear combinations. The artificial neural networks (ANNs), principal component analysis (PCA) and multimodal nested sampling (MultiNest) techniques are employed to allow a highly efficient sampling of posterior distribution and the calculation of Bayesian evidence. As a demonstration, we apply this tool to a sample of hyperluminous infrared galaxies (HLIRGs). The Bayesian evidences obtained for a pure Starburst, a pure AGN, and a linear combination of Starburst+AGN models show that the Starburst+AGN model have the highest evidence for all galaxies in this sample. The Bayesian evidences for the three models and the estimated contributions of starburst and AGN to infrared luminosity show that HLIRGs can be classified into two groups: one dominated by starburst and the other dominated by AGN. Other parameters and corresponding uncertainties about starburst and AGN are also estimated by using the model with the highest Bayesian evidence. We found that the starburst region of the HLIRGs dominated by starburst tends to be more compact and has a higher fraction of OB star than that of HLIRGs dominated by AGN. Meanwhile, the AGN torus of the HLIRGs dominated by AGN tend to be more dusty than that of HLIRGs dominated by starburst. These results are consistent with previous researches, but need to be tested further with larger samples. Overall, we believe that BayeSED could be a reliable and efficient tool for exploring the nature of complex systems such as dust-obscured starburst-AGN composite systems from decoding their SEDs.