BlaST -- A Machine-Learning Estimator for the Synchrotron Peak of Blazars

TL;DR

BlaST is a machine-learning tool that accurately estimates the synchrotron peak frequency in blazar spectral energy distributions, accounting for additional components and providing uncertainty measures, thus improving classification accuracy.

Contribution

The paper introduces BlaST, a novel machine-learning algorithm that automates and enhances the estimation of blazar synchrotron peaks, including uncertainty quantification and component separation.

Findings

BlaST improves peak estimation accuracy in complex SEDs.

The tool effectively accounts for host galaxy and disk emission.

Application to Fermi 4LAC-DR2 catalog demonstrates enhanced reliability.

Abstract

Active Galaxies with a jet pointing towards us, so-called blazars, play an important role in the field of high-energy astrophysics. One of the most important features in the classification scheme of blazars is the peak frequency of the synchrotron emission ($\nu_{\rm peak}^S$) in the spectral energy distribution (SED). In contrast to standard blazar catalogs that usually calculate the $\nu_{\rm peak}^S$ manually, we have developed a machine-learning algorithm - BlaST - that not only simplifies the estimation, but also provides a reliable uncertainty evaluation. Furthermore, it naturally accounts for additional SED components from the host galaxy and the disk emission, which may be a major source of confusion. Using our tool, we re-estimate the synchrotron peaks in the Fermi 4LAC-DR2 catalog. We find that BlaST, improves the $\nu_{\rm peak}^S$ estimation especially in those cases where the contribution of components not related to the jet is important.