From seagull to hummingbird: New diagnostic methods for resolving galaxy activity

¹ Physics Department, and Institute of Theoretical and Computational Physics, University of Crete, 71003 Heraklion, Greece
² Institute of Astrophysics, Foundation for Research and Technology-Hellas, 71110 Heraklion, Greece
³ Center for Astrophysics | Harvard & Smithsonian, 60 Garden St., Cambridge, MA 02138, USA
⁴ Astronomical Institute, Academy of Sciences, Boční II 1401, CZ-14131 Prague, Czech Republic
⁵ ALMA Sistemi Srl, Guidonia, (Rome) 00012, Italy
⁶ Quantum Innovation Pc, Chania 73100, Greece

^⋆ Corresponding author; cdaoutis@physics.uoc.gr

Received: 1 July 2024
Accepted: 7 November 2024

Abstract

Context. One of the principal challenges in astrophysics involves the classification of galaxies based on their activity. Currently, the characterization of galactic activity usually requires multiple diagnostics to fully cover the diverse spectrum of galaxy activity types. Additionally, the presence of multiple sources of excitation with similar observational signatures hinders the exploration of the activity of a galaxy.

Aims. In this study our objective is to develop an activity diagnostic tool that addresses the degeneracy inherent in the existing emission line diagnostics by identifying the underlying excitation mechanisms of the principal components of a mixed-activity galaxy (star formation, active nucleus, or old stellar populations) and identifying the dominant ones.

Methods. We utilized the random forest machine-learning algorithm, trained on three primary activity classes: star-forming, active galactic nucleus (AGN), and passive; these classes represent the three key gas excitation mechanisms. This diagnostic relies on four discriminating features: the equivalent widths of three spectral lines, [O III] λ5007, [N II] λ6584, and Hα, along with the D4000 continuum break index.

Results. We find that this classifier achieves almost perfect performance scores in the principal activity classes. In particular, the achieved overall accuracy is ∼99%, while the recall scores are ∼100% for star-forming, ∼98% for AGN, and ∼99% for passive. The nearly perfect scores achieved enable the decomposition of mixed-activity classes into the three primary gas excitation mechanisms with high confidence, thereby resolving the degeneracy inherent in current activity classification methods. Furthermore, we find that our classifier scheme can be simplified to a two-dimensional diagnostic diagram of D4000 index versus the log₁₀(EW([O III])²) line without significant loss of its diagnostic power.

Conclusions. We introduce a diagnostic capable of classifying galaxies based on their primary gas excitation mechanisms. Simultaneously, it can deconstruct the activity of mixed-activity galaxies into these principal components. This diagnostic encompasses the entire range of galaxy activity. Additionally, the D4000 index serves as a valuable indicator for resolving the degeneracy among various activity components by estimating the age of the stellar populations within a galaxy.