Supernova Remnants (SNRs) constitute a remarkably heterogeneous population that exhibits a wide variety of observational properties. These differences are likely determined by the progenitor star nature, inhomogeneities in the pre-SN circumstellar material, explosion dynamics and other factors difficult to constrain. Is there any underlying order in this apparent chaos? To systematically search for patterns in the Galactic SNR population, we employed an unsupervised approach that, feeding on multiwavelength imagery, allows for an unbiased clustering of SNRs in a multidimensional latent space. Here, we present a pipeline consisting of (i) a feature extraction stage that achieves a compact representation of the sources using infrared and radio continuum imagery; and (ii) a clustering stage that groups together objects that display similar features in the resulting latent space. By applying this pipeline to a representative sample of Galactic SNRs, we found several physically meaningful subsets sharing common observational features.
Bordiu, C., Bufano, F., Cecconello, T., Sciacca, E., Riggi, S., Vizzari, G. (2023). Patterns in the Chaos: An Unsupervised View of Galactic Supernova Remnants. In Machine Learning for Astrophysics Proceedings of the ML4Astro International Conference 30 May - 1 Jun 2022 (pp.61-65). Springer Science and Business Media B.V. [10.1007/978-3-031-34167-0_13].
Patterns in the Chaos: An Unsupervised View of Galactic Supernova Remnants
Cecconello, T.;Vizzari, G.Ultimo
2023
Abstract
Supernova Remnants (SNRs) constitute a remarkably heterogeneous population that exhibits a wide variety of observational properties. These differences are likely determined by the progenitor star nature, inhomogeneities in the pre-SN circumstellar material, explosion dynamics and other factors difficult to constrain. Is there any underlying order in this apparent chaos? To systematically search for patterns in the Galactic SNR population, we employed an unsupervised approach that, feeding on multiwavelength imagery, allows for an unbiased clustering of SNRs in a multidimensional latent space. Here, we present a pipeline consisting of (i) a feature extraction stage that achieves a compact representation of the sources using infrared and radio continuum imagery; and (ii) a clustering stage that groups together objects that display similar features in the resulting latent space. By applying this pipeline to a representative sample of Galactic SNRs, we found several physically meaningful subsets sharing common observational features.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
