| Issue |
A&A
Volume 683, March 2024
|
|
|---|---|---|
| Article Number | A185 | |
| Number of page(s) | 12 | |
| Section | Numerical methods and codes | |
| DOI | https://doi.org/10.1051/0004-6361/202348566 | |
| Published online | 19 March 2024 | |
Application of neural networks to synchro-Compton blazar emission models★
1
Department of Physics, National and Kapodistrian University of Athens, University Campus Zografos,
15784
Athens,
Greece
e-mail: tzavellas.anastasios@gmail.com; gevas@phys.uoa.gr
2
Institute of Accelerating Systems & Applications, University Campus Zografos,
Athens,
Greece
Received:
10
November
2023
Accepted:
21
December
2023
Context. Jets from supermassive black holes at the centers of active galaxies are the most powerful and persistent sources of electromagnetic radiation in the Universe. To infer the physical conditions in the otherwise out-of-reach regions of extragalactic jets, we usually rely on fitting their spectral energy distributions (SEDs). The calculation of radiative models for the jet’s non-thermal emission usually relies on numerical solvers of coupled partial differential equations.
Aims. In this work, we use machine learning to tackle the problem of high computational complexity to significantly reduce the SED model evaluation time, which is necessary for SED fittings carried out with Bayesian inference methods.
Methods. We computed the SEDs based on the synchrotron self-Compton model for blazar emission using the radiation code ATHEvA. We used them to train neural networks (NNs) to explore whether they can replace the original code, which is computationally expensive.
Results. We find that a NN with gated recurrent unit neurons (GRUN) can effectively replace the ATHEvA leptonic code for this application, while it can be efficiently coupled with Markov chain Monte Carlo (MCMC) and nested sampling algorithms for fitting purposes. We demonstrate this approach through an application to simulated data sets, as well as a subsequent application to observational data.
Conclusions. We present a proof-of-concept application of NNs to blazar science as the first step in a list of future applications involving hadronic processes and even larger parameter spaces. We offer this tool to the community through a public repository.
Key words: radiation mechanisms: non-thermal / radiative transfer / methods: numerical / methods: statistical
The results of our work are available in GitHub; https://github.com/tzavellas/blazar_ml. This includes: (a) the NN and accompanied code produced to train them, (b) code for visualization of results in python and jupyter notebooks with instructions, and (c) part of the ATHEvA datasets that can be used for evaluation and plotting examples.
© The Authors 2024
Open Access article, published by EDP Sciences, under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This article is published in open access under the Subscribe to Open model. Subscribe to A&A to support open access publication.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.