arXiv subjects

A. Casotto

Publications and source records attributed to A. Casotto.

3 recordsLinked to original sources

arXiv2026

AstroGenesis: A Domain-Specific Multi-Agent AI for Astrophysical Research

Modern astrophysical research requires the integration of rapidly expanding scientific literature, heterogeneous observational data, and increasingly complex physical models. We introduce AstroGenesis (https://astrogenai.com), a domain-specific multi-agent AI framework that integrates literature retrieval, multiwavelength data access and analysis, theoretical modeling, and research ideation within a unified research environment.The current implementation focuses on blazar research, with specialized agents coordinated by a Supervisor Agent and Planner/Replanner architecture. These agents provide capabilities for retrieving and synthesizing literature, accessing and analyzing multiwavelength observations, performing physical modeling, and identifying research directions. A central component is the Theoretical Modeling Agent, which uses pretrained neural-network surrogate models to enable efficient broadband and multimessenger modeling through natural-language interaction. The framework provides natural-language access to science-ready multiwavelength observational data and retrieval-grounded literature through a multi-stage retrieval and ranking pipeline. The literature-retrieval system was evaluated using two benchmarks: a single-paper benchmark, in which each question targets one publication, and a multi-paper benchmark, in which questions may require evidence from several publications. At least one relevant publication was retrieved among the top five results for 76.6% of the single-paper questions and 79.2% of the multi-paper questions. Representative workflows demonstrate how literature, observational data, physical modeling, and hypothesis generation can be combined within traceable analyses. The framework is extensible to additional astrophysical domains, with the goal of streamlining research workflows and enabling efficient, connected, and reproducible scientific investigations.

astro-ph.IM↗

arXiv2025

Modeling blazar broadband emission with convolutional neural networks -- III. proton synchrotron and hybrid models

Modeling the broadband emission of blazars has become increasingly challenging with the advent of multimessenger observations. Building upon previous successes in applying convolutional neural networks (CNNs) to leptonic emission scenarios, we present an efficient CNN-based approach for modeling blazar emission under proton synchrotron and hybrid lepto-hadronic frameworks. Our CNN is trained on extensive numerical simulations generated by SOPRANO, which span a comprehensive parameter space accounting for the injection and all significant cooling processes of electrons and protons. The trained CNN captures complex interactions involving both primary and secondary particles, effectively reproducing electromagnetic and neutrino emissions. This allows for rapid and thorough exploration of the parameter space characteristic of hadronic and hybrid emission scenarios. The effectiveness of the trained CNN is demonstrated through fitting the spectral energy distributions of two prominent blazars, TXS 0506+059 and PKS 0735+178, both associated with IceCube neutrino detections. The modeling is conducted under assumptions of constant neutrino flux across distinct energy ranges, as well as by adopting a fitting that incorporates the expected neutrino event count through a Poisson likelihood method. The trained CNN is integrated into the Markarian Multiwavelength Data Center (MMDC; https://www.mmdc.am), offering a robust tool for the astrophysical community to explore blazar jet physics within a hadronic framework.

astro-ph.HE↗

arXiv2024

Modeling blazar broadband emission with convolutional neural networks -- II. External Compton model

In the context of modeling spectral energy distributions (SEDs) for blazars, we extend the method that uses a convolutional neural network (CNN) to include external inverse Compton processes. The model assumes that relativistic electrons within the emitting region can interact and up-scatter external photon originating from the accretion disk, the broad-line region, and the torus, to produce the observed high-energy emission. We trained the CNN on a numerical model that accounts for the injection of electrons, their self-consistent cooling, and pair creation-annihilation processes, considering both internal and all external photon fields. Despite the larger number of parameters compared to the synchrotron self-Compton model and the greater diversity in spectral shapes, the CNN enables an accurate computation of the SED for a specified set of parameters. The performance of the CNN is demonstrated by fitting the SED of two flat-spectrum radio quasars, namely 3C 454.3 and CTA 102, and obtaining their parameter posterior distributions. For the first source, the available data in the low-energy band allowed us to constrain the minimum Lorentz factor of the electrons, $γ_{\rm min}$, while for the second source, due to the lack of these data, $γ_{\rm min} = 10^2$ was set. We used the obtained parameters to investigate the energetics of the system. The model developed here, along with one from Bégué et al. (2023), enables self-consistent, in-depth modeling of blazar broadband emissions within leptonic scenario.

astro-ph.HE↗