A research agent for blazars and a neural network for the photon ring

Astrophysics is a logistical discipline. It requires pulling together a fast-growing literature, heterogeneous data from across the spectrum, and increasingly complex physical models. Some teams are trying to organize that work rather than automate single tasks.

A multi-agent system for blazars

Authors from Armenia, Italy, and the United States, including David Pasham and Alessandro Tramacere, describe the AstroGenesis framework. It is a domain-specific multi-agent system that combines literature search and synthesis, access to and analysis of multi-frequency data, theoretical modeling, and the generation of research directions. A supervising agent works with a planner and replanner architecture. At the center sits a theoretical modeling agent that uses previously trained neural surrogate models. This allows complex broadband, multi-aspect modeling to proceed as a natural language conversation, without launching full computations every time.

The authors primarily measured the literature search layer on two test sets: one where each question concerns a single publication, and one where the answer requires evidence from several papers. At least one relevant publication appeared in the top five results for 76.6 percent of questions in the first set and 79.2 percent in the second. Example workflows show how literature, observational data, modeling, and hypothesis generation form a traceable analysis. The creators plan to expand the system to other fields of astrophysics.

A ring that reveals spin

A separate paper focuses on the measurement tool itself. Very long baseline interferometry could directly image black hole structures on the scale of the event horizon, including the photon ring. The n = 1 subring is especially valuable because its radial profile and azimuthal brightness modulation encode information about the black hole's spin. In observations, it overlaps with the emission of the n = 0 subring, so it must be isolated first.

Courtney Duong, Frank Myhre, and Joseph Farah trained a convolutional neural network on 111,000 simulated ring images generated with the eht-imaging tool. The network reconstructs the radial and angular profile of the n = 1 subring with a mean normalized cross-correlation of about 0.99. It reports no false detections when no ring is present in the data, and the extracted features allow recovery of the brightness modulation sensitive to spin. On images from KerrBAM simulations and magnetohydrodynamic ones, it reconstructs the general radial profile, though discrepancies appear in the intensity profile. The authors link this work to future missions, such as the planned Black Hole Explorer.