Interpreting Multi-band Galaxy Observations with Large Language Model-Based Agents

TL;DR

This paper introduces mephisto, a multi-agent framework using large language models to interpret multi-band galaxy observations, demonstrating near-human reasoning and accelerating astronomical research workflows.

Contribution

It presents a novel multi-agent system that emulates human reasoning in astronomy, integrating LLMs with spectral models for interpreting galaxy data, including new approaches for open-world learning.

Findings

Mephisto achieves near-human reasoning in galaxy interpretation.

The framework effectively handles new galaxy populations like 'Little Red Dot' galaxies.

Demonstrates potential for automating and accelerating astronomical research.

Abstract

Astronomical research traditionally relies on extensive domain knowledge to interpret observations and narrow down hypotheses. We demonstrate that this process can be emulated using large language model-based agents to accelerate research workflows. We propose mephisto, a multi-agent collaboration framework that mimics human reasoning to interpret multi-band galaxy observations. mephisto interacts with the CIGALE codebase, which includes spectral energy distribution (SED) models to explain observations. In this open-world setting, mephisto learns from its self-play experience, performs tree search, and accumulates knowledge in a dynamically updated base. As a proof of concept, we apply mephisto to the latest data from the James Webb Space Telescope. mephisto attains near-human proficiency in reasoning about galaxies' physical scenarios, even when dealing with a recently discovered population of "Little Red Dot" galaxies. This represents the first demonstration of agentic research in astronomy, advancing towards end-to-end research via LLM agents and potentially expediting astronomical discoveries.