Astrocook v2: Advancing Quasar Spectral Analysis with Improved Algorithms and AI Integration
Astrocook is a Python package designed for the analysis of quasar spectra, with a focus on modelling absorption features from intervening systems. Since its initial release, it has provided a flexible framework where each processing step—continuum fitting, line detection, profile fitting, and more—is implemented as a modular recipe, allowing users to construct reproducible, customizable workflows.
The upcoming v2 release marks a significant evolution of the package. It introduces improved algorithms for continuum normalization and Voigt profile fitting, with enhanced accuracy and stability across a wide range of spectral qualities. In addition, v2 emphasizes interoperability and AI-assisted workflows. Each recipe exposes a clean, standardized API designed for seamless integration with external tools—including large language models and intelligent agents. This shift enables interactive, human-in-the-loop data analysis and paves the way for autonomous workflows driven by AI reasoning.
Astrocook v2 aims to bridge the gap between domain expertise and automation, empowering researchers to process large spectral datasets with greater control and insight. We will present the architecture of the new version, key algorithmic updates, and examples of AI-guided spectral analysis using LLMs.
