ELISA: A Bayesian X/$γ$-ray Spectral Analysis Toolkit

Abstract

Forward-folded X/$γ$-ray spectral fitting increasingly benefits from Bayesian workflows to better capture counting statistics, background uncertainties, and complex parameter degeneracies, while fully reproducible end-to-end implementations are still not always straightforward to deploy in routine analyses. We present ELISA (Efficient Library for Spectral Analysis), an open-source Python toolkit that streamlines the full spectral analysis loop for X/$γ$-ray astronomy: data loading, model construction, inference, and model validation and comparison. ELISA couples a JAX-accelerated forward model with a unified fitting interface, supporting both rapid optimization and modern Bayesian sampling. Built-in diagnostics make convergence checks and predictive assessment a standard part of the workflow, motivating principled model refinement when needed. Using a spectral fitting example of the Insight-HXMT Crab observation, we find substantial improvements in sampling efficiency compared with XSPEC's default MCMC implementation. ELISA is available at https://github.com/wcxve/elisa and on PyPI at https://pypi.org/project/astro-elisa.