jaxspec : a fast and robust Python library for X-ray spectral fitting

TL;DR

jaxspec is a Python library that enables fast, robust, and fully Bayesian X-ray spectral fitting using differentiable likelihoods on CPUs and GPUs, significantly improving efficiency and scalability for high-resolution data analysis.

Contribution

The paper introduces jaxspec, a new Python package leveraging JAX for differentiable Bayesian spectral fitting, offering improved speed and robustness over existing software.

Findings

jaxspec achieves results identical to existing software.

It is approximately 10 times faster than current alternatives.

Variational inference on GPU produces reliable results in under 10 minutes.

Abstract

Context. Inferring spectral parameters from X-ray data is one of the cornerstones of high-energy astrophysics, and is achieved using software stacks that have been developed over the last twenty years and more. However, as models get more complex and spectra reach higher resolutions, these established software solutions become more feature-heavy, difficult to maintain and less efficient. Aims. We present jaxspec, a Python package for performing this task quickly and robustly in a fully Bayesian framework. Based on the JAX ecosystem, jaxspec allows the generation of differentiable likelihood functions compilable on core or graphical process units (resp. CPU and GPU), enabling the use of robust algorithms for Bayesian inference. Methods. We demonstrate the effectiveness of jaxspec samplers, in particular the No U-Turn Sampler, using a composite model and comparing what we obtain with the existing frameworks. We also demonstrate its ability to process high-resolution spectroscopy data and using original methods, by reproducing the results of the Hitomi collaboration on the Perseus cluster, while solving the inference problem using variational inference on a GPU. Results. We obtain identical results when compared to other softwares and approaches, meaning that jaxspec provides reliable results while being $\sim 10$ times faster than existing alternatives. In addition, we show that variational inference can produce convincing results even on high-resolution data in less than 10 minutes on a GPU. Conclusions. With this package, we aim to pursue the goal of opening up X-ray spectroscopy to the existing ecosystem of machine learning and Bayesian inference, enabling researchers to apply new methods to solve increasingly complex problems in the best possible way. Our long-term ambition is the scientific exploitation of the data from the newAthena X-ray Integral Field Unit (X-IFU).