BXA

Free plan Linux

What BXA is

BXA connects the Xspec and Sherpa X-ray spectral analysis environments to UltraNest for Bayesian parameter estimation and model comparison. It estimates parameters across arbitrary dimensions, finds best fits, and computes error bars and marginal probability distributions. Its model evidence calculations support Bayes factors and comparisons between models that are not nested. Quantile-Quantile plots show model-data deviations without binning. The fitting method explores parameter spaces without requiring good starting points and continues until convergence. The documentation highlights systematic work on large datasets, comparisons among multiple models, and analysis of low-count datasets with realistic models. With mpi4py installed, MPI parallelisation can extend inference from laptops to computing clusters. When used through pyxspec, BXA produces chain FITS files compatible with Xspec/pyxspec MCMC features and error propagation tools. It is free software under GNU GPL v3 or later and can be installed with pip or conda, but requires Sherpa or Xspec to be installed and configured. Supported operating systems follow CIAO/Sherpa or HEASoft/Xspec; release testing is stated to cover Ubuntu Linux.

Who it suits

It suits researchers working with X-ray spectra in Sherpa or Xspec who need Bayesian parameter estimation or model comparison. Its stated uses include large datasets, multiple models, and low-count data.

What it does well

Computes parameter estimates, error bars, and distributions.
Compares models that are not nested.
Quantile-Quantile plots do not require binning.
MPI can scale inference when mpi4py is installed.
Free under GNU GPL v3 or later.

What to know first

Requires Sherpa or Xspec to be installed and configured.
Release testing covers Ubuntu Linux only.
MPI scaling requires mpi4py.
Version 5 changes Xspec prior limits to soft limits.

Plans

Plan	Billed and limits	Price
BXA	Free software under GNU GPL v3 or later	Free

As johannesbuchner.github.io lists them, read 5 Oct 2026. The maker's own page has the final word.

Getting started

Install Sherpa or Xspec and configure it.
Install BXA with pip or conda.
Use BXA with Xspec or Sherpa for Bayesian analysis.

Verdict

BXA adds Bayesian inference and model comparison to Sherpa or Xspec workflows, with options for MPI processing. It requires one of those analysis environments, and its version 5 prior change may affect users relying on hard parameter limits.

Questions

How much does BXA cost?

The listed BXA plan costs nothing. BXA is free software under GNU GPL v3 or later.

Which platforms does BXA support?

Linux is listed, and platform coverage is desktop. BXA runs on operating systems supported by CIAO/Sherpa or HEASoft/Xspec; release testing covers Ubuntu Linux.

What does BXA need before installation?

BXA can be installed with pip or conda. It requires Sherpa or Xspec to be installed and configured.

Can BXA run inference in parallel?

Yes. MPI parallelisation is supported, and mpi4py enables scaling from laptops to computing clusters.

What license does BXA use?

BXA is free software under GNU GPL v3 or later.

Facts from BXA's own pages

Automated fitting: The fitting algorithm explores parameter spaces without requiring good starting points and runs until convergence. · johannesbuchner.github.io, 5 Oct 2026
Installation: BXA can be installed with pip or conda and requires either Sherpa or Xspec to be installed and configured. · johannesbuchner.github.io, 5 Oct 2026
Intended use: The documentation highlights systematic analysis of large datasets, comparing multiple models, and analyzing low-count datasets with realistic models. · johannesbuchner.github.io, 5 Oct 2026
Model comparison: It computes model evidence for Bayes factors and supports comparing models that are not nested. · johannesbuchner.github.io, 5 Oct 2026
Model discovery: It provides Quantile-Quantile plots to visualize deviations between a model and data without binning. · johannesbuchner.github.io, 5 Oct 2026
Notable change: In version 5, Xspec priors use soft parameter limits instead of hard parameter limits, which the release notes flag as a potentially breaking change. · johannesbuchner.github.io, 5 Oct 2026
Parallel processing: MPI parallelisation can scale inference from laptops to computing clusters when mpi4py is installed. · johannesbuchner.github.io, 5 Oct 2026
Parameter estimation: It estimates parameters in arbitrary dimensions, finds best fits, computes error bars and marginal probability distributions, and supports MPI parallelisation. · johannesbuchner.github.io, 5 Oct 2026
Purpose: BXA connects Xspec and Sherpa X-ray spectral analysis environments to UltraNest for Bayesian parameter estimation and model comparison. · johannesbuchner.github.io, 5 Oct 2026
Security and warranty: The repository publishes a GNU GPL license whose text disclaims warranties for the program. · github.com, 5 Oct 2026
Support: The project directs users to file bug reports and feedback as GitHub issues and asks bug reports to include operating system, Sherpa or Xspec version, setup details, and reproduction steps. · johannesbuchner.github.io, 5 Oct 2026
Supported systems: BXA runs on operating systems supported by CIAO/Sherpa or HEASoft/Xspec, while release testing is stated to cover Ubuntu Linux only. · johannesbuchner.github.io, 5 Oct 2026
Xspec compatibility: When run from pyxspec, BXA provides chain FITS files compatible with Xspec/pyxspec MCMC features and error propagation tools. · johannesbuchner.github.io, 5 Oct 2026

Where BXA stands on Freedom251

Astronomy Apps#12 of 35