Welcome to BXA’s documentation!

About Bayesian X-ray Analysis (BXA)

BXA connects the nested sampling algorithm MultiNest to the X-ray spectral analysis environments Xspec/Sherpa for Bayesian Parameter Estimation and Model comparison.

BXA provides the following features:

  • parameter estimation in arbitrary dimensions, which involves:
    • finding the best fit
    • computing error bars
    • computing marginal probability distributions
  • plotting of spectral model vs. the data:
    • for the best fit
    • for each of the solutions (posterior samples)
    • for each component
  • model selection:
    • computing the evidence for the considered model, ready for use in computing Bayes factors
    • unlike likelihood-ratios, not limited to nested models
  • model discovery:
    • visualize deviations between model and data with Quantile-Quantile (QQ) plots. QQ-plots do not require binning and are more comprehensive than residuals. This will give you ideas on when to introduce more complex models, which may again be tested with model selection

Usage

  • We start with some notes on best practice in currently common Maximum Likelihood analysis (fitting) and Bayesian analysis.

The usage is similar in Sherpa and Xspec conceptionally:

Get BXA

See the code repository page

Citing BXA correctly

Refer to the accompaning paper Buchner et al. (2014) which gives introduction and detailed discussion on the methodology and its statistical footing.

We suggest giving credit to the developers of Sherpa/Xspec, MultiNest and of this software. As an example:

For analysing X-ray spectra, we use the analysis software BXA (\ref{Buchner2014}),
which connects the nested sampling algorithm MultiNest (\ref{FerozHobson2010})
with the fitting environment CIAO/Sherpa (\ref{Fruscione2006}).

Where the BibTex entries are:

  • for BXA, PyMultiNest software, and the contributions to X-ray spectral analysis methodology (model comparison, model discovery, Experiment design, Model discovery through QQ-plots):

  • for MultiNest: see MultiNest

  • for Sherpa: see Sherpa

  • for Xspec: see Xspec

Installation

However, you need to have pymultinest and MultiNest installed and working.

You need to have Sherpa or Xspec installed and its environment loaded.

The following commands should not yield any error message:

$ python -c 'import pymultinest'
$ python -c 'import xspec'
$ sherpa

Install the required python packages, through your package manager, through pip or easy_install. For example:

$ yum install ipython python-matplotlib scipy numpy matplotlib
$ apt-get install python-numpy python-scipy python-matplotlib ipython

$ pip install progressbar --user

BXA itself can installed directly from PyPI using pip or easy_install:

$ pip install bxa # --user if you want to install locally

The source code is available from https://github.com/JohannesBuchner/BXA, so alternatively you can download and install it:

$ python setup.py install

Or if you only want to install it for the current user:

$ python setup.py install --user

Running

In Sherpa, load the package:

jbuchner@ds42 ~ $ sherpa
-----------------------------------------------------
Welcome to Sherpa: CXC's Modeling and Fitting Package
-----------------------------------------------------
CIAO 4.4 Sherpa version 2 Tuesday, June 5, 2012

sherpa-1> import bxa.sherpa as bxa
sherpa-2> bxa.run_model?

For Xspec, start python or ipython:

jbuchner@ds42 ~ $ ipython
In [1]: import xspec

In [2]: import bxa.xspec as bxa

In [3]: bxa.standard_analysis?

Now you can use bxa.

Indices and tables

Table Of Contents

Next topic

BXA/Sherpa

This Page