A Bayesian approach to estimating the power spectrum of the Epoch of Reionization (EoR) from interferometric observations.
BayesEoR provides a means of performing a joint Bayesian analysis of models for large-spectral-scale foreground emission and a stochastic signal from redshifted 21-cm emission emitted by neutral Hydrogen during the EoR.
For a detailed description of the methodology, see Sims et al. 2016 and Sims et al. 2019. For more detail on the methodology and demonstrations using simulated data, see Sims and Pober 2019 and Burba et al. 2023.
BayesEoR relies on GPUs to perform a Cholesky decomposition on large matrices using the Matrix Algebra on GPU and Multicore Architectures (MAGMA) library. As currently implemented, the following software dependencies must be installed to run BayesEoR:
BayesEoR has been succesfully run with:
- GPUs: NVIDIA P100, V100, and A100 architectures
- MAGMA: 2.4.0, 2.5.4, and 2.7.1
- MPI:
condainstallation (mpich) and OpenMPI 4.0.5 - CUDA: 9.1.85.1 and 11.1.1
- MultiNest:
condainstallation and a source installation
This is not an exhaustive list of software versions which are compatible with our analysis, just a guide of what versions we have used succesfully in our BayesEoR analyses.
While it is in principle possible to run BayesEoR on CPUs, we strongly suggest using GPUs due to their increased speed and precision relative to CPU-based methods.
BayesEoR is written primarily in python, with the exception of the MAGMA interface which is written in C (and wrapped in python). The required python dependencies are
- astropy
- astropy-healpix
- gcc_linux-64
- h5py
- jsonargparse
- mpi4py>=3.0.0
- numpy
- pip
- pycuda
- pymultinest
- python
- pyuvdata
- rich
- scipy
- setuptools
- setuptools_scm
- sphinx
If you with to install all of these dependencies with conda, you can do so using the included environment.yaml file via
conda env create -f environment.yaml
If you have pre-configured installations of CUDA or MPI, e.g. installations optimized/configured for a compute cluster, we suggest installing pycuda and/or mpi4py via pip (and commenting out pycuda and mpi4py in the environment.yaml file). If you install these dependencies with conda, conda will install its own CUDA and MPI binaries which may not be desirable. For pycuda, you need only have the path to your cuda binaries in your bash PATH variable prior to pip installation. For mpi4py, see this article to ensure mpi4py points to the desired MPI installation.
Similarly, if using a pre-configured implementation of MultiNest, pymultinest can also be installed with pip and forced to point to a particular installation by including the MultiNest installation in your LD_LIBRARY_PATH. See the pymultinest documentation for more details.
Documentation on how to estimate the power spectrum of redshifted 21-cm emission in a radio interferometric data set using BayesEoR is hosted on ReadTheDocs.
Users of the code are requested to cite the BayesEoR papers:
- https://ui.adsabs.harvard.edu/abs/2016MNRAS.462.3069S/abstract
- https://ui.adsabs.harvard.edu/abs/2019MNRAS.484.4152S/abstract
- https://ui.adsabs.harvard.edu/abs/2019MNRAS.488.2904S/abstract
- https://ui.adsabs.harvard.edu/abs/2023MNRAS.520.4443B/abstract
in their publications.
There are two ways to interface with variables in BayesEoR: command line arguments or config files. For a list of available command line arguments and their descriptions, run
python run-analysis.py --help
or see the documentation for more info on the analysis parameters used by BayesEoR.
The jsonargparse package allows for all of these command line arguments to be set via a yaml configuration file. An example yaml file has been provided (example-config.yaml). Any variable that can be set via a command line argument can also be set in this yaml configuration file (command line arguments containing dashes in the variable name must be replaced with underscores, i.e. the command line argument --data-path can be set in the configuration file via data_path: "/path/to/data.npy"). The example configuration file also specifies the minimally sufficient variables that must be set for a BayesEoR analysis.
run-analysis.py provides an example driver script for running BayesEoR. This file contains all of the necessary steps to set up the PowerSpectrumPosteriorProbability class and to run MultiNest and obtain power spectrum posteriors. If using a configuration file, this driver script can be run via
python run-analysis.py --config /path/to/config.yaml
BayesEoR is an open source project and contributions to this package in any form are very welcome (e.g. new features, feature requests, bug reports, documentation fixes). Please make such contributions in the form of an issue and/or pull request. For any additional questions or comments, please contact one of the BayesEoR project managers:
- Peter Sims - psims3 [at] asu.edu
- Jacob Burba - jacob.burba [at] manchester.ac.uk
- Jonathan Pober - jonathan_pober [at] brown.edu