This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| Bilby (software) | |
|---|---|
| Name | Bilby |
| Title | Bilby |
| Developer | Collaborative Science Teams |
| Released | 2015 |
| Latest release | 1.x |
| Programming language | Python |
| Operating system | Cross-platform |
| Genre | Bayesian inference, data analysis |
| License | Open-source |
Bilby (software) is an open-source Python package for Bayesian inference and probabilistic modeling developed to support parameter estimation and model selection in scientific research. It provides a modular framework that integrates sampling algorithms, likelihood construction, and post-processing utilities for analyses in fields such as gravitational-wave astronomy, astrophysics, and statistical signal processing. Bilby emphasizes extensibility, reproducibility, and interoperability with established scientific tools and libraries.
Bilby was created to address complex inference problems encountered in projects like the LIGO Scientific Collaboration, the Virgo Collaboration, and the KAGRA observatory programs. It builds upon practices used in pipelines such as LALSuite, PyCBC, and GStreamer workflows while aligning with data standards set by initiatives like the Open Science Framework and the International Virtual Observatory Alliance. Early adopters included teams associated with institutions such as the University of Cambridge, Caltech, MIT, and the Max Planck Institute for Gravitational Physics.
Bilby implements a layered architecture combining front-end problem specification with back-end samplers and post-processing. The core supports likelihood wrappers compatible with analysis modules from NumPy, SciPy, Astropy, and pandas. Sampling back ends interface with libraries and tools including emcee, PyMultiNest, dynesty, UltraNest, and PTMCMC implementations. For parallel execution and workflow management Bilby can integrate with systems like MPI, Dask, Kubernetes, and cluster schedulers employed at facilities such as CERN computing centers and national high-performance facilities like NERSC and PRACE resources. Output formats and metadata align with practices from projects like HDF5, FITS, and JSON Schema specifications maintained by the IETF community.
Bilby installs via Python package tools familiar to developers and researchers, including pip, conda, and containerized deployments using Docker or Singularity. Typical usage examples demonstrate model definition, prior selection, and likelihood construction interoperable with utilities from SciPy distributions and Astropy units. Tutorials and notebooks have been distributed through platforms like the Jupyter Project, coordinated teaching materials from organizations such as the Alan Turing Institute and training workshops held at conferences including American Astronomical Society meetings and NeurIPS satellite events.
Bilby has been applied extensively in gravitational-wave parameter estimation for events cataloged by collaborations such as GWTC-1, GWTC-2, and subsequent catalogs produced by LIGO and Virgo analyses. Beyond gravitational waves, Bilby workflows have been adapted for electromagnetic transient modeling in surveys like Zwicky Transient Facility and Pan-STARRS, population inference in programs tied to the Sloan Digital Sky Survey, and cosmological parameter constraints in projects collaborating with the Dark Energy Survey and Planck teams. Integrations exist with visualization libraries such as Matplotlib, Seaborn, and corner.py for posterior plotting, and with model repositories curated by groups at Imperial College London and Monash University.
Development is driven by contributors from academic groups, observatories, and research institutes including the University of Birmingham, Cardiff University, and the Australian National University. The project management model follows practices used by repositories hosted on platforms such as GitHub with issue tracking, pull requests, and continuous integration pipelines modeled after those used by NumPy and SciPy ecosystem projects. Community engagement occurs via workshops, conference presentations at venues like APS March Meeting and ESO schools, and collaborative efforts with initiatives such as the Open Source Initiative and the Reproducible Research movement.
Performance comparisons have been reported against established inference suites used in collaborations like LALInference and against contemporary sampler packages such as emcee and PyMultiNest. Benchmarks typically measure effective sample size, wall-clock time, and scalability on compute platforms used by the European Grid Infrastructure and national supercomputing centers like XSEDE. Results discussed in technical notes and community presentations indicate that Bilby’s performance depends on sampler choice, likelihood complexity, and parallelization strategy, with demonstrable scalability when using nested sampling implementations on distributed resources, as reported in workshops hosted by GraceDB-linked teams and consortium meetings.
Bilby is distributed under an open-source license aligned with academic reuse and collaborative development models; source code, issue trackers, and contribution guidelines are available through public code hosting services like GitHub and mirrors maintained by institutional repositories at member institutions such as UNSW and Monash University. Documentation, tutorials, and example analyses are published in conjunction with community-maintained archives and teaching portals from organizations like the Software Carpentry foundation.
Category:Bayesian software Category:Astrophysics software Category:Open-source scientific software