$ exttt{unimpeded}$: A Public Nested Sampling Database for Bayesian Cosmology

Our group is excited to announce a new paper, “$\texttt{unimpeded}$: A Public Nested Sampling Database for Bayesian Cosmology” 2511.05470, led by PhD student Dily Duan Yi Ong and Will Handley. This work introduces a powerful new Python library and data repository designed to significantly lower the computational barrier for advanced cosmological analysis.

The Challenge: The Bayesian Evidence Bottleneck

In modern cosmology, Bayesian inference is the cornerstone of our analytical toolkit. While methods like Markov Chain Monte Carlo (MCMC) are effective for estimating the parameters of a given model, they fall short when we need to compare different models or robustly quantify tensions between datasets. These more advanced tasks require the Bayesian evidence, a quantity that nested sampling algorithms like PolyChord are designed to calculate. However, running these calculations is computationally intensive, often requiring supercomputing resources that are not accessible to all researchers. This creates a significant bottleneck, slowing down the pace of discovery and reproducibility.

Our Solution: unimpeded

To address this challenge, we developed unimpeded, a publicly available tool that provides pre-computed nested sampling results for a wide array of cosmological models and datasets. By making these computationally expensive products freely accessible, we aim to democratize access to state-of-the-art Bayesian analysis. The chains were generated using the powerful Cobaya sampling framework (10.1088/1475-7516/2021/05/057) and the CAMB Boltzmann solver.

unimpeded offers three core features that streamline the process of cosmological data analysis:

• A Public Grid of Nested Sampling Chains: The repository contains a vast, pre-computed grid of results spanning eight cosmological models (the standard ΛCDM model and seven popular extensions) tested against 39 modern cosmological datasets and their combinations. This provides a standardized baseline for the community and saves thousands of hours of valuable CPU time. All chains are easily accessible via the unimpeded API and ready for analysis with our visualization package, anesthetic (10.21105/joss.01414).

• Archival and Reproducibility with Zenodo: All data products are permanently archived on Zenodo, each with a citable Digital Object Identifier (DOI). The unimpeded library provides a seamless interface to download these chains, ensuring that analyses are transparent, reproducible, and built on a solid, open-science foundation. This approach mirrors the success of resources like the Planck Legacy Archive but extends its principles to the nested sampling era.

• An Integrated Tension Calculator: Quantifying statistical tensions between different experimental results is one of the most pressing challenges in cosmology today. unimpeded includes a built-in calculator that computes six distinct tension metrics from the nested sampling chains. This allows researchers to rapidly assess the consistency of different datasets within various theoretical models, leveraging robust statistical methods such as the evidence ratio developed in previous work (10.1103/PhysRevD.100.043504).

By providing these resources, unimpeded not only accelerates research but also fosters a more collaborative and reproducible scientific environment. We invite the community to explore the database, utilize the tools for their own analyses, and build upon this foundation to push the frontiers of cosmological discovery.

Content generated by gemini-2.5-pro using this prompt.

Image generated by imagen-4.0-generate-001 using this prompt.