A Step Toward Model Comparison: Connecting Electroweak-Scale Observables to BSM through EFT and Bayesian Statistics

TL;DR

This paper develops a Bayesian framework to compare various BSM theories by connecting electroweak observables to EFT parameters, enabling model discrimination and parameter inference using precision data.

Contribution

It introduces a systematic method to relate BSM models to EFT operators and employs Bayesian inference to compare models based on electroweak data.

Findings

Bayesian inference can distinguish between different BSM models.

The approach links BSM parameters to observable effects via EFT.

Theoretical constraints effectively exclude large BSM parameter regions.

Abstract

Recognizing the potential of effective field theories to posit multiple BSM scenarios in similar footing, with a possibility to compare them, we inspect the effects of 11 single scalar-multiplet extensions of the SM on the combined set of electroweak precision observables and Higgs signal strength data, by systematically integrating out the heavy multiplets and computing the resulting SMEFT operators and Wilson coefficients (WCs) up to one-loop level. Noting that multiple BSM models give rise to a degenerate set of WCs, we then perform Bayesian statistical inference both directly on the BSM parameters and on the associated set of independent WCs. Using the posteriors of the BSM parameters, we infer the respective (correlated) WC-distributions and compare both the model-independent and dependent analyses by overlaying the 2-D marginal WC-posteriors from both processes, thus laying the ground for a data-driven attempt to compare diverse BSM theories of different origins, and hopefully, a possible way to approach the intractable inverse problem. We also demonstrate, with an example model, the crucial role of theoretical constraints to rule out large chunks of BSM parameter spaces. The entirety of numerical results is available in GitHub.