Comparing multi-field primordial feature models with the Planck data

TL;DR

This paper develops a comprehensive methodology to compare complex multi-field primordial feature models with Planck data, enabling detailed analysis of potential inflationary signals and setting a foundation for future cosmological data screening.

Contribution

It introduces a numerical prediction and Bayesian comparison framework for multi-field inflation models, addressing previous analytical limitations and classifying feature signals across frequency ranges.

Findings

No statistically significant primordial feature candidates found.

Methodology efficiently explores complex model spectra.

Framework supports future data-driven primordial feature analysis.

Abstract

In this paper, we use a complete model of classical primordial standard clocks as an example to develop a methodology of directly comparing numerical predictions from complicated multi-field feature models with the Planck data, including the Planck 2018 Plik unbinned likelihood and the statistically most powerful CamSpec 2020 likelihood for temperature and polarization data. As this two-field inflationary model offers a plethora of primordial feature spectra that represent combinations of sharp and resonance feature signals non-trivially distributed over extended cosmological scales, its data comparison has not been satisfactorily addressed by previous attempts using analytical templates. The method of this paper, consisting of numerical prediction, effective parameter construction and nested sampling data comparison, allows us to efficiently explore every possible spectra from the model. We classify the resulting feature candidates in three different frequency ranges. We use the Bayesian evidences to assess the statistical significance of the candidates over the baseline model, taking into account the effect of additional parameters and the look-elsewhere effect. Although none of the candidates is statistically significant, the methodology of this paper can be used to facilitate the future model-building and data-screening process of primordial features, and the candidates can be subjected to further tests with data from the upcoming cosmic microwave background polarization observations and galaxy surveys.