Investigating scalar-tensor-gravity with statistics of the cosmic large-scale structure

TL;DR

This paper assesses how future large-scale structure observations can constrain complex scalar-tensor theories of gravity, demonstrating that even intricate models can be tightly limited with advanced statistical methods.

Contribution

It introduces a comprehensive analysis of Horndeski theories using combined cosmological data and compares Fisher and MCMC methods to evaluate parameter constraints.

Findings

Complex models can be constrained to within 10% of Newtonian gravity.

Strong parameter correlations are identified.

MCMC results show larger uncertainties than Fisher forecasts.

Abstract

Future observations of the large-scale structure have the potential to investigate cosmological models with a high degree of complexity, including the properties of gravity on large scales, the presence of a complicated dark energy component, and the addition of neutrinos changing structures on small scales. Here we study Horndeski theories of gravity, the most general minimally coupled scalar-tensor theories of second order. While the cosmological background evolution can be described by an effective equation of state, the perturbations are characterised by four free functions of time. We consider a specific parametrisation of these functions tracing the dark energy component. The likelihood of the full parameter set resulting from combining cosmic microwave background primary anisotropies including their gravitational lensing signal, tomographic angular galaxy clustering and weak cosmic shear, together with all possible non-vanishing cross-correlations is evaluated; both with the Fisher-formalism as well as without the assumption of a specific functional form of the posterior through Monte-Carlo Markov-chains (MCMCs). Our results show that even complex cosmological models can be constrained and could exclude variations of the effective Newtonian gravitational coupling larger than 10\% over the age of the Universe. In particular, we confirm strong correlations between parameter groups. Furthermore, we find that the expected contours from MCMC are significantly larger than those from the Fisher analysis even with the vast amount of signal provided by stage IV experiments, illustrating the importance of a proper treatment of non-Gaussian likelihoods and the high level of precision needed for unlocking the sensitivity on gravitational parameters.