Models of f(R) Cosmic Acceleration that Evade Solar-System Tests

TL;DR

This paper investigates a class of f(R) gravity models that can explain cosmic acceleration without a cosmological constant and pass solar-system tests, emphasizing the importance of galactic halo structure and future observational constraints.

Contribution

It demonstrates that viable f(R) models can evade solar-system tests through high-curvature locking and highlights the need for cosmological simulations to assess their viability.

Findings

Solar-system tests alone weakly constrain f(R) models.

Galactic halo structure is crucial for model viability.

Future measurements can detect small deviations in f(R) models.

Abstract

We study a class of metric-variation f(R) models that accelerates the expansion without a cosmological constant and satisfies both cosmological and solar-system tests in the small-field limit of the parameter space. Solar-system tests alone place only weak bounds on these models, since the additional scalar degree of freedom is locked to the high-curvature general-relativistic prediction across more than 25 orders of magnitude in density, out through the solar corona. This agreement requires that the galactic halo be of sufficient extent to maintain the galaxy at high curvature in the presence of the low-curvature cosmological background. If the galactic halo and local environment in f(R) models do not have substantially deeper potentials than expected in LCDM, then cosmological field amplitudes |f_R| > 10^{-6} will cause the galactic interior to evolve to low curvature during the acceleration epoch. Viability of large-deviation models therefore rests on the structure and evolution of the galactic halo, requiring cosmological simulations of f(R) models, and not directly on solar-system tests. Even small deviations that conservatively satisfy both galactic and solar-system constraints can still be tested by future, percent-level measurements of the linear power spectrum, while they remain undetectable to cosmological-distance measures. Although we illustrate these effects in a specific class of models, the requirements on f(R) are phrased in a nearly model-independent manner.