Dark Energy in Ghost-free non-local Gravity

TL;DR

This paper evaluates ghost-free non-local gravity against observational data, finding it viable with some limitations, and compares it to generalized exponential $F(R)$ gravity and $ ext{Lambda}$CDM.

Contribution

It provides an observational confrontation of ghost-free non-local gravity, highlighting its successes and limitations relative to other dark energy models.

Findings

Ghost-free non-local gravity fits late-time data well but struggles with combined CMB data.

Generalized exponential $F(R)$ gravity performs better with full observational data sets.

The effective dark energy equation of state in the non-local model remains quintessence-like, unlike the evolving state in $F(R)$ models.

Abstract

Ghost-free non-local gravity is investigated with regards to its late-time dynamics. Viable solutions in this model are confronted with the observational data including the Pantheon+ catalogue of Type Ia supernovae, the Dark Energy Spectroscopic Instrument, the measurements of baryon acoustic oscillations and the Hubble parameter estimations $H(z)$. The ghost-free non-local gravity is found to be successful in these tests in comparison to the $\Lambda$CDM model and can be also comparable with the generalized exponential $F(R)$ gravity scenario. However the model encounters difficulties when the data from the above observations and the cosmic microwave background radiation data are combined. In tests with the whole set of Pantheon+, DESI, $H(z)$ and CMB data, the generalized exponential $F(R)$ model is essentially more successful. This success is related with the dynamical behavior of its effective dark energy equation of state evolving from a phantom to a quintessence phase during the late-time epoch, whereas the ghost-free non-local model demonstrates only a quintessence behavior. Hence the ghost-free non-local gravity scenario is successful only when the Pantheon+, DESI and $H(z)$ data are considered. The generalized exponential $F(R)$ model satisfies the viability conditions and in tests with all observational data including CMB surpasses the $\Lambda$CDM model in $\chi^2$ statistics and also with information criteria.