The prior dependence of the DESI results

TL;DR

This paper examines how the choice of prior assumptions influences the Bayesian comparison of dark energy models in DESI results, highlighting the importance of prior selection in cosmological inferences.

Contribution

It demonstrates that extending the prior bounds on dark energy parameters can significantly alter the preference between cosmological models, emphasizing caution in Bayesian interpretations.

Findings

Extending prior bounds favors ΛCDM over dynamical dark energy.

Shifting prior lower bounds can change model preference.

Results depend strongly on prior assumptions.

Abstract

In this short contribution, we explore the prior dependence of the recent DESI results, according to which a Bayesian comparison of $\Lambda$CDM with $w_0w_a$CDM cosmologies favors dynamical dark energy. It is a simple exercise to show that extending the prior lower bound of the dark energy parameters tilts the preference in favor of $\Lambda$CDM. Adopting the PantheonPlus supernovae catalog, for instance, a shift of the lower bound for $w_0,w_a$ from $-3$ to beyond $-4.6$ and $-5$, respectively, is sufficient to favor $\Lambda$CDM. This calls for caution when interpreting DESI results in the Bayesian context.