# Rapid late-time reionization: constraints and cosmological implications

## TL;DR

This paper constrains the reionization optical depth using multiple independent methods, finding a rapid, late reionization history largely consistent across datasets but with some tension in combined constraints.

## Contribution

It provides a comprehensive, model-insensitive analysis of reionization history and optical depth using Ly$eta$ constraints, CMB polarization, and small-scale anisotropies, highlighting the consistency and tensions among different data sets.

## Key findings

- Reionization occurred rapidly around z=7.
- Optical depth $ au$ is approximately 0.0492, largely model-independent.
- Some tension exists between Ly$eta$ bounds and combined CMB and BAO data.

## Abstract

We present constraints on the reionization optical depth, $\tau$, obtained using several independent methods. First, we perform a non-parametric reconstruction of the reionization history, using Lyman-$\alpha$ constraints on the evolution of the volume-averaged neutral hydrogen fraction, $x_\mathrm{HI}(z)$, including recent results from the James Webb Space Telescope. When combined with baryon acoustic oscillation (BAO) measurements from DESI and Big Bang nucleosynthesis constraints, these data imply a rapid reionization history ($z_\mathrm{mid}=7.00^{+0.12}_{-0.18}$ and $\Delta z_{50}=1.12^{+0.12}_{-0.29}$) and a value of $\tau=0.0492^{+0.0014}_{-0.0030}$, which is largely insensitive to the assumed cosmological model and independent of cosmic microwave background (CMB) data. The optical depth can also be measured from large-scale $(\ell<30)$ CMB polarization data, yielding constraints that are similarly model-insensitive and consistent with the Ly$\alpha$ bound. Third, $\tau$ may be constrained from the attenuation of small-scale $(\ell>30)$ CMB anisotropies, but the results are sensitive to the choice of cosmological model. Assuming $\Lambda$CDM and combining small-scale CMB data with CMB lensing and type 1a supernovae (SNe) yields tight constraints that are compatible with the Ly$\alpha$ bound. Adding galaxy clustering and lensing measurements brings the constraints further into agreement with the Ly$\alpha$ bound. These independent results reinforce a consensus picture in which reionization is rapid and late. However, the combination of small-scale CMB, CMB lensing, and BAO data yields $\tau=0.094\pm0.011$, which is in $4\sigma$ tension with our Ly$\alpha$ bound. Non-standard reionization scenarios can reconcile some but not all constraints. Concordance is restored in alternative cosmological models, such as models with dynamical dark energy favoured by BAO, CMB, and SNe data.