Cosmic Reionization On Computers: Biases and Uncertainties in the Measured Mean Free Path at the End Stage of Reionization

Abstract: Recent observations and analyses of absorption in quasar spectra suggest a rapid drop in the mean free path (MFP) at the late stage of reionization at $z\sim6$. We use the Cosmic Reionization on Computers simulation to examine potential biases in observed measurements of the MFP at the late stage of reionization, particularly in the presence of a quasar. We analyze three snapshots surrounding the `ankle' point of reionization history, when extended neutral patches of the intergalactic medium disappeared in the simulation box. Specifically, these are $z=6.8$ (true MFP $\approx 0.4$ pMpc), in addition to $z=6.1$ (true MFP $\approx 2$ pMpc) and $z=5.4$ (true MFP $\approx 6$ pMpc). We compare the inferred MFP $\lambda_{\rm mfp}$ from synthetic spectra fits to the true MFP. We find that the mean Lyman continuum (LyC) profile at $z=6.8$ changes significantly with quasar lifetime $t_Q$. We attribute this sensitivity to $t_Q$ to a combination of extended neutral IGM patches and the prevalence of small-scale dense clumps. Consequently, the inferred MFP can be biased by a factor of few depending on $t_Q$. On the other hand, for the $z=6.1$ and $z=5.4$ snapshots, the mean LyC profile shows minimal sensitivity to variation in $t_Q\gtrsim 1$ Myr. The inferred MFP in these two cases is accurate to the $\lesssim 30\%$ level. Our results highlight how modeling systematics can affect the inferred MFP, particularly in the regime of small true MFP ($\lesssim 0.5$ pMpc). We also discuss the potential of this regime to provide a testing ground for constraining quasar lifetimes from LyC profiles.