Figure 9.
Error ellipses for the BAO parameters α and ϵ from the mocks (grey), the non-Gaussian (fit) models (blue), and noisy realizations of those models (dashed). The discrepancies between the mocks and the Gaussian (fit) model are reduced, though a 3 per cent discrepancy persists in σϵ. The similarity between the two error ellipses indicates that choice of likelihood (|$\mathcal {L}_{1}$| or |$\mathcal {L}_{2}$|) for fitting the non-Gaussian model has very little impact on BAO error estimation.
