Free Access

Table B.1.

Tier 1 (left) and tier 2 (right) test results for COSEBIs, band powers (BP) and 2PCFs.

Data split log10R(trad.) log10R(import.) lnS Δ(S8, AIA) Δ(S8) Δ(AIA)
COSEBIs z-bin 1 vs. all others 0.57 1.12 0.22 0.7σ 1.4σ 0.1σ
z-bin 2 vs. all others −1.89 −1.56 −8.82 2.2σ 2.7σ 2.1σ
z-bin 3 vs. all others 1.69 2.47 0.14 0.1σ 0.1σ 0.5σ
z-bin 4 vs. all others 0.95 1.82 −2.06 1.2σ 1.1σ 1.4σ
z-bin 5 vs. all others 0.82 1.47 −2.77 1.3σ 1.3σ 1.2σ

BP z-bin 1 vs. all others 0.05 0.65 −0.25 1.6σ 2.0.σ 0.4σ
z-bin 2 vs. all others −2.46 −1.76 −9.61 2.8σ 3.0σ 1.5σ
z-bin 3 vs. all others 1.56 2.26 0.23 0.1σ 0.2σ 0.1σ
z-bin 4 vs. all others 0.05 0.75 −3.64 1.5σ 1.6σ 1.5σ
z-bin 5 vs. all others 1.25 2.00 −0.75 1.0σ 0.8σ 1.4σ

2PCFs z-bin 1 vs. all others 1.20 2.00 0.99 0.3σ 1.1σ 0.1σ
z-bin 2 vs. all others −2.07 −1.23 −9.92 2.2σ 2.4σ 2.1σ
z-bin 3 vs. all others 2.35 3.13 0.42 0.7σ 1.2σ 0.5σ
z-bin 4 vs. all others 0.69 1.14 −4.65 0.9σ 1.1σ 1.2σ
z-bin 5 vs. all others 0.78 1.61 −3.93 1.0σ 0.7σ 1.5σ

Notes. The data split is done by separating a tomographic bin and all its cross-correlations from the rest of the data. The sampled parameters are duplicated and each part of the data is allowed to constrain one set, while the cross-correlations within the data are taken into account through their covariance matrix. The δz parameters are fixed to the mean of their prior for both parts of the chain (see the discussion in the text). The error on the estimated Bayes factor log10R and the suspiciousness lnS is about 0.05 for all cases. We show log10R for both the traditional (trad.) and importance nested sampling (import.) methods. The tier 2 results measure the significance of the differences between the marginalised distributions of the duplicated parameters indicated in the table heading.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.