Ensemble mean of the effective survey response operator (left panel) and corresponding standard deviation map (right panel). The ensemble mean is renormalized by the highest pixel value, as the absolute value does not have a meaning independent of the mean density and the radial selection function. The two above maps should be compared to the north galactic cap of the map in the lower right panel of Fig. 2. The ensemble mean is quite different owing to the introduced star contamination, which could introduce contamination in targets. This manifests itself by an over-completeness on the edge of the map. The right map shows a similar trend, but touching the uncertainty on the selection this time.

Univariate posterior distributions of power spectrum amplitudes for a test without (left panel) and with (right panel) foreground corrections over the full range of Fourier modes considered in this work. Red lines correspond to the true underlying cosmological power spectrum from which mock data sets were generated. The left panel clearly shows that uncorrected foreground effects yield excessive power for large-scale modes and also introduce an overall biased result. In contrast, the right panel shows results obtained from our test with foreground corrections. Clearly, a detailed treatment of all foreground effects permits us to obtain an unbiased measurement of power spectrum amplitudes over the full range of Fourier modes.