Illustration of the Wiener filter code example showing (left to right) a Gaussian random signal a), d), g), the data including noise (b,e,h), and the reconstructed map (c,f,i). The additive Gaussian white noise has a variance that sets a signal-to-noise ratio ⟨σ_s⟩_Ω/σ_n of roughly 2. The same code has been applied to three different spaces (top to bottom), namely a 1D regular grid with 512 pixels (a,b,c), a 2D regular grid with 256 × 256 pixels (d,e,f), and a HEALPix grid with n_side = 128 corresponding to 196   608 pixels on the sphere (g,h,i). (All figures have been created by NIFTy using the field.plot method.)