Tests of the BORG algorithm on a N-body simulation. Top-left panel: center slice of the density contrast computed at a resolution of 32³ from the N-body cosmological simulation used in our test. The simulation itself is obtained with 256³ dark matter particles with ΛCDM cosmology as used in Jasche & Lavaux (2019). Top-right panel: central slice of grid obtained by assigning objects of the mock catalog to a 32³ grid with the nearest grid point filter. Bottom-left panel: density slice computed from one posterior sample obtained by BORG. We note the clear spatial correlations between the density field of the fiducial simulation and the inferred sample, despite the very low sampling rate of the mock catalog. Bottom-right panel: difference between the BORG reconstruction and the simulation truth divided by the standard deviation per voxel estimated from the posterior distribution. The BORG posterior covers the truth and gives a conservative estimate of the uncertainty.