Cosmological model dependence of the galaxy luminosity function: far-infrared results in the Lemaitre-Tolman-Bondi model

Aims. This is the first paper of a series aiming at investigating galaxy formation and evolution in the giant-void class of the Lemaitre-Tolman-Bondi (LTB) models that best fits current cosmological observations. Here we investigate the luminosity function (LF) methodology, and how its estimates would be affected by a change on the cosmological model assumed in its computation. Are the current observational constraints on the allowed cosmology enough to yield robust LF results? Methods. We used the far-infrared source catalogues built on the observations performed with the Herschel/PACS instrument and selected as part of the PACS evolutionary probe (PEP) survey. Schechter profiles were obtained in redshift bins up to z approximate to 4, assuming comoving volumes in both the standard model, that is, the Friedmann-Lemaitre-Robertson-Walker metric with a perfect fluid energy-momentum tensor, and non-homogeneous LTB dust models, parametrized to fit the current combination of results stemming from the observations of supernovae Ia, the cosmic microwave background, and baryonic acoustic oscillations. Results. We find that the luminosity functions computed assuming both the standard model and LTB void models show in general good agreement. However, the faint-end slope in the void models shows a significant departure from the standard model up to redshift 0.4. We demonstrate that this result is not artificially caused by the used LF estimator which turns out to be robust under the differences in matter-energy density profiles of the models. Conclusions. The differences found in the LF slopes at the faint end are due to variation in the luminosities of the sources that depend on the geometrical part of the model. It follows that either the standard model is over-estimating the number density of faint sources or the void models are under-estimating it.