Dark energy explained by geometrical considerations

Approximating a real manifold by an idealized one requires to calibrate the parameters characterizing the idealized manifold in function of the real one. This calibration is a purely conventional process and can generally be done in several ways, leading to different fittings. In practice, however, all possible fittings cannot be considered as representative of the real manifold. Approximating the real metric by the FLRW metric would be adequate only if both corresponding structures, defined by the space-time interval, are equivalent on large scales. This requirement puts some constraints on what would be a representative FLRW metric. We show that the way how measurements on SNIa are interpreted to determine the evolution of the scale factor implicitly define the calibration process, and that this one is compatible, on a theoretical point of view, with the aforementionned constraints, indicating that the as fitted FLRW metric would be representative of the real one. On a practical point of view, however, we show that a bias in the measurements could invalidate this conclusion. The bias comes from the fact that SNIa are not randomly distributed over space, but are probably mostly located in regions were matter is largely present, i.e., in overdense regions. We explain how this bias could account for the apparent accelerated expansion of the universe, without needing to introduce the dark energy assumption