Abstract :
[en] Context: Type Ia Supernovae (SNIa) observations in the late 90’s were the first hints for an accelerated expansion of our Universe. Today, hundreds of objects have been observed and seem to confirm the flat LambdaCDM model as the cosmological model best representing
our Universe.
Aims: We study the SNIa observations gathered in the Union 2.1 and in the JLA compilations. By analyzing correlations and different ways of comparing cosmological models to the data, we bring to light some statistical biases, due to the current way of computing SNIa luminosity corrections for light-curve shape, color and host galaxy mass.
Methods: We suggest an alternative, safer and model-independent methodology to calibrate the luminosity corrections, using only nearby SNIa.
Results: With our recalibrated data, biases are strongly reduced. Moreover, open cosmological models are shown to be favoured over flat models (Omega_m,0 = 0.26+-0.08, Omega_Lambda,0 = 0.66+-0.12 for the SCP compilation and Omega_m,0 = 0.20+-0.08, Omega_Lambda,0 = 0.56+- 0.13 for the JLA one).
Conclusions: The usual method to process SNIa data, i.e. simultaneously determining the parameters of the cosmological model and of the luminosity corrections on the full sample, is prone to bias the data in favour of the assumed cosmology, currently a flat LambdaCDM model, as well as to bias the cosmological parameters of the assumed model.
