Probing Dark Energy with Supernovae : Bias from the time evolution of the equation of state

Observation of thousands of type Ia supernovae should offer the most direct approach to probe the dark energy content of the universe. This will be undertaken by future large ground-based surveys followed by a space mission (SNAP/JDEM). We address the problem of extracting the cosmological parameters from the future data in a model independent approach, with minimal assumptions on the prior knowledge of some parameters. We concentrate on the comparison between a fiducial model and the fitting function and adress in particular the effect of neglecting (or not) the time evolution of the equation of state. We present a quantitative analysis of the bias which can be introduced by the fitting procedure. Such bias cannot be ignored as soon as the statistical errors from present data are drastically improved.