Nonparametric determination of redshift evolution index of Dark Energy

We propose a nonparametric method to determine the sign of \gamma - the redshift evolution index of dark energy. This is important for distinguishing between positive energy models, a cosmological constant, and what is generally called ghost models. Our method is based on geometrical properties and is more tolerant to uncertainties of other cosmological parameters than fitting methods in what concerns the sign of \gamma. The same parameterization can also be used for determining $\gamma$ and its redshift dependence by fitting. We apply this method to SNLS supernovae and to gold sample of re-analyzed supernovae data from Riess et al. 2004. Both datasets show strong indication of a negative \gamma. If this result is confirmed by more extended and precise data, many of dark energy models, including simple cosmological constant, standard quintessence models without interaction between quintessence scalar field(s) and matter, and scaling models are ruled out.