Supernovae type Ia data favour coupled phantom energy

We estimate the constraints that the recent high-redshift sample of supernovae type Ia put on a phenomenological interaction between dark energy and dark matter. The interaction can be interpreted as arising from the time variation of the mass of dark matter particles. We find that the coupling correlates with the equation of state: roughly speaking, a negative coupling (in our sign convention) implies phantom energy ($w_{\phi}<-1$) while a positive coupling implies ``ordinary'' dark energy. The constraints from the current supernovae Ia Hubble diagram favour a negative coupling and an equation of state $w_{\phi}<-1$. A zero or positive coupling is in fact unlikely at 99% c.l. (assuming constant equation of state); at the same time non-phantom values ($w_{\phi}>-1$) are unlikely at 95%. We show also that the usual bounds on the energy density weaken considerably when the coupling is introduced: values as large as $\Omega_{m0}=0.7$ become acceptable for as concerns SNIa. We find that the rate of change of the mass $\dot{m}/m$ of the dark matter particles is constrained to be $\delta_{0}$ in a Hubble time, with $-10<\delta_{0}<-1$ to 95% c.l.. We show that a large positive coupling might in principle avoid the future singularity known as ``big rip'' (occurring for $w_{\phi}<-1$) but the parameter region for this to occur is almost excluded by the data. We also forecast the constraints that can be obtained from future experiments, focusing on supernovae and baryon oscillations in the power spectra of deep redshift surveys. We show that the method of baryon oscillations holds the best potential to contrain the coupling.