Dark matter direct detection signals inferred from a cosmological N-body simulation with baryons

International audienceWe extract at redshift z = 0 a Milky Way sized object including gas, stars and dark matter (DM) from a recent, high-resolution cosmological N-body simulation with baryons. Its resolution is sufficient to witness the formation of a rotating disk and bulge at the center of the halo potential, therefore providing a realistic description of the birth and the evolution of galactic structures in the Lambda CDM cosmology paradigm. The phase-space structure of the central galaxy reveals that, throughout a thick region, the dark halo is co-rotating on average with the stellar disk. At the Earth's location, the rotating component, sometimes called dark disk in the literature, is characterized by a minimum lag velocity v(lag) similar or equal to 75 km/s, in which case it contributes to around 25% of the total DM local density, whose value is rho(DM) similar or equal to 0.37 GeV/cm(3). The velocity distributions also show strong deviations from pure Gaussian and Maxwellian distributions, with a sharper drop of the high velocity tail. We give a detailed study of the impact of these features on the predictions for DM signals in direct detection experiments. In particular, the question of whether the modulation signal observed by DAMA is or is not excluded by limits set by other experiments (CDMS, XENON and CRESST...) is re-analyzed and compared to the case of a standard Maxwellian halo. We consider spin-independent interactions for both the elastic and the inelastic scattering scenarios. For the first time, we calculate the allowed regions for DAMA and the exclusion limits of other null experiments directly from the velocity distributions found in the simulation. We then compare these results with the predictions of various analytical distributions. We find that the compatibility between DAMA and the other experiments is improved. In the elastic scenario, the DAMA modulation signal is slightly enhanced in the so-called channeling region, as a result of several effects that include a departure from a Maxwellian distribution and anisotropies in the velocity dispersions due to the dark disk. For the inelastic scenario, the improvement of the fit is mainly attributable to the departure from a Maxwellian distribution at high velocity. It is correctly modeled by a generalized Maxwellian distribution with a parameter alpha similar or equal to 1.95, or by a Tsallis distribution with q similar or equal to 0.75