Directional statistics for WIMP direct detection

The direction dependence of the event rate in WIMP direct detection experiments provides a powerful tool for distinguishing WIMP events from potential backgrounds. We use a variety of (non-parametric) statistical tests to examine the number of events required to distinguish a WIMP signal from an isotropic background when the uncertainty in the reconstruction of the nuclear recoil direction is included in the calculation of the expected signal. We consider a range of models for the Milky Way halo, and also study rotational symmetry tests aimed at detecting non-sphericity/isotropy of the Milky Way halo. Finally we examine ways of detecting tidal streams of WIMPs. We find that if the senses of the recoils are known then of order ten events will be sufficient to distinguish a WIMP signal from an isotropic background for all of the halo models considered, with the uncertainties in reconstructing the recoil direction only mildly increasing the required number of events. If the senses of the recoils are not known the number of events required is an order of magnitude larger, with a large variation between halo models, and the recoil resolution is now an important factor. The rotational symmetry tests require of order a thousand events to distinguish between spherical and significantly triaxial halos, however a deviation of the peak recoil direction from the direction of the solar motion due to a tidal stream could be detected with of order a hundred events, regardless of whether the sense of the recoils is known.