Evidence for dark matter modulation in CoGeNT?

Abstract

We investigate the question of whether the recent modulation signal claimed by CoGeNT is best explained by the dark matter (DM) hypothesis from a Bayesian model comparison perspective. We consider five phenomenological explanations for the data: no modulation signal, modulation due to DM, modulation due to DM compatible with the total CoGeNT rate, and a signal coming from other physics with a free phase but annual period, or with a free phase and a free period. In each scenario, we assign to the free parameters physically motivated priors. We find that while the DM models are weakly preferred to the no modulation model, but when compared to models where the modulation is due to other physics, the DM hypothesis is favoured with odds ranging from 185:1 to 560:1. This result is robust even when astrophysical uncertainties are taken into account and the impact of priors assessed. Interestingly, the odds for the DM model in which the modulation signal is compatible with the total rate against a DM model in which this prior is not implemented is only 5:8, in spite of the former's prediction of a modulation amplitude in the energy range 0.9 to 3.0 keVee that is significantly smaller than the value observed by CoGeNT. Classical hypothesis testing also rules out the null hypothesis of no modulation at the 1.6 sigma to 2.3 sigma level, depending on the details of the alternative. Lastly, we investigate whether anisotropic velocity distributions can help to mitigate the tension between the CoGeNT total and modulated rates, and find encouraging results.