Speaker
Description
High-energy neutrinos provide a unique probe of dark matter (DM) from remote environments. This study investigates the hypothesis that the DM density 'spike' region formed in the vicinity of supermassive black holes (SMBHs) has an enhanced annihilation rate. To overcome the lack of observed halo parameters ($M_{vir}, r_s$) of extragalactic DM halos and bridge the gap with the observable $M_{BH}$, we developed a machine learning framework trained on the MultiDark cosmological simulation to infer the virial masses and scale radii of potential candidates. We performed a stacking likelihood analysis on a population of 47 nearby, $10^{6}-10^{9}M_{\odot}$ SMBHs using a 10-year dataset of neutrino track-like events by IceCube. Constraints are derived for the $W^{+}W^{-}$ and $b\bar{b}$ annihilation channels for DM masses ranging from 100 GeV to 10 TeV. Our results yield 90% C.L. upper limits on the thermally averaged annihilation cross-section ($\langle\sigma v\rangle$) that are competitive with current constraints from dwarf spheroidal galaxies stacking. This work demonstrates the efficacy of integrating cosmological simulations with machine learning to extend the reach of indirect dark matter searches into the extragalactic sky.