How to observe initial state momentum anisotropies in nuclear collisions

We show that the correlation between the elliptic momentum anisotropy, $v_2$, and average transverse momentum, $[p_T]$, in small system nuclear collisions carries information on the origin of the observed momentum anisotropy. A calculation using a hybrid IP-Glasma+Music+UrQMD model that includes contributions from final state response to the initial geometry as well as initial state momentum anisotropies of the Color Glass Condensate, predicts a characteristic sign change of the correlator $\rho(v_2^2,[p_T])$ as a function of charged particle multiplicity in p+Au and d+Au collisions at $\sqrt{s}=200\,{\rm GeV}$, as well as p+Pb collisions at $\sqrt{s}=5020\,{\rm GeV}$. This sign change is absent in calculations without initial state momentum anisotropies. The model also predicts a clear difference between the centrality dependence of $\rho(v_2^2,[p_T])$ in Au+Au collisions at $\sqrt{s}=200\,{\rm GeV}$ and Pb+Pb collisions at $\sqrt{s}=5020\,{\rm GeV}$, with only the latter showing a sign change in peripheral events. Experimental observation of these distinct qualitative features of $\rho(v_2^2,[p_T])$ in small and large systems would constitute strong evidence for the presence and importance of initial state momentum anisotropies predicted by the Color Glass Condensate.