Inconsistencies in, and short pathlength correction to, $R_{AA}(p_T)$ in $\mathrm{A}+\mathrm{A}$ and $\mathrm{p} + \mathrm{A}$ collisions

We present the first leading hadron suppression predictions in $\mathrm{Pb}+\mathrm{Pb}$ and $\mathrm{p}+\mathrm{Pb}$ collisions from a convolved radiative and collisional energy loss model in which partons propagate through a realistic background and in which the inelastic energy loss receives a short pathlength correction. We find that the short pathlength correction is small for $D$ and $B$ meson $R_{AA}(p_T)$ in both $\mathrm{Pb}+\mathrm{Pb}$ and $\mathrm{p}+\mathrm{Pb}$ collisions. However the short pathlength correction leads to a surprisingly large reduction in suppression for $\pi$ mesons in $\mathrm{p}+\mathrm{Pb}$ and even $\mathrm{Pb}+\mathrm{Pb}$ collisions. We systematically check the consistency of the assumptions used in the radiative energy loss derivation - such as collinearity, softness, and large formation time - with the final numerical model. While collinearity and softness are self-consistently satisfied in the final numerics, we find that the large formation time approximation breaks down at modest to high momenta $p_T \gtrsim 30$ GeV. We find that both the size of the small pathlength correction to $R_{AA}(p_T)$ and the $p_T$ at which the large formation time assumption breaks down are acutely sensitive to the chosen distribution of scattering centers in the plasma.