b-dependent-x-profile
Diagnosis: While we seem to be able to describe underlying event quite comfortably, our min bias in AMISIC (and in SHRiMPS) always is too active and tends to have a p_T distribution that is too hard.
Suspicion: I suspect this has to do with the fact that min bias is quite often more of a "glancing collision", i.e. the protons just touching each other. Having the pdf's uniformly distributed in impact parameter space does not give us a bias towards softer interactions at larger impact parameter, but only less interactions. This is a bit in conflict with the old notion of "wee" partons, based on the uncertainty principle.
Solution: We (i.e. I) should try and make the pdf's aware of the impact parameter, i.e. not uniform. In other words we need to modify the pdf's f_i(x,Q^2) to become f_i(x,Q^2,b_perp), allowing us to have a "hard" core of high-energy (i.e. large-x) partons and a "soft" cloud of low-energy (i.e. small-x) partons around it.