Fitting density models to observational data

We consider the general problem of fitting a parametric density model to discrete observations, taken to follow a non-homogeneous Poisson point process. This class of models is very common, and can be used to describe many astrophysical processes, including the distribution of protostars in molecular clouds. We give the expression for the likelihood of a given spatial density distribution of protostars and apply it to infer the most probable dependence of the protostellar surface density on the gas surface density. Finally, we apply this general technique to model the distribution of protostars in the Orion molecular cloud and robustly derive the local star formation scaling (Schmidt) law for a molecular cloud. We find that in this cloud the protostellar surface density, ΣYSO, is directly proportional to the square gas column density, here expressed as infrared extinction in the K-band, AK: more precisely, ΣYSO = (1.65 ± 0.19) (AK/mag)2.03 ± 0.15 stars pc-2.