Fitting Methods for Probability Distribution Functions in Turbulent Star-forming Clouds

Abstract

We use a suite of 3D simulations of star-forming molecular clouds, with and without stellar feedback and magnetic fields, to investigate the effectiveness of different fitting methods for volume and column density probability distribution functions (PDFs). The first method fits a piecewise lognormal and power-law (PL) function to recover PDF parameters such as the PL slope and transition density. The second method fits a polynomial spline function and examines the first and second derivatives of the spline to determine the PL slope and the functional transition density. The first PL (set by the transition between lognormal and PL function) can also be visualized in the derivatives directly. In general, the two methods produce fits that agree reasonably well for volume density but vary for column density, likely due to the increased statistical noise in the column density PDFs as compared to the volume density PDFs. We test a well-known conversion for estimating volume density PL slopes from column density slopes and find that the spline method produces a better match (χ2 of 3.34 versus χ2 of 5.92), albeit with a significant scatter. Ultimately, we recommend the use of both fitting methods on column density data to mitigate the effects of noise.