Probability Distribution Function of Cosmological Density Fluctuations from Gaussian Initial Condition: Comparison of One- and Two-point Log-normal Model Predictions with N-body Simulations

Abstract

We quantitatively study the probability distribution function (PDF) of cosmological nonlinear density fluctuations from N-body simulations with Gaussian initial condition. In particular, we examine the validity and limitations of one-point and two-point log-normal PDF models against those directly estimated from the simulations. We find that the one-point log-normal PDF describes very accurately the cosmological density distribution even in the nonlinear regime (the rms variance \sigma_{nl} \simlt 4 and the over-density \delta \simlt 100). Furthermore the two-point log-normal PDFs are also in good agreement with the simulation data from linear to fairly nonlinear regime, while slightly deviate from them for \delta \simlt -0.5. Thus the log-normal PDF can be used as a useful empirical model for the cosmological density fluctuations. While this conclusion is fairly insensitive to the shape of the underlying power spectrum of density fluctuations P(k), models with substantial power on large scales, i.e., n\equiv d\ln P(k)/d \ln k \simlt -1, are better described by the log-normal PDF. On the other hand, we note that the one-to-one mapping of the initial and the evolved density fields consistent with the log-normal model does not approximate the broad distribution of their mutual correlation even on average. Thus the origin of the phenomenological log-normal PDF approximation still remains to be understood.