Biased power spectrum and bispectrum for an ensemble of three-dimensional scale-free numerical simulations

Abstract

We examine the effect of a threshold bias on the power spectrum and the bispectrum in an ensemble of numerical simulations (Gaussian initial perturbations with power law spectra P(k) \sim k^n, n=+1, 0, -1, -2) and compare our results with theoretical predictions. Our simulations are evolved sufficiently that on the scale where we apply the threshold the rms fluctuation has developed significantly into the nonlinear regime. Thus, predictions based on perturbation theory do not necessarily apply. Nevertheless, we find our results for the power spectrum, biased power simply amplified by a numerical factor, follow predicted trends, far beyond the regime where perturbation theory is expected to be valid. We find that the biased bispectrum continues to follow the so-called hierarchical form, with reduced three-point amplitude Q \approx 1 in the strongly nonlinear regime, independent of initial spectrum. In the quasi-linear perturbative regime the three-point amplitude depends on configuration shape, a behavior that is found to give useful information about the amount of bias without information about the unbiased matter distribution.