Halo sampling, local bias, and loop corrections

We develop a new test of local bias, by constructing a locally biased halo density field from sampling the dark matter-halo distribution. Our test differs from conventional tests in that it preserves the full scatter in the bias relation and it does not rely on perturbation theory. We put forward that bias parameters obtained using a smoothing scale R can only be applied to computing the halo power spectrum at scales k similar to 1/R. Our calculations can automatically include the running of bias parameters and give vanishingly small loop corrections at low-k. Our proposal results in much better agreement of the sampling and perturbation theory results with simulations. In particular, unlike the standard interpretation of local bias in the literature, our treatment of local bias does not generate a constant power in the low-k limit. We search for extra noise in the Poisson corrected halo power spectrum at wave numbers below its turnover and find no evidence of significant positive noise (as predicted by the standard interpretation) while we find evidence of negative noise coming from halo exclusion for very massive halos. Using perturbation theory and our non-perturbative sampling technique we also demonstrate that nonlocal bias effects discovered recently in simulations impact the power spectrum only at the few percent level in the weakly nonlinear regime.