Stochastic nonlinear galaxy biasing

We propose a general formalism for galaxy biasing and apply it to methods for measuring cosmological parameters, such as regression of light versus mass, the analysis of redshift distortions, measures involving skewness, and the cosmic virial theorem. The common linear and deterministic relation g = b delta between the density fluctuation fields of galaxies g and mass delta is replaced by the conditional distribution P(g\delta) of these random fields, smoothed at a given scale at a given time. The nonlinearity is characterized by the conditional mean [g\delta] = b(delta)delta, while the local scatter is represented by the conditional variance sigma(b)(2)(delta) and higher moments. The scatter arises from hidden factors affecting galaxy formation and from shot noise unless it has been properly removed. For applications involving second-order local moments, the biasing is defined by three natural parameters: the slope (b) over cap of the regression of g on delta, a nonlinearity (b) over tilde, and a scatter sigma(b). The ratio of variances b(var)(2) and the correlation coefficient r mix these parameters. The nonlinearity and the scatter lead to underestimates of order (b) over tilde(2)/(b) over cap(2) and sigma(b)(2)/(b) over cap in the different estimators of beta (similar to Omega(0.6)/b). The nonlinear effects are typically smaller. Local stochasticity affects the redshift-distortion analysis only by limiting the useful range of scales, especially for power spectra. In this range, for linear stochastic biasing, the analysis reduces to Kaiser's formula for (b) over cap (not b(var))(,) independent of the scatter. The distortion analysis is affected by nonlinear properties of biasing but in a weak way. Estimates of the nontrivial features of the biasing scheme are made based on simulations and toy models, and strategies for measuring them are discussed. They may partly explain the range of estimates for beta.