Generalized Riemann hypothesis and stochastic time series

Using the Dirichlet theorem on the equidistribution of residue classes modulo q and the Lemke Oliver-Soundararajan conjecture on the distribution of pairs of residues on consecutive primes, we show that the domain of convergence of the infinite product of Dirichlet L-functions of non-principal characters can be extended from R(s) > 1 down to R(s) > 1/2, without encountering any zeros before reaching this critical line. The possibility of doing so can be traced back to a universal diffusive random walk behavior C-N = O(N-1/2) of a series C-N over the primes which underlies the convergence of the infinite product of the Dirichlet functions. The series C-N presents several aspects in common with stochastic time series and its control requires to address a problem similar to the single Brownian trajectory problem in statistical mechanics. In the case of the Dirichlet functions of non principal characters, we show that this problem can be solved in terms of a self-averaging procedure based on an ensemble epsilon of block variables computed on extended intervals of primes. Those intervals, called inertial intervals, ensure the ergodicity and stationarity of the time series underlying the quantity C-N. The infinity of primes also ensures the absence of rare events which would have been responsible for a different scaling behavior than the universal law C-N = O(N-1/2) of the random walks.