Point asymptotics for probabilities of large deviations of the \(\omega^2\) statistics in verification of the symmetry hypothesis (Q1776205)

Let \(F\) be a continuous probability distribution function, \(x_1,\dots, x_n\) be the sample obtained from this distribution and \(F_n\) be the corresponding empirical distribution function. For the \(\omega^2\) statistics \[ \omega^2_n=n\int_{-\infty}^\infty[F_n(x)+F_n(-x)-1]^2\,dF_n(x) \] used for symmetry hypothesis testing, an exact asymptotics \(P(\omega^2_n>nv)\), \(n\to\infty\), \(0<v<1/3\), is obtained, i.e., both the exponential decay rate (large deviations) and the non-exponential prefactor. The proof involves an application of Laplace's method and analysis of a certain extreme value problem.