Cramér type moderate deviation theorems for self-normalized processes (Q726728)

For the self-normalized process \(T_n=(W_n+D_{1n})/V_n(1+D_{2n})^{1/2}\), where \(W_n=\sum_{i=1}^n \xi_i\), \(V_n=\big(\sum_{i=1}^n\xi_i^2\big)^{1/2}\) and \(\{\xi_i\}\) are zero-mean independent random variables satisfying \(\sum_{i=1}^n E\xi_i^2 =1\), a general ``Cramér type moderate deviation theorem quantify(ing) the accuracy of the relative error of the normal approximation'' is proved. ``In particular, a sharp moderate deviation theorem under optimal moment conditions is established for Studentized \(U\)-statistics.'' (Citations taken from the authors' abstract).