On the Rates of the Other Law of the Logarithm
Li-Xin Zhang,You-You Chen
DOI: https://doi.org/10.1007/s10114-011-0082-z
2012-01-01
Abstract:Let X, X 1 , X 2 , … be i.i.d. random variables, and set S n = X 1 + … + X n , M n = max k ≤ n | S k |, n ≧ 1. Let a_n = o( √(n)/ . -log n) . By using the strong approximation, we prove that, if E X = 0, Var X = σ 2 > 0 and E| X | 2+ ε < ∞ for some ε > 0, then for any r > 1, lim_ε↗ 1/ . -√(r - 1)[ ε ^ - 2 - ( r - 1)]∑_n = 1^∞n^r - 2 P{M_n ⩽εσ√(π ^2 n/ . -( 8log n)) + a_n } = 4/π. We also show that the widest a n is o( √(n)/ . -log n) .