A nonclassical law of the iterated logarithm for self-normalized partial sums (Q2250849)

Let \{\(X_n:n\geq 1\)\} be a sequence of nondegenerate and symmetric i.i.d.\ random variables in the domain of attraction of the normal law with zero means and possibly infinite variances. Let \( S_n=\sum_{i=1}^{n} X_i\) and \(V_n^2=\sum_{i=1}^n X_i^2\). Let \{\(b_n:n\geq 1\)\} be a sequence of positive constants, as defined by \textit{O. Klesov and A. Rosalsky} [Stochastic Anal. Appl. 19, 627--641 (2001; Zbl 0990.60027)], and such that it is monotonically approaching infinity and is not asymptotically equivalent to \(\log\log n\). The authors prove that for such a sequence, under some suitable growth conditions such as the truncated second moment grows at certain rate etc., \(\limsup_{n\to \infty} \frac{| S_n| } {\sqrt{2\, V_{n}^2\, b(n)}}=1\) almost surely. This is a nonclassical law of the iterated logarithm for self-normalized partial sums.