Two limit theorems for NA random variables (Q1370169)

This paper reviews the definition of negatively associated random variables and proves a maximal inequality for the partial sums \((S_n)\) of a negatively associated, not necessarily stationary, sequence. Consequences of this result include \[ S_n= o(n^{1/2}\ln^\alpha n)\text{ a.s.}\quad\text{and}\quad \sum^\infty_{n=1} n^{n-2} P\Biggl(\max_{1\leq k\leq n} |S_n|\geq n\varepsilon\Biggr)<\infty \] under suitable conditions. The conditions require that the individual variables be stochastically bounded by a single variable \(X\) with finite \(p\)th moment.