Abstract:
We find necessary and sufficient conditions for convergences of series of weighted probabilities of large deviations for combinatorial sums $\sum_i X_{ni\pi_n(i)}$, where $||X_{nij}||$ is a matrix of order $n$ of independent random variables and $(\pi_n(1), \pi_n(2), \ldots , \pi_n(n))$ is a random permutation with the uniform distribution on the set of permutations of numbers $1, 2, \ldots , n$, independent with $X_{nij}$. We obtain combinatorial variants of results on convergence rates in the strong law of large numbers and the law of the iterated logarithm under conditions closed to optimal ones. We discuss applications to rank statistics.
Keywords:combinatorial sums, convergence rate, law of the iterated logarithm, strong law of large numbers, Baum - Katz bounds, combinatorial strong law of large numbers, combinatorial law of the iterated logarithm, rank statistics, Spearman's coefficient of rank correlation.
Fulltext:
PDF file (315 kB)