On the probabilities of large deviations of combinatorial sums of independent random variables that satisfy the Linnik condition (Q6060233): Difference between revisions

Let \((\{X_{nij}:1\leq i,j\leq n\})_{n=2,3,\ldots}\) be a sequence of matrices of independent random variables such that \(\sum_{i=1}^nE[X_{nij}]=\sum_{j=1}^nE[X_{nij}]=0\) for each \(n\) and all moments exist, and let \((\pi_n)_{n=2,3,\ldots}\) be a sequence such that \(\pi_n\) is a permutation of \(\{1,2,\ldots,n\}\) chosen uniformly at random from the set of all such permutations. Letting \(S_n=\sum_{i=1}^nX_{ni\pi_n(i)}\), the author shows that under a Linnik-type condition on the \(X_{nij}\) there are sequences \((u_n)_{n=2,3,\ldots}\) such that \[ P\left(S_n\geq u_n\sqrt{\frac{1}{n}\sum_{i,j=1}^nE[X_{nij}^2]}\right)\sim1-\Phi(u_n)\, \] as \(n\to\infty\), where \(\Phi\) is the distribution function of a standard normal random variable. The proof uses the truncation method.