Method of moments and sums of random indicators (Q2135134)

Given a sum of \(n\) dependent indicator random variables \(\xi_1,\ldots,\xi_n\), the authors establish conditions for convergence to a normal distribution as \(n\to\infty\) (in terms of both the distribution function and moments), and a bound on the corresponding uniform distance. These results are similar in flavour to analogous results for Poisson approximation by \textit{B. A. Sevast'yanov} [Theory Probab. Appl. 17, 695--699 (1972; Zbl 0299.60024)], using quantities that compare the distribution of the \(\xi_j\) with that of independent copies \(\xi_1^\prime,\ldots,\xi_n^\prime\) having the same marginal distributions. Proofs use the method of moments. Applications are given to a particle allocation problem and estimating the number of solutions of nonlinear inclusions.