Tail Asymptotics of Randomly Weighted Sums of Dependent Strong Subexponential Random Variables
Qian, Huan,Geng, Bingzhen,Wang, Shijie
DOI: https://doi.org/10.1007/s10986-021-09546-5
2021-01-01
Lithuanian Mathematical Journal
Abstract:Following [C. Yu, Y. Wang, and D. Cheng, Tail behavior of the sums of dependent and heavy-tailed random variables, J. Korean Stat. Soc., 44(1):12–27, 2015], we revisit the asymptotic formulas of tail probabilities of randomly weighted sums and their maxima. Concretely, let {Xn, n ≥ 1} be a sequence of nonnegative random variables with common strong subexponential distribution, which are dependent according to a pretty weak structure, and let {θn, n ≥ 1} be another sequence of nonnegative and arbitrarily dependent random variables but independent of {Xn, n ≥ 1}. For any fixed n ≥ 1, under some mild conditions, we derive the asymptotics of the tail probability of randomly weighted sums $$ {S}_n^{\theta } $$ = $$ {\sum}_{i=1}^n{\theta}_i{X}_i $$ and their maxima $$ {M}_n^{\theta } $$ = max1≤m≥n $$ {S}_m^{\theta } $$ . Our result substantially extends some existing ones in the literature.