Large deviations for nonstationary arrays and sequences (Q1320929)

Several results on large deviations that apply to empirical distributions and empirical processes for nonstationary sequences are proved. The first result which deals with triangular arrays applies in particular to arrays of independent random variables. A new criterion for the large deviation principle to hold is given; this is based on some earlier work of the first two authors [Proc. Am. Math. Soc. 103, No. 4, 1235-1240 (1988; Zbl 0659.60047)]. Among applications of this criterion is a large deviation principle for nonstationary hypermixing sequences. Finally, it is shown that the results of this paper can be applied to the case of an independent sequence whose distributions are quasi-regular, in particular when the distributions are generated by a stationary random process.