Abstract:
Let ξ1,…,ξn be a sequence of independent random variables. Form another sequence
ηn=ξ1+⋯+ξnBn−An.\eqno(1)
Suppose that for any nξn has one of τ absolutely continuous distributions
F1(x),F2(x),…,Fτ(x)
The following assertion is proved.
For the sequence of the densities pn(x) of the sums (1) to converge uniformly to the density of a limit law for some Bn>0, An it is necessary and sufficient that
1. P{ηn<x}→G(x) weakly (G is the limit law).
2. There exists such an N that pN(x) is bounded.
Citation:
V. M. Kruglov, “A local limit theorem for unequally distributed random variables”, Teor. Veroyatnost. i Primenen., 13:2 (1968), 348–351; Theory Probab. Appl., 13:2 (1968), 332–334