Poisson approximations for sequences of random variables
Let [xi]1, [xi]2, ... be a sequence of independent identically distributed lattice variables, E[xi]12+[delta] x [infinity], Sn = [xi]1 + [xi]2 + ··· + [xi]n. The classical approach is to approximate the distribution of Sn by the normal law. We show that for a suitably centered Sn the Poisson approximation also can be applied. Moreover, this approximation holds for all Borel sets which is impossible for the normal distribution. The rate of accuracy is determined by Ibragimov's necessary and sufficient conditions.
Year of publication: |
1998
|
---|---|
Authors: | Cekanavicius, V. |
Published in: |
Statistics & Probability Letters. - Elsevier, ISSN 0167-7152. - Vol. 39.1998, 2, p. 101-107
|
Publisher: |
Elsevier |
Keywords: | Poisson approximation Total variation norm Integer centering |
Saved in:
Saved in favorites
Similar items by person
-
Signed Poisson approximations for Markov chains
Cekanavicius, V., (1999)
-
On multivariate Le Cam theorem and compound Poisson measures
Cekanavicius, V., (1996)
-
On the convergence of Markov binomial to Poisson distribution
Cekanavicius, V., (2002)
- More ...