Bounds on the constant in the mean central limit theorem

Abstract: Let

X_{1}, ..., X_{n}

be independent with zero means, finite variances

s i g m a_{1}^{2}, ..., s i g m a_{n}^{2}

and finite absolute third moments. Let

F_{n}

be the distribution function of

(X_{1} + ... + X_{n}) / s i g m a

, where

s i g m a^{2} = s u m_{i = 1}^{n} s i g m a_{i}^{2}

, and

P h i

that of the standard normal. The

L^{1}

-distance between

F_{n}

and

P h i

then satisfies [Vert F_n-PhiVert_1lefrac{1}{sigma^3}sum_{i=1}^nE|X_i|^3.] In particular, when

X_{1}, ..., X_{n}

are identically distributed with variance

s i g m a^{2}

, we have [Vert F_n-PhiVert_1lefrac{E|X_1|^3}{sigma^3sqrt{n}}qquad for all

n i n m a t h b b N

,] corresponding to an

L^{1}

-Berry--Esseen constant of 1.

Recommendations

Cites Work

Cited In (29)

This page was built for publication: Bounds on the constant in the mean central limit theorem