Abstract:
Let $X_1,\dots,X_n$ be independent random vectors with zero mean vectors. Let
\begin{gather*}
\Lambda_i=\mathbf E|X_i|^2,\quad M_i=\mathbf E|X_i|^3,\quad\Lambda=\frac1n\sum_{i=1}^n\Lambda_i,\quad M=\frac1n\sum_{i=1}^nM_i,
\\
Y_n=\frac1{\sqrt n}(X_1+\dots+X_n)
\end{gather*}
We prove the following
Theorem.There exist absolute constants $K_1$ and $K_2$ such that for any$x>0$ $$
\mathbf P(|Y_n|\ge x)\le4\exp(-K_1x^2/\Lambda)+K_2M/\sqrt nx^3
$$