Let $Z_1, Z_2, \ldots, Z_k$ of $Z$ i.i.d. in $\mathbb{R}^n$. A sample mean

\begin{equation} \bar{Z}_k = \frac{1}{k} \sum_{j=1}^k Z_j \end{equation}

by the strong law of large numbers is

\begin{equation} \bar{Z}_k \to x \end{equation}

almost surely as $k \to \infty$.

The variance of $\bar{Z}_k$ is given by

\begin{equation} \frac{1}{k^2} \mathbb{E} \left( \left\| \sum_{j=1}^k (Z_j -x ) \right\|_2^2 \right) \end{equation}

in which cases can obtain the equality?

\begin{equation} \frac{1}{k^2} \mathbb{E} \left( \left\| \sum_{j=1}^k (Z_j - x ) \right\|_2^2 \right) = \frac{1}{k^2} \sum_{j=1}^k \mathbb{E} \left\| Z_j - x \right\|_2^2 \end{equation}


It is true by independence of $Z_j$'s. Note that $Z_i-x, 1\leq i \leq n$ are orthogonal since they have mean $0$ and they are independent. For orthogonal vectors $(v_i)$ we have $\|\sum v_i\|^{2}=\sum \|v_i\|^{2}$.

  • $\begingroup$ How can we relate orthogonality with zero mean? Do we assume mean 0, since $\bar{Z_k} \to x$ almost surely for large enough $k$ ? Can you please elaborate a little more about it? $\endgroup$
    – Lin
    Jan 22 '20 at 7:57
  • $\begingroup$ If $X$ and $Y$ are independent random vectors with mean $0$ then $E \langle X, Y \rangle=E\sum X_iY_i=\sum EX_iY_i=\sum EX_i EY_i=0$. @Lin $\endgroup$ Jan 22 '20 at 7:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.