In this question , I am unable to follow the hints given by @leonbloy in the second answer . Can somebody please write down a complete solution . I am stuck at the very first step .


I will not write down a complete solution, but I will get you started. The answer suggested you write $(\overline{X},A)$ as $CX$ for some square matrix $C$. Here I assume $X=(X_1,\dots,X_n)$. Expanding this, you want to fill in the square matrix below so that the following is true: $$\begin{pmatrix} \overline{X}\\X_2-\overline{X} \\ X_3-\overline{X} \\ \vdots \\ X_n-\overline{X}\end{pmatrix} = \begin{pmatrix} * & * & * & \cdots & * \\ * & * & * & \cdots & * \\ * & * & * & \cdots & * \\ \vdots &\vdots & \vdots & \ddots & \vdots \\ * & * & * & \cdots & * \\ \end{pmatrix}\begin{pmatrix} X_1\\X_2 \\ X_3 \\ \vdots \\ X_n\end{pmatrix} $$ So, for instance, $\overline{X}$ should be the dot product of the first row of $C$ with the vector $X$. But you know that $$\overline{X}=\frac{X_1+X_2+\cdots+X_n}{n}$$ so what must the first row of $C$ be? Similarly, you need to figure out how to write each element of the vector $(\overline{X},A)$ as the dot product of a row of $C$ and the vector $X$.

Now the covariance matrix of $(\overline{X},A)$ is given by $CC^T$. So once you have found $C$, compute the $(1,i)$-entry of $CC^T$ to find the covariance of $\overline{X}$ and $A_i$.

  • $\begingroup$ please give the form of $C$ . I have tried a lot and it still deceives me ! $\endgroup$ – John Apr 30 at 6:18
  • $\begingroup$ Say the first row of $C$ is $(c_{11}, c_{12},\dots,c_{1n})$. Then the dot product of this row with the vector $X$ is $$c_{11}X_1+c_{12}X_2+\cdots + c_{1n}X_n.$$ On the other hand, this dot product should be equal to $$\overline{X} = \frac{1}{n}X_1+\frac{1}{n}X_2+\cdots+\frac{1}{n}X_n$$ so what are the $c_{1i}$'s? Do the same for every row of $C$. $\endgroup$ – kccu Apr 30 at 12:18

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.