Calculating covariance for 10 dice rolls

46 Views Asked by At

I have two variables, X and Y. I also have a dice. I roll the dice 10 times. X is defined as the number of times I get a result greater than 3, meaning 4,5,6 and Y is defined as the number of times I get a roll less than 3, meaning 1,2. I noticed that X and Y are Geometricly distributed, with n = 10 and p = 0.5 for X and 0.33 for Y. As a result I know E(X) and E(Y). I have sat and tried but I have no idea how to calculate E(XY). I would very much appreciate it if someone could give a hint or direction as to how I would calculate E(XY). I have thought of trying to sum over all x and y of xyP(x,y) but that seems way too long for this question.

Thank you in advance.

Edit: Thanks to everyone who pointed out they are not geometrically distributed but rather binomial. I meant to write binomial distribution but made a mistake as English is not my first language and the names are different in my language.

1

There are 1 best solutions below

3
On BEST ANSWER

Well, unfortunately they are not geometrically distributed.. in fact, since the 10 rolls are indipendents between each other we have: $X\sim B(10,\frac{1}{2})$ and $Y\sim B(10,\frac{1}{3})$ , where $B(n,p)$ is the binomial distribution It would have been nicer if $X,Y$ were indipendents (to istantly conclude that $\mathbb{E}(XY) = \mathbb{E}(X)\mathbb{E}(Y)$), but they are not.

We can write instead $X = \sum_{i=1}^{10}X_i$ and $Y = \sum_{j=1}^{10}Y_j$ where $X_i\sim B(1,\frac{1}{2})$ and $Y_j\sim(1,\frac{1}{3})$ so that: $$ \mathbb{E}(XY) = \mathbb{E}(\Bigr(\sum_{i=1}^{10}X_i\Bigr)\cdot\Bigr(\sum_{j=1}^{10}Y_j\Bigr)) = \mathbb{E}\Bigl(\sum_{i,j=1}^{10}X_iY_j\Bigr) = \sum_{k=1}^{10}\mathbb{E}(X_kY_k) +\sum_{i\not= j}^{10}\mathbb{E}(X_iY_j) $$ $$ \mathbb{E}(XY) = 10\mathbb{E}(X_1Y_1) + 90\mathbb{E}(X_1Y_2) = 10a+90b $$ Since $\mathbb{E}(X_kY_k) = a\in\mathbb{R}$ for all $k\in\{1,\dots,10\}$ and similarly

$\mathbb{E}(X_iY_j) = b\in\mathbb{R}$ for all $(i,j)\in\{1,\dots,10\}^2$ with $i \not= j$.

Now: $$ a = \mathbb{E}(X_1Y_1) = \sum_{x,y=0}^{1}xyf_{(X_1,Y_1)}(x,y) = f_{(X_1,Y_1)}(1,1) = P(X_1=1,Y_1=1) = 0 $$ because it's impossible that the first roll is a number less than 3 and at the same time bigger than 3.

$$ b = \mathbb{E}(X_1Y_2) = \sum_{x,y=0}^{1}xyf_{(X_1,Y_2)}(x,y) = f_{(X_1,Y_1)}(1,1) = P(X_1=1,Y_2=1) = P(X_1 = 1)P(Y_2 = 1) = \frac{1}{2}\cdot\frac{1}{3} = \frac{1}{6} $$

because $X_1$ and $Y_2$ are indipendents! (different roll). Finally we can conclude that:

$$ \mathbb{E}(XY) = 10a+90b = 10\cdot 0 + 90\cdot \frac{1}{6} = 15 $$ Just for fun we can compute the covariance between $X$ and $Y$ as following: $$ Cov(X,Y) = \mathbb{E}(XY)-\mathbb{E}(X)\mathbb{E}(Y) = 15 - (10\cdot\frac{1}{2})(10\cdot\frac{1}{3}) = -\frac{5}{3}. $$