Let table has shape of $n$ regular polygon and at each side is sitting one person. Each person is flipping a fair coin once (results of $n$ independent tosses are independent). Person is happy iff he and his neighbours flips a head. Let $X$ means amount of happy people. Find $Var(X)$.

My attempt:

Let $X_i=1$ if $i$-th person is happy and $0$ otherwise then $X=X_1+X_2+…+X_n$ then $Var(X)=Var(X_1+…+X_n)=Var(X_1)+…+Var(X_n)+2\cdot\displaystyle\sum_{i<j}Cov(X_i,X_j)$ we have $Var(X_i)=E(X_i^2)-(E(X_i))^2=1^2*P(X_i=1)+0^2*P(X_i=0)-(1*P(X_i=1)+0*P(X_i=0))^2=\frac{1}{8}-\frac{1}{64}=\frac{7}{64}$

so $Var(X_1)+…+Var(X_n)=\frac{7n}{64}$ and $Cov(X_i,X_j)=E(X_iX_j)-E(X_i)E(X_j)$ but if $i,j$ don't have common neighbours then $X_i,X_j$ are independent so $Cov(X_i,X_j)=0$ if $i,j$ have one common neighbour then $E(X_iX_j)=\frac{1}{32}$(since 5 people need to have head) if $i,j$ are neighbours then $E(X_iX_j)=\frac{1}{16}$(since $4$ people need to have head) thus $2\cdot\displaystyle\sum_{i<j}Cov(X_i,X_j)=\frac{n}{8}$ hence $Var(X)=\frac{15n}{64}$

is my reasoning correct ? I'm not sure about fragment with covariances. Thanks in advance


The analysis is structurally correct. Some details (and the answer) change in the cases $n=3$ and $n=4$. Let $n\ge 5$.

For the sum $\sum_{i,j}\text{Cov}(X_i,X_j)$, there are $2n$ ordered pairs with $E(X_iX_j)=\frac{1}{16}$, and $2n$ with $E(X_iX_j)=\frac{1}{32}$, for a total of $\frac{3n}{16}$. From this we must subtract $\frac{4n}{64}$ for the $E(X_i)E(X_j)$. That agrees with your $\frac{n}{8}$ for which some calculation details were omitted.


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.