$(X,Y)$ is uniform over the triangular region with vertices at $(0,0)$,$(\theta, 0)$, and $(0,\theta)$, $\theta$ is unknown. Let $(X_i,Y_i)$ be iid as $(X,Y)$. Find a one dimensional sufficient statistic $T$ for $\theta$, and prove it's sufficient. Find an unbiased estimate of $\hat{\theta}$ which is a function of $T$.


$f_{X,Y}(x,y) = \frac{2}{\theta^2}$

$L(\theta| \mathbf{X},\mathbf{Y}) = (\frac{\theta^2}{2})^{-n} \prod I_{x_i+y_i < \theta}$

Sufficient statistic: $T = max_i (X_i+Y_i)$

$\hat{\theta} = ?$


Denote $Z =T= \max \limits_i (X_i + Y_i)$. Note that the joint pdf is $$ f_{X,Y}(x,y) = \frac{2}{\theta^2} \mathbf{1}_{0 \le x \le \theta} \cdot \mathbf{1}_{0 \le y \le \theta} \cdot \mathbf{1}_{0 \le x+y \le \theta} $$ and so we find the CDF of $Z$: \begin{align*} P(X + Y \le z) = \begin{cases} \int_0^{z } \int_0^{x} \frac{2}{\theta^2} \; dy \; dx = \frac{z^2}{\theta^2}& 0 \le z \le \theta \\ 1 & z \ge \theta \end{cases} \end{align*} From the fact that $X_i + Y_i$ is iid, we get \begin{align*} P(Z \le z) = P(\max_i (X_i + Y_i) \le z) &= P(X_1 + Y_1 \le z,\dots,X_n + Y_n \le z) \\ &= P(X_1 + Y_1 \le z) \cdots P(X_n + Y_n \le z) \\ &= P(X+Y \le z)^n\\ &= \left( \frac{z^2}{\theta^2} \right)^n \cdot \mathbf{1}_{0 \le z \le \theta} \end{align*}

Finally, calculate $E[Z]$ . . . I'm getting $\frac{2n}{2n+1} \theta$, and therefore $\frac{2n+1}{2n} Z$ is unbiased for $\theta. $


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.