Consider the normal distribution when $\theta=\sigma^2$: $N(\theta,\theta)$ with probability density function:

  • $f(x;\theta)=\frac{1}{\sqrt{2\pi\theta}}e^{-\frac{1}{2\theta}(x-\theta)^2}$

I was asked to find the maximum likelihood estimator of the law. In other words, how to find the maximum value of:

  • $L(\theta)=\frac{1}{(2\pi\theta)^{n/2}}e^{-\frac{1}{2\theta}\sum(X_i-\theta)^2}$

where $X_1,...,X_n$ are iid following $N(\theta,\theta)$.

  • $\begingroup$ It would be nice to know what have you tried. The maximum value of the likelihood occurs when the first derivative is equal to zero, given some mild conditions. $\endgroup$ – Tony Hellmuth May 29 '18 at 12:58
  • $\begingroup$ @TonyHellmuth I did not achieve to solve $L'(\theta)=0$, maybe there is a trick or am I just bad at algebra? $\endgroup$ – Jean Dupuis May 29 '18 at 13:01
  • $\begingroup$ How about the natural log of $L(\theta)$. Same holds at least here. $\endgroup$ – Tony Hellmuth May 29 '18 at 13:03


It would be much easier if you just used the log-likelihood $l(\theta)$ giving $$l(\theta)=-\frac n2\log2\pi\theta-\frac1{2\theta}\sum_i(X_i-\theta)^2$$ Then solve $$\frac{\partial l}{\partial \hat{\theta}}=0.$$

  • $\begingroup$ Should come out nicely :) $\endgroup$ – Tony Hellmuth May 29 '18 at 13:06

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.