The negative loglikelihood is as follows:

$$\dfrac{nd}{2} \log 2\pi + \dfrac{n}{2} \log |\Sigma| + \dfrac{1}{2}\sum_{i=1}^n(x_i-\mu)^T\Sigma^{-1}(x_i-\mu) \tag{1}$$

If I take differentiation with respect to $\mu$ on $(x_i-\mu)^T\Sigma^{-1}(x_i-\mu)$, the result becomes as follows: $$2\Sigma^{-1}\mu - 2\Sigma^{-1} = 2\Sigma^{-1}(\mu-x_i) \tag{2}$$


$$\dfrac{\partial l(u, \Sigma)}{\partial \mu} = \dfrac{1}{2}\sum_{i=1}^n2\Sigma^{-1}(\mu-x_i) = \Sigma^{-1}\sum_{i=1}^n(\mu-x_i) \tag{3}$$

But what I can't do with the algebraic steps for the next step to get the following result:

$$\mu_{MLE}^{*} = \dfrac{1}{n}\sum_{i=1}^n x_i \tag{4}$$

How can I go from $(3)$ to $(4)$? Hope to get algebraic steps for it.


It's quite easy. Just equate the equation 3 to zero and solve for mu. Have a try and lets see what you get.

Have a look at this https://en.m.wikipedia.org/wiki/Maximum_likelihood_estimation

Example part. I think you get some mistake in your equations.

  • $\begingroup$ I have missed some part in the equation. 3. I have included $\sum_{i=1}^n$. I am horrible at doing algebraic steps... $\endgroup$ – user122358 Mar 2 '17 at 8:32
  • $\begingroup$ Hope it is ok. $$\Sigma^{-1}\sum_{i=1}^n (\mu-x_i) = 0$$ $$n\mu-\sum_{i=1}^n x_i = 0$$ $$n\mu = \sum_{i=1}^n x_i$$ $$\mu = \dfrac{\sum_{i=1}^n x_i}{n}$$ $$\therefore \mu_{MLE}^{*} = \dfrac{1}{n}\sum_{i=1}^n x_i$$ $\endgroup$ – user122358 Mar 2 '17 at 8:42
  • $\begingroup$ The last equation is correct and once you equate it to zero you will gt the answer, but I need to check all you equation to make sure they are correct. $\endgroup$ – user151245 Mar 2 '17 at 8:44
  • $\begingroup$ You have some typo in third step. You miss x_i $\endgroup$ – user151245 Mar 2 '17 at 8:45
  • $\begingroup$ Thank you for having me think about the problem. Even though I don't get the straight answer, I still get the support and it feels a lot better than being alone facing the wall. Thank you for taking your time, too. $\endgroup$ – user122358 Mar 2 '17 at 8:50

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.