The Hessian is derived for the log-likelihood log p(y|X,β) (parts (a)-(c)). Whereas part (d) asks for the convexity of the "negative" of the log-likelihood, so one would be looking at -H instead of H, I believe. It would be nice if someone else could confirm this?
mock 2014 positive definite even though negative
Hello,
For the 2rd question on poisson regression: I don't understand why the negative of a positive definite matrix would still be positive definite?
Thanks in advance for the help
1
The Hessian is derived for the log-likelihood log p(y|X,β) (parts (a)-(c)). Whereas part (d) asks for the convexity of the "negative" of the log-likelihood, so one would be looking at -H instead of H, I believe. It would be nice if someone else could confirm this?
3
I'm not a TA but I confirm :)
1
Add comment