Homework 3 : Optimizing a Likelihood Function

This homework asks you to think about a relatively simple case of nonlinear optimization: the well known Probit model. Of course there are plenty of canned solutions to this problem, however, here we want to hone our skills a bit when it comes to actually implementing a nonlinear optimization problem. We will see that

  1. providing gradient and/or hessian information to an algorithm changes the speed and quality of convergence

  2. Different algorithms reach slightly different optima

  3. there are several ways to obtain standard errors in a likelihood estimation setting.

Get the notebook here