I used CVXOPT and Python for this homework set. I had never used CVXOPT before this. Here are some observations based on my experience.

- Yes, the negative of the identity matrix works to deal with the fact that cvxopt.solvers.qp wants constraints that go the opposite direction from what we need.
- Make sure all your matrices are floating point. For most of them you'll get an error if they are integer, but I think there was one where it did not give an error but rather just silently gave inaccurate results.
- Keep in mind that the results aren't exact. In particular, the 's might sometimes be negative despite the constraint so you need to include some fuzziness in how you interpret the results.