Non-Linear Programming
A
AY23/24 S1
This course is all about solving non-linear optimisation problems. Mostly quadratic ones, but you'll occasionally have the trigonometric or exponential functions. To put it simply, you solve problems like these by finding the minimum or maximum of a function, given a set of constraints. The course starts off with the basics, like what a feasible region is, and how to prove that a function is convex (using Hessian Matrices). Then, we learn how to find the optimal solution for an unbounded problem. We then move on to more advanced topics, like how to find the Lagrange multipliers, and how to solve the KKT conditions. These are all very powerful concepts, and they can be used to solve a wide variety of problems.
The problem is that these processes are so long and convoluted, all to solve a simple problem. For example, first you need to find the gradient of the function, then you need to find the Hessian matrix, then you need to find the eigenvalues of the Hessian matrix, then you need to find the Lagrange multipliers, then you need to find the KKT conditions. It's a very long and tedious process but luckily, your answers are more verifiable, as the solutions are actually simple just by doing some accurate diagram drawings.
Workload is normal, with a few assignments, and exams. The questions are all about applying the concepts and processes properly. All pretty standard questions that as long as you can follow the instructions on your cheatsheet, you should be okay. I'm quite satisfied with my A.