The Algorithmic Showdown:
Gradient Descent vs Newton's Method
14D002: Deterministic Models and Optimization.
By: Reid Falconer
This project seeks to determine how two line search methods perform relative to each other (e.g. in convergence rate or the number of iterations needed) and to further assess the differences in the execution of first order and second order optimisation as the algorithms scales with n.
- Gradient descent
- Newton's method
Animation
- Gradient Descent vs Newton's Method, contour plots with sample paths, scrolling through rho = - 0:9 : 0:9
