This repository contains work done as a part of the course taught by the fabulous Prof. Harish Guruprasad during the July - Nov 2019 semester at IIT Madras. It primarily contains the programming assignments completed as a part of the course.
PA_constrained_opt_student.ipynb: A jupyter notebook with implementations and visualizations of constrained optimization aglorithms for the LASSO problem. The algorithms presented are projected gradient descent and the Frank Wolfe algorithm over the L2 norm ball and L1 norm ball. A discussion about the relative performance of the algorithms as the radius of the L1 or L2 norm ball is changed is presented.PA_unconstrained_opt_student.ipynb: A jupyter notebook with implementations and visualizations of gradient descent and gradient descent with adaptive learning rate (determined using Armijio Goldstein and Full relaxation criteria) and Conjugate Gradient Descent (CGD). A discussion about the convergence rate of CGD for quadratic functions is presented.