Convex Optimization Chapter 1 Introduction What, Why and How What is convex optimization Why study convex optimization How to study convex optimization What is Convex Optimization? Mathematical Optimization Nonlinear Optimization Convex Optimization Least-squares LP Mathematical Optimization Convex Optimization Least-squares Analytical Solution of Least-squares f 0 (x) = jjAx ¡ bjj 2 = (Ax ¡ b) > (Ax ¡ b) 2 @f 0 ( x ) @x = 2A > (Ax ¡ b) = 0 x = (A > A) ¡ 1 A > b Linear Programming (LP) Why Study Convex Optimization? Solving Optimization Problems Mathematical Optimization Nonlinear Optimization Convex Optimization Least-squares LP Mathematical Optimization Nonlinear Optimization Convex Optimization Least-squares LP • • • • Analytical solution Good algorithms and software High accuracy and high reliability Time complexity: C n 2 k A mature technology! Mathematical Optimization Nonlinear Optimization Convex Optimization Least-squares LP • • • • No analytical solution Algorithms and software Reliable and efficient Time complexity: C n 2 m Also a mature technology! Mathematical Optimization Nonlinear Optimization Convex Optimization Least-squares LP • • • • No analytical solution Algorithms and software Reliable and efficient Time complexity (roughly) max{ n3 , n 2 m, F} Almost a mature technology! Mathematical Optimization Nonlinear Optimization Convex Optimization Least-squares • • • • • LP Sadly, no effective methods to solve Only approaches with some compromise Local optimization: “more art than technology” Global optimization: greatly compromised efficiency Help from convex optimization 1) Initialization 2) Heuristics 3) Bounds Far from a technology! (something to avoid) Why Study Convex Optimization If not, …… there is little chance you can solve it. -- Section 1.3.2, p8, Convex Optimization How to Study Convex Optimization? Two Directions As potential users of convex optimization As researchers developing convex programming algorithms Recognizing least-squares problems Straightforward: verify the objective to be a quadratic function the quadratic form is positive semidefinite Standard techniques increase flexibility Weighted least-squares Regularized least-squares Recognizing LP problems Example: t i = jr i j Sum of residuals approximation Chebyshev or minimax approximation t = maxi ja> x ¡ bi j i Recognizing Convex Optimization Problems An Example C 10 P Adding mP linear constraints????? 10 p · 1 m p 8f j 1 ; j 2 ; ¢¢¢; j 10 g j k= 1 j=1 j 2 k Summary From the book, we expect to learn To recognize convex optimization problems To formulate convex optimization problems To (know what can) solve them!