This preview shows pages 1–3. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Math 171A: Linear Programming Lecture 15 Testing for Optimality Philip E. Gill c 2011 http://ccom.ucsd.edu/~peg/math171a Wednesday, February 9th, 2011 Recap: Farkas’ lemma c T p ≥ 0 for all p such that A a p ≥ if and only if c = A T a λ * a for some λ * a ≥ UCSD Center for Computational Mathematics Slide 2/30, Wednesday, February 9th, 2011 Farkas’ lemma states that if a feasible x * is optimal, then there is at least one λ * a ≥ 0 such that A T a λ * a = c Unfortunately, if x * is optimal there may be infinitely many vectors λ a such that A T a λ a = c . Many of which will have [ λ a ] s < 0 for some index s ! UCSD Center for Computational Mathematics Slide 3/30, Wednesday, February 9th, 2011 Consider an LP with objective vector c = 2 2 . Suppose that ¯ x is a vertex with activeset matrix A a = 1 2 1 3 1 1 1 1 Observe that ¯ x is a degenerate vertex. The equations A T a λ a = c have infinitely many solutions . UCSD Center for Computational Mathematics Slide 4/30, Wednesday, February 9th, 2011 The equations A T a λ a = c are 1 1 1 1 2 3 1 1 λ 1 λ 2 λ 3 λ 4 λ 5 = 2 2 For example, ¯ λ = solve(Aa’,c) gives ¯ λ = 1 1 , but another solution is λ * = 2 2 ⇒ ¯ x is optimal. UCSD Center for Computational Mathematics Slide 5/30, Wednesday, February 9th, 2011 A FUNDAMENTAL QUESTION! How do we find a nonnegative λ * a hidden among the infinitely many solutions of A T a λ a = c ? The answer to the fundamental question. . . A feasible x * with activeset matrix A a is a solution of an LP if and only if there is a nonnegative basic solution of A T a λ a = c . Proof: Later! UCSD Center for Computational Mathematics Slide 6/30, Wednesday, February 9th, 2011 There are a finite number of basic solutions of A T a λ a = c ....
View
Full
Document
This note was uploaded on 03/19/2012 for the course MATH 171a taught by Professor Staff during the Spring '08 term at UCSD.
 Spring '08
 staff
 Linear Programming

Click to edit the document details