This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: 1 IOE 510 Linear Programming I Wednesday February 17, 2010 Professor Amy Cohn 2 Announcements Exam solutions posted to Ctools Graded exams to be handed back on Monday 3 Recap from Last Time Proved optimality of extreme points Motivating the simplex method for solving linear programs 4 Candidate Direction Vectors Let x B represent the vector of variables basic at point x, and x N represent the nonbasic variables Given a BFS, we have n m candidate direction vectors those corresponding to increasing one element of x N Choose some p N what does the corresponding direction vector d p look like? 5 Ensuring d Is Feasible Let A B be the columns of A corresponding with the variables in x B We know that Ax = b Therefore A B x B = b (why?) Therefore x B = A B1 b (why?) We want to have A(x + d p ) remain feasible for some &gt; 0 Therefore Ad p = 0 6 What Is d p ? Let d p i represent the i th element of the vector d p d p p is 1 by convention d p i is 0 if i is in N but not p we are only relaxing a single active constraint What is d p j if j is in B? 7 What Is d p ? Let d p B be the vector in R m containing the elements of vector d p corresponding to the elements of the basis B Ad p = 0 implies A B d p B + A p = 0 Therefore d p B = A B1 A p 8 What Is the Cost Impact of d p ? If we move one step in direction d p , our change in cost will be cd p = c p c B A B1 A p (why?) For a given nonbasic variable x p , the corresponding candidate vector d p is improving if its reduced cost c p = c p c B A B1 A p is strictly negative 9...
View
Full
Document
This note was uploaded on 04/11/2010 for the course IOE 510 taught by Professor Staff during the Fall '08 term at University of Michigan.
 Fall '08
 STAFF

Click to edit the document details