MIT15_053S13_rec07sol

# Since this solution is not better than the incumbent

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: solution to this relaxation is x5 = 1/2 with an objective value of 14. Since this solution is not better than the incumbent, node 13 gets fathomed. Problem 2 Consider the knapsack problem with the following decision variables for i = 1 to 4: � 1 if item i is selected; xi = 0 otherwise. The knapsack problem is formulated as follows: 2 max 19x1 + 23x2 + 30x3 + 40x4 s.t. 6x1 + 8x2 + 10x3 + 13x4 ≤ 25, (1) xi ∈ {0, 1}, for i = 1, . . . , 4. Apply the branch-and-bound algorithm to solve the problem (notice that the the LP relax­ ation of a knapsack problem can be easily solved by selecting the items. Solution. Let’s ﬁrst review the branch and bound algorithm: At each iteration, we select an active node j and make it inactive. Suppose that LP(j ) represents the integer program corresponding to node j , in which the binary variables are relaxed to be fractional. Let x(j ) and ZLP (j ) be the optimal solution and the optimal value of LP. There are three cases: Case: 1 If ZLP (j ) ≤ Z ∗ , the node j gets pruned; Case: 2 If ZLP (j ) > Z ∗ , and x(j ) is integral, then update the incumbent solution by setting Z ∗ := ZLP (j ) and x∗ := x(j ); In addition, node j gets pruned, Case: 3 If ZLP (j ) > Z ∗ and x(j ) is not integr...
View Full Document

## This note was uploaded on 03/18/2014 for the course MGMT 15.053 taught by Professor Jamesorli during the Spring '07 term at MIT.

Ask a homework question - tutors are online