An Algorithm for the 0-1 Equality Knapsack Problem

The 0-1 knapsack problem is a linear integer-programming problem with a single constraint and binary variables. The knapsack problem with an inequality constraint has been widely studied, and several efficient algorithms have been published. We consider the equality-constraint knapsack problem, which has received relatively little attention. We describe a branch-and-bound algorithm for this problem, and present computational experience with up to 10,000 variables. An important feature of this algorithm is a least-lower-bound discipline for candidate problem selection.