A lifted-space dynamic programming algorithm for the Quadratic Knapsack Problem

The Quadratic Knapsack Problem (QKP) is a well-known combinatorial optimization problem which amounts to maximizing a quadratic function of binary variables, subject to a single linear constraint. It has many applications in finance, logistics, telecommunications, facility location, etc. The QKP is...

Full description

Saved in:
Bibliographic Details
Published in:Discrete Applied Mathematics Vol. 335; pp. 52 - 68
Main Author: Djeumou Fomeni, Franklin
Format: Journal Article
Language:English
Published: Elsevier B.V 15.08.2023
Subjects:
Binary quadratic optimization
Dynamic programming
Integer programming
Knapsack problems
Local search
Integer programming
Local search
Knapsack problems
Binary quadratic optimization
Dynamic programming
ISSN:0166-218X, 1872-6771
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The Quadratic Knapsack Problem (QKP) is a well-known combinatorial optimization problem which amounts to maximizing a quadratic function of binary variables, subject to a single linear constraint. It has many applications in finance, logistics, telecommunications, facility location, etc. The QKP is NP-hard in the strong sense and the state-of-the-art algorithm for solving the QKP can only handle problems of small and moderate sizes. In this paper, we present a novel deterministic heuristic algorithm for finding good QKP feasible solutions. This algorithm consists of combining the dynamic programming approach with a local search procedure, with the novelty that both are adapted and implemented in the space of lifted variables of the QKP. The algorithm runs in O(n3c) time and is able to find optimal solutions to more than 97% of standard instances, about 80% of some well-known hard QKP instances, as well as optimality gaps of 0.1% or less for other instances.
ISSN:0166-218X
1872-6771
DOI:10.1016/j.dam.2023.02.003