Pivot, Cut, and Dive: a heuristic for 0-1 mixed integer programming

Authors:
Jonathan Eckstein;Mikhail Nediak
Affiliations:
Business School and RUTCOR, Rutgers University, Piscataway, USA 08854;Queen's School of Business, Queen's University, Kingston, Canada K7L 3N6
Venue:
Journal of Heuristics
Year:
2007

Citing 9
Cited 4

A practical anti-cycling procedure for linearly constrained optimization

Mathematical Programming: Series A and B
Normal conical algorithm for concave minimization over polytopes

Mathematical Programming: Series A and B
Tabu Search

Tabu Search
General Purpose Heuristics for Integer Programming—Part II

Journal of Heuristics
Octane: A New Heuristic for Pure 0-1 Programs

Operations Research
Topics in parallel integer optimization

Topics in parallel integer optimization
Exploring relaxation induced neighborhoods to improve MIP solutions

Mathematical Programming: Series A and B
The feasibility pump

Mathematical Programming: Series A and B
Introduction to Global Optimization (Nonconvex Optimization and Its Applications)

Introduction to Global Optimization (Nonconvex Optimization and Its Applications)

Metaheuristic Search with Inequalities and Target Objectives for Mixed Binary Optimization-Part II: Exploiting Reaction and Resistance

International Journal of Applied Metaheuristic Computing
A New Local Search Algorithm for Binary Optimization

INFORMS Journal on Computing
Feasibility Pump-like heuristics for mixed integer problems

Discrete Applied Mathematics
BDD-based heuristics for binary optimization

Journal of Heuristics

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper describes a heuristic for 0-1 mixed-integer linear programming problems, focusing on "stand-alone" implementation. Our approach is built around concave "merit functions" measuring solution integrality, and consists of four layers: gradient-based pivoting, probing pivoting, convexity/intersection cutting, and diving on blocks of variables. The concavity of the merit function plays an important role in the first and third layers, as well as in connecting the four layers. We present both the mathematical and software details of a test implementation, along with computational results for several variants.