A feasibility pump algorithm embedded in an annealing framework

06/14/2019
by   Nicolas Pradignac, et al.
0

The feasibility pump algorithm is an efficient primal heuristic for finding feasible solutions to mixed-integer programming problems. The algorithm suffers mainly from fast convergence to local optima. In this paper, we investigate the effect of an alternative approach to circumvent this challenge by designing a two-stage approach that embeds the feasibility pump heuristic into an annealing framework. The algorithm dynamically breaks the discrete decision variables into two subsets based on the fractionality information obtained from prior runs, and enforces integrality on each subset separately. The feasibility pump algorithm iterates between rounding a fractional solution to one that is integral and projecting an infeasible integral solution onto the solution space of the relaxed mixed-integer programming problem. These two components are used in a Monte Carlo search framework to initially promote diversification and focus on intensification later. The computational results obtained from solving 91 mixed-binary problems demonstrate the superiority of our new approach over Feasibility Pump 2.0.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
07/07/2023

Scylla: a matrix-free fix-propagate-and-project heuristic for mixed-integer optimization

We introduce Scylla, a primal heuristic for mixed-integer optimization p...
research
07/30/2022

PUSH: a primal heuristic based on Feasibility PUmp and SHifting

This work describes PUSH, a primal heuristic combining Feasibility Pump ...
research
01/31/2019

On the statistical evaluation of algorithmic's computational experimentation with infeasible solutions

The experimental evaluation of algorithms results in a large set of data...
research
02/18/2021

Smart Feasibility Pump: Reinforcement Learning for (Mixed) Integer Programming

In this work, we propose a deep reinforcement learning (DRL) model for f...
research
10/16/2021

Finding Backdoors to Integer Programs: A Monte Carlo Tree Search Framework

In Mixed Integer Linear Programming (MIP), a (strong) backdoor is a "sma...
research
05/12/2019

Continuous-Time Systems for Solving 0-1 Integer Linear Programming Feasibility Problems

The 0-1 integer linear programming feasibility problem is an important N...
research
02/15/2022

Confidence Threshold Neural Diving

Finding a better feasible solution in a shorter time is an integral part...

Please sign up or login with your details

Forgot password? Click here to reset