Approximation schemes for non-separable non-linear boolean programming problems under nested knapsack constraints

•A model suitable for handling take-or-leave decisions is introduced.•The objective is a non-linear non-separable function.•The constraints are nested linear knapsack constraints.•Fully polynomial-time approximation schemes are presented and analysed.•Approaches: geometric rounding and K-approximati...

Full description

Saved in:
Bibliographic Details
Published in:European journal of operational research Vol. 270; no. 2; pp. 435 - 447
Main Authors: Halman, Nir, Kellerer, Hans, Strusevich, Vitaly A.
Format: Journal Article
Language:English
Published: Elsevier B.V 16.10.2018
Subjects:
Combinatorial optimization
FPTAS
Geometric rounding
K-approximation sets and functions
Non-linear boolean programming
K-approximation sets and functions
Non-linear boolean programming
Geometric rounding
Combinatorial optimization
FPTAS
ISSN:0377-2217, 1872-6860
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•A model suitable for handling take-or-leave decisions is introduced.•The objective is a non-linear non-separable function.•The constraints are nested linear knapsack constraints.•Fully polynomial-time approximation schemes are presented and analysed.•Approaches: geometric rounding and K-approximation of sets and functions. We consider a fairly general model of “take-or-leave” decision-making. Given a number of items of a particular weight, the decision-maker either takes (accepts) an item or leaves (rejects) it. We design fully polynomial-time approximation schemes (FPTASs) for optimization of a non-separable non-linear function which depends on which items are taken and which are left. The weights of the taken items are subject to nested constraints. There is a noticeable lack of approximation results on integer programming problems with non-separable functions. Most of the known positive results address special forms of quadratic functions, and in order to obtain the corresponding approximation algorithms and schemes considerable technical difficulties have to be overcome. We demonstrate how for the problem under consideration and its modifications FPTASs can be designed by using (i) the geometric rounding techniques, and (ii) methods of K-approximation sets and functions. While the latter approach leads to a faster scheme, the running times of both algorithms compare favorably with known analogues for less general problems.
ISSN:0377-2217
1872-6860
DOI:10.1016/j.ejor.2018.04.013