.png){kind=icon}
.png){kind=icon}
Return
Multi-objective optimization based algorithms for solving mixed integer linear minimum multiplicative programs
.png){kind=icon}
PRE
AI
DOI：10.1016/j.cor.2020.105178.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
We present two new algorithms for a class of single-objective non-linear optimization problems, the socalled Mixed Integer Linear minimum Multiplicative Programs (MIL-mMPs). This class of optimization problems has a desirable characteristic: a MIL-mMP can be viewed as a special case of the problem of optimization over the efficient set in multi-objective optimization. The proposed algorithms exploit this characteristic and solve any MIL-mMP from the viewpoint of multi-objective optimization. A computational study on 960 instances demonstrates that the proposed algorithms outperform a generic purpose solver, SCIP, by a factor of more than 10 on many instances. We numerically show that selecting the best algorithm among our proposed algorithms highly depends on the class of instances used. Specifically, since one of the proposed algorithms is a decision space search algorithm and the other one is a criterion space search algorithm, one can significantly outperform the other depending on the dimension of decision space and criterion space. Although it is possible to linearize some instances of MIL-mMPs, we show that a commercial solver, CPLEX, struggles to directly solve such linearized instances because linearization introduces additional constraints and binary decision variables. (c) 2020 Elsevier Ltd. All rights reserved.
Keywords:
Multi-objective optimization
Minimum multiplicative programming
Optimization over the efficient set
Decision space search algorithm
Criterion space search algorithm
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
C
IF:
4.3
Papers:
6.5K
Citations:
1.8W
