A partitioning gradient based (PGB) algorithm for solving nonlinear goal pogramming problems

Hussein M. Saber, Arunachalam Ravindran

Research output: Contribution to journalArticle

Abstract

This paper presents an efficient and reliable partitioning Gradient Based (PGB) algorithm for solving Nonlinear Goal Programming (NLGP) Problems. The PGB algorithm uses the partitioning technique development for linear GP problems, to decompose the NLGP problem into a series of single objective nonlinear problems. It begins by considering only those constraints associated with the first priority, and uses the Generalized Reduced Gradient (GRG) method to solve the first subproblem. If a unique optimal solution is reached, the algorithm terminates with the current point as the optimal solution to the entire NLGP; otherwise, the second subproblem is solved by adding the second priority goal constraints and a linear constraint to retain the first priority and minimizing the second priority objective. This procedure is continued until all the subproblems are solved or unique optimal solution is detected at any subproblem. A sufficient test is developed to detect unique solutions for nonlinear problems. The PGB algorithm is tested against the Modified Pattern Search (MPS) method, currently available for solving NLGP problems. The results indicate that the PGB algorithm always outperforms the MPS method except for more small problems. In addition, the PGB method found the optimal solution for all test problems proving its robustness and reliability, while the MPS method failed in more than half of the test problems by converging to a nonoptimal point.

Original languageEnglish (US)
Pages (from-to)291-294
Number of pages4
JournalComputers and Industrial Engineering
Volume23
Issue number1-4
DOIs
StatePublished - Jan 1 1992

Fingerprint

Nonlinear programming
Partitioning
Goal Programming
Gradient
Nonlinear Programming
Pattern Search
Optimal Solution
Search Methods
Test Problems
Nonlinear Problem
Gradient methods
Gradient Method
Terminate
Linear Constraints
Unique Solution
Entire
Sufficient
Robustness
Decompose
Series

All Science Journal Classification (ASJC) codes

  • Information Systems and Management
  • Management Science and Operations Research
  • Industrial and Manufacturing Engineering
  • Applied Mathematics

Cite this

Saber, Hussein M. ; Ravindran, Arunachalam. / A partitioning gradient based (PGB) algorithm for solving nonlinear goal pogramming problems. In: Computers and Industrial Engineering. 1992 ; Vol. 23, No. 1-4. pp. 291-294.
@article{517bae0927604d4b880e4c238ec605b5,
title = "A partitioning gradient based (PGB) algorithm for solving nonlinear goal pogramming problems",
abstract = "This paper presents an efficient and reliable partitioning Gradient Based (PGB) algorithm for solving Nonlinear Goal Programming (NLGP) Problems. The PGB algorithm uses the partitioning technique development for linear GP problems, to decompose the NLGP problem into a series of single objective nonlinear problems. It begins by considering only those constraints associated with the first priority, and uses the Generalized Reduced Gradient (GRG) method to solve the first subproblem. If a unique optimal solution is reached, the algorithm terminates with the current point as the optimal solution to the entire NLGP; otherwise, the second subproblem is solved by adding the second priority goal constraints and a linear constraint to retain the first priority and minimizing the second priority objective. This procedure is continued until all the subproblems are solved or unique optimal solution is detected at any subproblem. A sufficient test is developed to detect unique solutions for nonlinear problems. The PGB algorithm is tested against the Modified Pattern Search (MPS) method, currently available for solving NLGP problems. The results indicate that the PGB algorithm always outperforms the MPS method except for more small problems. In addition, the PGB method found the optimal solution for all test problems proving its robustness and reliability, while the MPS method failed in more than half of the test problems by converging to a nonoptimal point.",
author = "Saber, {Hussein M.} and Arunachalam Ravindran",
year = "1992",
month = "1",
day = "1",
doi = "10.1016/0360-8352(92)90119-5",
language = "English (US)",
volume = "23",
pages = "291--294",
journal = "Computers and Industrial Engineering",
issn = "0360-8352",
publisher = "Elsevier Limited",
number = "1-4",

}

A partitioning gradient based (PGB) algorithm for solving nonlinear goal pogramming problems. / Saber, Hussein M.; Ravindran, Arunachalam.

In: Computers and Industrial Engineering, Vol. 23, No. 1-4, 01.01.1992, p. 291-294.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A partitioning gradient based (PGB) algorithm for solving nonlinear goal pogramming problems

AU - Saber, Hussein M.

AU - Ravindran, Arunachalam

PY - 1992/1/1

Y1 - 1992/1/1

N2 - This paper presents an efficient and reliable partitioning Gradient Based (PGB) algorithm for solving Nonlinear Goal Programming (NLGP) Problems. The PGB algorithm uses the partitioning technique development for linear GP problems, to decompose the NLGP problem into a series of single objective nonlinear problems. It begins by considering only those constraints associated with the first priority, and uses the Generalized Reduced Gradient (GRG) method to solve the first subproblem. If a unique optimal solution is reached, the algorithm terminates with the current point as the optimal solution to the entire NLGP; otherwise, the second subproblem is solved by adding the second priority goal constraints and a linear constraint to retain the first priority and minimizing the second priority objective. This procedure is continued until all the subproblems are solved or unique optimal solution is detected at any subproblem. A sufficient test is developed to detect unique solutions for nonlinear problems. The PGB algorithm is tested against the Modified Pattern Search (MPS) method, currently available for solving NLGP problems. The results indicate that the PGB algorithm always outperforms the MPS method except for more small problems. In addition, the PGB method found the optimal solution for all test problems proving its robustness and reliability, while the MPS method failed in more than half of the test problems by converging to a nonoptimal point.

AB - This paper presents an efficient and reliable partitioning Gradient Based (PGB) algorithm for solving Nonlinear Goal Programming (NLGP) Problems. The PGB algorithm uses the partitioning technique development for linear GP problems, to decompose the NLGP problem into a series of single objective nonlinear problems. It begins by considering only those constraints associated with the first priority, and uses the Generalized Reduced Gradient (GRG) method to solve the first subproblem. If a unique optimal solution is reached, the algorithm terminates with the current point as the optimal solution to the entire NLGP; otherwise, the second subproblem is solved by adding the second priority goal constraints and a linear constraint to retain the first priority and minimizing the second priority objective. This procedure is continued until all the subproblems are solved or unique optimal solution is detected at any subproblem. A sufficient test is developed to detect unique solutions for nonlinear problems. The PGB algorithm is tested against the Modified Pattern Search (MPS) method, currently available for solving NLGP problems. The results indicate that the PGB algorithm always outperforms the MPS method except for more small problems. In addition, the PGB method found the optimal solution for all test problems proving its robustness and reliability, while the MPS method failed in more than half of the test problems by converging to a nonoptimal point.

UR - http://www.scopus.com/inward/record.url?scp=0026946391&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0026946391&partnerID=8YFLogxK

U2 - 10.1016/0360-8352(92)90119-5

DO - 10.1016/0360-8352(92)90119-5

M3 - Article

AN - SCOPUS:0026946391

VL - 23

SP - 291

EP - 294

JO - Computers and Industrial Engineering

JF - Computers and Industrial Engineering

SN - 0360-8352

IS - 1-4

ER -