## Lagrangean decomposition using an improved Nelder–Mead approach for Lagrangean multiplier update (2005)

Citations: | 1 - 0 self |

### BibTeX

@MISC{Wu05lagrangeandecomposition,

author = {Dan Wu and Marianthi Ierapetritou},

title = {Lagrangean decomposition using an improved Nelder–Mead approach for Lagrangean multiplier update},

year = {2005}

}

### OpenURL

### Abstract

Lagrangean decomposition has been recognized as a promising approach for solving large-scale optimization problems. However, Lagrangean decomposition is critically dependent on the method of updating the Lagrangean multipliers used to decompose the original model. This paper presents a Lagrangean decomposition approach based on Nelder–Mead optimization algorithm to update the Lagrangean multipliers. The main advantage of the proposed approach is that it results in improved objective function values for the majority of iterations. The efficiency of the proposed approach is illustrated with examples from the literature and the solution of scheduling problems.

### Citations

1371 |
A simplex method for function minimization
- Nelder, Mead
- 1965
(Show Context)
Citation Context ...s, only the promising directions need to be searched. Thus, the computational complexity decreases and the objective of the Lagrangean decomposition is improved at each iteration. Nelder–Mead method (=-=Nelder & Mead, 1965-=-) is used to determine the promising search directions since it is proven to be a very efficient direct search algorithm. For a function of n variables, the algorithm maintains a set of n + 1 points f... |

811 |
Programming and Extensions
- Dantzig, Linear
- 1963
(Show Context)
Citation Context ... optimal solution. Therefore, a number of publications have been focused on decomposition techniques that reduce the problem size and achieve solution in affordable computational time (Benders, 1962; =-=Dantzig, 1963-=-; Fisher, 1981). Lagrangean relaxation was original developed by Held and Karp (1970) and successfully applied to many optimization problems such as production scheduling (Fisher, 1973), planning (Gra... |

226 |
The lagrangian relaxation method for solving integer programming problems
- Fisher
- 1981
(Show Context)
Citation Context ...on. Therefore, a number of publications have been focused on decomposition techniques that reduce the problem size and achieve solution in affordable computational time (Benders, 1962; Dantzig, 1963; =-=Fisher, 1981-=-). Lagrangean relaxation was original developed by Held and Karp (1970) and successfully applied to many optimization problems such as production scheduling (Fisher, 1973), planning (Graves, 1982; Gup... |

171 | The Travelling Salesman Problem and Minimum Spanning Trees, Part II - Held, Karp - 1971 |

142 |
Linear programming under uncertainty
- Dantzig
- 2004
(Show Context)
Citation Context ...s of attention in the research of scheduling and planning problems. A commonly used approach is a multi-stage optimization that utilizes a set of scenarios to represent the possible parameter values (=-=Dantzig, 1955-=-). This however, increases the problem size due to the scenarios introduced. Lagrangean decomposition is thus an ideal technique to use in order to efficiently address the problem of scheduling under ... |

81 |
The cutting-plane method for solving convex programs
- Kelley
- 1960
(Show Context)
Citation Context ...the bundle method allows to move and the bound improvement. Realizing these deficiencies a number of techniques have been developed to update the Lagrangean multipliers. Constraint generation method (=-=Kelley, 1960-=-) considers a family of k existing solution (x k , y k ) and generates a new Lagrangean multiplier u by solving a restricted LP master problem (MP k ) min u,η η s.t. η ≥ cxk + u(yk − xk ), k = 1, ...,... |

78 |
A dual-based procedure for uncapacitated facility location
- Erlenkotter
- 1978
(Show Context)
Citation Context .... Typically the direction of the steepest descent is chosen and the step size t k minimizes V(LDuk+tdk). Although this method is reported to work better than the subgradient method for some problems (=-=Erlenkotter, 1978-=-), the set of directions to choose from may involve specific knowledge of the problem in order to reduce the set of directions examined. A good review of the methods solving for Lagrangean multipliers... |

70 |
Lagrangean decomposition: a model yielding strong Lagrangean bounds
- Guignard, Kim
- 1987
(Show Context)
Citation Context ...lems have overlapping constraints. However, this increases the problem complexity and thus the solution computational time. Fig. 3 provides a geometric interpretation of the Lagrangean decomposition (=-=Guignard & Kim, 1987-=-). The main difficulty in optimizing Lagrangean decomposition is that it is non-differentiable since the inner maximization probFig. 2. Lagrangean decomposition.780 D. Wu, M. Ierapetritou / Computers... |

60 |
Partitioning Procedures for Solving Mixed Variables Programming Problems,” Numerische Mathematics
- Benders
- 1962
(Show Context)
Citation Context ...vergence to the optimal solution. Therefore, a number of publications have been focused on decomposition techniques that reduce the problem size and achieve solution in affordable computational time (=-=Benders, 1962-=-; Dantzig, 1963; Fisher, 1981). Lagrangean relaxation was original developed by Held and Karp (1970) and successfully applied to many optimization problems such as production scheduling (Fisher, 1973)... |

43 | Effective continuous-time formulation for short-term scheduling. 3. Multiple intermediate due dates
- Ierapetritou, Hené, et al.
- 1999
(Show Context)
Citation Context ... whereas the timing constraints enforce the correct duration and sequence of tasks to align with the recipe of different products. Although the above formulation has been proved to perform very well (=-=Ierapetritou & Floudas 1998-=-), the simultaneous solution of this problem for all three periods is computationally expensive due to the large number of binary variables involved. With Lagrangean decomposition, the problem is deco... |

37 |
Lagrangean relaxation and its uses in integer programming
- Geoffrion
- 1974
(Show Context)
Citation Context ...roblem. Let operator Co(.) denotes the convex hull and (P * ) represents the following problem: (P∗) max z = cx s.t.Ax≤ b x ∈ Co{Cx ≤ d, x ∈ X} It can be further shown that (P * ) and (LR) are duals (=-=Geoffrion, 1974-=-). Thus: V (LP) ≥ V (LR) = V (P ∗ ) ≥ V (P) 0098-1354/$ – see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.compchemeng.2005.12.008D. Wu, M. Ierapetritou / Computers and Chemic... |

27 |
A general method of solving extremum problems
- Polyak
- 1967
(Show Context)
Citation Context ...ally has multiple optimal solutions. The most common approach to deal with this lack of differentiability is the subgradient method (Equi, Giorgio, Maziale, & Weintraub, 1997; Marin & Pelegrin, 1998; =-=Polyak, 1967-=-; Rana & Vickson, 1991). The subgradient method utilizes the distance between the objective value at the current iteration Z k and the estimated optimum Z * to calculate a step size tk which is used t... |

26 | Lower Bounds and Efficient Algorithms for the Simple Plant Location Problem - BILDE, KRARUP, et al. - 1977 |

23 |
Optimal Solution of Scheduling Problem Using Lagrange Multipliers: Part I
- Fisher
(Show Context)
Citation Context ...Benders, 1962; Dantzig, 1963; Fisher, 1981). Lagrangean relaxation was original developed by Held and Karp (1970) and successfully applied to many optimization problems such as production scheduling (=-=Fisher, 1973-=-), planning (Graves, 1982; Gupta & Maranas, 1999) and lot-sizing problems (Diaby, Bahl, Karwan, & Zionts, 1992; Thizy & Wassenhove, 1985). The idea of Lagrangean relaxation is based on the existence o... |

20 | Database location in computer networks - Fisher, Hochbaum - 1980 |

16 |
Lagrangean Relaxation
- Guignard
- 2003
(Show Context)
Citation Context ..., u(x − y) = 0), which results in the optimal solution of the original optimization problem. However, in practice using subgradient method is reported to result in unpredictable convergence behavior (=-=Guignard, 2003-=-). For some problems, subgradient method generates monotonic improvement in Lagrangean objective function and the convergence is quick and reliable. While other problems result in erratic multiplier s... |

15 | A combined transportation and scheduling problem
- Equi, Gallo, et al.
- 1997
(Show Context)
Citation Context ... 3. Geometric interpretation of Lagrangean decomposition. lem generally has multiple optimal solutions. The most common approach to deal with this lack of differentiability is the subgradient method (=-=Equi, Giorgio, Maziale, & Weintraub, 1997-=-; Marin & Pelegrin, 1998; Polyak, 1967; Rana & Vickson, 1991). The subgradient method utilizes the distance between the objective value at the current iteration Z k and the estimated optimum Z * to ca... |

15 |
Using Lagrangean Techniques to Solve Hierarchical Production Planning Problems
- Graves
- 1982
(Show Context)
Citation Context ...963; Fisher, 1981). Lagrangean relaxation was original developed by Held and Karp (1970) and successfully applied to many optimization problems such as production scheduling (Fisher, 1973), planning (=-=Graves, 1982-=-; Gupta & Maranas, 1999) and lot-sizing problems (Diaby, Bahl, Karwan, & Zionts, 1992; Thizy & Wassenhove, 1985). The idea of Lagrangean relaxation is based on the existence of “hard” constraints in t... |

14 |
A Lagrangean relaxation approach for very-large-scale capacitated lot-sizing
- Diaby, Bahl, et al.
- 1992
(Show Context)
Citation Context ...ed by Held and Karp (1970) and successfully applied to many optimization problems such as production scheduling (Fisher, 1973), planning (Graves, 1982; Gupta & Maranas, 1999) and lot-sizing problems (=-=Diaby, Bahl, Karwan, & Zionts, 1992-=-; Thizy & Wassenhove, 1985). The idea of Lagrangean relaxation is based on the existence of “hard” constraints in the mathematic description of the optimization problems that increase the computationa... |

12 | An algorithm for minimizing convex functions - Lemaréchal - 1974 |

11 |
A hierarchical Lagrangean relaxation procedure for solving midterm planning problem
- Gupta, Maranas
- 1999
(Show Context)
Citation Context ...981). Lagrangean relaxation was original developed by Held and Karp (1970) and successfully applied to many optimization problems such as production scheduling (Fisher, 1973), planning (Graves, 1982; =-=Gupta & Maranas, 1999-=-) and lot-sizing problems (Diaby, Bahl, Karwan, & Zionts, 1992; Thizy & Wassenhove, 1985). The idea of Lagrangean relaxation is based on the existence of “hard” constraints in the mathematic descripti... |

8 | An application of Lagrangean decomposition to the resource-constrained minimum weighted arborescence problem - Guignard, Rosenwein - 1990 |

7 |
Routing container ships using lagrangean relaxation and decomposition
- Rana, Vickson
- 1991
(Show Context)
Citation Context ...ple optimal solutions. The most common approach to deal with this lack of differentiability is the subgradient method (Equi, Giorgio, Maziale, & Weintraub, 1997; Marin & Pelegrin, 1998; Polyak, 1967; =-=Rana & Vickson, 1991-=-). The subgradient method utilizes the distance between the objective value at the current iteration Z k and the estimated optimum Z * to calculate a step size tk which is used to update the Lagrangea... |

5 |
Lagrangean relaxation: A short course
- Guignard
- 1995
(Show Context)
Citation Context ...P k ) min u,η η s.t. η ≥ cxk + u(yk − xk ), k = 1, ..., K The resulting u is then used in (LDu) and a new cut of (x k+1 , y k+1 ) is obtained from solving (LDu), which is added to the master problem (=-=Guignard, 1995-=-). The process terminates when V(MP k )=V(LDu). However, there is no guarantee that the new Lagrangean multiplier will generate an improved solution, thus the issue of cycling needs to be resolved. Th... |

3 |
The return plant location problem: Modeling and resolution.European
- Marin, Pelegrin
- 1998
(Show Context)
Citation Context ...decomposition. lem generally has multiple optimal solutions. The most common approach to deal with this lack of differentiability is the subgradient method (Equi, Giorgio, Maziale, & Weintraub, 1997; =-=Marin & Pelegrin, 1998-=-; Polyak, 1967; Rana & Vickson, 1991). The subgradient method utilizes the distance between the objective value at the current iteration Z k and the estimated optimum Z * to calculate a step size tk w... |

3 |
A combination of the genetic algorithm and lagrangian relaxation decomposition techniques for the generation unit commitment problem.Electric Power Systemic Research
- Orero, Irving
- 1997
(Show Context)
Citation Context ...ion is that the dualized variable set must be appropriately chosen so that the resulting subproblems are easy to solve and the solution converges fast, which is case-dependent (Gupta & Maranas, 1999; =-=Orero & Irving, 1997-=-). Bundle method is an extension of subgradient method presented by Lemarechal (1974) and Zowe (1985). This method considers improving the Lagrangean function as well as staying close to the approxima... |

3 | Decomposition approaches for the efficient solution of short-term scheduling problems - Wu, Ierapetritou - 2003 |

1 |
Computational improvements for subgradient optimization. Symposia mathematica: vol. XIX
- Crowder
- 1976
(Show Context)
Citation Context ... drawbacks of subgradient method are (a) the lack of convergence criterion, (b) the lack of estimate of the optimal objective Z * and (c) dependency to the heuristic choice of the step size sequence (=-=Crowder, 1976-=-). Due to these problems the subgradient method is not stable when applied to large-scale problems. Another issue related to Lagrangean decomposition is that the dualized variable set must be appropri... |

1 | Optimal solution of set sconvering/partitioning problems using dual heuristics - Fisher, Kedia - 1990 |

1 | Iterative methods for optimization. Philadelhia: Society for Industrial and Applied Mathematics - Kelley - 1999 |

1 |
Lagrangean relaxation for the multi-iterm capacitated lot-sizing problem: A heuristic implementation
- Thizy, Wassenhove
- 1985
(Show Context)
Citation Context ...sfully applied to many optimization problems such as production scheduling (Fisher, 1973), planning (Graves, 1982; Gupta & Maranas, 1999) and lot-sizing problems (Diaby, Bahl, Karwan, & Zionts, 1992; =-=Thizy & Wassenhove, 1985-=-). The idea of Lagrangean relaxation is based on the existence of “hard” constraints in the mathematic description of the optimization problems that increase the computational complexity of the soluti... |