TY - JOUR
T1 - An application of Lagrangean decomposition to the capacitated multi-item lot sizing proble
AU - Millar, Harvey H.
AU - Yang, Minzhu
N1 - Funding Information:
t This research has been partially supported by the National Science and Engineering Research Council of Canada under grant GPINOZO. $ Author for correspondence. #Harvey H. Millar is an Assistant Professor of Operations Management in the Department of Finance and Management Science at Saint Mary’s University in Halifax, Nova Scotia. He obtained a BSc. in Industrial Engineering from the University of the West Indies and M.A.Sc. and Ph.D. degrees from the Technical University of Nova Scotia. His current research interests include: production planning, manpower scheduling, managing industrial fishing operations, and fisheries surveillance and patrol. c Minzhu Yang is a Ph.D. student in the Department of Management Engineering in Xian Jiao-tong University in China. Mr Yang was a visiting student at Saint Mary’s University as part of a Canadian International Development Agency joint Canada-China Ph.D. program.
PY - 1993/5
Y1 - 1993/5
N2 - In this paper we present a Lagrangean decomposition technique for solving the capacitated multi-item lot sizing problem (CMLSP). The approach decomposes the problem into a transportation problem and N independent single-item uncapacitated lot sizing problems. The algorithm applies subgradient optimization to update the Lagrangean multipliers. Each iteration of the Lagrangean procedure generates two primal feasible solutions: one from the transportation subproblem, and the other from the network flow problem resulting from a primal partition produced by the solution to the N single-item problems. The Lagrangean algorithm proves to be quite effective and stable in producing good primal and dual solutions to the CMLSP.
AB - In this paper we present a Lagrangean decomposition technique for solving the capacitated multi-item lot sizing problem (CMLSP). The approach decomposes the problem into a transportation problem and N independent single-item uncapacitated lot sizing problems. The algorithm applies subgradient optimization to update the Lagrangean multipliers. Each iteration of the Lagrangean procedure generates two primal feasible solutions: one from the transportation subproblem, and the other from the network flow problem resulting from a primal partition produced by the solution to the N single-item problems. The Lagrangean algorithm proves to be quite effective and stable in producing good primal and dual solutions to the CMLSP.
UR - https://www.scopus.com/pages/publications/0027588850
U2 - 10.1016/0305-0548(93)90085-W
DO - 10.1016/0305-0548(93)90085-W
M3 - Article
AN - SCOPUS:0027588850
SN - 0305-0548
VL - 20
SP - 409
EP - 420
JO - Computers and Operations Research
JF - Computers and Operations Research
IS - 4
ER -