1993 Lagrangean heuristics for location problems教程.pdfVIP

European Journal of Operational Research 65 (1993) 383-399 383 North-Holland Theory and Methodology Lagrangean heuristics for location problems J.E. Beasley The Management School, Imperial College, London S W 7 2AZ, England, UK Received September 1988; revised March 1991 Abstract: In this paper we present a framework for developing Lagrangean heuristics (heuristics based upon Lagrangean relaxation and subgradient optimisation) with respect to location problems. Computa- tional results are given for four different location problems: p-median, uncapacitated warehouse location, capacitated warehouse location and capacitated warehouse location with single source con- straints. These results indicate that the framework presented in this paper is robust, i.e. it gives good quality solutions for each of these different location problems. Keywords: Lagrangean heuristic; Location I. Introduction A lower bound for the above program can be found by introducing a Lagrange multiplier ma- It is well-known that in recent years a popular trix s ( 0) for (2) to get the Lagrangean lower technique for the optimal solution of structured bound program (LLBP) given by: pure (or mixed) zero-one integer programs has been the use of tree search procedures with minimise ( c - sA ) x + sb (4) bounds to curtail the search being found via subject to x ~ (0, 1). (5) Lagrangean relaxation and subgradient optimisa- tion (e.g. see Fisher [26,27]).