M. Laguna, J. P. Kelly, J. L. Gonzalez Velarde and F. Glover
European Journal of Operational Research, vol. 82, pp. 176-189 (1995)
The multilevel generalized assignment problem (MGAP) differs from the classical GAP in that agents can perform tasks at more than one efficiency level. Important manufacturing problems, such as lot sizing, can be formulated as MGAPs; however, the large number of variables in the related 0-1 integer program makes the use of commercial optimization packages impractical. In this paper, we present a heuristic approach to the solution of the MGAP, which consists of a novel application of tabu search (TS). Our TS method employs neighborhoods defined by ejection chains, that produce moves of greater power without significantly increasing the computational effort.
