Solving large scale generalized assignment problems-An aggregation/disaggregation approach

In this paper, large scale generalized assignment problems are considered. The authors propose an approximation technique, where the original (full size) programming problem is aggregated into a problem of more moderate size. Feasible solutions are generated through a disaggregation approach coupled with some complementing heuristics. An upper bound on the loss of accuracy is also calculated. A small test problem is solved in order to illustrate the idea and computational results from two larger problems is presented.