Global and local search for scheduling job shop with parallel machines

Classical Job Shop Scheduling problem describes operation in several industries. In this classical model, there is only one machine available for each group of tasks, and the precedence relations between the tasks are restricted to serial-parallel. Here we suggest a generalization, which allows parallel unrelated machines and arbitrary precedence relations between the tasks. The feasible solution space of the generalized problem (which is significantly larger than that of a corresponding version without parallel machines) is efficiently reduced to an essentially smaller subset. The use of this reduced subset, instead of the whole feasible set, is beneficial. We propose global and local search algorithms which start from the reduced solution set.