We consider job shop scheduling problems with release and due-dates, as well as various tardiness objectives. To date, no efficient general-purpose heuristics have been developed for these problems. genetic algorithms can be applied almost directly, but come along with apparent weaknesses.

TWe perform a novel analysis of the fitness landscape of thejob-shop scheduling problem (JSP). In contrast to other well-known combinatorial optimization problems, we show that the landscape of the JSP is non-regular, in that the connectivity of solutions is variable.