12.8. Summary 363 details." As an example of such a technology, they suggest Least Squares. Heuristics, today, are no such technology. A difficulty is the number of decisions that we have to take. For tra- jectory methods, for instance, we need to think about how to choose a neighborhood; for Genetic Algorithms, we need to decide what mutation operators we include, and so on. If heuristics are to become a mature tech- nology, an important question is not which parameter values are optimal for a given problem class (or more often, a single instance of a problem), but how sensitive a method's solutions are to specific choices. To put it the other way around, we would like to know classes of problems for which a given method with a given implementation works robustly across different parameter settings. 12.8. Summary This chapter has given a brief overview of different heuristic methods. The presentation has not been complete; it could not even include the innumer- able variations of the discussed techniques. Different methods can also be combined into hybrid techniques, further increasing the number of available