12.7. General considerations 359 In sum, a simple strategy like local search yielded very good results for a problem whose exact solution could never be computed. The quality of the solution was evidently determined by the chosen method, by the computational resources (how many steps?), and by the particular settings (how many assets to change?). As a rule, computational resources have much more influence on the results than parameter settings. It will not always be so simple. In later chapters, we will discuss how we can solve more-difficult problems, in particular when we have constraints. 12.7. General considerations 12.7.1. What technique to choose? It is difficult to give general advice on this question. Different methods may be able to solve a given problem class, so one possibility would be to test which technique is best. We do not suggest pursuing such a strategy; "best" is certainly not required. What we need is a technique that provides sufficiently good solutions. This is, again, the satisficing rule: we define a desired quality for a solution, and once we have found a solution that meets this requirement, we stop searching. Admittedly, this is not too helpful, so