Tuning the Algorithm to the Problem

Abstract

Almost every practical heuristic search algorithm is controlled by some set of parameters. In simulated annealing, for example, there’s a temperature parameter, and what’s more, you have to decide the schedule for reducing the temperature over time. In hill-climbing, there’s a parameter that controls the size of the local neighborhood in which you’ll look for improvements. In tabu search, you must determine how to implement the rules for the memory structure. None of these algorithms comes neatly wrapped in a gift box where all you have to do is open the box and receive your nice surprise!

He who wishes to be rich in a day will be hanged in a year.

Leonardo da Vinci, Notebooks