Algorithms For Dummies

 

»

Metaheuristics: Heuristics that help you determine (or even generate) the 

right heuristic for your problem. Among metaheuristics, the most widely 

known are genetic algorithms, inspired by natural evolution. Genetic algorithms 

start with a pool of possible problem solutions and then generate new 

solutions using mutation (they add or remove something in the solution) and 

cross-over (they mix parts of different solutions when a solution is divisible). 

For instance, in the n-Queen problem (Chapter 18), you see that you can split 

a checkerboard vertically into parts because the Queens do not move 

horizontally, making it a problem suitable for cross-over. When the pool is 

enough large, genetic algorithms select the surviving solutions by ruling out 

those that don’t work or lack promise. The selected pool then undergoes 

another iteration of mutation, cross-over, and selection. After enough time 

and iterations, genetic algorithms can find solutions that perform better and 

are completely different from the initial ones.