03 June 2008

genetic algorithms

Biology employs what computer scientists know as genetic algorithms. Sometimes biology simply mutates the candidate solutions, but other times it combines them (which we call sexual reproduction).
Biology has adopted the sexual reproduction technique so widely that it must have something going for it, even if we don’t yet fully understand what it is. In experiments, local search algorithms that employ only mutations (i.e. “asexual reproduction”) consistently do better at solving NP-complete problems than do algorithms that incorporate sex.
The best theory people currently have as to why nature relies so heavily on sex is that nature isn’t trying to solve a fixed problem; the constraints are always changing. For example, the features that were ideal for an organism living in the dinosaur age are different from the features that are ideal for modern-day organisms. This is still very much an open issue.

0 comments: