Local Maximum

What you find if you wander around in the fog, searching for the peak of a mountain.

I mean, the ground slopes downhill for thirty feet in every direction, right? Must be the top, right?

SimulatedAnnealing tries not to get stuck on a local maximum by initially bouncing around the landscape and then gradually settling down. At some point, though, it just starts walking up hill.

GeneticAlgorithms have some other search strategies up their sleeves. But if you choose your representation and recombination operators wrong, you may seal off large portions of the search space.