Great Deluge algorithm

The deluge algorithm (english great deluge algorithm ) is a heuristic optimization method of computer science. It is related to the simulated annealing and is usually used for optimization problems that preclude their high complexity, the complete testing every possibility and simple mathematical method.

The idea is to restrict a random search in the search space by a rising water table with time. For this purpose, a threshold value (water level ) and a constant ( rain) can be defined. From a random start value starting a new value is iteratively generated in the search space and just be accepted if it is above, ie he must be better than. But it must be worse than. is thereby increased regularly to. Figuratively reduce thereby the accessible regions of the search space, so that the algorithm can overcome Although initially local optima, while crossing lower regions with time but goes into a climber algorithm.

As the simulated annealing algorithm is the Flood usually found with regard to the (local) optimum, less efficient than about evolution strategies, but not so expensive.