Simulated annealing is an optimization method for finding the global optimum of a function. The algorithm is similar to a Hill climbing algorithm.
Simulated annealing, just like hill climbing, keeps track of one current solution. In each iteration, that solution takes a random step and either improves, stays the same, or becomes worse.
The ability to move from the current solution to a potentially worse solution gives this algorithm a chance to get out of local minima.
Tags: Numerical optimization Algorithm