Introduction - If you have any usage issues, please Google them yourself
Simulated Annealing attempts to mimic the process of annealing in slowly cooled metal. The atoms rearrange themselves into a near-optimal energy state despite the apparent improbability of this occuring. Simulated Annealing abstracts this with a kind of biased random walk: some solution is subjected to a perturbation with probability dependent on the cost of doing so, and the current "temperature". Specifically, the method mimics the Boltzmann distribution: the probability of accepting a pertubation is exp(cost/T). In other words, more costly exploration can occur at high temperatures. The annealing schedule specifies how T changes over time -- this is generally decreasing.