This is a random search optimization technique such has the capability to escape from local optima along with the employ of probability function. Simulated Annealing draws its inspiration from physical annealing of solids, whereas a metal is carried to its lowest energy state via first heating this to an extremely high temperature usually re-crystallization temperature of metal and then cooling at an extremely slow rate, to an extremely low temperature. It might result in quenching, that is not desirable, if the cooling is not slowing sufficient. Simulated annealing algorithm is iteration by enhancement method such produced random displacement from any type of feasible solution. Basic step-wise detail of the algorithm is described in following figure.
Get an initial solution S
Get an initial temperature T > 0
While not frozen, do:
Perform the following loop n times Pick a random neighbor S* of S Let, ? = f(S*) - f(S)
If ?<0 (downhill move) then S = S*
If Δ ≥ 0
(uphill move), then S = S*
With probability P (?, T)
If f(S) ≥ f(Sbest )
then Sbest= S
T=New value of temperature
Sbest is the best solution.
Program: Generic Procedure of Simulated Annealing