Why it matters
Simulated annealing tackles hard optimization. Understanding shapes NP-hard approach.
Advertisement
The architecture
Move to neighbor: always if better, probability e^(-ΔE/T) if worse.
Cool T over time.
Advertisement
How it works end to end
Cooling schedule: geometric (T *= 0.95) or logarithmic.
Neighbor generation: problem-specific.
Termination: fixed iterations or temp threshold.
Applications: TSP, scheduling, layout.