DataTerminology.comNEW
A probabilistic metaheuristic that explores solutions with a temperature-controlled acceptance of worse moves to escape local minima and approximate a global optimum.
DESCRIPTION
Simulated annealing is inspired by physical annealing, using a temperature parameter to stochastically accept worse solutions early and gradually reduce randomness. It is effective for combinatorial optimization and continuous problems with rugged landscapes. The cooling schedule critically affects performance and convergence. The algorithm commonly uses the Metropolis criterion to accept moves. Applications include TSP, scheduling, and design optimization. SA balances exploration and exploitation and can be hybridized with local search for efficiency.
ALTERNATIVE DEFINITIONS
Wikipedia
Simulated annealing (SA) is a probabilistic technique for approximating the global optimum of a given function. Specifically, it is a metaheuristic to approximate global optimization in a large search space.
Stresses metaheuristic nature and large search space global optimization.
NIST Dictionary of Algorithms and Data Structures
Simulated annealing is a probabilistic strategy for combinatorial optimization that allows occasional uphill moves according to a temperature parameter, modeled after the Metropolis algorithm.
Connects SA to the Metropolis acceptance rule and combinatorial focus.
MIT OpenCourseWare
Simulated annealing is an optimization method inspired by annealing in metallurgy, using a cooling schedule to reduce the probability of accepting worse solutions over time.
Emphasizes physical analogy and cooling schedules.
Wolfram MathWorld
Simulated annealing is a random-search technique designed to avoid entrapment in local minima by admitting uphill moves controlled by a temperature parameter that decreases with time.
Highlights uphill moves and temperature-controlled search.
RELATED TERMS
Hill Climbing Metropolis Algorithm Cooling Schedule Objective Function Local Minima Stochastic Optimization Traveling Salesman Problem Global Optimization
COMPETING TERMS
Genetic Algorithm Tabu Search Gradient Descent
KNOWLEDGE AREAS
TARGET AUDIENCE
Data Scientist AI Engineer Software Developer
EXTERNAL SOURCES
- Wikipedia - Simulated annealing
- NIST DADS - Concise algorithm definition
- MIT OCW - Lecture notes on SA
- Wolfram - Reference entry on SA
TOP QUESTIONS
What is the origin of Simulated Annealing?
It derives from the Metropolis algorithm and physical annealing. Kirkpatrick, Gelatt, and Vecchi popularized it in 1983 for optimization.
Category: Origin
How does Simulated Annealing escape local minima?
It probabilistically accepts worse solutions early, controlled by a temperature parameter that decays over time.
Category: Example
What is the biggest mistake when using Simulated Annealing?
Choosing an inappropriate cooling schedule can stall convergence or waste time. Tune temperature decay to balance exploration and exploitation.
Category: Mistakes
Where is Simulated Annealing commonly applied?
In TSP, scheduling, circuit design, and other combinatorial or nonconvex optimization problems.
Category: Example
What are risks of Simulated Annealing?
Poor parameter choices may yield suboptimal solutions or slow convergence. Stochastic variability requires multiple runs.
Category: Risks
How to choose a cooling schedule in Simulated Annealing?
Use geometric or logarithmic schedules; start high to explore, then reduce temperature steadily while monitoring solution quality.
Category: Requirements
What competes with Simulated Annealing for optimization?
Genetic Algorithms and Tabu Search are common alternatives for similar problem classes.
Category: Alternatives
Why is Simulated Annealing still relevant?
Its simplicity, flexibility, and robustness make it effective for hard optimization problems and hybrid metaheuristics.
Category: Relevancy
What’s a striking analogy for Simulated Annealing?
Like cooling metal: high heat allows atoms to rearrange freely; gradual cooling locks in a stable configuration.
Category: Example