Simulated annealing

class SA(domain: Domain, fitness_function: Callable[[Solution], float], warmup_iterations: int = 5, max_iterations: int = 20, alteration_limit: int = 1, initial_temp: float = 50.0, cooling_rate: float = 0.99, neighbor_population_size: int = 1, distributed=False, log_dir: str = 'logs/SA')

Bases: Metaheuristic

Simulated Annealing (SA) algorithm for optimization problems.

This class implements the Simulated Annealing metaheuristic which uses temperature-based probabilistic acceptance of worse solutions to escape local optima. The temperature gradually decreases according to a cooling schedule, reducing the probability of accepting worse solutions over time.

Parameters:

  • domain (Domain) – The problem domain that defines the solution space

  • fitness_function (Callable[[Solution], float]) – Function to evaluate solutions

  • max_iterations (int, optional) – Maximum number of iterations to run, defaults to 20

  • alteration_limit (float, optional) – Maximum proportion of solution to alter when generating neighbors, defaults to 0.1

  • initial_temp (float, optional) – Initial temperature for annealing process, defaults to 50.0

  • cooling_rate (float, optional) – Rate at which temperature decreases, defaults to 0.99

  • neighbor_population_size (int, optional) – Number of neighbors to generate in each iteration, defaults to 1

  • distributed (bool, optional) – Whether to use distributed computation, defaults to False

  • log_dir (str, optional) – Directory for logging, defaults to “logs/SA”

Variables:

  • max_iterations (int) – Maximum number of iterations

  • alteration_limit (float) – Maximum proportion of solution to alter

  • initial_temp (float) – Current temperature in the annealing process

  • cooling_rate (float) – Rate of temperature decrease

  • neighbor_population_size (int) – Number of neighbors per iteration

Initialize the Simulated Annealing algorithm.

Parameters:

  • domain (Domain) – The problem domain that defines the solution space

  • fitness_function (Callable[[Solution], float]) – Function to evaluate solutions

  • max_iterations (int, optional) – Maximum number of iterations to run, defaults to 20

  • alteration_limit (float, optional) – Maximum proportion of solution to alter when generating neighbors, defaults to 0.1

  • initial_temp (float, optional) – Initial temperature for annealing process, defaults to 50.0

  • cooling_rate (float, optional) – Rate at which temperature decreases, defaults to 0.99

  • neighbor_population_size (int, optional) – Number of neighbors to generate in each iteration, defaults to 1

  • distributed (bool, optional) – Whether to use distributed computation, defaults to False

  • log_dir (str, optional) – Directory for logging, defaults to “logs/SA”

initialize(num_solutions: int = 1) → Tuple[List[Solution], Solution]

Initialize the Simulated Annealing algorithm with random solutions.

Parameters:

num_solutions (int, optional) – Number of initial solutions to generate, defaults to 1

Returns:

A tuple containing the list of solutions and the best solution found

Return type:

Tuple[List[Solution], Solution]

iterate(solutions: List[Solution]) → Tuple[List[Solution], Solution]

Execute one iteration of the Simulated Annealing algorithm.

In each iteration, multiple neighbor solutions are generated, and the best one is selected. The Metropolis criterion determines whether the selected neighbor replaces the current solution based on the current temperature. The temperature is then decreased according to the cooling schedule.

Parameters:

solutions (List[Solution]) – Current population of solutions (expected to have one solution in SA)

Returns:

A tuple containing the updated population and the best solution found

Return type:

Tuple[List[Solution], Solution]

stopping_criterion() → bool

Check if the algorithm should stop.