Gradient-Free Optimizers#

The GFO backend provides optimization algorithms from the gradient-free-optimizers package. These implement various metaheuristic and numerical optimization algorithms.

Simulated Annealing#

Probabilistic technique for approximating the global optimum by allowing occasional moves to worse solutions to escape local optima.

SimulatedAnnealing([search_space, ...])

Simulated annealing optimizer.

Direct Methods#

Direct search methods that do not require gradient information and work directly with function evaluations.

DownhillSimplexOptimizer([search_space, ...])

Downhill simplex optimizer.

PowellsMethod([search_space, initialize, ...])

Powell's method optimizer.

PatternSearch([search_space, initialize, ...])

Pattern search optimizer.

LipschitzOptimizer([search_space, ...])

Lipschitz optimizer.

DirectAlgorithm([search_space, initialize, ...])

Direct optimizer.

Population-Based#

Optimization algorithms that maintain and evolve populations of candidate solutions.

ParallelTempering([search_space, ...])

Parallel tempering optimizer.

ParticleSwarmOptimizer([search_space, ...])

Particle swarm optimizer.

SpiralOptimization([search_space, ...])

Spiral optimizer.

GeneticAlgorithm([search_space, initialize, ...])

Genetic algorithm optimizer.

EvolutionStrategy([search_space, ...])

Evolution strategy optimizer.

DifferentialEvolution([search_space, ...])

Differential evolution optimizer.

Sequential Model-Based#

Algorithms that build surrogate models of the objective function to guide the search towards promising regions.

BayesianOptimizer([search_space, ...])

Bayesian optimizer.

TreeStructuredParzenEstimators([...])

Tree structured parzen estimators optimizer.

ForestOptimizer([search_space, initialize, ...])

Forest optimizer.