CADETProcess.optimization.BotorchModular

class CADETProcess.optimization.BotorchModular(acquisition_fn, surrogate_model, early_stopping_improvement_window, early_stopping_improvement_bar, n_init_evals, n_max_evals, seed, progress_frequency, x_tol, f_tol, cv_bounds_tol, cv_lineqcon_tol, cv_lincon_tol, cv_nonlincon_tol, n_max_iter, similarity_tol, parallelization_backend)

Modular bayesian optimization algorithm.

BotorchModular takes 2 optional arguments and uses the BOTORCH_MODULAR API of Ax to construct a Model which connects both components with the respective transforms necessary.

Attributes:

acquisition_fn: type, optional

AcquisitionFunction class. The default is LogExpectedImprovement.

surrogate_model: type, optional

Model class. The default is SingleTaskGP.

Methods

check_optimization_problem(optimization_problem)	Check if problem is configured correctly and supported by the optimizer.
check_required_parameters()	Verify if all required parameters are set.
check_x0(optimization_problem, x0)	Check the initial guess x0 for an optimization problem.
load_results(checkpoint_path[, ...])	Load optimization results from a checkpoint file.
optimize(optimization_problem[, x0, ...])	Solve OptimizationProblem.
run_final_processing()	Run post processing at the end of the optimization.
run_post_processing(X_transformed, ...[, ...])	Run post-processing of generation.
train_model()	Train model.