pandakota.input.methods module

Methods

DAKOTA Methods

class pandakota.input.methods.ColinyCobylaOptimize(max_iterations: int, max_function_evaluations: int, convergence_tolerance: float, initial_delta: float | None = None, variable_tolerance: float | None = None, solution_target: float | None = None)

Bases: Optimize

Constrained Optimization BY Linear Approximations

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

refinements

Methods

add_refinement
to_string

optimize_type = 'coliny_cobyla'

to_string() → str

class pandakota.input.methods.JegaOptimize(max_iterations: int, max_function_evaluations: int, convergence_tolerance: float, population_size: int, seed: int, replacement_type: str | None = None, convergence_type: str | None = None, initialization_type: str | None = None, crossover_type: str | None = None, crossover_rate: float | None = None, mutation_type: str | None = None, mutation_rate: float | None = None, num_parents: int | None = None, num_offspring: int | None = None, flat_file_path: str | None = None)

Bases: Optimize

Abstract Base Class for JEGA methods

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

population_size

refinements

seed

Methods

add_refinement
check_convergence_type
check_crossover_type
check_initialization_type
check_mutation_type
check_replacement_type
to_string

classmethod check_convergence_type(ct: str)

classmethod check_crossover_type(cot: str)

classmethod check_initialization_type(it: str, flat_file_path: str)

classmethod check_mutation_type(mt: str)

classmethod check_replacement_type(rt: str)

convergence_types = {}

crossover_types = {'multi_point_binary', 'multi_point_parameterized_binary', 'multi_point_real', 'shuffle_random'}

initialization_types = {'flat_file', 'simple_random', 'unique_random'}

mutation_types = {'bit_random', 'offset_cauchy', 'offset_normal', 'offset_uniform', 'replace_uniform'}

optimize_type = "jega  # TODO: Change to 'soga' or 'moga'."

property population_size

replacement_types = {'elitist', 'roulette_wheel', 'unique_roulette_wheel'}

property seed

to_string() → str

type_rate_pairs = {'crossover_type': 'crossover_rate', 'mutation_type': 'mutation_rate'}

class pandakota.input.methods.LatinHypercubeSampling(nsamples: int, seed: int)

Bases: Sampling

LHS

Attributes:

nsamples

refinements

Methods

add_refinement
to_string

add_refinement(refinement_samples: int | None = None)

sample_type = 'lhs'

class pandakota.input.methods.Method

Bases: ABC

Abstract Base Class for all methods

Attributes:

refinements

Methods

add_refinement
to_string

add_refinement(refinement_samples: int)

property refinements

requires_gradients = False

requires_hessians = False

abstract to_string() → str

class pandakota.input.methods.MogaOptimize(*args, niching_type=None, postprocessor_type=None, **kwargs)

Bases: JegaOptimize

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

population_size

refinements

seed

Methods

add_refinement
check_convergence_type
check_crossover_type
check_initialization_type
check_mutation_type
check_replacement_type
to_string

convergence_types = {'metric_tracker', 'num_generations', 'percent_change'}

optimize_type = 'moga'

replacement_types = {'below_limit', 'elitist', 'roulette_wheel', 'unique_roulette_wheel'}

class pandakota.input.methods.MonteCarloSampling(nsamples: int, seed: int)

Bases: Sampling

Random sampling

Attributes:

nsamples

refinements

Methods

add_refinement
to_string

sample_type = 'random'

class pandakota.input.methods.NcsuDirectOptimize(max_iterations: int, max_function_evaluations: int, convergence_tolerance: int, solution_target: float | None = None, min_boxsize_limit: float | None = None, volume_boxsize_limit: float | None = None)

Bases: Optimize

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

min_boxsize_limit

refinements

solution_target

volume_boxsize_limit

Methods

add_refinement
to_string

property min_boxsize_limit

optimize_type = 'ncsu_direct'

property solution_target

to_string() → str

property volume_boxsize_limit

class pandakota.input.methods.NlpqlSqpOptimize(max_iterations: int, max_function_evaluations: int, convergence_tolerance: float)

Bases: Optimize

NLPQL Sequential Quadratic Program

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

refinements

Methods

add_refinement
to_string

optimize_type = 'nlpql_sqp'

class pandakota.input.methods.Optimize(max_iterations: int, max_function_evaluations: int, convergence_tolerance: float)

Bases: Method

Abstract Base Class for all optimization methods

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

optimize_type

refinements

Methods

add_refinement
to_string

property convergence_tolerance

function_key = 'objective_functions'

property max_function_evaluations

property max_iterations

optimize_type = None

to_string() → str

class pandakota.input.methods.OptppCgOptimize(max_iterations: int, max_function_evaluations: int, convergence_tolerance: float, max_step: float | None = None, gradient_tolerance: float | None = None, speculative: bool = False)

Bases: OptppOptimize

Conjugate gradient optimization

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

refinements

Methods

add_refinement
check_merit_function
to_string

optimize_type = 'optpp_cg'

requires_gradients = True

class pandakota.input.methods.OptppFdNewtonOptimize(*args, search_method: str | None = None, centering_parameter: float | None = None, steplength_to_boundary: float | None = None, **kwargs)

Bases: OptppNewtonOptimize

Finite Difference Newton optimization method

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

refinements

Methods

add_refinement
check_merit_function
check_search_method
to_string

optimize_type = 'optpp_fd_newton'

requires_gradients = True

requires_hessians = False

class pandakota.input.methods.OptppGNewtonOptimize(*args, search_method: str | None = None, centering_parameter: float | None = None, steplength_to_boundary: float | None = None, **kwargs)

Bases: OptppNewtonOptimize

Newton optimization method based on least-squares calibration

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

refinements

Methods

add_refinement
check_merit_function
check_search_method
to_string

optimize_type = 'optpp_g_newton'

requires_gradients = False

requires_hessians = False

class pandakota.input.methods.OptppNewtonOptimize(*args, search_method: str | None = None, centering_parameter: float | None = None, steplength_to_boundary: float | None = None, **kwargs)

Bases: OptppOptimize

The optpp_newton method, and base class for other Opt++ Newton methods

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

refinements

Methods

add_refinement
check_merit_function
check_search_method
to_string

classmethod check_search_method(sm)

optimize_type = 'optpp_newton'

requires_gradients = True

requires_hessians = True

search_methods = {'gradient_based_line_search', 'tr_pds', 'trust_region', 'value_based_line_search'}

to_string() → str

class pandakota.input.methods.OptppOptimize(max_iterations: int, max_function_evaluations: int, convergence_tolerance: float, max_step: float | None = None, gradient_tolerance: float | None = None, speculative: bool = False, merit_function: str | None = None)

Bases: Optimize

Abstract Base Class for Opt++ family of local optimizers

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

refinements

Methods

add_refinement
check_merit_function
to_string

classmethod check_merit_function(mf: str)

merit_functions = {'argaez_tapia', 'el_bakry', 'van_shanno'}

optimize_type = 'optpp  # TODO: change to one of the optpp_* methods'

to_string() → str

class pandakota.input.methods.OptppPdsOptimize(max_iterations: int, max_function_evaluations: int, convergence_tolerance: float, search_scheme_size=None)

Bases: OptppOptimize

Simplex-based derivative-free optimization

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

refinements

Methods

add_refinement
check_merit_function
to_string

optimize_type = 'optpp_pds'

to_string() → str

class pandakota.input.methods.OptppQNewtonOptimize(*args, search_method: str | None = None, centering_parameter: float | None = None, steplength_to_boundary: float | None = None, **kwargs)

Bases: OptppNewtonOptimize

Quasi-Newton optimization method

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

refinements

Methods

add_refinement
check_merit_function
check_search_method
to_string

optimize_type = 'optpp_q_newton'

requires_gradients = True

requires_hessians = False

class pandakota.input.methods.Sampling(nsamples: int, seed: int)

Bases: Method

Abstract Base Class for all sampling methods

Attributes:

nsamples

refinements

sample_type

Methods

add_refinement
to_string

function_key = 'response_functions'

property nsamples

sample_type = None

to_string() → str

class pandakota.input.methods.SogaOptimize(max_iterations: int, max_function_evaluations: int, convergence_tolerance: float, population_size: int, seed: int, replacement_type: str | None = None, convergence_type: str | None = None, initialization_type: str | None = None, crossover_type: str | None = None, crossover_rate: float | None = None, mutation_type: str | None = None, mutation_rate: float | None = None, num_parents: int | None = None, num_offspring: int | None = None, flat_file_path: str | None = None)

Bases: JegaOptimize

Attributes:

convergence_tolerance

max_function_evaluations

max_iterations

population_size

refinements

seed

Methods

add_refinement
check_convergence_type
check_crossover_type
check_initialization_type
check_mutation_type
check_replacement_type
to_string

convergence_types = {'average_fitness_tracker', 'best_fitness_tracker'}

optimize_type = 'soga'

replacement_types = {'elitist', 'favor_feasible', 'roulette_wheel', 'unique_roulette_wheel'}