topologicpy.GA module

class topologicpy.GA.GA(num_genes: ~typing.Optional[int] = None, gene_space: ~typing.Optional[~typing.Any] = None, fitness_function: ~typing.Optional[~typing.Callable[[~typing.Any, ~numpy.ndarray, int], ~typing.Union[int, float, ~typing.Sequence[~typing.Union[int, float]]]]] = None, *, sol_per_pop: int = 50, num_generations: int = 100, num_parents_mating: ~typing.Optional[int] = None, parent_selection_type: str = 'sss', keep_parents: int = 1, crossover_type: ~typing.Optional[str] = 'single_point', mutation_type: ~typing.Optional[str] = 'random', mutation_num_genes: ~typing.Optional[int] = None, mutation_percent_genes: ~typing.Union[int, str] = 'default', gene_type: ~typing.Optional[~typing.Union[type, ~typing.Sequence[type]]] = <class 'float'>, init_range_low: ~typing.Union[int, float] = -1.0, init_range_high: ~typing.Union[int, float] = 1.0, allow_duplicate_genes: bool = True, random_seed: ~typing.Optional[int] = None, stop_criteria: ~typing.Optional[~typing.Union[str, ~typing.List[str]]] = None, on_generation: ~typing.Optional[~typing.Callable[[~typing.Any], ~typing.Any]] = None, pygad_kwargs: ~typing.Optional[~typing.Dict[str, ~typing.Any]] = None, silent: bool = False)

Bases: object

A TopologicPy-style wrapper around PyGAD with ergonomic hyperparameters, result accessors, and Plotly Pareto visualizations.

Key ideas: - You provide a fitness function (maximization). PyGAD calls it as:

fitness(ga_instance, solution, solution_idx) -> float OR sequence of floats

(single-objective or multi-objective). :contentReference[oaicite:1]{index=1}

For multi-objective runs, set parent_selection_type=”nsga2” (or another MO selector).
This wrapper caches fitness, solutions, and convenient Pareto-front access.

Attributes

BestFitness
BestIndex
BestSolution
GA: Direct access to the underlying pygad.GA instance (after Build()).
Population: Population of the last generation (if available).
PopulationFitness: Fitness array for the last generation (if available).
Ran

Methods

`Build`()	Build the internal pygad.GA instance.
`EnableCheckpointing`(directory[, interval, ...])	Enable periodic checkpointing.
`Load`(filepath)	Load a saved PyGAD instance and attach it to this wrapper.
`Params`()	Return a shallow copy of current hyperparameters.
`ParetoFront`()	Returns (pareto_solutions, pareto_fitness) from the last generation.
`ParetoFrontIndices`([fitness])	Return indices of non-dominated solutions in the provided fitness array (or the cached last-generation fitness).
`Results`()	Returns a compact results dictionary designed to be easy to print/log/store.
`ResumeFromLatestCheckpoint`()	Load the latest checkpoint (if available) into this GA wrapper.
`Run`([target_generations])	Execute the genetic algorithm and cache commonly-used results.
`Save`(filepath)	Save the PyGAD instance (requires Build() or Run() beforehand).
`SetParams`(**kwargs)	Set/override hyperparameters before calling Run().

LatestCheckpoint
PlotParetoFront

property BestFitness: Optional[Union[int, float, Sequence[Union[int, float]]]]

property BestIndex: Optional[int]

property BestSolution: Optional[ndarray]

Build() → GA: Build the internal pygad.GA instance. You normally don’t need to call this; Run() will call it if needed.

EnableCheckpointing(directory: str, interval: int = 5, keep_last: int = 5, prefix: str = 'ga_checkpoint', save_final: bool = True, atomic: bool = True, silent: bool = False) → GA

Enable periodic checkpointing.

Parameters

directorystr: Directory where checkpoints will be written.
intervalint, optional: Save every N generations. Default is 5.
keep_lastint, optional: Keep only the most recent K checkpoints. Default is 5.
prefixstr, optional: File prefix. Default is “ga_checkpoint”.
save_finalbool, optional: If True, save a final checkpoint at the end of Run(). Default is True.
atomicbool, optional: If True, write to a temp file then move into place. Default is True.
silentbool, optional: If True, suppress checkpoint log messages. Default is False.

property GA: Any: Direct access to the underlying pygad.GA instance (after Build()).

LatestCheckpoint() → str | None

Load(filepath: str) → GA: Load a saved PyGAD instance and attach it to this wrapper.

Params() → Dict[str, Any]: Return a shallow copy of current hyperparameters.

ParetoFront() → Tuple[ndarray, ndarray]: Returns (pareto_solutions, pareto_fitness) from the last generation.

ParetoFrontIndices(fitness: Optional[ndarray] = None) → ndarray: Return indices of non-dominated solutions in the provided fitness array (or the cached last-generation fitness). Assumes maximization objectives (PyGAD convention). :contentReference[oaicite:3]{index=3}

PlotParetoFront(*, title: str = 'Pareto Front', objective_names: list[str] = None, show_all_points: bool = True, show_pareto_points: bool = True, connect_pareto: bool = False, pareto_color: str = '#000000', population_color: str = '#B0B0B0', pareto_size: int = 8, population_size: int = 5, background_color: str = 'white', font_family: str = 'Arial', font_size: int = 14, width: int = 800, height: int = 600, show_grid: bool = True, grid_color: str = '#E6E6E6', grid_width: float = 1.0, show_legend: bool = True, max_points: int = None)

property Population: Optional[ndarray]: Population of the last generation (if available).

property PopulationFitness: Optional[ndarray]: Fitness array for the last generation (if available).

property Ran: bool

Results() → Dict[str, Any]: Returns a compact results dictionary designed to be easy to print/log/store.

ResumeFromLatestCheckpoint() → bool

Load the latest checkpoint (if available) into this GA wrapper.

Returns

bool: True if a checkpoint was found and loaded, False otherwise.

Run(target_generations: int = None) → GARunSummary

Execute the genetic algorithm and cache commonly-used results.

Parameters

target_generationsint, optional: If provided, ensures the run continues until this TOTAL number of generations (useful when resuming from checkpoints). If current generations_completed is already >= target_generations, no evolution is run.

Save(filepath: str) → None: Save the PyGAD instance (requires Build() or Run() beforehand).

SetParams(**kwargs) → GA: Set/override hyperparameters before calling Run(). Example: ga.SetParams(num_generations=200, mutation_percent_genes=10)

class topologicpy.GA.GARunSummary(generations_completed: 'int', best_solution: 'Optional[np.ndarray]', best_fitness: 'Optional[Fitness]', best_index: 'Optional[int]')

Bases: object

best_fitness: Optional[Union[int, float, Sequence[Union[int, float]]]]

best_index: Optional[int]

best_solution: Optional[ndarray]

generations_completed: int