MaxDivSolver

MaxDivSolver(
    vectors: ndarray,
    k: int,
    distance_metric: DistanceMetric,
    diversity_metric: DiversityMetric,
    diversity_tie_breakers: list[DiversityMetric],
    constraints: list[Constraint],
    solver_steps: list[SolverStep],
    seed: int = 42,
)

Solver that combines a maximum diversity problem with a solver configuration.

Use MaxDivSolverBuilder to create instances -- it provides convenient defaults, presets and validation.

Initialize the MaxDivSolver with the given configuration.

Parameters:

Name	Type	Description	Default
`vectors`	`ndarray`	(n x d ndarray) A set of n vectors in d dimensions.	required
`k`	`int`	(int) The number of vectors to be selected from the input set ('universe').	required
`distance_metric`	`DistanceMetric`	(DistanceMetric) The distance metric to use.	required
`diversity_metric`	`DiversityMetric`	(DiversityMetric) The diversity metric to use.	required
`diversity_tie_breakers`	`list[DiversityMetric]`	(list[DiversityMetric]) A list of diversity tie-breaker metrics to use.	required
`constraints`	`list[Constraint]`	(list[Constraint]) A list of m constraints to try to satisfy during solving.	required
`solver_steps`	`list[SolverStep]`	(list[SolverStep]) A list of solver steps to execute, the first of which needs to be an InitializationStep, while all latter ones need to be OptimizationSteps.	required
`seed`	`int`	(int) Random seed for the solver.	`42`

solve

solve(verbosity: int = 10) -> MaxDivSolution

Solve the maximum diversity problem with the given configuration.

Parameters:

Name	Type	Description	Default
`verbosity`	`int`	(int) The verbosity level. 0 = silent, 10 = tqdm progress bar per solver step 2x = progress table with iteration count, metrics, elapsed time, ... 20 --> slowest updates (spacing increasing with 10%) 21 --> slower updates (spacing increasing with 5%) 22 --> faster updates (spacing increasing with 2%) 23 --> fastest updates (spacing increasing with 1%) `25 --> debug mode (1% spacing + debug info column)`	`10`

Returns:

Type	Description
`MaxDivSolution`	A MaxDivSolution object representing the solution found.