platypus.weights module
- platypus.weights.chebyshev(solution, ideal_point, weights, min_weight=0.0001)
Chebyshev (Tchebycheff) fitness of a solution with multiple objectives.
This function is designed to only work with minimized objectives.
Parameters
- solutionSolution
The solution.
- ideal_pointlist of float
The ideal point.
- weightslist of float
The weights.
- min_weightfloat
The minimum weight allowed.
- platypus.weights.normal_boundary_weights(nobjs, divisions_outer, divisions_inner=0)
Returns weights generated by the normal boundary method.
The weights produced by this method are uniformly distributed on the hyperplane intersecting:
[(1, 0, ..., 0), (0, 1, ..., 0), ..., (0, 0, ..., 1)].
Parameters
- nobjsint
The number of objectives.
- divisions_outerint
The number of divisions along the outer set of weights.
- divisions_innerint (optional)
The number of divisions along the inner set of weights.
- platypus.weights.pbi(solution, ideal_point, weights, theta)
Penalty-based boundary intersection fitness of a solution with multiple objectives.
Requires numpy. This function is designed to only work with minimized objectives.
Callers need to set the theta value by using:
functools.partial(pbi, theta=0.5)
Parameters
- solutionSolution
The solution.
- ideal_point: list of float
The ideal point.
- weightslist of float
The weights.
- thetafloat
The theta value.
- platypus.weights.random_weights(nobjs, population_size)
Returns a set of randomly-generated but uniformly distributed weights.
Simply producing N randomly-generated weights does not necessarily produce uniformly-distributed weights. To help produce more uniformly-distributed weights, this method picks weights from a large collection of randomly- generated weights such that the distances between weights is maximized.
Parameters
- nobjsint
The number of objectives.
- population_sizeint
The number of weights to generate.