UnconstrainedSetGeneticAlgorithm#

class pybrops.opt.algo.UnconstrainedSetGeneticAlgorithm.UnconstrainedSetGeneticAlgorithm(ngen=250, mu=100, lamb=100, M=1.5, rng=RandomState(MT19937) at 0x7F8D0DF4C840, **kwargs)[source]#

Bases: UnconstrainedOptimizationAlgorithm

Class implementing an NSGA-II genetic algorithm adapted for subset selection optimization. The search space is discrete and nominal in nature.

Constructor for NSGA-II set optimization algorithm.

Parameters:

ngen (int) – Number of generations to evolve population.
mu (int) – Number of parental candidates to keep in population.
lamb (int) – Number of progeny to generate.
M (float) – Length of the chromosome genetic map, in Morgans.
rng (numpy.random.Generator, numpy.random.RandomState, None) – Random number generator source.
kwargs (dict) – Additional keyword arguments.

Methods

`cxSet`	Set crossover operator.
`mutSet`	Set mutation operator.
`optimize`	Optimize an objective function.
`selRandomReplacement`
`selTournamentReplacement`

Attributes

`M`	Length of the genetic map in Morgans.
`lamb`	Number of progenies to generate from the main chromosome population.
`mu`	Number of individuals in the main chromosome population.
`ngen`	Number of generations.
`rng`	Random number generator source.

property M: float#: Length of the genetic map in Morgans.

cxSet(ind1, ind2, indpb)[source]#

Set crossover operator.

Parameters:

ind1 (numpy.ndarray) – Chromosome of the first parent (modified in place).
ind2 (numpy.ndarray) – Chromosome of the second parent (modified in place).
indpb (float) – Probability of initiating a crossover at a specific chromosome locus.

Returns:

out – A tuple of length 2 containing progeny resulting from the crossover.

Return type:

tuple

property lamb: int#: Number of progenies to generate from the main chromosome population.

property mu: int#: Number of individuals in the main chromosome population.

mutSet(ind, sspace, indpb)[source]#

Set mutation operator.

Parameters:

ind (numpy.ndarray) – Individual chromosome to mutate (modified in place).
sspace (numpy.ndarray) – Array representing the set search space.
indpb (float) – Probability of mutation at a single locus.

Returns:

ind – A mutated chromosome.

Return type:

array_like

property ngen: int#: Number of generations.

optimize(objfn, k, sspace, objfn_wt, **kwargs)[source]#

Optimize an objective function.

Parameters:

objfn (callable) – Objective function which to optimize.
k (int) – Number of decision variables in the search space. A vector is formed as sspace^k
sspace (numpy.ndarray) – Search space that the OptimizationAlgorithm searches in.
objfn_wt (numpy.ndarray) – Weight(s) applied to output(s) from the objfn.

Returns:

out – A tuple of length 3 containing (frontier, sel_config, misc).

Where:

frontier is a matrix of the Pareto frontier points.
sel_config is an array of corresponding decision variables for the corresponding Pareto frontier points.
misc is a dictionary of miscellaneous output.

Return type:

tuple

property rng: Generator | RandomState#: Random number generator source.