# Copyright (c) 2022 AIRBUS and its affiliates.
# This source code is licensed under the MIT license found in the
# LICENSE file in the root directory of this source tree.
from typing import Optional
from discrete_optimization.generic_tools.ea.alternating_ga import AlternatingGa
from discrete_optimization.generic_tools.ea.ga import Ga
from discrete_optimization.generic_tools.ea.ga_tools import (
DeapCrossover,
ParametersAltGa,
ParametersGa,
)
from discrete_optimization.rcpsp.problem import RcpspProblem
from discrete_optimization.rcpsp.solvers import RcpspSolver
[docs]
class GaRcpspSolver(RcpspSolver):
problem: RcpspProblem
hyperparameters = Ga.copy_and_update_hyperparameters(
crossover=dict(
choices=[
DeapCrossover.CX_UNIFORM_PARTIALY_MATCHED,
DeapCrossover.CX_ORDERED,
DeapCrossover.CX_PARTIALY_MATCHED,
]
)
)
# RCPSP needs permutation encoding, not all crossover are available.
[docs]
def solve(self, parameters_ga: Optional[ParametersGa] = None, **args):
if parameters_ga is None:
parameters_ga = ParametersGa.default_rcpsp()
args = self.complete_with_default_hyperparameters(args)
for key in args:
setattr(parameters_ga, key, args[key])
ga_solver = Ga(
problem=self.problem,
encoding=parameters_ga.encoding,
objective_handling=parameters_ga.objective_handling,
objectives=parameters_ga.objectives,
objective_weights=parameters_ga.objective_weights,
mutation=parameters_ga.mutation,
max_evals=parameters_ga.max_evals,
crossover=parameters_ga.crossover,
selection=parameters_ga.selection,
pop_size=parameters_ga.pop_size,
mut_rate=parameters_ga.mut_rate,
crossover_rate=parameters_ga.crossover_rate,
tournament_size=parameters_ga.tournament_size,
deap_verbose=parameters_ga.deap_verbose,
)
return ga_solver.solve()
[docs]
class GaMultimodeRcpspSolver(RcpspSolver):
problem: RcpspProblem
hyperparameters = Ga.hyperparameters
[docs]
def solve(
self, parameters_ga: ParametersAltGa = ParametersAltGa.default_mrcpsp(), **args
):
if parameters_ga is None:
parameters_ga = ParametersAltGa.default_mrcpsp()
args = self.complete_with_default_hyperparameters(args)
for key in args:
setattr(parameters_ga, key, args[key])
ga_solver = AlternatingGa(
problem=self.problem,
encodings=parameters_ga.encodings,
objective_handling=parameters_ga.objective_handling,
objectives=parameters_ga.objectives,
objective_weights=parameters_ga.objective_weights,
mutations=parameters_ga.mutations,
selections=parameters_ga.selections,
crossovers=parameters_ga.crossovers,
max_evals=parameters_ga.max_evals,
sub_evals=parameters_ga.sub_evals,
pop_size=parameters_ga.pop_size,
mut_rate=parameters_ga.mut_rate,
crossover_rate=parameters_ga.crossover_rate,
tournament_size=parameters_ga.tournament_size,
deap_verbose=parameters_ga.deap_verbose,
)
return ga_solver.solve()