:py:mod:`desdeo_emo.EAs.RVEA` ============================= .. py:module:: desdeo_emo.EAs.RVEA Module Contents --------------- Classes ~~~~~~~ .. autoapisummary:: desdeo_emo.EAs.RVEA.RVEA .. py:class:: RVEA(problem: desdeo_problem.MOProblem, population_size: int = None, population_params: Dict = None, initial_population: desdeo_emo.population.Population.Population = None, alpha: float = 2, lattice_resolution: int = None, selection_type: str = None, interact: bool = False, use_surrogates: bool = False, n_iterations: int = 10, n_gen_per_iter: int = 100, total_function_evaluations: int = 0, time_penalty_component: Union[str, float] = None, keep_archive: bool = False, save_non_dominated: bool = False) Bases: :py:obj:`desdeo_emo.EAs.BaseEA.BaseDecompositionEA` The python version reference vector guided evolutionary algorithm. Most of the relevant code is contained in the super class. This class just assigns the APD selection operator to BaseDecompositionEA. NOTE: The APD function had to be slightly modified to accomodate for the fact that this version of the algorithm is interactive, and does not have a set termination criteria. There is a time component in the APD penalty function formula of the type: (t/t_max)^alpha. As there is no set t_max, the formula has been changed. See below, the documentation for the argument: penalty_time_component See the details of RVEA in the following paper R. Cheng, Y. Jin, M. Olhofer and B. Sendhoff, A Reference Vector Guided Evolutionary Algorithm for Many-objective Optimization, IEEE Transactions on Evolutionary Computation, 2016 :param problem: The problem class object specifying the details of the problem. :type problem: MOProblem :param population_size: The desired population size, by default None, which sets up a default value of population size depending upon the dimensionaly of the problem. :type population_size: int, optional :param population_params: The parameters for the population class, by default None. See desdeo_emo.population.Population for more details. :type population_params: Dict, optional :param initial_population: An initial population class, by default None. Use this if you want to set up a specific starting population, such as when the output of one EA is to be used as the input of another. :type initial_population: Population, optional :param alpha: The alpha parameter in the APD selection mechanism. Read paper for details. :type alpha: float, optional :param lattice_resolution: The number of divisions along individual axes in the objective space to be used while creating the reference vector lattice by the simplex lattice design. By default None :type lattice_resolution: int, optional :param selection_type: One of ["mean", "optimistic", "robust"]. To be used in data-driven optimization. To be used only with surrogate models which return an "uncertainity" factor. Using "mean" is equivalent to using the mean predicted values from the surrogate models and is the default case. Using "optimistic" results in using (mean - uncertainity) values from the the surrogate models as the predicted value (in case of minimization). It is (mean + uncertainity for maximization). Using "robust" is the opposite of using "optimistic". :type selection_type: str, optional :param a_priori: A bool variable defining whether a priori preference is to be used or not. By default False :type a_priori: bool, optional :param interact: A bool variable defining whether interactive preference is to be used or not. By default False :type interact: bool, optional :param n_iterations: The total number of iterations to be run, by default 10. This is not a hard limit and is only used for an internal counter. :type n_iterations: int, optional :param n_gen_per_iter: The total number of generations in an iteration to be run, by default 100. This is not a hard limit and is only used for an internal counter. :type n_gen_per_iter: int, optional :param total_function_evaluations: Set an upper limit to the total number of function evaluations. When set to zero, this argument is ignored and other termination criteria are used. :type total_function_evaluations: int, optional :param penalty_time_component: The APD formula had to be slightly changed. If penalty_time_component is a float between [0, 1], (t/t_max) is replaced by that constant for the entire algorithm. If penalty_time_component is "original", the original intent of the paper is followed and (t/t_max) is calculated as (current generation count/total number of generations). If penalty_time_component is "function_count", (t/t_max) is calculated as (current function evaluation count/total number of function evaluations) If penalty_time_component is "interactive", (t/t_max) is calculated as (Current gen count within an iteration/Total gen count within an iteration). Hence, time penalty is always zero at the beginning of each iteration, and one at the end of each iteration. Note: If the penalty_time_component ever exceeds one, the value one is used as the penalty_time_component. If no value is provided, an appropriate default is selected. If `interact` is true, penalty_time_component is "interactive" by default. If `interact` is false, but `total_function_evaluations` is provided, penalty_time_component is "function_count" by default. If `interact` is false, but `total_function_evaluations` is not provided, penalty_time_component is "original" by default. :type penalty_time_component: Union[str, float], optional .. py:method:: _time_penalty_constant() Returns the constant time penalty value. .. py:method:: _time_penalty_original() Calculates the appropriate time penalty value, by the original formula. .. py:method:: _time_penalty_interactive() Calculates the appropriate time penalty value. .. py:method:: _time_penalty_function_count() Calculates the appropriate time penalty value.