emulator_parameter_evolution: Evolve parameter sets that meet a desired ensemble outcome

Description Usage Arguments Value

Description

This method takes a user specified fitness function and runs the nsga2 algorithm on an ensemble using the nsga2 implementation provided in the mco package, in an attempt to locate parameters that achieve a desired response (determined by the fitness function). The method outputs a list describing the values for each simulation output measure, (or objective, res), an evolved set of parameter inputs (par), and a boolean stating whether the candidate is pareto optimal (pareto.optimal)

Usage

1
2
emulator_parameter_evolution(function_to_evaluate,
  nsga2_user_set_parameters, nsga2_settings)

Arguments

function_to_evaluate

A user-defined function that NSGA2 seeks to minimise

nsga2_user_set_parameters

An object containing the emulator input and output names, the input parameters for function to evaluate, minimum and maximum values for emulator inputs. These should be set using the function that creates that object prior to running this method

nsga2_settings

An object containing the population size, number of generations, crossover probability and mutation probability to be assessed. Again see the function nsga2_settings to set these values before running this function

Value

List containing evolved parameter sets, the output for the ensemble using those sets, and whether these sets are pareto optimal


kalden/spartan documentation built on May 31, 2019, 11:52 p.m.