algorithm_runner.jl

include("./representation.jl")
include("./parent_selection.jl")
include("./recombination.jl")
include("./mutation.jl")
include("./metrics.jl")
include("./functions_collection.jl")
include("./utils.jl")
include("./survivor_selection.jl")
include("./graphing_utilities.jl")

using .Representation, .ParentSelection, .Recombination, .Mutation, .Metrics, .FunctionsCollection
using .SurvivorSelection, .Utils, Statistics, .GraphingUtilities

parameters =  Utils.get_algo_params()
x_values = Utils.generate_linespace(parameters["x_range"][1], parameters["x_range"][2], parameters["population_size"])
true_values = parameters["evaluate_function"](x_values)
population = Representation.generate_population(parameters["population_size"], parameters["n_kernels"], parameters["σ_initial"])
extra_params_error = Dict("threshold"=>parameters["threshold"],
                        "lower_weight"=>parameters["lower_weight"],
                        "upper_weight"=>parameters["upper_weight"])

Metrics.calc_population_fitness!(population,
                                 true_values,
                                 parameters["error_function"],
                                 extra_params_error)

all_fitness_values = Array{Array{Float64, 1}, 1}[]
all_hit_ratios = Array{Array{Float64, 1}, 1}[]
for run in 1:parameters["num_runs"]
    fitness_values = Array{Float64, 1}[]
    hit_ratios = Array{Float64, 1}[]
    for i in 1:parameters["num_generations"]
        println("Generation: ", i)
        println("Selecting parents...")
        parents_groups = ParentSelection.random_parent_selection(population, parameters["λ_children"],
                                                                parameters["μ_parents"])
        println("Applying recombination...")
        offspring = Recombination.apply_recombination(parents_groups, parameters["μ_parents"],
                                                    parameters["recombination_type"])
        println("Applying mutation")
        offspring_population = Representation.Population(offspring)
        mutated_population = parameters["mutation_type"](offspring_population)
        println("Calculating fitness...")
        Metrics.calc_population_fitness!(mutated_population,
                                         true_values,
                                         parameters["error_function"],
                                         extra_params_error)
        println("Selecting survivors...")
        next_generation = nothing
        if parameters["survivor_selection"] == "children"
            next_generation = SurvivorSelection.rank_selection(mutated_population, parameters["μ_parents"])
        elseif parameters["survivor_selection"] == "all"
            next_generation = SurvivorSelection.rank_selection(mutated_population, population, parameters["μ_parents"])
        else
            throw(ArgumentError("Survivor selection requested not available. Options are: all, children"))
        end
        push!(fitness_values, [member.fitness for member in next_generation.members])
        push!(hit_ratios, [member.hit_ratio for member in next_generation.members])
        global population = next_generation
    end
    push!(all_fitness_values, fitness_values)
    push!(all_hit_ratios, hit_ratios)
end
println("Finished training...")
mean_runs_fitness = mean(all_fitness_values)
mean_hit_ratio = mean(all_hit_ratios)

output_name = nothing
if parameters["output_file"] != nothing
    output_name = "outputs/" * parameters["output_file"]
end

GraphingUtilities.plot_all(mean_runs_fitness,
                            parameters["num_generations"],
                            population,
                            true_values,
                            x_values,
                            mean_hit_ratio,
                            parameters["population_size"],
                            output_name)


open(output_name * ".txt", "w") do file
    write(file, string([mean(values) for values in mean_runs_fitness]))
    write(file, "\n")
    write(file, string([mean(values) for values in mean_hit_ratio]))
    write(file, "\n")
end