module gmle.salsa.framework;

import java.io.BufferedWriter;
import java.io.File;
import java.io.FileWriter;

public behavior GeneticSearch {

	/**
	 *	Variables to control the genetic search
	 */
	int elite_population_size = 5;
	int population_size = 20;
	double mutation_chance = 0.3;
	double mutation_strength = 1.0;
	double reproduce_chance = 0.7;

	/**
	 *	Variables for stopping conditions
	 */
	int current_iteration = 0;
	int maximum_iterations = 10000;

	double probability_cutoff = 0.001;
	int stopping_population_size = 20;
	double stopping_population_probability_threshold = 0.00001;


	MLEvaluator evaluator;

	public void init_data(String evaluatorClass, String theaters_file) {
		this.evaluator = new MLEvaluator(evaluatorClass, theaters_file);
	}

	public void init_data(String evaluatorClass, int number_actors) {
		this.evaluator = new MLEvaluator(evaluatorClass, number_actors);
	}

        public void initialize(Object[] init_params) {
                evaluator<-initialize(init_params) @ currentContinuation;
        }

	double[] parameter_min;
	double[] parameter_max;

	double[] genetic_search(double[] parameter_min, double[] parameter_max) {
		this.parameter_min = parameter_min;
		this.parameter_max = parameter_max;

		current_population_likelihood = new double[parameter_min.length];
		previous_population_likelihood = new double[parameter_min.length];
		current_population = new double[population_size][parameter_min.length];
		previous_population = new double[population_size][parameter_min.length];
		for (int i = 0; i < current_population.length; i++) {
			for (int j = 0; j < current_population.length; j++) {
				current_population[i][j] = parameter_min[j] + ((parameter_max[j]-parameter_min[j])*Math.random());
			}
		}

		evaluator<-init_data() @
		evaluate_current_population() @
		search() @
		currentContinuation;
	}

	double[] current_population_likelihood;
	double[][] current_population;
	double[] previous_population_likelihood;
	double[][] previous_population;

	void evaluation_current_population() {
		join {
			for (int i = 0; i < current_population.length; i++) {
				evaluator<-evaluate(current_population[i]);
			}
		} @
		set_current_likelihood(token) @
		currentContinuation;
	}

	void set_current_likelihood(Object[] results) {
		for (int i = 0; i < results.length; i++) {
			current_population_likelihood[i] = ((Double)results[i]).doubleValue();
		}
	}

	double[] search() {
		for (int i = 0; i < current_population_likelihood.length; i++) {
			previous_population_likelihood[i] = current_population_likelihood[i];
			for (int j = 0; j < current_population[i].length; j++) {
				previous_population[i][j] = current_population[i][j];
			}
		}

		//generate new population


		return new double[10];
	}




	BufferedWriter evaluation_file;
	void set_evaluation_output(String filename) {
		try {
			this.evaluation_file = new BufferedWriter(new FileWriter(new File(filename)));
		} catch (Exception e) {
			System.err.println("Unable to open evaluation output file: " + filename);
			System.err.println(e);
			e.printStackTrace();
			System.exit(0);
		}
	}

	int evaluations_done = 0;
	double track_evaluate(double[] parameters) {
		token result = evaluator<-evaluate(parameters);
		write_result(result, parameters) @
		currentContinuation;
	}

	double write_result(double result, double[] parameters) {
		if (evaluation_file != null) {
			evaluations_done++;
			try {
				evaluation_file.write(evaluations_done + " : " + result + " :");
				for (int i = 0; i < parameters.length; i++) {
					evaluation_file.write(" " + parameters[i]);
				}
				evaluation_file.write("\n");
				evaluation_file.flush();
			} catch (Exception e) {
				System.err.println("Unable to write to evaluation output file");
				System.err.println(e);
				e.printStackTrace();
				System.exit(0);
			}
		}
		return result;
	}
}