package ch.idsia.scenarios.test;
import ch.idsia.ai.SRN;
import ch.idsia.ai.agents.Agent;
import ch.idsia.ai.agents.ai.LargeSRNAgent;
import ch.idsia.tools.EvaluationOptions;
import ch.idsia.tools.CmdLineOptions;
import ch.idsia.tools.Evaluator;
import ch.idsia.tools.EvaluationInfo;
/**
* Created by IntelliJ IDEA.
* User: koutnij
* Date: Jul 27, 2009
* Time: 4:34:37 PM
*/
public class EvaluateJLink {
/** returns {in, rec, out} array. Just to make math and java codes fully independent. */
public static int[] getDimension() {
return new int[]{getInputSize()*getInputSize()*2+3, 6, 6};
}
/** returns length of an edge of the input window square*/
public static int getInputSize() {
return 7;
}
public double evaluateLargeSRN (double[][] inputs, double[][] recurrent, double[][] output, int level, int seed) {
// System.out.println(inputs.length+" "+inputs[0].length);
// System.out.println(recurrent.length+" "+recurrent[0].length);
// System.out.println(output.length+" "+output[0].length);
SRN srn = new SRN (inputs, recurrent, output, recurrent.length, output[0].length);
Agent agent = new LargeSRNAgent(srn);
EvaluationOptions options = new CmdLineOptions(new String[0]);
final int startingSeed = 0;
options.setLevelRandSeed(seed);
options.setMaxAttempts(1);
options.setVisualization(false);
options.setMaxFPS(true);
options.setLevelDifficulty(level);
options.setPauseWorld(false);
agent.reset();
options.setAgent(agent);
Evaluator evaluator = new Evaluator (options);
EvaluationInfo result = evaluator.evaluate().get(0);
// System.out.print(".");
double score = result.computeDistancePassed();
System.out.println("score: " +score);
return score;
}
}