Examples of Reward_observation_terminal

org.rlcommunity.rlglue.codec.types.Reward_observation_terminal
I'm iteratively updating these methods to have getters and setters for all components because eventually we might like to make the actual members private. @author btanner

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal

    }
    
    public void start()
    {
        Action action = RLGlue.RL_agent_start(RLGlue.RL_env_start());
        Reward_observation_terminal rot = RLGlue.RL_env_step(action);
        reward = rot.getReward();
        observation = rot.getObservation();
        
        if(observation == null)
        {
            throw new RuntimeException("observation should not be null");
        }

View Full Code Here

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal

        {
            throw new RuntimeException("observation should not be null");
        }
        
        Action action = RLGlue.RL_agent_step(reward,observation);
        Reward_observation_terminal rot = RLGlue.RL_env_step(action);
        reward = rot.getReward();
        observation = rot.getObservation();
        
        if(rot.isTerminal())
        {
            RLGlue.RL_agent_end(reward);
            RLGlue.RL_cleanup();
        }
    }

View Full Code Here

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal

    public Reward_observation_terminal env_step(Action action)
    {
        // Translate the action to a GroundFact
        GroundFact fact = parseAction(action);
        applyAction(fact);
        Reward_observation_terminal ret = new Reward_observation_terminal();
        ret.setObservation(createObservation(state));
        ret.setReward(state.getReward(role.toString()));
        ret.setTerminal(state.isTerminal());
        return ret;
    }

View Full Code Here

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal

        // Se asigna a la posicion 1 el numero de estados
        returnObservation.intArray[1] = numStates;
        // Se crea un clase Reward_observation_terminal
        // que contiene la recompensa, el estado actual y la bandera de fin de
        // episodio
        Reward_observation_terminal returnRewardObs = new Reward_observation_terminal(
                theReward, returnObservation, episodeOver);
        return returnRewardObs;
    }

View Full Code Here

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal

        }
        return theAction;
    }
    
    public synchronized Reward_observation_terminal RL_env_step(Action theAction) {
        Reward_observation_terminal RO = E.env_step(theAction);
        if (RO == null) {
            System.err.println("RO came back as null from RL_step");
        }
        if (RO.getObservation() == null) {
            System.err.println("Ro.o came back as null from RL_step");
        }


        totalReward += RO.getReward();


        if (RO.isTerminal()) {
            numEpisodes++;
        } else {
            numSteps++;
        }
        return RO;

View Full Code Here

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal


    public synchronized Reward_observation_action_terminal RL_step() {
        if (lastAction == null) {
            System.err.println("lastAction came back as null from RL_step");
        }
        Reward_observation_terminal RO=RL_env_step(lastAction);
       


        if (RO.isTerminal()) {
            RL_agent_end(RO.getReward());
        } else {
           lastAction = RL_agent_step(RO.getReward(), RO.getObservation());
        }
        return new Reward_observation_action_terminal(RO.getReward(), RO.getObservation(), lastAction, RO.terminal);
    }

View Full Code Here

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal

            updateObservers(firstAction);


        } else {
            Reward_observation_action_terminal ROAT = new Reward_observation_action_terminal();


            Reward_observation_terminal ROT = RLGlue.RL_env_step(lastAction);


            ROAT.o = ROT.getObservation();
            ROAT.r = ROT.getReward();
            boolean isTerminal=ROT.isTerminal();
            ROAT.terminal=0;
            if(isTerminal){
                ROAT.terminal=1;
            }






            synchronized (this) {
                totalSteps++;
                timeStep++;
                lastObservation = ROAT.getObservation();
                lastReward = ROAT.getReward();


                returnThisEpisode += lastReward;
                totalReturn += lastReward;
            }


            updateObservers(ROT);


            if (ROT.isTerminal()) {
                RLGlue.RL_agent_end(ROT.getReward());
            } else {
                ROAT.a = RLGlue.RL_agent_step(ROT.getReward(), ROT.getObservation());
            }


            synchronized (this) {
                if (!ROAT.isTerminal()) {
                    lastAction = ROAT.getAction();

View Full Code Here

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal

        theWorld.updatePosition(thisAction.getInt(0));




        Observation theObservation = new Observation(1, 0, 0);
        theObservation.setInt(0, theWorld.getState());
        Reward_observation_terminal RewardObs = new Reward_observation_terminal();
        RewardObs.setObservation(theObservation);
        RewardObs.setTerminal(theWorld.isTerminal());
        RewardObs.setReward(theWorld.getReward());


        return RewardObs;
    }

View Full Code Here

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal

            theReward=1.0d;
        }
        Observation returnObservation=new Observation(1,0,0);
        returnObservation.intArray[0]=currentState;
        
        Reward_observation_terminal returnRewardObs=new Reward_observation_terminal(theReward,returnObservation,episodeOver);
        return returnRewardObs;
    }

View Full Code Here

Examples of org.rlcommunity.rlglue.codec.types.Reward_observation_terminal

public abstract class EnvironmentBase implements EnvironmentInterface {


  abstract protected Observation makeObservation();
  
  protected Reward_observation_terminal makeRewardObservation(double reward, boolean isTerminal){
    Reward_observation_terminal RO=new Reward_observation_terminal();
    RO.o=makeObservation();
    RO.r=reward;
    
    RO.terminal=1;
    if(!isTerminal)

View Full Code Here

0 1

TOP

All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.