Examples of org.rlcommunity.rlglue.codec.taskspec.TaskSpecVRLGLUE3


        //Create a task spec programmatically.  This task spec encodes that state, action, and reward space for the problem.
        //You could forgo the task spec if your agent and environment have been created specifically to work with each other
        //ie, there is no need to share this information at run time.  You could also use your own ad-hoc task specification language,
        //or use the official one but just hard code the string instead of constructing it this way.
        TaskSpecVRLGLUE3 theTaskSpecObject = new TaskSpecVRLGLUE3();
        theTaskSpecObject.setEpisodic();
        theTaskSpecObject.setDiscountFactor(1.0d);


        //Specify that there will be an integer observation [0,107] for the state
        theTaskSpecObject.addDiscreteObservation(new IntRange(0, theWorld.getNumStates() - 1));
        //Specify that there will be an integer action [0,3]
        theTaskSpecObject.addDiscreteAction(new IntRange(0, 3));
        //Specify the reward range [-100,10]
        theTaskSpecObject.setRewardRange(new DoubleRange(-100.0d, 10.0d));


        theTaskSpecObject.setExtra("SampleMinesEnvironment(Java) by Brian Tanner.");


        String taskSpecString = theTaskSpecObject.toTaskSpec();
        TaskSpec.checkTaskSpec(taskSpecString);


        return taskSpecString;
    }

  
  //Create a task spec programatically.  This task spec encodes that state, action, and reward space for the problem.
  //You could forgo the task spec if your agent and environment have been created specifically to work with each other
  //ie, there is no need to share this information at run time.  You could also use your own ad-hoc task specification language,
  //or use the official one but just hard code the string instead of constructing it this way.
      TaskSpecVRLGLUE3 theTaskSpecObject = new TaskSpecVRLGLUE3();
        theTaskSpecObject.setEpisodic();
        theTaskSpecObject.setDiscountFactor(1.0d);
  //Specify that there will be an integer observation [0,20] for the state
        theTaskSpecObject.addDiscreteObservation(new IntRange(0, 20));
  //Specify that there will be an integer action [0,1]
        theTaskSpecObject.addDiscreteAction(new IntRange(0, 1));
  //Specify the reward range [-1,1]
        theTaskSpecObject.setRewardRange(new DoubleRange(-1, 1));


        String taskSpecString = theTaskSpecObject.toTaskSpec();
        TaskSpec.checkTaskSpec(taskSpecString);


    //This actual string this makes is:
    //VERSION RL-Glue-3.0 PROBLEMTYPE episodic DISCOUNTFACTOR 1.0 OBSERVATIONS INTS (1 0 20)  ACTIONS INTS (1 0 1)  REWARDS (1 -1.0 1.0)  EXTRA

Examples of org.rlcommunity.rlglue.codec.taskspec.TaskSpecVRLGLUE3

Related Classes of org.rlcommunity.rlglue.codec.taskspec.TaskSpecVRLGLUE3