R/qLearnQ2.R

In kalinn/iqLearn: Interactive Q-Learning

qLearnQ2 <-
function (object, h2){

  if (length (object$s2ints) > 0){
    h2pos = c (1, h2, 1, h2[object$s2ints]);
    q2Pos = sum (h2pos*c (object$betaHat20, object$betaHat21));
    h2neg = c (1, h2, -1, -h2[object$s2ints]);
    q2Neg = sum (h2neg*c (object$betaHat20, object$betaHat21));
  }
  else{
    h2pos = c (1, h2, 1);
    q2Pos = sum (h2pos*c (object$betaHat20, object$betaHat21));
    h2neg = c (1, h2, -1);
    q2Neg = sum (h2neg*c (object$betaHat20, object$betaHat21));
  }
    
  q2opt = sign (q2Pos - q2Neg);
  return (list ("q2Pos"=q2Pos, "q2Neg"=q2Neg, "q2opt"=q2opt));
}