//Initiate S
node.state = start;
myPath.add(node.state);

//Given the state select action and get statePrime
//Beware that action & statePrime are the same
statePrime = myActions.epsilonGreedy(node.state, 0.1f, qMatrix);

while(node.state != goal){

    actionPrime = myActions.epsilonGreedy(statePrime, 0.1f, qMatrix);

    //- - - - - - - - - - Cliff - - - - - - - - - - - - -
    if(rMatrix[node.state][statePrime] == -100){
        //Walked into the cliff!
        //Update Q Matrix
        myQ = qMatrix[statePrime][actionPrime];

        qMatrix[node.state][statePrime] = alpha*(rMatrix[node.state][statePrime] + myQ)+ (1-alpha)*qMatrix[node.state][statePrime];

        //Agent is returned to the start
        node.state = start;
        statePrime = myActions.epsilonGreedy(node.state, 0.1f, qMatrix);

        myPath.add(node.state);
        }

    else{
        //No cliff, continue
        myReward.add(rMatrix[node.state][statePrime]);
        //Update Q Matrix
        myQ = qMatrix[statePrime][actionPrime];

        qMatrix[node.state][statePrime] = alpha*(rMatrix[node.state][statePrime] + myQ)+ (1-alpha)*qMatrix[node.state][statePrime];

        node.state = statePrime;
        statePrime = actionPrime;

        myPath.add(node.state);
        }

public Integer epsilonGreedy(Integer state, Float epsilon, Float[][] qMatrix){
    Random random = new Random();
    Float value = random.nextFloat();
    Integer statePrime = 0;

    if(value < epsilon){
        myNode.state = state;
        statePrime = random(myNode.neighbors());
    }
    else{
        statePrime = greedy(state, qMatrix);
    }
    return statePrime;
}

(state,action)==(state,statePrime)
(state', action')==(statePrime,actionPrime)

000000000000
111111111111
100000000001
100000000001

111111111111
100000000001
100000000001
100000000001