# The total collected rewards in the q learning?

In the q learning assignment, I am asked to get the total collected rewards in one episode,

For example, $Q(s,t) = Q(s,t) + a(R + rQ(s',t') - Q(s,t))$.

Which one is it in the following choice?

1. the sum of all the $Q(s,a)$ values

2. or all the updated part - sum ( $a(R + rQ(s',t') - Q(s,t))$ )?

3. or all the updated part in the policy - all the walking steps from the beginning to the end ?

-
 There is the total reward $\sum_{t}R_t$ and there is the total discounted reward $\sum_{t}\gamma^tR_t$. – Don Reba Sep 25 '12 at 8:50