Bellman expectation equation
Value defined in terms of itself one step later: a reward now, plus the discounted value of wherever you land. Every value-based algorithm is some way of solving this fixed point.
- The next state.
- The immediate reward.
- The discount factor.