Of Notes
Richard Bellman
- RAND mathematician who named it dynamic programming, a deliberately boring name so the Secretary of Defense would not ask what he was researching.
- "Dynamics" because he was studying multi-stage decision processes; "programming" in the military-planning sense.
- The Bellman Equation: value of a state is the immediate reward plus the optimal value you can reach from the next state.
- It is the backbone of optimal control, and is the exact form that latent prediction with actions takes in JEPA's
.