I am working on a reinforcement learning strategy for parameter control of a local search heuristic.
The complete state for this RL problem can be defined as $S = \{s, p\}$, where $s$ and $p$ are respectively the current solution and the vector of current parameter values.
An action correponds to raplacing or maintaining the vector of parameters $p$ with a new configuration.
Since the above formulation of the state space is impractical, would it be reasonable to represent the state only as $S = p$? In this case, the state space basically becomes equivalent to the action space.
