State-action value function (Q-function)

expected \gls{DEF_return} when starting from state $\mathbf{x}$ at time $t$, taking action $\mathbf{a}$, and then following policy \gls{policy}