I have a transition matrix $P=(p_{x,y})_{x,y\in S}$ which defines some Markov chain $X$ on some state space $S$. Now, I consider for $A,B\subset S$, $A\cap B=\emptyset$ the probability that $X$ transitions in one step from $A$ to $B$, i.e., $P[X_{t+1} \in B| X_t\in A]$. I want to find a lower bound on this transition probability and I know that I have a uniform bound $p_{x,y}\geq \alpha$ for some constant $\alpha\in (0,1)$ and all $x,y\in S$ with $p_{x,y}>0$ .
Denoting by $E_{A,B}$ the number of possible transitions between $A$ and $B$, i.e., $$E_{A,B}=|\{x\in A, y\in B| p_{x,y}>0\}|$$ intuitively, I only consider each transition probability for each $x\in A$ to $y\in B$, i.e., $\sum_{x\in A,y\in B}p_{x,y}$ and I then find the estimate $$\sum_{x\in A,y\in B}p_{x,y}\geq E_{A,B}\alpha.$$
But, this does not really work, since by rules of conditional probability I have $P[X_{t+1} \in B| X_t\in A]\neq \sum_{x\in A}P[X_{t+1} \in B| X_t = x]$. Is the intuition correct and how can I prove it in this case? Or is it incorrect and what can I do?
PS: Basically, when I calculate it explicitly I am stuck with $$\alpha\sum_{x\in A,y\in B, p_{x,y}>0} P[X_t = x| X_t \in A]$$ as a lower bound.