In Sipser's Introduction to the theory of computation (3rd edition), I found the following claim.
Consider the grammar:
$$ \begin{align*} &R \to XRX \mid S \\ &S \to aTb \mid bTa \\ &T \to XTX \mid X \mid \epsilon \\ &X \to a \mid b \end{align*} $$
In this grammar, it holds that $T \Rightarrow^* T$.
Can anyone explain how this is true?
We say that $\alpha \Rightarrow^* \beta$ if $\beta$ can be derived from $\alpha$ in zero or more steps. (More fancily, $\Rightarrow^*$ is the reflexive-transitive closure of $\Rightarrow$.) In particular, it is always that case that $\alpha \Rightarrow^* \alpha$, for every $\alpha$, due to a derivation of zero steps.
In contrast, in your case it doesn't hold that $T \Rightarrow^+ T$. (This is derivation in one or more steps, or the transitive closure of $\Rightarrow$.)
