Let $f: \mathbb{R} \to \mathbb{R}$ be a function.
Prove that: $$\forall (x_n)_n \subseteq \mathbb{R}: (x_n \to x \land \forall n: x_n \leq x_{n+1} \implies f(x_n) \to a)$$ implies that $\lim_{y \to x-}f(y) = a$
My attempt:
Suppose $\lim_{y\to x-}f(y) \neq a$. This means,
$$\exists \epsilon >0: \forall \delta >0: \exists y \in \mathbb{R}: 0 < x-y < \delta \land |f(y) - a | \geq \epsilon$$
Choosing this $\epsilon > 0$ and letting $\delta := 1/n$, we can find a sequence $(y_n)_n$ in the reals such that $0 < x-y_n< 1/n$ and $|f(y_n)-a| \geq \epsilon$
This implies that $y_n \to x$ and $f(y_n) \not\to a$ and it is also clear that $(y_n)_n$ is increasing. The result follows by contraposition.
Is this correct?
This is not quite right: it is not "clear" that $(y_n)$ is increasing, and in fact that may not be true. There is absolutely no reason you couldn't have $y_{n+1}<y_n$, if the only restriction in choosing $y_n$ is that $0 < x-y_n< 1/n$ and $|f(y_n)-a| \geq \epsilon$.
To fix this, you need to change how you choose the sequence $y_n$ in order to guarantee that it is increasing.