I'm reading the definition of cluster point in Royden's Real Analysis. It is as follows.
$\ell$ is a cluster point of $<x_n>$ if, given $\epsilon>0$ and given $N$, $\exists n \geq N$ such that $|x_n- \ell| < \epsilon$.
Prior to reading this definition, my understanding of cluster point $\ell$ was that there should be INFINITELY MANY points of the sequence in ANY neighborhood of $\ell$. However, with the definition given in Royden's book, I don't see any conditions that enforce the existence of infinitely many points of a sequence and any neighborhood? In fact, I think one should reword the definition as follows.
$\ell$ is a cluster point of $<x_n>$ if, for any $\epsilon>0$, there exists infinitely many $x_n$ such that $|x_n- \ell| < \epsilon$.
Is there anything wrong with my rationale?
The definition does secretly enforce the existence of infinitely many $x_n$. Notice that $N$ is given. Thus, if you find some $x_k$ such that $|x_k - l| < \varepsilon$, then you can again apply the definition using $N' > k$. You can then find $x_{k+a}$ where $|x_{k+a} - l| < \varepsilon$.