I need to show the following assertion: if $(a_n)$ is a Cauchy sequence in $\mathbb{N}$, then $a_n$ is constant starting from a certain index.
In other words: $\exists (n_o, N \in \mathbb{N)}, \forall n \geq n_o: a_n = N $
I don't understand how this can be true. This implies that a Cauchy sequence will eventually become constant, which is not true. It converges to a certain value, but never actually reach it. So there clearly is something else I'm not taking into account but I can't figure out what.
Thank you very much. Any help is highly appreciated
You may be slightly mixed up on the difference between being Cauchy and convergent. (Though maybe not; it's hard to tell.) Whether or not a Cauchy sequence must converge depends on the space you're working in. In $\mathbb{R}$, every Cauchy sequence converges, and this is a very special property of the reals. In $\mathbb{Q}$, there are Cauchy sequences that fail to converge, because what they "should" converge to is not rational. (They will still converge to a real number, though, but that's "outside" your space.) The good example is the increasingly-long decimal approximations to $\sqrt{2}$.
In your problem, you have a Cauchy sequence of natural numbers. And yes, in your case, this will necessarily converge (to the eventual constant). Let $\{a_n\}$ be your Cauchy sequence of naturals. Let $\epsilon = 0.5$. Then there is an index $k$ such that $m,n \geq k$ implies that $|a_m-a_n| < 0.5$. But the closest any two distinct natural numbers can ever be is $1$: this occurs only when they are consecutive, of course. Since $a_m$ and $a_n$ are natural, this implies that we must have $a_m=a_n$ for all indices $m,n \geq k$. In particular, the sequence must be constant starting at index $k$ (maybe earlier) and beyond. Hence the sequence is eventually constant with value $a_k$, and that will be what it converges to.
It is good to work out what goes wrong with this argument in $\mathbb{Q}$ or $\mathbb{R}$, because as you pointed out, it is definitely not true there.