(I am aware this question may be considered subjective and somewhat nitpicky. I hope it is not too inappropriate.)

When doing proofs in analysis, for example, one may encounter some convergent sequence $(x_n)_{n\in\mathbb{N}}$ in some metric space $(X,d)$. Symbolically, one may write: $\forall \varepsilon >0 \exists N\in\mathbb{N} \forall n\geqslant N:d(x,x_n)<\varepsilon $, where $x\in X$ is the limit of $(x_n)_{n\in\mathbb{N}}$. Some argument using the convergence of $(x_n)_{n\in\mathbb{N}}$ may start like this:

"Let $\varepsilon >0$. Since $(x_n)_{n\in\mathbb{N}}$ converges to $x$, we can find $N\in\mathbb{N}$ such that for all $n\geqslant N$ we have $d(x,x_n)<\varepsilon$."

It is possible that the natural number $N$ (such that $\forall n\geqslant N: d(x,x_n)<\varepsilon$) is needed in some further part of the argument, so we want to fix the letter $N$ as being that specific natural number. However the above line does not explicitly specify that the letter $N$ should represent this natural number. It really just says that such a number exists.

When writing proofs, is it considered good custom to add another line explicitly specyfing $N$? For example by adding a line such as

"Let $N\in\mathbb{N}$ be such that $\forall n\geqslant N: d(x,x_n)<\varepsilon$."


"Let $N\in\mathbb{N}$ such that the above holds."

Or, is it more customary to just leave the specification implicit? (That is, write "$\ldots$we can find $N\in\mathbb{N}$ such that$\ldots$", and just use the letter $N$ assuming that it has been appropriately specified.) We could also write:

"Let $\varepsilon >0$. Let $N\in\mathbb{N}$ such that $\forall n \geqslant N: d(x,x_n)<\varepsilon$, which we can find because $(x_n)_{n\in\mathbb{N}}$ converges.

This somewhat combines the existence and specification of $N$ into one line. Is there any other way of handling this nicely?

Thank you for your time.

  • 1
    $\begingroup$ The only way to "formally" disambigaute such cases is with the scope of the quantifiers; if we write: $∀ε > 0 \ ∃N ∈ \mathbb N \ ∀n ⩾ N \ d(x,x_n)<ε$, we have no problem. If we have a longer semi-formal expression like e.g.: "$∀ε > 0 \ ∃N ∈ \mathbb N \ ∀n ⩾ N(d(x,x_n)<ε$ and $N > 5$", we may have the doubt that the last occurrence of $N$ is outside the scope of the initial quantifier. In cases like this, we have to add parentheses. $\endgroup$ Sep 2, 2016 at 11:37
  • 1
    $\begingroup$ Existential instantiation is definitely a thing. The tricky matter is when you have existentials which are not at top level, such as your $N $. It can be helpful to Skolemize in this case, which in your example means introducing a function $N (\varepsilon) $. In my class we did this for understanding uniform continuity and uniform convergence. Skolemization can have bad effects with respect to obfuscation foundations sometimes; for example it can make usage of the axiom of choice less explicit. $\endgroup$
    – Ian
    Sep 2, 2016 at 11:42

1 Answer 1


It's very rare for people to add a whole extra sentence just to introduce formally something whose existence was already proven. Here are some more constructions that can be used instead:

Since $(x_n)_{n\in\mathbb{N}}$ converges to $x$, we can find $N\in\mathbb{N}$ such that for all $n\geqslant N$ we have $d(x,x_n)<\varepsilon$. Given such an $N$, ...

Since $(x_n)_{n\in\mathbb{N}}$ converges to $x$, we can (and do) choose $N\in\mathbb{N}$ such that for all $n\geqslant N$ we have $d(x,x_n)<\varepsilon$.

Since $(x_n)_{n\in\mathbb{N}}$ converges to $x$, pick $N\in\mathbb{N}$ such that for all $n\geqslant N$ we have $d(x,x_n)<\varepsilon$.

The last is the briefest, although perhaps a bit ungrammatical. Alternatively it is acceptable to, as you say, "leave the specification implicit".


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .