4

Here they say that the emptyset is compact. Nevertheless, in $\mathbb R$, I know that compact sets are closed and bounded. So, indeed $\varnothing $ is closed, but we have that $$\inf_{\mathbb R}\varnothing =+\infty \quad \text{and}\quad \sup_{\mathbb R}\varnothing =-\infty ,$$ in partuclar, it doesn't seem bounded. So, who is correct ?

Bruce
  • 299

2 Answers2

12

Boundedness is not defined in terms of $\inf$ and $\sup$ (it's actually the other way around; $\inf$ and $\sup$ are defined in terms of bounds).

A subset of $\Bbb R$ is bounded if there is a positive real number $M$ such that each element of the subset is contained in the interval $(-M, M)$. The empty set fulfills this vacuously.

Arthur
  • 204,511
  • With same vacuous logic empty set is unbounded also. – zkutch Sep 14 '20 at 11:17
  • 1
    @zkutch Only if your definition of "unbounded" is something other than "not bounded". And it really shouldn't be. What is your definition, by the way? – Arthur Sep 14 '20 at 12:01
  • Let's speak about unboundedness only from right. May be I am wrong, but I think about following, let call it, "strong" unboundedness: $\forall M \in \mathbb{R}, \forall x \in \emptyset, x>M$. As you see I change $\exists$ with $\forall$ for $x \in \emptyset$. Now if this sentence we can consider as correct, then we can obtain by it usual unboundedness, because it use $\forall \Rightarrow \exists$. By the way, unique set, which I can imagine with "strong" unboundedness is ${ +\infty }$. – zkutch Sep 14 '20 at 20:14
  • 1
    @zkutch I think $\forall x \in A,P(x) \implies \exists x \in A, P(x)$ only if $A$ is nonempty. – somitra Aug 07 '21 at 06:35
7

The empty set is a subset of any set.

In particular it is included in open balls $\varnothing\subset B(0,r)$ so it is bounded.

But you do not really need a metric, since it is included in any open set, for a given open covering you can just take any set in it and it becomes a finite covering, making the empty set compact.

zwim
  • 29,833