I am having some trouble wrapping my head around the formalization to this. I'm working in $\mathbb{R}$ here for simplicity.

We say a set $S$ is closed if the set contains all its limit points. A limit point is a point where any positive-radius ball we draw around it must include points inside $S$ (or does it suffice to say "at least one point in $S$"? Anyway...). I also don't know if balls include points on the border of the radius or not. Anyhow if $S$ is open then this means $S$ includes the points on its "border" since such a point will have a ball that must overlap $S$'s territory.

An open set, in contrast, is a set $S$ such that any point belonging to $S$ we can make a positive-radius ball around it that is fully inside $S$. This means it does not include the "border" points since part of their ball will be inside $S$ and the other part will be outside of it.

Now the complement of $S$ is everything outside of $S$, which I believe is denoted $S^C$. Let's assume $S$ is an open set. That means every point belonging to $S$ has a ball fully inside $S$. Therefore these balls do not contain anything in $S^C$, which by definition means they are not limit points of $S^C$. That means any limit point of $S^C$ must belong to $S^C$. And if $S^C$ contains all its limit points that must mean it is closed by definition.

This seems right to me but is there a way to formally describe everything that's going on so there's a more rigorous way to express all these concepts rather than relying on this sort of mental picture I am using? I don't really know what it means formally to be a ball, or to "overlap", or to be "in" a set, or "on" a border, or to be a limit point, or a point with a ball inside the set, etc etc etc.

  • $\begingroup$ Of course there is a way to be formal about it. In metric spaces (such as the set of reals with the standard metric), you define an open ball of radius $r$ centered at a point $p$ as the set of elements of the space that are less than distance $r$ from $p$. The closed ball replaces "less than" with "at most," i.e. "less than or equal to." Etc. All this is covered formally in any book on real analysis. $\endgroup$ – symplectomorphic Mar 22 '18 at 5:12

since I taught you those definitions, perhaps I should formalize them:

1) A metric space $X$ is a set of element, and a "rule" called a metric to define what the "distance" between any two point is. $d(x,y) =$ the "distance" between $x$ and $y$.

$d(x,y)$ must follow the following rules:

1) $d(x,y) \ge 0$. A distance is always non-negative real value.

2) $d(x,y) = 0 \iff x=y$. Every two distinct points are always some real positive distance apart. There is never any distance between a point and itself.

3) $d(x,y) = d(x,y)$. Going from point $a$ to point $b$ is the same distance as going from point $b$ to point $a$.

4) $d(x,z) \le d(x,y) + d(y, z)$. The triangle inequality. Two legs of a triangle add up to more than the third leg. Going for point $x$ and $z$ directly is shorter than taking a side trip for $x$ to $y$ and then $y$ to $z$ (unless $y$ is directly in line with $x$ and $z$ in which case it is equal.)

In $\mathbb R$, $d(x,y) = |x - y|$ and in $\mathbb R^2$, $d((x_1,y_1),(x_2,y_2)) = \sqrt{(x_1 - x_2)^2 + (y_1 + y_2)^2}$ are examples of metrics but they aren't the only ones. Anything that satifies the rules is a metric space.

For a point $x$ in the metric space an "open ball" or a "neighborhood" of $x$ is the set of all points $y$ so that $d(x,y) < r$ for some $r > 0$. (In $\mathbb R$ with the euclidean metric this is all points $y$ so that $|x-y| < r$.)

An "interior" point of a set $A$ is a point $a \in A$ so that there exist some neighbor hood (in other words a distance $r$) so that all the points in the neighborhood (in other words all $y$ so that $d(x,y) < r$) are in $A$. (in other words so that $\{y|d(x,y) < r \} \subset A$.

A "limit point" of a set $A$ is a point $y$ (maybe in $A$; maybe not) so that for every neighborhood of $y$ (no matter how small an $r$ we choose), every neighborhood of $y$ will contain a point, not equal to $y$, that is in the set $A$.

A set is called "open" if every point is an interior point.

A set is called "closed" if all its limit points are in the set. Or alternatively, if none of the points not in the set are limit points.

Theorem: A set $A$ is open, if and only if $A^c = \{y \in X| y \not \in A\}$ is closed.

Pf: If $A$ is open then all points in $A$ are interior points. So for every point $a\in A$ there is a neighborhood that is entirely a subset of $A$. That means the neighborhood does not contain any points not in $A$. So it does not contain any point of $A^c$. So it is not a limit point of $A^c$. So none of the limit points of $A^c$ are in $A$. So all the limit point os $A^c$ (if any) are in $A^c$. So $A^c$ is closed.

If $A^c$ is closed then none of the points of $A$ are limit points of $A^c $. So and if $a \in A$ then $a$ is not a limit point. That means not every neighborhood around $a$ contains a point of $A^c$. THat means there is some neighborhood of $a$ that doesn't have any points in $A^c$. That means all the points of the neighborhood of $a$ are all in $A$. So $a$ is an interior point of $A$. That is true of all points of $A$. So $A$ is open.

  • $\begingroup$ Does this also imply that if $S$ is neither open or closed, then likewise is true for its complement? $\endgroup$ – user525966 Mar 22 '18 at 13:32
  • $\begingroup$ Yep............. $\endgroup$ – fleablood Mar 22 '18 at 20:46

Let $S$ be open, and consider a limit point $x$ of $ S^C$. Then $x\in S^C$, for if instead $x\in S$ we would have a ball $B(x;r) \subset S$ with $r>0$. But since $x$ is a limit point of $S^C$ we know the ball also contains elements of $S^C$, a contradiction. So $S^C$ contains all its limit points and so is closed.

A ball is usually defined by $B(a;r)=\{ x: d(x,a) < r\}$ so it is open unless otherwise defined.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.