Unpacking Quantifiers in Topological Definitions

Definitions as Quantified Statements

Topological definitions often appear to ask for an entire collection of neighborhoods, open sets, or points. Writing the quantifiers explicitly makes the proof task more concrete: a positive claim usually asks us to find a witness after an arbitrary choice, while a negative claim often asks us to find one choice that defeats every possible witness.

Throughout, $(X,\mathcal{T})$ and $(Y,\mathcal{S})$ are topological spaces. A neighborhood in this tutorial means an open set containing the point in question. We use the standard topology on $\mathbb{R}$ when discussing real intervals. For $A\subseteq X$, recall that $x\in\overline{A}$ means that $x$ belongs to the closure of $A$.

Definition (closure membership, in quantified form): A point $x\in X$ belongs to $\overline{A}$ if and only if $$ \text{for every open neighborhood }U\text{ of }x,\text{ there exists }a\in A\text{ such that }a\in U. $$ Its logical negation is: there exists an open neighborhood $U$ of $x$ such that $U\cap A=\varnothing$.

The order of the quantifiers matters. Closure membership says that after any neighborhood $U$ has been chosen, a point $a$ of $A$ can be found inside it. To disprove membership, find one neighborhood that contains no point of $A$. The witness $a$ may depend on $U$; the definition does not ask for one point of $A$ that lies in every neighborhood.

Closure: Find a Point Inside Each Neighborhood

Theorem (neighborhood test for closure): For any topological space $X$, any $A\subseteq X$, and any $x\in X$, $x\in\overline{A}$ if and only if every open neighborhood of $x$ intersects $A$.

Proof. Suppose $x\in\overline{A}$. If some open neighborhood $U$ of $x$ did not intersect $A$, then $X\setminus U$ would be a closed set containing $A$ but not containing $x$. Since the closure is contained in every closed set containing $A$, this would imply $x\notin\overline{A}$, a contradiction. Conversely, suppose every open neighborhood of $x$ intersects $A$. If $x\notin\overline{A}$, then $X\setminus\overline{A}$ is an open neighborhood of $x$, because $\overline{A}$ is closed. It misses $A$, since $A\subseteq\overline{A}$, contradicting the assumption. Therefore $x\in\overline{A}$. $\square$

Worked Example: Testing Two Points Against a Set

In $\mathbb{R}$ with its standard topology, let $A=(0,1)$. Test whether $0$ and $2$ belong to $\overline{A}$.

At $0$. Let $U$ be any open neighborhood of $0$. Because $U$ is open, there exists $\varepsilon>0$ such that $(-\varepsilon,\varepsilon)\subseteq U$. Set $a=\min(\varepsilon,1)/2$. Since $\varepsilon>0$ and $1>0$, we have $0<a\le 1/2<1$, so $a\in(0,1)=A$. Also $a<\varepsilon$, so $a\in(-\varepsilon,\varepsilon)\subseteq U$. Thus every such $U$ meets $A$, and $0\in\overline{A}$.

At $2$. The interval $U=(3/2,5/2)$ is an open neighborhood of $2$. Every point of $U$ is greater than $3/2>1$, so $U\cap(0,1)=\varnothing$. This one neighborhood is a counter-witness, and $2\notin\overline{A}$.

Check: the same reasoning works for every $x\in(0,1)$: any open neighborhood of $x$ contains $x$ itself, which is in $A$. In particular the closure contains $A$ as well as $0$; the point $2$ is excluded by the exhibited neighborhood.

There is a small distinction to keep in view: closure membership allows the point of $A$ found in a neighborhood to be $x$ itself. The definition of a limit point instead requires a point of $A$ different from $x$. Confusing these two quantifier statements can lead to errors at isolated points.

Continuity: One Target Neighborhood, Then a Source Neighborhood

Definition (continuity at a point): A map $f:X\to Y$ is continuous at $x\in X$ if for every open neighborhood $V$ of $f(x)$, there exists an open neighborhood $U$ of $x$ such that $f(U)\subseteq V$. Equivalently, every point $z\in U$ satisfies $f(z)\in V$.

The quantifier order is “for every $V$, there exists $U$.” The source neighborhood may depend on the target neighborhood. Negating the definition gives a useful disproof test: there must be one open neighborhood $V$ of $f(x)$ such that every open neighborhood $U$ of $x$ contains some $z$ with $f(z)\notin V$.

Proposition: A map $f:X\to Y$ is continuous at every point of $X$ if and only if $f^{-1}(V)$ is open in $X$ for every open set $V$ in $Y$.

Proof. Suppose first that $f$ is continuous at every point, and let $V$ be open in $Y$. For each $x\in f^{-1}(V)$, the set $V$ is an open neighborhood of $f(x)$. Pointwise continuity supplies an open neighborhood $U_x$ of $x$ with $f(U_x)\subseteq V$, hence $U_x\subseteq f^{-1}(V)$. Thus every point of $f^{-1}(V)$ has an open neighborhood contained in it, so $f^{-1}(V)$ is open. This also covers $f^{-1}(V)=\varnothing$, which is open. Conversely, suppose preimages of open sets are open. Given $x\in X$ and an open neighborhood $V$ of $f(x)$, the set $U=f^{-1}(V)$ is open, contains $x$, and satisfies $f(U)\subseteq V$. So $f$ is continuous at $x$. If $X$ is empty, pointwise continuity holds vacuously and every preimage is empty and open, so the equivalence still holds. $\square$

Worked Example: A Continuity Witness and a Counter-Witness

First consider $f:\mathbb{R}\to\mathbb{R}$, $f(x)=x^2$, at $x=2$. Let $V=(4-\varepsilon,4+\varepsilon)$ be a target neighborhood of $f(2)=4$, where $\varepsilon>0$. Choose $\delta=\min(1,\varepsilon/5)>0$ and let $U=(2-\delta,2+\delta)$.

If $x\in U$, then $|x-2|<\delta\le1$, so $1<x<3$ and $3<x+2<5$. Therefore $$ |f(x)-f(2)|=|x^2-4|=|x-2||x+2|<5\delta\le\varepsilon. $$ Thus $f(U)\subseteq V$. The chosen $U$ is the required witness for this $V$, and the argument works for every $\varepsilon>0$.

Now define $g:\mathbb{R}\to\mathbb{R}$ by $g(x)=0$ when $x\le0$, and $g(x)=1$ when $x>0$. At $0$, take $V=(-1/2,1/2)$, an open neighborhood of $g(0)=0$. Let $U$ be any open neighborhood of $0$. There exists $\delta>0$ with $(-\delta,\delta)\subseteq U$. Set $t=\min(\delta/2,1/2)$. Then $0<t<\delta$, so $t\in U$, while $g(t)=1\notin V$. This single $V$ defeats every $U$, proving that $g$ is not continuous at $0$.

Check: in the first calculation, $\delta\le\varepsilon/5$ gives $5\delta\le\varepsilon$, and the preceding estimate is strict. In the second, $t$ is positive and lies in $U$, so the failing value really is $g(t)=1$.

Compactness: A Finite Subcover Must Exist

Definition (compactness, with quantifiers): A space $X$ is compact if, for every indexed family $(U_j)_{j\in J}$ of open subsets of $X$ that covers $X$, there is a finite set $J_0\subseteq J$ such that $(U_j)_{j\in J_0}$ still covers $X$. The cover condition is $\forall x\in X\,\exists j\in J\ (x\in U_j)$; the finite-subcover conclusion is $\exists J_0\subseteq J\text{ finite}\,\forall x\in X\,\exists j\in J_0\ (x\in U_j)$.

To disprove compactness, it is not enough to exhibit an open cover with many sets. The counter-witness must be a cover for which every finite selection fails to cover. The empty space is compact: any open cover, including the empty family, has the empty finite subcover.

Worked Example: An Open Cover of the Real Line With No Finite Subcover

Give $\mathbb{R}$ its standard topology and, for each $n\in\mathbb{N}=\{1,2,\dots\}$, let $U_n=(-n,n)$. Each $U_n$ is open. For any $x\in\mathbb{R}$, choose a positive integer $n>|x|$; then $-n<x<n$, so $x\in U_n$. Hence $(U_n)_{n\in\mathbb{N}}$ covers $\mathbb{R}$.

Take any finite selection of these sets. If the selection is nonempty, let $N$ be its largest index. For each selected $n$, $n\le N$, so $(-n,n)\subseteq(-N,N)$. Their union is therefore contained in $(-N,N)$ and misses $N+1$, since $N+1>N$. If the selection is empty, its union is empty and misses $0$. Thus no finite selection covers $\mathbb{R}$, so $\mathbb{R}$ is not compact.

Check: the argument uses the same finite subfamily throughout: its largest index $N$ bounds every selected interval, and the point $N+1$ lies outside all of them. The fact that each individual real number belongs to some $U_n$ proves the cover condition, not the finite-subcover condition.

Hausdorffness: Separate Every Distinct Pair

Definition (Hausdorff property): A space $X$ is Hausdorff if for every pair $x,y\in X$ with $x\ne y$, there are open neighborhoods $U$ of $x$ and $V$ of $y$ with $U\cap V=\varnothing$. Its negation says that there exist distinct $x,y\in X$ such that every open neighborhood $U$ of $x$ intersects every open neighborhood $V$ of $y$.

For a positive Hausdorff proof, choose arbitrary distinct points and construct two disjoint neighborhoods. For a negative proof, one pair is enough, but the claim about that pair is universal: every choice of neighborhoods must intersect. A space with fewer than two points is Hausdorff, since there are no distinct pairs to test.

Worked Example: Separating Points in the Real Line

Let $x,y\in\mathbb{R}$ be distinct. If $x<y$, put $d=y-x>0$ and choose $U=(x-d/3,x+d/3)$, $V=(y-d/3,y+d/3)$. Both are open neighborhoods of their indicated points. The right endpoint of $U$ is $x+d/3$, and the left endpoint of $V$ is $y-d/3=x+2d/3$. Since $x+d/3<x+2d/3$, the intervals are disjoint. If $y<x$, interchange the roles of $x$ and $y$ in this construction. Thus $\mathbb{R}$ is Hausdorff.

For contrast, let $X=\{a,b\}$ have the indiscrete topology $\{\varnothing,X\}$. The only open neighborhood of $a$ is $X$, and the only open neighborhood of $b$ is also $X$. Their intersection is $X\ne\varnothing$. Hence this particular distinct pair has no disjoint neighborhoods, and $X$ is not Hausdorff.

Check: in the real-line construction, the distance between the two interval centers is $d$, while their radii add to $2d/3<d$. In the indiscrete example there is no alternative neighborhood to choose, so the failure checks every possible choice.

A Quantifier Audit Before You Write

Before proving or disproving a topological claim, identify what is arbitrary and what must be produced. In these four definitions, the key patterns are:

Closure: start with an arbitrary neighborhood; find a point of the set inside it. To disprove, exhibit one neighborhood that misses the set.
Continuity at a point: start with an arbitrary target neighborhood; find a source neighborhood mapping into it. To disprove, exhibit one target neighborhood that every source neighborhood fails to map into.
Compactness: start with an arbitrary open cover; produce a finite subcover. To disprove, produce one open cover and show every finite selection fails.
Hausdorffness: start with an arbitrary distinct pair; produce disjoint neighborhoods. To disprove, produce a distinct pair for which every pair of neighborhoods intersects.

A frequent mistake is to reverse the order of quantifiers. For instance, closure does not ask for one neighborhood that contains a point of $A$; that would only establish one local fact. Nor does compactness ask for each point to lie in some finite subcollection chosen separately for that point. A single finite subcollection must cover all of $X$.

Common pitfall. From the fact that each $x\in\mathbb{R}$ lies in some interval $U_n=(-n,n)$, one might incorrectly conclude that a finite number of these intervals cover $\mathbb{R}$. The calculation shows why this is wrong: any finite selection has a largest index $N$, all its intervals lie inside $(-N,N)$, and $N+1$ is not covered. Pointwise existence of an index $n$ does not provide one finite set of indices that works for every point.

Check Your Understanding

For each question, write the quantifiers in order and identify the witness or counter-witness the claim requires.

State the neighborhood test for $x\in\overline{A}$, then write its logical negation.
What must be chosen first to disprove continuity of $f$ at $x$: a source neighborhood or a target neighborhood? State the full negated condition.
For the cover $U_n=(-n,n)$ of $\mathbb{R}$, what point is missed by a nonempty finite selection whose largest index is $N$?
Write the negation of the Hausdorff condition using distinct points and neighborhoods.
Why does the indiscrete topology on a two-point set fail to be Hausdorff, and what must be checked to verify the failure?