Tutorials › Topology (Undergraduate) › Common Errors in Basis and Subbasis Arguments

Bases and Subbases · Tutorial 147 of 799

Common Errors in Basis and Subbasis Arguments

Use the basis and subbasis criteria to spot common logical errors and replace them with correct arguments.

Intermediate 10 min read

What You'll Learn

  • Check the covering condition before treating a collection as a basis.
  • Test intersections point by point rather than relying on nonemptiness alone.
  • Distinguish individual basis elements from the open sets formed by their unions.
  • Build a subbasis-generated topology using all finite intersections, including the empty intersection.
  • Verify basis and subbasis claims with explicit examples.

Why These Errors Matter

A basis is a concise way to describe a topology, but a short description is useful only if its conditions are checked correctly. Two separate requirements govern a basis: its members must cover the space, and intersections must have enough basis members to refine them around each point. A subbasis works differently: its finite intersections form the basis, and arbitrary unions of those intersections form the topology. Confusing these steps can produce a collection that does not describe the topology claimed.

Throughout, \((X,\mathcal{T})\) denotes a topological space. A basis for \(\mathcal{T}\) is a collection \(\mathcal{B}\) of open sets such that every open set is a union of members of \(\mathcal{B}\). A subbasis is a collection \(\mathcal{S}\) of open sets whose finite intersections form a basis. The empty union is \(\varnothing\), and the intersection of no sets is \(X\).

Error 1: Forgetting That Basis Members Must Cover

It is not enough for a collection to have a well-behaved intersection pattern. Every point of \(X\) must belong to at least one basis member. In the usual basis criterion, this covering requirement is written \(\bigcup_{B\in\mathcal{B}}B=X\).

Worked Example: A Collection That Misses a Point

Take \(X=\{a,b\}\) and \(\mathcal{B}=\{\{a\}\}\). The only basis member is \(\{a\}\), so the union of all members is \(\{a\}\), not \(X\). Thus \(\mathcal{B}\) does not cover \(X\) and cannot be a basis for a topology on \(X\).

The intersection-refinement condition alone does not catch the problem: the only pair to test is \(\{a\}\) with itself. Their intersection is \(\{a\}\), and at its point \(a\), the member \(\{a\}\) is a suitable refinement. But the collection of all unions of members of \(\mathcal{B}\) is just \(\{\varnothing,\{a\}\}\). It lacks \(X\), so it is not a topology on \(X\).

Check: The missing point is \(b\), and no member of \(\mathcal{B}\) contains \(b\). This directly contradicts the covering requirement.

For \(X=\varnothing\), the empty collection does cover \(X\), because its union is \(\varnothing\); it can be a basis for the unique topology on the empty space. Thus the covering condition also handles the empty-space case without requiring a basis member to exist.

Error 2: Checking Intersections Without Refining at Each Point

The second condition is more specific than saying that two basis members have a nonempty intersection. Whenever \(x\in B_1\cap B_2\), there must be some \(B_3\in\mathcal{B}\) with \(x\in B_3\subseteq B_1\cap B_2\). The refinement may depend on the point \(x\). Merely observing that \(B_1\cap B_2\ne\varnothing\) does not supply such a member.

Worked Example: A Nonempty Intersection Is Not Enough

On \(X=\mathbb{R}\) with its standard topology, let \(\mathcal{B}=\{\mathbb{R},(-1,1),(0,2)\}\). These sets cover \(\mathbb{R}\), because \(\mathbb{R}\in\mathcal{B}\). The two bounded members intersect in \((0,1)\), which is nonempty.

A mistaken argument might say that the intersection condition is satisfied because \((0,1)\ne\varnothing\). Test instead the point \(x=\tfrac12\). The members of \(\mathcal{B}\) containing \(x\) are \(\mathbb{R}\), \((-1,1)\), and \((0,2)\). None is contained in \((0,1)\): the first contains points outside \((0,1)\), the second contains negative points, and the third contains points greater than \(1\). There is no required refinement at \(x\), so \(\mathcal{B}\) is not a basis.

Indeed, unions of members of \(\mathcal{B}\) cannot produce \((0,1)\): any union containing \(\mathbb{R}\) is \(\mathbb{R}\), and unions using only the two bounded members are \(\varnothing\), one of those members, or their union \((-1,2)\). The intersection of the open sets \((-1,1)\) and \((0,2)\) is \((0,1)\), so this collection of unions is not a topology.

Check: The failure is pointwise: at \(x=\tfrac12\), no member of \(\mathcal{B}\) contains \(x\) and lies inside the intersection.

Basis criterion: A collection \(\mathcal{B}\subseteq\mathcal{P}(X)\) is a basis for a topology on \(X\) if and only if (1) its members cover \(X\), and (2) for every \(B_1,B_2\in\mathcal{B}\) and every \(x\in B_1\cap B_2\), there is \(B_3\in\mathcal{B}\) such that \(x\in B_3\subseteq B_1\cap B_2\). The topology it generates consists of all unions of members of \(\mathcal{B}\).

Proof. First suppose the two conditions hold, and let \(\mathcal{T}_{\mathcal{B}}\) be the collection of all unions of members of \(\mathcal{B}\). It contains \(\varnothing\) as the empty union and \(X\) by the covering condition. An arbitrary union of members of \(\mathcal{T}_{\mathcal{B}}\) is still a union of members of \(\mathcal{B}\). For finite intersections, write two such unions as \(U=\bigcup_{i\in I}B_i\) and \(V=\bigcup_{j\in J}C_j\). A point \(x\in U\cap V\) lies in some \(B_i\cap C_j\). The refinement condition gives a basis member containing \(x\) and contained in \(B_i\cap C_j\), hence contained in \(U\cap V\). Taking all such refinements shows that \(U\cap V\) is a union of basis members. Thus \(\mathcal{T}_{\mathcal{B}}\) is a topology containing \(\mathcal{B}\). Any topology containing \(\mathcal{B}\) contains every union of its members, so \(\mathcal{T}_{\mathcal{B}}\) is the smallest such topology. Conversely, if \(\mathcal{B}\) is a basis, \(X\) is a union of basis members. If \(x\in B_1\cap B_2\), that intersection is open, so its expression as a union of basis members includes one containing \(x\); that member is the required refinement. \(\square\)

Error 3: Treating Every Open Set as One Basis Member

A basis member is open, but an open set need not itself be a basis member. The definition permits an open set to be a union of many basis members. The basis criterion’s refinement condition is local: it guarantees a member around each point of an intersection, not that the entire intersection is one member.

Worked Example: Rational Intervals on the Real Line

Let \(X=\mathbb{R}\) with its standard topology and let \(\mathcal{B}=\{(p,q):p,q\in\mathbb{Q},\ p<q\}\). Consider the open set \(U=(0,\sqrt{2})\). It is not a member of \(\mathcal{B}\), because its right endpoint \(\sqrt{2}\) is irrational.

Nevertheless, \(U\) is a union of members of \(\mathcal{B}\): take all rational-endpoint intervals \((p,q)\) with \(0<p<q<\sqrt{2}\). Each such interval lies in \(U\). Conversely, for any \(x\in U\), density of \(\mathbb{Q}\) in \(\mathbb{R}\) gives rationals \(p,q\) with \(0<p<x<q<\sqrt{2}\). Then \(x\in(p,q)\subseteq U\). Therefore the union of those basis members is exactly \(U\).

Check: Every point of \(U\) lies in one of the stated intervals, and no such interval contains a point outside \(U\). Thus \(U\) is open even though it is not a single basis member.

The same distinction matters when using the basis criterion: for an intersection \(B_1\cap B_2\), it is enough to find a suitable basis member around each point in the intersection. Demanding that the whole intersection itself belong to \(\mathcal{B}\) is unnecessarily strong and is not part of the criterion.

Error 4: Omitting Finite Intersections of Subbasic Sets

A subbasis is not generally a basis as it stands. To use a subbasis \(\mathcal{S}\), first form all finite intersections of its members. This collection includes \(X\), the intersection of no subbasic sets. The topology generated by \(\mathcal{S}\) consists of unions of these finite intersections.

Theorem (Subbasis construction): For a collection \(\mathcal{S}\subseteq\mathcal{P}(X)\), let \(\mathcal{B}_{\mathcal{S}}\) consist of all finite intersections of members of \(\mathcal{S}\), including \(X\). Then \(\mathcal{B}_{\mathcal{S}}\) is a basis, and its unions form the smallest topology containing \(\mathcal{S}\).

Proof. The collection \(\mathcal{B}_{\mathcal{S}}\) covers \(X\), since \(X\) itself is one of its members. The intersection of two finite intersections is again a finite intersection of members of \(\mathcal{S}\), so it belongs to \(\mathcal{B}_{\mathcal{S}}\). In particular, for every point of that intersection, this same member is a refinement. The basis criterion therefore shows that unions of members of \(\mathcal{B}_{\mathcal{S}}\) form a topology. This topology contains every \(S\in\mathcal{S}\), since \(S\) is a finite intersection with one factor. Any topology containing \(\mathcal{S}\) contains its finite intersections and their unions, so the constructed topology is the smallest one containing \(\mathcal{S}\). \(\square\)

Worked Example: Two Open Rays Produce an Open Interval

In \(\mathbb{R}\) with its standard topology, take the subbasis \(\mathcal{S}=\{(-\infty,1),(0,\infty)\}\). Its finite intersections include \(X=\mathbb{R}\), each of the two rays, and their intersection \((-\infty,1)\cap(0,\infty)=(0,1)\). The empty set is also a union, namely the empty union.

Consequently \((0,1)\) belongs to the topology generated by \(\mathcal{S}\), even though it is not itself one of the two subbasic sets. A mistaken calculation that lists only the members of \(\mathcal{S}\) misses this open set. The correction is to form finite intersections first and then take arbitrary unions.

Check: A real number lies in the intersection exactly when it is greater than \(0\) and less than \(1\), so the intersection is precisely \((0,1)\), with both endpoints excluded.

A Reliable Checklist

When checking a basis or subbasis argument, write down the construction before drawing a conclusion. For a proposed basis, verify coverage and test intersections at an arbitrary point \(x\) in the intersection. For a proposed subbasis, list the finite intersections before taking unions. These checks prevent a nonempty intersection from being mistaken for a refinement, and prevent a generating family from being mistaken for the full collection of open sets.

Common pitfall. For the collection \(\{\mathbb{R},(-1,1),(0,2)\}\), the claim “the two bounded members intersect, so the collection is a basis” is wrong: at \(x=\tfrac12\), no member lies inside their intersection \((0,1)\) while containing \(x\). The corrected test checks the required containment point by point. In a subbasis argument, the analogous correction is to include intersections such as \((-\infty,1)\cap(0,\infty)\), not just the original rays.

Check Your Understanding

For each question, identify which condition or construction is relevant before deciding.

  1. On \(X=\{a,b\}\), why does \(\{\{a\}\}\) fail to be a basis, even though its only member intersects itself?
  2. For \(\mathcal{B}=\{\mathbb{R},(-1,1),(0,2)\}\), identify a point where the refinement condition fails and state the intersection that must be refined.
  3. With rational open intervals as a basis for \(\mathbb{R}\), can an open interval with an irrational endpoint be open without being a basis member? Explain using unions.
  4. For the subbasis \(\{(-\infty,1),(0,\infty)\}\), what finite intersection gives \((0,1)\)?
  5. In the subbasis construction, why must the intersection of no subbasic sets be included?