Topological Closure¶

Enlarge a subset to the least closed set containing it in a specified topology, equivalently including every point whose neighborhoods meet the subset.

Version: v1 · 2026-10-07 · History
Domain-specific #: 14035
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomain: General Topology → Mathematics
Aliases: Closure Topology

Core Idea¶

For a subset \(A\) of a specified topological space \(X\), its topological closure \(\operatorname{cl}_X(A)\) is the smallest closed subset of \(X\) containing \(A\). Equivalently, a point \(x\) belongs to the closure exactly when every open neighborhood of \(x\) meets \(A\). Both descriptions use the topology on \(X\); changing that topology can change the answer even when the underlying points and \(A\) stay fixed.^[1]

Closure turns an open-ended approach relation into a definite subset. It includes points already in \(A\) and points that cannot be separated from \(A\) by an open neighborhood. No distance function is required: a metric supplies one possible topology, while the Zariski topology supplies another with different closed sets and therefore different closures.^[1]^[2]

Structural Signature¶

Signature: topological space \((X,\tau)\) + subset \(A\subseteq X\) → intersection of all \(\tau\)-closed supersets of \(A\) → the points every \(\tau\)-open neighborhood of which intersects \(A\).^[1]

Ambient topology. The open-set system fixes which subsets count as closed and which sets are neighborhoods. It is constitutive, not a decorative choice of notation.^[1]
Input subset. Closure acts on a selected \(A\). The same space can give different answers for different subsets.^[1]
Least closed extension. Intersecting all closed supersets produces the unique smallest one. A larger closed set containing \(A\) is not automatically its closure.^[1]
Neighborhood characterization. Every open neighborhood of a closure point meets \(A\). This is an equivalent test, not a second independent operation or a demand for numerical distance.^[1]
Operator laws. On all subsets of one \(X\), the operation preserves the empty set, contains its input, is idempotent, and preserves finite unions. Conversely, a set operator with these Kuratowski laws determines a topology whose closure it is. These laws characterize the topological operator; they are not separately chosen actions by a user.^[1]

What It Is Not¶

Topological closure is not the statement that a carrier remains inside itself after a designated algebraic operation. That is the central sense of the live Prime Closure. The Prime also discusses closure operators, including topological ones; this entry isolates the specific least-closed-superset and neighborhood relation induced by a topology, rather than denying that overlap.^[1]

It is not every operation called a “hull.” For two distinct points in the Euclidean plane, their topological closure is still those two points, while their convex hull also contains the segment between them. Nor may one replace the neighborhood test by a sequence-only or distance-zero rule in an arbitrary topological space; those descriptions need additional conditions on the space.^[1]

Scope of Application¶

The construction applies to any topological space and any subset of it. In the Euclidean real line, the rationals are dense, so their closure is all real numbers. In the Zariski topology on \(\operatorname{Spec}(k[x])\), where \(k\) is a field, the generic point \((0)\) also has the whole space as its closure, because \(\overline{\{(0)\}}=V(0)=\operatorname{Spec}(k[x])\). These are unlike subsets and topologies, yet each uses the same least-closed-superset rule.^[3]^[4]

The Zariski example is about a scheme point, not an ordinary closed coordinate point. In affine-space presentations, closed sets are polynomial zero loci; an arbitrary such closed set need not be a single irreducible variety. The result always depends on the selected topology and carrier.^[5]^[4]

Clarity¶

The phrase “approaches \(A\)” is potentially misleading. In a general topological space, closure membership says that every open neighborhood meets \(A\); it does not state how quickly a sequence converges or measure a distance. The two equivalent tests—least closed superset and neighborhood intersection—let a reader prove the same membership fact from whichever side is easier.^[1]

It also makes density precise: \(A\) is dense in \(X\) exactly when \(\operatorname{cl}_X(A)=X\). Thus \(\mathbb Q\) is dense in the usual real topology. For the generic point in \(\operatorname{Spec}(k[x])\), the closure fills the spectrum for a different reason: the Zariski closed sets containing \((0)\) include only the whole spectrum. The shared rule does not erase the difference between the two topologies.^[3]^[4]

Manages Complexity¶

The closed-superset formulation compresses a potentially large family of closed sets into one least result. The neighborhood formulation converts a global closed-set calculation into a local exclusion test: one open neighborhood disjoint from \(A\) is enough to show a point lies outside the closure.^[1]

The Kuratowski laws let the topology be recovered from the closure operator. A set is closed precisely when applying closure changes nothing, so one can reason with the operator rather than enumerate every open set afresh. This is a formal equivalence, not a claim that all hull operators are topological closures.^[1]

Abstract Reasoning¶

The recurring move is fix the topology, then close the subset. First type the carrier and its open sets; next state the input subset; then use either closed supersets or neighborhoods to determine which points are forced in. If the ambient topology changes, rerun the test. A conclusion proved in a metric topology does not automatically hold for the Zariski topology on the same coordinate set.^[1]^[2]

The counterfactual identifies the abstraction's boundary. If an operator adds points for a rule unrelated to the selected topology—such as convex combinations—it may be an extensive, idempotent hull, but it is not this topological closure unless it satisfies the topological operator laws for an appropriate topology. If no topology is specified, “the closure of \(A\)” is underdetermined.^[1]

Knowledge Transfer¶

The definition travels literally among analysis, general topology, and algebraic geometry because those fields can each specify a topological space and a subset. What transfers is the method: name the topology, compute closed supersets or test neighborhoods, and state whether the result equals the carrier. Particular distance, sequence, or polynomial calculations are domain-specific ways of carrying out that method.^[1]^[4]

A useful transfer check is to ask what the closed sets are in the destination setting. Over \(\mathbb R\) with its usual topology, density of \(\mathbb Q\) follows from rational points in every open interval. Over \(\operatorname{Spec}(k[x])\), the closure of the generic point is computed with \(V(I)\) sets. The answer “the whole space” happens in both cases, but for different local structures.^[3]^[4]

Examples¶

Rationals in the real line. Set \(X=\mathbb R\) with its usual topology and \(A=\mathbb Q\). For every real \(x\) and every open interval around it, some rational lies inside. Hence the neighborhood test includes every \(x\), giving \(\operatorname{cl}_{\mathbb R}(\mathbb Q)=\mathbb R\). The source proves rational density; the closure conclusion follows from the general topological criterion.^[3]^[1]

Generic point of an affine scheme. Let \(X=\operatorname{Spec}(k[x])\) with its Zariski topology and \(A=\{(0)\}\). Since \(k[x]\) is a domain, \((0)\) is a prime ideal and therefore a point of \(X\). The Stacks Project gives the closure of a singleton prime \(\mathfrak p\) as \(V(\mathfrak p)\), so this closure is \(V(0)=X\). The instance uses a singleton that is not a closed point, unlike the many-point Euclidean dense subset above.^[4]

Structural Tensions¶

No intrinsic two-objective tension is required to define topological closure. Coarsening or refining a topology can alter the closed supersets and closure of a fixed subset, but that comparison alone does not assert a universal optimization pressure. The diagnostic question is mathematical: which topology is fixed, and which points pass its neighborhood test?^[1]^[2]

Structural–Framed Character¶

Topological closure is structural. Its defining test is a relation among a set, its open subsets, an input subset, and a least closed extension. No human role, institution, or judgment determines membership once those data are fixed. Its evaluative weight is low: “closed” is a formal status, not praise. People choose the topology for a problem, but the resulting closure is determined by that choice. The relation is recognized in any specified topology rather than imported by analogy. The vocabulary transfers literally between Euclidean and Zariski settings; applying “closure” metaphorically to finishing a meeting would not instantiate this operator. Its character: a topology-specific mathematical structure, distinct from algebraic containment under an operation.^[1]^[4]

Structural Core vs. Domain Accent¶

The structural core is the least closed extension of a subset, equivalently the neighborhood-intersection set. The domain accent is a topological space with its chosen open/closed system. Euclidean distance or polynomial zero loci can implement that system but are not part of the all-instance definition. Remove the topology and the claimed operator is undefined; substitute an unrelated hull and the topological identity is lost.^[1]^[4]

This is a domain-specific formal construct even though topology is abstract mathematics. The live Prime Closure has a broader operation-containment core and explicitly mentions topological closure operators as a related extension. The portable extensive and idempotent closure-operator skeleton is already described in live Prime Closure Structural Signature item 4; a new standalone Prime for that generic operator would need its own admission review. The present entry earns a separate identity only through its narrower all-case Kuratowski and neighborhood conditions, which also keep it from becoming a second generic Closure page.^[1]

This entry presupposes Topological Space.

Typed prerequisite: Topological Space. Topological closure presupposes a chosen topological space; without that space's open and closed sets, neither the least-closed-superset nor the neighborhood test is determined. A topological space can exist before a particular subset is closed, so this is a strict composition/presupposes edge, not subsumption.^[1]

Prime Closure is a related broader vocabulary node, but its core is operation containment and its description also notes topological operators. That overlap does not itself prove a strict all-instance parent edge for this topology-specific operator. Prime Topology supplies broad conceptual context; the more concrete Topological Space is the necessary typed prerequisite recorded in the DAG.^[1]

Relationships to Other Abstractions¶

Current abstraction Topological Closure Domain-specific

Parents (1) — more general patterns this builds on

Topological Closure presupposes Topological Space Domain-specific

Topological closure requires the open and closed subsets of a specified topological space.

Hierarchy paths (5) — routes to 3 parentless roots

Topological Closure → Topological Space → Closure

Show alternative paths (4)

Neighborhood in Abstraction Space¶

Topological Closure sits in a sparse region of the domain-specific corpus (68^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Point-Set Topology Foundations (17 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With¶

Closure under an algebraic operation asks whether the operation sends carrier elements back into the carrier. Convex hull adds combinations even when the input set is already topologically closed. Boundary is a related subset, but closure is not simply “the boundary”: it also contains \(A\). Sequential closure can fail to capture all closure points in spaces where sequences do not characterize the topology. Zariski closure is an instance of topological closure using polynomially defined closed sets, not a separate universal rule.^[1]^[4]

References¶

[1] Ronald Freiwald, Topological Spaces, Chapter 3, university-authored topology text, Definitions 2.7 and Theorem 2.8, PDF pp.3–4; Theorems 5.18–5.19, PDF pp.18–19; Chapter III §10 Example 10.7, PDF pp.43–44, on sequence closure outside first-countable settings. Full text consulted for closure definitions, neighborhood test, and operator laws. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t ↩u ↩v ↩w ↩x

[2] MIT OpenCourseWare, Algebraic Geometry I Lecture Notes, Lecture 3, PDF p.7; distinguishes the Zariski and classical topologies. registry ↩a ↩b ↩c

[3] MIT Mathematics, 18.095 IAP 2015 Lecture 1, Lemma 1.9; university lecture notes proving \(\mathbb Q\) dense in \(\mathbb R\). registry ↩a ↩b ↩c ↩d

[4] The Stacks Project, The Stacks Project, Definitions 10.17.1 and 10.17.3 and Lemma 10.17.2 on \(\operatorname{Spec}(R)\) and Zariski closed sets \(V(I)\); Lemma 10.26.1(1) on the closure of a singleton prime \(\mathfrak p\) as \(V(\mathfrak p)\). One living work, with two section locators. registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i

[5] Anand Deopurkar, Algebraic Geometry Notes, §§1.5.3 and 1.6.1–1.6.7; instructor-authored exposition of Zariski versus Euclidean closed sets, vanishing ideals, and algebraic closed sets. registry ↩