Topological Spaces — Mathematics Knowledge System

Overview

A topological space is, roughly speaking, a space in which 'closeness' is defined but cannot necessarily be measured by a numeric distance. Formally, it is a set X together with a collection τ of subsets of X, called the open sets, satisfying three axioms: the empty set and X itself are open, arbitrary unions of open sets are open, and finite intersections of open sets are open. This single structure — a topology — is enough to define continuity, convergence, connectedness, and compactness without ever mentioning a distance function, making topology the natural generalization of the geometry of the real line and Euclidean space to arbitrarily abstract sets.

Intuition

In ordinary geometry, 'close' means 'small distance apart' — you need a ruler. Topology asks: what if you throw away the ruler, and instead just declare, once and for all, which subsets of your space count as 'open' (roughly, subsets that contain a little wiggle-room around every one of their points)? It turns out that just knowing the open sets is enough to talk about limits, continuity, and connectedness, without ever measuring a distance. Two very different rulers (metrics) on the same set can generate the exact same collection of open sets — the same topology — meaning they encode exactly the same notion of 'nearness,' even though the numeric distances they assign are different.

Formal Definition

Definition

A topology on a set X is a collection τ of subsets of X (called the open sets of X) satisfying three axioms:

\emptyset \in \tau \text{ and } X \in \tau

The empty set and the whole space are both open

Axiom 1

U_\alpha \in \tau \text{ for } \alpha \in A \implies \bigcup_{\alpha \in A} U_\alpha \in \tau

Any union of open sets (even an infinite or uncountable family) is open

Axiom 2

U_1, \ldots, U_n \in \tau \implies \bigcap_{i=1}^{n} U_i \in \tau

Only FINITE intersections of open sets are guaranteed to be open — infinite intersections need not be

Axiom 3

(X, \tau)

The pair of the underlying set and its topology is called a topological space

Notation

Notation	Meaning
$(X, \tau)$	A topological space: a set X together with a topology τ (a designated collection of open subsets)
$\tau$	The topology — the collection of subsets of X declared to be open
$\wp(X)$	The power set of X — the discrete topology sets τ = ℘(X), making every subset open
$U \in \tau$	U is an open set of the topological space
$\tau = \{\emptyset, X\}$	The indiscrete (trivial) topology — the coarsest possible topology on any set

Derivation

Verifying that the standard topology on ℝ (generated by open intervals) satisfies the three axioms, connecting the abstract definition back to familiar calculus.

\tau_{\text{std}} = \{ U \subseteq \mathbb{R} : \forall x \in U,\ \exists \varepsilon>0 \text{ with } (x-\varepsilon, x+\varepsilon) \subseteq U \}

Defining U as open exactly when every point has a surrounding open interval fully contained in U — this recovers the familiar ε-neighborhoods from the ε-δ definition

\emptyset \in \tau_{\text{std}} \text{ (vacuously)},\quad \mathbb{R} \in \tau_{\text{std}} \text{ (any } \varepsilon \text{ works)}

Axiom 1 holds trivially

x \in \bigcup_\alpha U_\alpha \implies x \in U_{\alpha_0} \text{ for some } \alpha_0 \implies \exists \varepsilon: (x-\varepsilon,x+\varepsilon)\subseteq U_{\alpha_0} \subseteq \bigcup_\alpha U_\alpha

Axiom 2: the witnessing interval for membership in one set of the union also witnesses membership in the whole union

x \in U_1 \cap U_2 \implies \exists \varepsilon_1, \varepsilon_2: (x-\varepsilon_i,x+\varepsilon_i)\subseteq U_i \implies (x-\varepsilon,x+\varepsilon) \subseteq U_1\cap U_2 \text{ for } \varepsilon = \min(\varepsilon_1,\varepsilon_2)

Axiom 3: taking the minimum of finitely many radii gives a single interval inside the intersection — this step is exactly why the axiom is restricted to FINITE intersections (an infinite min of shrinking radii could be 0)

Proofs

An infinite intersection of open sets need not be open (in the standard topology on ℝ)

$\text{Let } U_n = \left(-\frac{1}{n}, \frac{1}{n}\right) \text{ for } n=1,2,3,\ldots$ (Each Uₙ is an open interval, hence open in the standard topology)
$\bigcap_{n=1}^{\infty} U_n = \{0\}$ (Any nonzero x fails |x|<1/n once n>1/|x|, so only x=0 survives every intersection)
$\{0\} \text{ is not open in } \tau_{\text{std}}, \text{ since no interval } (-\varepsilon,\varepsilon) \text{ is contained in } \{0\}.$ (Any positive-radius interval around 0 contains points other than 0)
$\therefore \text{an infinite intersection of open sets can fail to be open.}$ (This is exactly why Axiom 3 of the topology axioms is restricted to finite intersections)

Properties

Discrete topology

\tau = \wp(X)

Condition: The finest possible topology — every subset is open; every function out of a discretely-topologized space is continuous.

Indiscrete (trivial) topology

\tau = \{\emptyset, X\}

Condition: The coarsest possible topology — only ∅ and X are open.

Comparing topologies

\tau_1 \text{ is finer than } \tau_2 \iff \tau_2 \subseteq \tau_1

Example: The discrete topology is finer than every other topology on X; the indiscrete topology is coarser than every other.

Basis generation

\text{A collection } \mathcal{B} \subseteq \wp(X) \text{ generates a topology } \tau_{\mathcal{B}} \text{ as all unions of members of } \mathcal{B}, \text{ provided } \mathcal{B} \text{ covers } X \text{ and is closed under finite intersection within itself.}

Example: Open intervals form a basis generating the standard topology on ℝ.

Applications

Domain theory in programming-language semantics models computation and approximation using topological spaces (Scott topology), where 'open sets' correspond to observable/verifiable properties of program states.

Worked Examples

Check Axiom 1: ∅ and X are both present.
$\emptyset, X \in \tau \checkmark$
Check Axiom 2 (unions): {1}∪{2,3} = X ∈ τ; all other unions reduce to sets already listed.
$\{1\} \cup \{2,3\} = X \in \tau \checkmark$
Check Axiom 3 (finite intersections): {1}∩{2,3} = ∅ ∈ τ.
$\{1\} \cap \{2,3\} = \emptyset \in \tau \checkmark$

Answer: Yes, τ is a valid topology on X

Practice Problems

Difficulty 5/10

Which of the following is NOT one of the three topology axioms?

∅ and X are openArbitrary unions of open sets are openArbitrary intersections of open sets are openFinite intersections of open sets are open

Difficulty 4/10

How many topologies exist on a 1-element set X = {a}?

Difficulty 7/10

Prove that in the discrete topology (τ = ℘(X)), every singleton {x} is open.

Common Mistakes

Common Mistake

Assuming arbitrary (infinite) intersections of open sets are always open, by analogy with arbitrary unions.

Only FINITE intersections are guaranteed open. The infinite intersection ∩ₙ(-1/n,1/n) = {0} in the standard topology on ℝ is not open, showing the axiom genuinely requires finiteness.

Common Mistake

Thinking a topology must come from a metric (a distance function) — that 'open set' always means 'ball around a point.'

Topologies are more general than metrics. The cofinite topology, the Zariski topology in algebraic geometry, and many others have no underlying metric at all, yet still satisfy the three open-set axioms.

Common Mistake

Confusing 'open' with 'not closed' — assuming every set is either open or closed but not both, or neither.

A set can be open, closed, both (called clopen — e.g. ∅ and X always are), or neither (e.g. the half-open interval [0,1) in the standard topology on ℝ is neither open nor closed).

Quiz

Which collection of subsets is guaranteed to be open under the topology axioms, for ANY topology (X, τ)?

Every singleton {x} for x ∈ X∅ and X themselvesEvery finite subset of XThe complement of every open set

Why does the third topology axiom restrict to FINITE intersections rather than arbitrary intersections?

Because infinite intersections are undefined in set theoryBecause allowing infinite intersections would make every topology equal the discrete topologyBecause an infinite intersection of open sets can produce a non-open set, as with ∩(-1/n,1/n) = {0} in ℝThere is no real reason — it is just a historical convention

Flashcards

1 / 4

Historical Background

Topology grew out of 19th-century analysis, where mathematicians like Cantor and Weierstrass needed increasingly abstract notions of 'nearness' to handle pathological functions and infinite sets. Henri Poincaré's Analysis Situs (1895) is often cited as founding the subject, initially focused on what is now algebraic topology. The abstract, axiomatic notion of a topological space — defined purely via open sets, detached from any metric — was crystallized by Felix Hausdorff in his 1914 book Grundzüge der Mengenlehre, and refined into the modern open-set axioms by Kazimierz Kuratowski (via closure operators) and others through the 1920s–1930s. This abstraction let mathematicians study 'shape' and 'continuity' in settings — function spaces, spaces of measures, even finite sets — with no natural distance at all.

1895
Poincaré publishes Analysis Situs, founding systematic topology (with an early combinatorial/algebraic flavor)
Henri Poincaré
1906
Fréchet introduces metric spaces, generalizing distance beyond Euclidean space
Maurice Fréchet
1914
Hausdorff's Grundzüge der Mengenlehre gives the first widely adopted axiomatic definition of a topological space via neighborhoods
Felix Hausdorff
1922
Kuratowski formalizes the closure-operator axioms, an equivalent route to defining a topology
Kazimierz Kuratowski

Summary

A topological space (X, τ) is a set X with a topology τ — a collection of designated open subsets — satisfying: ∅,X ∈ τ; arbitrary unions of open sets are open; finite intersections of open sets are open.
Topology generalizes the geometry of ℝⁿ without requiring a distance function — the same underlying set can carry many different topologies, from the discrete (finest, every set open) to the indiscrete (coarsest, only ∅ and X open).
The standard topology on ℝ (open sets = unions of open intervals) recovers exactly the ε-neighborhoods used in the ε-δ definition of limits and continuity.
Only FINITE intersections of open sets are guaranteed open; infinite intersections can fail, e.g. ∩ₙ(-1/n,1/n) = {0}.
A basis (like the open intervals) can generate an entire topology by taking all possible unions of basis elements.

References

WebsiteWikipedia — Topological space
BookMunkres, J. Topology, 2nd ed. Ch. 2.
BookHausdorff, F. (1914). Grundzüge der Mengenlehre.