Skip to content

Cover

Definition

Consider a (topological) space X.X\text{.} Let EE be a subset of X,X\text{,} and let C={Gi ⁣:i∈I}\mathcal{C}=\{G_i\colon i\in I\} be a collection of subsets GiG_i of XX indexed by an arbitrary index set I.I\text{.} The collection C\mathcal{C} is called a cover of EE if E⊆⋃i∈IGi.\displaystyle E \subseteq \bigcup_{i\in I} G_i\text{.}

  • An open cover is a cover where each "patch" GiG_i is open.
  • A finite cover consists of a collection {Gi ⁣:i∈I}\{G_i\colon i\in I\} where II is finite. A cover is called a countable cover when II is a countable set, e.g., N.\mathbb{N}\text{.}
  • A subcover is a cover that is a subset of another cover.

The Union of Sets collects all points lying in at least one patch GiG_i; the inclusion requires every point of EE to occur in that union.


Interpretation

A cover is just a collection of sets that together contain the set you care about. Think of it as laying patches over the set so that every point is under at least one patch; the patches may overlap and may stick out beyond the set.

The interval E from 0 to 1 drawn as a segment with five overlapping open intervals G1 to G5 above it that together cover the whole segment.

Five open intervals G1,…,G5G_1,\dots,G_5 lie over E=[0,1]E=[0,1], and every point of the segment is under at least one of them. The outer two reach beyond 00 and 11, as they must: an open interval containing 00 also contains points to the left of 00. Since each GiG_i is open and there are finitely many, this is a finite open cover.

The index set II only labels the patches and may be infinite. A subcover keeps some of the patches, but it must still cover EE: patches can be dropped only as long as no point of EE is left uncovered.

Covers matter because of compactness: a set is compact if every open cover has a finite subcover. The first Example shows how this fails for (0,1)(0,1). The intervals (1n,1−1n)(\tfrac{1}{n},1-\tfrac{1}{n}) increase with nn, so finitely many of them cover only the largest one, (1N,1−1N)(\tfrac{1}{N},1-\tfrac{1}{N}), and no finite subcover exists. For [0,1][0,1], by contrast, every open cover can be reduced to finitely many patches, as the Heine-Borel theorem guarantees.

A common slip is to read "every open cover has a finite subcover" as "some finite open cover exists". The second is always true, because the space XX itself covers EE, as the Key Results note.


Key Results

  • Any set E⊆XE\subseteq X is covered by the space XX. So covers always exist.

Examples

  • The set (0,1)(0,1) in R\mathbb{R} can be covered by the collection of open sets {(1/n,1−1/n) ⁣:n≥3}\{(1/n,1-1/n)\colon n\geq 3\}. This is an infinite open cover.
  • The set {1,2,3}\{1,2,3\} is covered by the three singletons {1}\{1\}, {2}\{2\}, {3}\{3\}. That is a finite cover.
  • The set of rational numbers Q\mathbb{Q} in R\mathbb{R} can be covered by open intervals centred at each rational with radius 11. That's a countable cover because Q\mathbb{Q} is countable.

Prerequisites

Sources

Symbols: element of, natural numbers, rational numbers, real numbers, subset

Last updated: 20 October 2026 03:02