Altanis

3.3Components and Local Connectedness

Updated 11 Jul 2026Chapter (PDF)

This section seeks to formalize the idea that, for any topological space XX, there is a cover of connected and path-connected subsets of XX.

[3.3.1]Definition(Connected Components)#

Suppose XX is a topological space. Define \sim to be an equivalence relation on XX such that, if xyx \sim y, then there is some connected subspace CXC \subseteq X such that x,yCx, y \in C. The equivalence classes for \sim are called the connected components of XX.

Proof.
  1. Reflexivity. xxx \sim x since {x}X\{x\} \subseteq X is a connected subspace.

  2. Symmetry. Immediate.

  3. Transitivity. If xyx \sim y and yzy \sim z, then there are connected subsets A,BXA, B \subseteq X where x,yAx, y \in A and y,zBy, z \in B. Then ABA \cup B is a union that shares the point yy, so it is connected, and it contains x,zx, z. Thus xzx \sim z.

[3.3.2]Theorem#

The connected components of XX are connected, disjoint subspaces of XX whose union is XX. Every nonempty connected subspace of XX is fully contained in one equivalence class (i.e., it intersects exactly one equivalence class).

Proof.

Since the connected components of XX are equivalence classes, they form a partition for XX. If CXC \subseteq X were connected and were such that C[x1]C \cap [x_1]_{\sim} \ne \emptyset and C[x2]C \cap [x_2]_{\sim} \ne \emptyset, where \sim is the equivalence relation that induces the components, then x1x2x_1 \sim x_2 and so [x1]=[x2][x_1]_{\sim} = [x_2]_{\sim}—thus CC intersects exactly one equivalence class and is also fully contained in one.

Fix some x0Xx_0 \in X, and note [x0][x_0]_{\sim} is a component. For any x[x0]x \in [x_0]_{\sim}, there is some connected subspace CxC_x for which x,x0Cxx, x_0 \in C_x. With the previous part in mind, note Cx[x0]C_x \subseteq [x_0]_{\sim}, and so [x0]=x[x0]Cx[x_0]_{\sim} = \bigcup_{x \in [x_0]_{\sim}} C_x, which is connected since each CxC_x has the shared point x0x_0. Thus every component is connected.

[3.3.3]Remark#

In a topological space, a component is referred to as a “maximal” connected subspace because it is completely saturated: any connected subspace belongs to exactly one component, and adjoining any outside element to a component would strictly break its connectedness.

An analogous process may be applied to split a space into path components.

[3.3.4]Definition(Path Components)#

Suppose XX is a topological space. Define \sim to be an equivalence relation on XX such that, if xyx \sim y, then there is some path from xx to yy contained in XX. The equivalence classes for \sim are called the path components of XX.

Proof.
  1. Reflexivity. xxx \sim x since γ:[a,a]X\gamma: [a, a] \to X, where γ(a)=x\gamma(a) = x, is a path from xx to xx.

  2. Symmetry. Immediate.

  3. Transitivity. Suppose xyx \sim y and yzy \sim z. Then there are continuous maps γ1:[a,b]X\gamma_1: [a, b] \to X and γ2:[b,c]X\gamma_2: [b, c] \to X such that γ1\gamma_1 parameterizes a path from xx to yy and γ2\gamma_2 parameterizes a path from yy to zz. By the pasting lemma, note that γ:[a,c]X\gamma: [a, c] \to X defined by “pasting” γ1,γ2\gamma_1, \gamma_2 is a path from xx to zz, and so xzx \sim z.

[3.3.5]Theorem#

The path components of XX are path-connected, disjoint subspaces of XX whose union is XX. Every nonempty path-connected subspace of XX is fully contained in one equivalence class (i.e., it intersects exactly one equivalence class).

Proof.

The proof is identical to the proof about connected components. The proof for showing path components are path-connected is an immediate application of transitivity.

[3.3.6]Remark#

Note that the connected components CC of a topological space XX are closed. Note that if a subspace AXA \subseteq X is connected, then so is its closure A\bar{A}. Since all connected subspaces must be the subset of a connected component CC, and the closure C\bar{C} of a connected component is connected, a connected component CC equals its closure C\bar{C}. Put another way, we established that connected components CC are maximal, and so they must equal their closures. If there are finitely many connected components C1,,CnC_1, \dots, C_n of XX, then each CiC_i is open, since the complement of one connected component CiC_i is the finite union of closed sets jiCj\bigcup_{j \ne i} C_j, which is closed.

To summarize, the connected components of a topological space are all closed, and if there are finitely many of them, they're all open.

[3.3.7]Definition(Locally Connected, Locally Path Connected)#

Suppose XX is a topological space. We say XX is locally connected at xx (resp. locally path-connected at xx) if, for every neighborhood UXU \subseteq X of xx, there is some open, connected (resp. path connected) neighborhood VXV \subseteq X of xx fully contained in UU. If all points in XX are locally connected (resp. locally path connected), then we say XX is locally connected (resp. locally path connected).

[3.3.8]Theorem#

A space XX is locally connected if and only if, for every open UXU \subseteq X, the connected components of UU are open.

Proof.

()(\Longrightarrow): Suppose XX is locally connected. Let UXU \subseteq X be open, and let CUC \subseteq U be some connected component of UU. For each xCx \in C, there is some open, connected neighborhood VxV_x of xx such that xVxUx \in V_x \subseteq U. Since VxCV_x \cap C \ne \emptyset and CC is a connected component, it follows that VxCV_x \subseteq C since VxV_x is connected too. Thus C=xCVxC = \bigcup_{x \in C} V_x, the arbitrary union of open sets, which is open.

()(\Longleftarrow): Let UXU \subseteq X be an arbitrary open subset. Then note, for each xUx \in U, there is some connected component CUC \subseteq U such that xCUx \in C \subseteq U. By hypothesis, CC is open, completing the proof.

[3.3.9]Theorem#

A space XX is locally path-connected if, for each open UXU \subseteq X, the path components of UU are open.

Proof.

The proof is identical to the proof of the previous theorem.

[3.3.10]Theorem#

If XX is a topological space, then each path component of XX is fully contained in one connected component of XX. If XX is locally path-connected, then the path components and connected components coincide.

Proof.

Let PP be a path component of XX. Since PP is path-connected, it is also connected, so it is fully contained in some connected component CXC \subseteq X, proving the first statement.

Now assume XX is locally path-connected. To prove the second statement, we show CPC \subseteq P. For the sake of contradiction, suppose PCP \subsetneq C. Let QQ be the union of all path components of XX that are different from PP and intersect CC. Since all path components are contained in exactly one connected component, it follows that QCQ \subseteq C. Then note

C=CX=C(PPc)=(CP)(CPc)=PQ.C = C \cap X = C \cap (P \cup P^c) = (C \cap P) \cup (C \cap P^c) = P \cup Q.

Since every path component of a locally path-connected set is open, we have that P,QP, Q are disjoint, nonempty open subsets whose union is CC, meaning CC is separated (a contradiction, since CC is connected). Thus CPC \subseteq P, and so P=CP = C.

3.3.1Exercises#

[3.3.11]Problem#

What are the connected components and path components of R\bR_\ell? What are the continuous maps f:RRf: \bR \to \bR_\ell?

Proof.

Suppose CRC \subseteq \mathbb{R}_\ell contains at least two distinct points, x<yx < y. Then C=((,y)C)([y,)C)C = \big((-\infty, y) \cap C\big) \cup \big([y, \infty) \cap C\big) constitutes a separation of CC, so CC is disconnected. Thus, the only nonempty connected subspaces of R\mathbb{R}_\ell are singletons, which are therefore the connected components. Since every path-connected space is connected, the path components must also be exactly the singletons.

Note that for some f:RRf: \bR \to \bR_\ell to be continuous, the image of a connected set must be connected. Since R\bR is connected in its standard topology, it follows that f(R)f(\bR) is connected, but the only connected subsets of R\bR_\ell are singletons, so it follows that f(R)={}f(\bR) = \{*\}, so all continuous maps are constant maps.

[3.3.12]Problem#
  1. What are the connected components and path components of Rω\bR^\omega in the product topology?

  2. Consider Rω\bR^\omega in the uniform topology. Show that x,y\vb{x}, \vb{y} lie in the same component of Rω\bR^\omega if and only if the sequence (xy)(\vb{x} - \vb{y}) is bounded above.

Solution.
  1. Note that the arbitrary product of connected and path-connected sets is connected and path-connected in the product topology. Thus Rω\bR^\omega is the only connected and path-connected component in Rω\bR^\omega with the product topology.

  2. ()(\Longrightarrow): Suppose x,y\vb{x}, \vb{y} are in the same connected component. Then there is some connected CRωC' \subseteq \bR^\omega such that x,yC\vb{x}, \vb{y} \in C'. Now consider a translation T:RωRωT: \bR^\omega \to \bR^\omega defined by T(x)=xyT(\vb{x}) = \vb{x} - \vb{y}. This is a homeomorphism, so C:=T(C)C := T(C') is also a connected component. Thus we have xy,0C\vb{x - y}, \vb{0} \in C.

    First, we show Rω\bR^\omega can be separated by the subspaces of bounded and unbounded sequences. Let ORωO \subseteq \bR^\omega be the subspace of all bounded sequences. For each bounded sequence xOx \in O, define Bx=Bρ(x,1/2)B_x = B_{\bar{\rho}}(x, 1/2), a ball about xx defined by the uniform metric of radius 1/21/2. For each yBxy \in B_x, note that xnyn<1/2|x_n - y_n| < 1/2 for every nZ+n \in \bZ_+, and so yy is bounded as well, meaning yOy \in O. Thus xBxOx \in B_x \subseteq O, so O=xOBxO = \bigcup_{x \in O} B_x. An analogous procedure can be performed to show OcO^c is open, since drawing a ball of radius 1/21/2 about an unbounded sequence still lies in OcO^c. Thus Rω=OOc\bR^\omega = O \cup O^c, a separation.

    We know 0\vb{0} is bounded. If xy\vb{x - y} were unbounded, then C=(OC)(OcC)C = (O \cap C) \cup (O^c \cap C) would be a valid separation for a supposedly connected set (a contradiction), so xy\vb{x - y} is forced to be bounded.

    ()(\Longleftarrow): Suppose xy\vb{x} - \vb{y} is a bounded sequence. Let ORωO \subseteq \bR^\omega be the subspace of all bounded sequences. We will show OO is connected, which will complete the proof.

    We proceed by showing OO is path-connected. Note the translation T:RωRωT: \bR^\omega \to \bR^\omega given by T(z)=zyT(\vb{z}) = \vb{z} - \vb{y} is a homeomorphism. Since xy\vb{x} - \vb{y} and yy=0\vb{y} - \vb{y} = \vb{0} are both bounded sequences, we have T(x),T(y)OT(\vb{x}), T(\vb{y}) \in O. We can show there is a path from h=xy\vb{h} = \vb{x} - \vb{y} to 0\vb{0} in OO. We may define a map γ:[0,1]O\gamma: [0, 1] \to O by γ(t)=th\gamma(t) = t\vb{h}. Indeed, γ(0)=0\gamma(0) = \vb{0} and γ(1)=h\gamma(1) = \vb{h}, so to show γ\gamma is a path, we simply show γ\gamma is continuous.

    Since hO\vb{h} \in O, there is some MRM \in \bR such that hn<M|h_n| < M for every nZ+n \in \bZ_+. Fix some t0[0,1]t_0 \in [0, 1] and some 1>ε>01 > \epsilon > 0. Let δ=ε/M\delta = \epsilon/M, then let t[0,1]t \in [0, 1] be such that tt0<δ|t - t_0| < \delta. Then

    ρ(γ(t),γ(t0))=supnZ+{min(thnt0hn,1)}=supnZ+{tt0hn}<supnZ+{(ε/M)(M)}=supnZ+{ε}=ε.\begin{align*} \bar{\rho}(\gamma(t), \gamma(t_0)) &= \sup_{n \in \bZ_+} \left\{ \min\left(\abs{th_n - t_0 h_n}, 1\right) \right\} \\ &= \sup_{n \in \bZ_+} \left\{ \abs{t - t_0} \abs{h_n} \right\} \\ &< \sup_{n \in \bZ_+} \left\{ (\epsilon/M)(M) \right\} \\ &= \sup_{n \in \bZ_+} \{\epsilon\} = \epsilon. \end{align*}

    If ε1\epsilon \ge 1, then δ\delta may be anything since the uniform metric is at most 11. Thus γ\gamma is continuous, and we conclude OO is path-connected, which implies OO is also connected. Because T(x)T(\vb{x}) and T(y)T(\vb{y}) both belong to the connected subspace OO, they are in the same component. Since T1T^{-1} is a homeomorphism, it maps components to components, meaning x\vb{x} and y\vb{y} must also belong to the same component in Rω\bR^\omega, completing the proof.

[3.3.13]Problem#

Let XX be locally path-connected. Show all open, connected subsets of XX are path-connected.

Solution.

Let OXO \subseteq X be open and connected. Fix xOx \in O and any open neighborhood UOU \subseteq O of xx. Since UU is an open subset of the locally path-connected space XX, all of UU's path components are open. Thus there is some open, path component (which is thus path-connected) [x][x]_\sim such that x[x]Ux \in [x]_\sim \subseteq U. Thus OO is locally path-connected, and so its path components and connected components coincide. Since OO is connected, its only connected component is OO, and so OO is also path-connected, completing the proof.