Connectedness
When is a space held together as one piece, and when is it secretly separable into parts?
What does it mean for one-piece-ness to be topological?
A closed interval and two separated closed intervals can both be compact. Finite control alone does not say whether the space is one piece.
[0, 1] one piece
[0, 1] union [2, 3] two pieces
The second space has a visible gap. A ruler can measure the gap, but topology wants to express the distinction without measuring distance. The question becomes: can the open-set structure split the space into independent parts?
Connectedness answers that question.
A topological space is disconnected when it can be written as the union of two nonempty disjoint open sets. Such a pair is called a separation of the space.
X = U union V
U nonempty
V nonempty
U and V open
U intersection V = empty
If no such separation exists, the space is connected.
This definition is deliberately phrased in terms of open sets. Connectedness is about whether the topology itself can divide the whole space into two open regions that are internally visible and mutually separate.
disconnected:
U V
| |
open open
no overlap
together cover X
The definition may look negative, but the structural meaning is positive:
A connected space is one whose topology holds the whole space together as a single piece.
The closed interval [0, 1] is connected in the usual topology. If it could be split into two nonempty disjoint open parts, then moving from a point in one part to a point in the other would require a first place where the split changes. The order and completeness of the real line prevent such a clean open break.
That fact is often encountered through the Intermediate Value Theorem. A continuous real-valued function on an interval cannot jump from negative to positive without passing through zero. Structurally, this happens because continuous images of connected spaces remain connected, and the connected subsets of the real line are intervals.
connected interval --continuous map--> connected subset of R
So connectedness explains why continuous functions on intervals cannot create jumps. A jump would split the image into separated pieces.
The union [0, 1] union [2, 3] is different. Inside that subspace, the pieces [0, 1] and [2, 3] are both open relative to the whole space. They are nonempty, disjoint, and together cover the space. The topology can see the split.
X = [0, 1] union [2, 3]
U = [0, 1]
V = [2, 3]
U and V are open in X
This example also shows why "open" must be read relative to the space under discussion. The set [0, 1] is not open as a subset of the whole real line with its usual topology. But it is open inside X = [0, 1] union [2, 3], because there is an open region of the real line whose intersection with X is exactly [0, 1].
Topology always asks which space supplies the open sets.
Connectedness can also be expressed through clopen sets. A set is clopen when it is both open and closed. If a nonempty proper subset U of X is clopen, then its complement is also nonempty and open, so U and its complement separate the space.
U clopen and proper
|
v
X = U union (X \ U)
Therefore, a space is connected exactly when its only clopen subsets are the empty set and the whole space.
A nontrivial clopen set is a region the topology can isolate completely from its complement. Connectedness says that no such complete isolation exists inside the space.
connected:
only clopen sets are empty and X
disconnected:
some proper nonempty clopen set exists
This makes connectedness visibly topological. It uses only open sets, complements, and the whole-space structure. No lengths, angles, coordinates, or metrics appear.
Connectedness is preserved by homeomorphism. If X and Y are homeomorphic, a separation of Y would pull back to a separation of X. Conversely, a separation of X would push forward to a separation of Y. A reversible continuous change of description cannot turn one connected piece into two disconnected pieces.
X ----homeomorphism----> Y
separation in Y
|
| pull back
v
separation in X
So connectedness is a topological invariant. It is also preserved by continuous images. If X is connected and f: X -> Y is continuous, then f(X) is connected. If the image could be separated into two open parts, their inverse images would separate X.
X connected ----f----> f(X)
separation of f(X)
|
| pull back along f
v
separation of X
Since no separation of X exists, no separation of the image can exist. This preservation law explains why connectedness is so effective. Continuous maps may bend, stretch, identify, or collapse parts of a space. They may turn a whole interval into a point. But they cannot send a connected input onto a disconnected image.
connected input
|
| continuous map
v
connected image
A disconnected space can map continuously onto a connected one. For example, a constant map sends any space to a single point, and a single point is connected. Connectedness is preserved by continuous image, not reflected by arbitrary continuous maps. This keeps the invariant precise. Connectedness survives allowed deformation forward through continuous maps and exactly through homeomorphism. It does not say that every map from a disconnected space remembers the disconnection.
Connectedness also separates two ideas that are easy to confuse. A space can be connected without every pair of points being joined by a visible path. Path-connectedness is stronger: it asks for a continuous path between any two points. Every path-connected space is connected, but connectedness itself only asks whether the space can be separated by open sets.
path-connected
|
v
connected
For familiar intervals, disks, circles, and many geometric regions, the two ideas often coincide. In general topology they diverge. That divergence is useful because it shows that connectedness is the more basic open-set invariant, while paths introduce extra structure about continuous movement through the space.
At the present level, the open-set invariant is the point.
Compactness asked whether local covers can be finitely controlled. Connectedness asks whether the topology can split the space into independent open parts. Both are global properties expressed through open sets, and both survive homeomorphism. They reveal different aspects of the same structural discipline.
compactness:
local coverage has finite global control
connectedness:
the whole space cannot be separated into open pieces
The objects are topological spaces. The allowed transformations are continuous maps, with homeomorphisms giving equivalence. What composes are continuous maps, and connectedness is stable under their images. The invariant is one-piece-ness as seen by open sets. The defining relation is the absence of a separation: no two nonempty disjoint open sets cover the whole space. Equality is replaced by homeomorphic equivalence, under which connectedness is preserved. What remains outside the frame is a stronger question about deformation itself: not only whether spaces are equivalent, but whether maps between spaces can be continuously transformed into one another.
Connectedness says that the topology holds the space together as one piece.
Compactness and connectedness are properties of spaces. They tell us what survives continuous change of the space itself. But topology also studies transformations as objects in their own right.
When should two continuous maps count as the same because one can be deformed into the other?