- For hyper-connectivity in node-link graphs, see Connectivity_(graph_theory)#Super-_and_hyper-connectivity.
This article includes a list of references, related reading or external links, but its sources remain unclear because it lacks inline citations. (June 2015) (Learn how and when to remove this template message)
In mathematics, a hyperconnected space is a topological space X that cannot be written as the union of two proper closed sets (whether disjoint or non-disjoint). The name irreducible space is preferred in algebraic geometry.
For a topological space X the following conditions are equivalent:
- No two (nonempty) open sets are disjoint.
- X cannot be written as the union of two proper closed sets.
- Every (nonempty) open set is dense in X.
- The interior of every proper closed set is empty.
- Every subset is dense or nowhere dense in X.
A space which satisfies any one of these conditions is called hyperconnected or irreducible.
An irreducible set is a subset of a topological space for which the subspace topology is irreducible. Some authors do not consider the empty set to be irreducible (even though it vacuously satisfies the above conditions).
In algebraic geometry, taking the spectrum of a ring whose reduced ring is an integral domain is an irreducible topological space — applying the lattice theorem to the nilradical, which is within every prime, to show the spectrum of the quotient map is a homeomorphism, this reduces to the irreducibility of the spectrum of an integral domain. For example, the schemes
are irreducible since in both cases the polynomials defining the ideal are irreducible polynomials (meaning they have no non-trivial factorization). A non-example is given by the normal crossing divisor
since the underlying space is the union of the affine planes , , and . Another non-example is given by the scheme
where is an irreducible degree 4 homogeneous polynomial. This is the union of the two genus 3 curves (by the genus–degree formula)
Hyperconnectedness vs. connectedness
Note that in the definition of hyper-connectedness, the closed sets don't have to be disjoint. This is in contrast to the definition of connectedness, in which the open sets are disjoint.
For example, the space of reals with the standard topology is connected but not hyperconnected. This is because it cannot be written as a union of two disjoint open sets, but it can be written as a union of two (non-disjoint) closed sets.
The (nonempty) open subsets of a hyperconnected space are "large" in the sense that each one is dense in X and any pair of them intersects. Thus, a hyperconnected space cannot be Hausdorff unless it contains only a single point.
Since the closure of every nonempty open set in a hyperconnected space is the whole space, which is an open set, every hyperconnected space is extremally disconnected.
The continuous image of a hyperconnected space is hyperconnected. In particular, any continuous function from a hyperconnected space to a Hausdorff space must be constant. It follows that every hyperconnected space is pseudocompact.
Every open subspace of a hyperconnected space is hyperconnected. A closed subspace need not be hyperconnected, however, the closure of any hyperconnected subspace is always hyperconnected.
An irreducible component in a topological space is a maximal irreducible subset (i.e. an irreducible set that is not contained in any larger irreducible set). The irreducible components are always closed.
Unlike the connected components of a space, the irreducible components need not be disjoint (i.e. they need not form a partition). In general, the irreducible components will overlap. Since every irreducible space is connected, the irreducible components will always lie in the connected components.
The irreducible components of a Hausdorff space are just the singleton sets.
Every subset of a Noetherian topological space is Noetherian, and hence has finitely many irreducible components.