Bishop-compact space

Bishop-compact spaces


The classical notion of compact space (defined e.g. using the finite subcover property) is not very useful in point-based constructive topology (as opposed to locale theory or formal topology). In particular, it is impossible to prove constructively that familiar spaces like the unit interval [0,1][0,1]\subseteq \mathbb{R} are compact in this sense. Since classically, a metric space (or more generally a uniform space) is compact if and only if it is complete and totally bounded, Errett Bishop decided to simply take the conjunction of the latter two properties as a definition of compactness in his constructive mathematics.


A uniform space is Bishop-compact if it is both complete and totally bounded.


Equivalence to open-cover compactness

In classical mathematics (more precisely, assuming the ultrafilter principle), a uniform space is Bishop-compact if and only if its underlying topological space is compact in the usual sense. In fact, more can be said: a compact Hausdorff space admits a unique compatible uniformity, which is complete and totally bounded. So Bishop-compactness is really equivalent to ordinary compactness in a strong sense. One can also use Bishop-compactness to develop various parts of the theory of compactness directly; for instance, Shulman constructs the Stone-Cech compactification using Bishop-compactness for gauge spaces.

Topological invariance

Completeness and total boundedness are not individually topological invariants. In other words, if two uniform spaces are topologically isomorphic (i.e. their underlying topological spaces are homemorphic), it doesn’t follow that if one is complete or totally bounded then so is the other. For example, (0,1)\mathbb{R} \cong (0,1) topologically, but the former is complete and not totally bounded, while the latter is totally bounded and not complete.

However, the fact that Bishop-compactness is classically equivalent to open-cover compactness means that it is topologically invariant: if two uniform spaces are topologically isomorphic, and one is both complete and totally bounded, then so is the other.

In constructive mathematics, though, even this much cannot be proven: the statement that Bishop-compactness is a topological invariant implies the fan theorem, which is (classically true but) unprovable constructively. This is essentially the content of section 3.2 of Diener’s thesis, though it is phrased there in contrapositive form (assuming a counterexample to the fan theorem, we can construct a counterexample to the topological invariance of Bishop-compactness).

This constructive failure of topological invariance for Bishop-compactness has been one of the impetuses for the constructive study of proximity spaces, in the hope of finding a structure at least somewhat weaker than uniformity under which a point-based notion of compactness is invariant.


  • Hannes Diener, Compactness under constructive scrutiny (Ph.D. thesis), 2008, pdf

  • Mike Shulman, The shape of infinity, arxiv

Created on July 12, 2018 at 01:38:01. See the history of this page for a list of all contributions to it.