1
$\begingroup$

In the book "Categorical Structure of Closure Operators with Applications to Topology" by Dikranjan and Tholen a Katětov closure operator is defined in terms of filter covergence:

$k_X(M):=\{x \in X: \exists \mathcal{F} \; \mathcal{F} \to x \land M \in \mathcal{F}\}$,

is there an axiomatic definition of closures that correspond to the convergence spaces similar to that of axioms for Kuratowski (for topological spaces) and Čech (for pretopological spaces) closures?

$\endgroup$
4
  • $\begingroup$ I'm not familiar with Dikranjan and Tholen's book, but I think you are looking for the Moore closure spaces which are are an equivalent description of the pretopologiacal convergence spaces. $\endgroup$
    – Tyrone
    Commented Feb 21 at 3:44
  • $\begingroup$ @Tyrone, from what I can see the Moore closure can not describe pretopological spaces (that are described by the Čech closure) as it is idempotent (3 in definition 2.4 from the link that you've provided). $\endgroup$ Commented Feb 21 at 10:51
  • $\begingroup$ You're right. The discussion on nlab is really not what I thought it was. What I had in mind is the results of this paper by Kasahara (see Theorem 3). Dolecki and Nogura use the term `Moore closure space' in their article in the Encyclopedia of General Topology, and I was probably wrong to use it. $\endgroup$
    – Tyrone
    Commented Feb 21 at 11:49
  • $\begingroup$ @Tyrone, thank you very much, the paper seems to be what I was looking for. $\endgroup$ Commented Feb 21 at 12:15

2 Answers 2

1
$\begingroup$

There is a discussion of the relation between convergence spaces and closeness spaces in the paper

S. Kasahara, Closeness spaces and convergence spaces, Proc. Japan Acad., 50(4) (1974), 303-308.

The structure of a closeness spaces is determined by a family of semiclosure operators - which are weaker than Cech closure operators. On the other hand, if a closure space is determined by family consisting of a single semiclosure operator, then this must be a Cech closure operator.

$\endgroup$
0
$\begingroup$

This is a partial answer. Let $X$ be a compact Hausdorff space and let $\mathcal{U}$ be an ultrafilter on a set $I$. Then any function $\phi: I \to X$ induces an ultrafilter on $X$, namely $\{A \subseteq X: \phi^{-1}(A) \in \mathcal{U}\}$. In a compact Hausdorff space every ultrafilter converges to exactly one point. So for each $I$ and $\mathcal{U}$ we get an infinitary "operation" $f_{I,\mathcal{U}}$ which takes an $I$-tuple of elements of $X$ (i.e., a function from $I$ into $X$) to a single element of $X$.

Compact Hausdorff spaces can be axiomatized in terms of the existence of these operations $f_{I,\mathcal{U}}$ by the three axioms

  1. (triviality) if $I = \{i\}$ consists of a single point then $f_{I,\mathcal{U}}(\phi) = \phi(i)$

  2. (restriction) if $A \in \mathcal{U}$ then $f_{A, \mathcal{U}|_A}(\phi|_A) = f_{I,\mathcal{U}}(\phi)$

  3. (iteration) if $\{I_i: i \in I\}$ are disjoint sets equipped with ultrafilters $\mathcal{U}_i$, $J = \bigcup I_i$, and $\mathcal{U}$ is an ultrafilter on $I$, then we get an ultrafilter $\mathcal{V}$ on $J$ consisting of the sets whose intersection with each $I_i$ belongs to $\mathcal{U}_i$; then for any $\phi: J \to X$ we have $f_{J,\mathcal{V}}(\phi) = f_{I,\mathcal{U}}(f_{I_i,\mathcal{U}_i}(\phi))$.

I guess that is something like what you're asking for. This is proven in my paper "The variety of CH-algebras", Acta Mathematica Hungarica 69, 221-232 (1995). I guess I wrote this when I was in grad school.

$\endgroup$
2

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .