Kuratowski's closure-complement problem

From HandWiki

In point-set topology, Kuratowski's closure-complement problem asks for the largest number of distinct sets obtainable by repeatedly applying the set operations of closure and complement to a given starting subset of a topological space. The answer is 14. This result was first published by Kazimierz Kuratowski in 1922.[1] It gained additional exposure in Kuratowski's fundamental monograph Topologie (first published in French in 1933; the first English translation appeared in 1966) before achieving fame as a textbook exercise in John L. Kelley's 1955 classic, General Topology.[2]

Proof

Letting [math]\displaystyle{ S }[/math] denote an arbitrary subset of a topological space, write [math]\displaystyle{ kS }[/math] for the closure of [math]\displaystyle{ S }[/math], and [math]\displaystyle{ cS }[/math] for the complement of [math]\displaystyle{ S }[/math]. The following three identities imply that no more than 14 distinct sets are obtainable:

  1. [math]\displaystyle{ kkS=kS }[/math]. (The closure operation is idempotent.)
  2. [math]\displaystyle{ ccS=S }[/math]. (The complement operation is an involution.)
  3. [math]\displaystyle{ kckckckcS=kckcS }[/math]. (Or equivalently [math]\displaystyle{ kckckckS=kckckckccS=kckS }[/math], using identity (2)).

The first two are trivial. The third follows from the identity [math]\displaystyle{ kikiS=kiS }[/math] where [math]\displaystyle{ iS }[/math] is the interior of [math]\displaystyle{ S }[/math] which is equal to the complement of the closure of the complement of [math]\displaystyle{ S }[/math], [math]\displaystyle{ iS=ckcS }[/math]. (The operation [math]\displaystyle{ ki=kckc }[/math] is idempotent.)

A subset realizing the maximum of 14 is called a 14-set. The space of real numbers under the usual topology contains 14-sets. Here is one example:

[math]\displaystyle{ (0,1)\cup(1,2)\cup\{3\}\cup\bigl([4,5]\cap\Q\bigr), }[/math]

where [math]\displaystyle{ (1,2) }[/math] denotes an open interval and [math]\displaystyle{ [4,5] }[/math] denotes a closed interval. Let [math]\displaystyle{ X }[/math] denote this set. Then the following 14 sets are accessible:

  1. [math]\displaystyle{ X }[/math], the set shown above.
  2. [math]\displaystyle{ cX=(-\infty,0]\cup\{1\}\cup[2,3)\cup(3,4)\cup\bigl((4,5)\setminus\Q\bigr)\cup(5,\infty) }[/math]
  3. [math]\displaystyle{ kcX=(-\infty,0]\cup\{1\}\cup[2,\infty) }[/math]
  4. [math]\displaystyle{ ckcX=(0,1)\cup(1,2) }[/math]
  5. [math]\displaystyle{ kckcX=[0,2] }[/math]
  6. [math]\displaystyle{ ckckcX=(-\infty,0)\cup(2,\infty) }[/math]
  7. [math]\displaystyle{ kckckcX=(-\infty,0]\cup[2,\infty) }[/math]
  8. [math]\displaystyle{ ckckckcX=(0,2) }[/math]
  9. [math]\displaystyle{ kX=[0,2]\cup\{3\}\cup[4,5] }[/math]
  10. [math]\displaystyle{ ckX=(-\infty,0)\cup(2,3)\cup(3,4)\cup(5,\infty) }[/math]
  11. [math]\displaystyle{ kckX=(-\infty,0]\cup[2,4]\cup[5,\infty) }[/math]
  12. [math]\displaystyle{ ckckX=(0,2)\cup(4,5) }[/math]
  13. [math]\displaystyle{ kckckX=[0,2]\cup[4,5] }[/math]
  14. [math]\displaystyle{ ckckckX=(\infty,0)\cup(2,4)\cup(5,\infty) }[/math]

Further results

Despite its origin within the context of a topological space, Kuratowski's closure-complement problem is actually more algebraic than topological. A surprising abundance of closely related problems and results have appeared since 1960, many of which have little or nothing to do with point-set topology.[3]

The closure-complement operations yield a monoid that can be used to classify topological spaces.[4]

References

  1. Kuratowski, Kazimierz (1922). "Sur l'operation A de l'Analysis Situs". Fundamenta Mathematicae (Warsaw: Polish Academy of Sciences) 3: 182–199. doi:10.4064/fm-3-1-182-199. ISSN 0016-2736. http://matwbn.icm.edu.pl/ksiazki/fm/fm3/fm3121.pdf. 
  2. Kelley, John (1955). General Topology. Van Nostrand. p. 57. ISBN 0-387-90125-6. 
  3. Hammer, P. C. (1960). "Kuratowski's Closure Theorem". Nieuw Archief voor Wiskunde (Royal Dutch Mathematical Society) 8: 74–80. ISSN 0028-9825. 
  4. Schwiebert, Ryan (2017). "The radical-annihilator monoid of a ring". Communications in Algebra 45 (4): 1601–1617. doi:10.1080/00927872.2016.1222401. 

External links