4
$\begingroup$

Problem 2-2 in Lee's Introduction to Topological Manifolds reads:

Let $X = \{1, 2, 3\}$. Give a list of topologies on $X$ such that every topology on $X$ is homeomorphic to exactly one on your list.

Below is my attempted solution. Is there a more efficient way to solve a problem like this one? My method felt a bit tedious.


The topologies can be partitioned into classes according to how many 1- and 2-element sets they contain. Let us write $(m, n) \in \{0, \dotsc, 3\}^2$ to represent each such class, with $m$ being the number of 1-element sets (singletons) and $n$ the number of 2-element sets in the topology, and go through each of them. Note that homeomorphisms preserve class membership.

  • (0,0) This class contains only the trivial topology $\{X, \varnothing\}$.

  • (1,0) If we add a single singleton to the trivial topology we obtain a topology homeomorphic to $\{X, \varnothing, \{1\}\}$.

  • (2,0) This class is empty, since if we have two singletons we must also have the 2-element set given by their union.

  • (3,0) Similarly empty.

  • (0,1) Adding a 2-element set to the trivial topology we obtain a topology homeomorphic to $\{X, \varnothing, \{1, 2\}\}$.

  • (1,1) There are two distinct ways we can add a singleton and a 2-element set to the trivial topology; either they overlap or they do not. Hence we either get a topology homeomorphic to $\{X, \varnothing, \{1\}, \{1, 2\}\}$, or one homeomorphic to $\{X, \varnothing, \{1\}, \{2, 3\}\}$. Note that these are not homeomorphic to each other. (A homeomorphism would have to map $\{1\}$ to $\{1\}$, but simultaneously map $\{1,2\}$ to $\{2,3\}$, which is impossible.)

  • (2,1) All topologies in this class are hoemorphic to $\{X, \varnothing, \{1\}, \{2\}, \{1, 2\}\}$, since a topology with two singletons must also contain their union.

  • (3,1) All three singletons together generate the discrete topology, so this class is empty.

  • (0,2) If we have two 2-element sets they must overlap on a single element, thus giving us also a singleton via closure under finite intersection. Hence this class is empty.

  • (1,2) By the logic above, the topologies in this class are all homeomorphic to $\{X, \varnothing, \{1\}, \{1, 2\}, \{1, 3\}\}$.

  • (2,2) These topologies must have a (1,2)-topology as a subset, by the logic above. Hence they are homeomorphic to $\{X, \varnothing, \{1\}, \{2\}, \{1, 2\}, \{1, 3\}\}$.

  • (3,2) Again, only the discrete topology contains all three singletons.

  • (0,3) With all three 2-element sets we get all the singletons via intersections, hence this class is empty.

  • (1,3) Similarly empty.

  • (2,3) Similarly empty.

  • (3,3) Here we have only the discrete topology, $\mathcal P(X)$.

Hence every topology on $X$ is homeomorphic to exactly one of the following topologies: $$\{X, \varnothing\},$$ $$\{X, \varnothing, \{1\}\},$$ $$\{X, \varnothing, \{1, 2\}\},$$ $$\{X, \varnothing, \{1\}, \{1, 2\}\},$$ $$\{X, \varnothing, \{1\}, \{2, 3\}\},$$ $$\{X, \varnothing, \{1\}, \{2\}, \{1, 2\}\},$$ $$\{X, \varnothing, \{1\}, \{1, 2\}, \{1, 3\}\},$$ $$\{X, \varnothing, \{1\}, \{2\}, \{1, 2\}, \{1, 3\}\},$$ $$\mathcal P(X).$$

$\endgroup$
1
  • 1
    $\begingroup$ Its good to include a subject in your tags, general-topology in this case. Also solution-verification questions of the form "is my proof correct" are off-topic (see tag description for more details). I've deleted the question about your proof being correct, since the tag problem-solving still applies (i.e. what's the right method to solve the question), I see no further problems though. $\endgroup$
    – Jakobian
    Commented Feb 26 at 16:25

1 Answer 1

4
$\begingroup$

I suspect the repeated string of symbols $\varnothing\, \{1\}$ should be $\varnothing, \{1\}$. Otherwise it looks fine to me.

For a discussion of the difficulties of this problem in the general case see this math.stackexchange post. In the comments of that post you'll see a link there for the OEIS sequence that gives number of different topologies on an $n$-element set up to homeomorphism; for the case $n=3$ that list agrees with your count of 9 topologies.

$\endgroup$
4
  • $\begingroup$ Oh, yes of course. Fixed that typo now. $\endgroup$
    – ummg
    Commented Feb 26 at 15:12
  • $\begingroup$ Is there a less "brute force" method? $\endgroup$
    – ummg
    Commented Feb 26 at 15:36
  • 1
    $\begingroup$ Seems unlikely. This is a hard, combinatorial problem. Not deep, particularly... but hard in the computational sense. $\endgroup$
    – Lee Mosher
    Commented Feb 26 at 15:50
  • 1
    $\begingroup$ I added a bit on the "difficulty" issue, with a link to a related post. $\endgroup$
    – Lee Mosher
    Commented Feb 26 at 16:19

You must log in to answer this question.

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