Talk:Directed set

Template:WikiProject banner shell Why not use ≤ for the partial order? -- Miguel

Well, I want to make the real numbers into a directed set by directing them towards a number x₀, and then the notation ≤ would clash with the ordinary order of the reals. Ideally, we would have a separate symbol, such as a round less-that symbol, or a less-than symbol with a tilde underneath. AxelBoldt

I extended the page to include the notion of a directed subset, that appears quite frequently in order theory. I leaft the rest as it was. In a first change I made an error that is now corrected. -- Markus (no login)

I added the condition that a directed set should be nonempty. This is often forgotten, but should be there: many of the convenient properties of directed sets are false without it. For instance, take the statement that in a compact space, every net (= map from a directed set) has a convergent subnet. This is only true if you insist that a directed set must be nonempty. Another reason to build nonemptiness into the definition is that it makes the following true: a directed set is a preordered set A in which every finite subset has an upper bound. (This is perhaps the best definition of directed set.) By induction, this is equivalent to saying that every subset of cardinality 0 or 2 has an upper bound. The "cardinality 0" part says that A contains an element, i.e. is nonempty. The "cardinality 2" part says that for all a, b, there exists c bigger than both a and b. -- Tom (no login)

preorder and base of filters

As you allude to "(partial) orders", you could mention that a refexive and transitive relation is called a preorder.

For the type of preorder occuring here, I knew the terminology of "right filtering preorder".

Concerning limits, I knew them rather defined in terms of bases of filters instead of directed sets. Note that given a directed set A, you have canonically associated a base of filters (the set of all elements "bigger than" x, where x runs through A ; for a left filtering relation these are just the "balls of radius x"), but the converse is not true. MFH 23:15, 8 Mar 2005 (UTC)

right filtering

The definition at the top says that a directed set is a "right filtering preorder". If "right filtering" is a general term, there should be a link to a definition. I'm not sure where to look for one. Also, right at the top is a description of how, starting with a and b with a ≤ b, you can create a sequence with a ≤ b ≤ c ≤ d …. Since c, d, … can all be equal to b, this statement holds for any preorder, and doesn't seem the least bit interesting. I'm removing it. Dfeuer 18:41, 2 October 2007 (UTC)

Upper bound of empty set

@ShirAko1 It seems to me that your edit https://en.wikipedia.org/w/index.php?title=Directed_set&diff=prev&oldid=1295101846 is incorrect.

By definition, $y$ is an upper bound of $X$ when $x\le y$ holds for all $x\in X$. So any real number is an upper bound of the empty set in ℝ, because the condition is vacuously true. Jean Abou Samra (talk) 15:23, 14 June 2025 (UTC)

Real line directed towards a point

@mgkrupa I do not understand the point of the example Directed set#Directed towards a point with ℝ∖{x₀}, which you added in the edit https://en.wikipedia.org/w/index.php?title=Directed_set&diff=prev&oldid=987313074. This preordered set is equivalent (I mean, in the categorical sense of having isomorphic poset reflections) to (ℝ, ≤). In what context is it natural and why is it important to consider this specific preorder with each point doubled instead of just the order? Jean Abou Samra (talk) 22:05, 14 June 2025 (UTC)