Talk:Join and meet

Changed redirect
I changed this from a redirect to Lattice (order) to an article. Cf. Talk:Join (mathematics) for further discussion. JoergenB 20:05, 21 September 2006 (UTC)

Some examples would greatly enhance this entry. Harry Metcalfe 12:39, 19 January 2007 (UTC)

Explain these names?
Could someone please explain how these common English words "join" and "meet" came to stand for "least upper bound" and "greatest lower bound"? On the face of it, they don't seem the slightest bit intuitive as names for these concepts. I have some hope that an explanation for this naming might aid in remembering what they refer to. Gwideman (talk) 00:12, 23 August 2012 (UTC)


 * Probably from the Boolean algebra: if the elements of your poset are sets, the join is the union (what you get when you join them together) and the meet is the intersection (where they meet). The notations (which are otherwise regrettable) reinforce this.  I do not know off-hand of a reliable source for this. --JBL (talk) 00:53, 23 August 2012 (UTC)


 * Ah, that sounds plausible for some particular cases, as you point out. Thanks. Gwideman (talk) 09:23, 24 August 2012 (UTC)

Subsets
This article talks about sets when it should be talking about subsets. The concepts of suprema and infima only make sense when you are talking about subsets, otherwise you may as well be talking about maximal and minimal elements. (Of course, a subset of a set may be the set itself but that is a special case.) — Preceding unsigned comment added by Eptified (talk • contribs) 02:24, 22 June 2013 (UTC)
 * Nonsense. As usual, when you attempt to talk about mathematical concepts.  — Arthur Rubin  (talk) 07:19, 24 June 2013 (UTC)
 * Please comment on the content not the contributor. Spectral sequence (talk) 18:48, 5 July 2013 (UTC)

In the introduction it says: "The join/meet of a subset of a totally ordered set is simply its maximal/minimal element.", what's the point of that? supremum and infimum are stronger 192.76.172.10 (talk) 17:47, 16 May 2014 (UTC)

Idempotency vs reflexivity
It seems that the standard terminology for an operation with the property x ∧ x = x is idempotence : in this context it is a consequence of the reflexivity of the partial order. Spectral sequence (talk) 18:48, 5 July 2013 (UTC)


 * I donated all my lattice theory textbooks in one of the few moves, so I can assert is that I have never seen x ∧ x = x referred to has reflexivity. As I noted on one of the talk pages, there is a difference between an element being idempotent (that is, a specific x satisfies x op x = x), and the idempotency of the operation op (that is, for x in the domain, x op x = x), but the same term is used for both.  — Arthur Rubin  (talk) 21:41, 5 July 2013 (UTC)


 * Spectral sequence and Arthur Rubin are correct. "Idempotent" is the proper way to describe combining x with x to obtain x. Notice that such a combination is an instance of a ternary relation. "Reflexive" is a possible property of binary relations, not ternary relations. JRSpriggs (talk) 05:57, 6 July 2013 (UTC)


 * Some reliable sources for the use of "idempotent" or "idempotency" in this context:
 * "x ∩ x = x, x ∪ x = x (idempotent laws)"
 * "x ∨ x = x (idempotency laws)"
 * Spectral sequence (talk) 11:01, 6 July 2013 (UTC)
 * "x ∨ x = x (idempotency laws)"
 * Spectral sequence (talk) 11:01, 6 July 2013 (UTC)

Link back to ordered sets
The text says Partially ordered sets are categorized by the existence/non-existence of joins/meets for each pair of elements. Should this read charatcterised, and is it correct? The assertion appears to be that if a set has binary operations join and meet satisfying certain properties, then a partial order can be defined on it. What are those conditions, and what is the reference? Spectral sequence (talk) 11:23, 6 July 2013 (UTC)


 * See Semilattice:
 * Join and meet are associative, commutative, idempotent binary operations, and any such operation induces a partial order (and the respective inverse order) such that the result of the operation for any two elements is the least upper bound (or greatest lower bound) of the elements with respect to this partial order.
 * And in Semilattice:
 * An order theoretic meet-semilattice &lang;S, ≤&rang; gives rise to a binary operation ∧ such that &lang;S, ∧&rang; is an algebraic meet-semilattice. Conversely, the meet-semilattice &lang;S, ∧&rang; gives rise to a binary relation ≤ that partially orders S in the following way: for all elements x and y in S, x ≤ y if and only if x = x ∧ y.
 * The relation ≤ introduced in this way defines a partial ordering from which the binary operation ∧ may be recovered. Conversely, the order induced by the algebraically defined semilattice &lang;S, ∧&rang; coincides with that induced by ≤.
 * See e.g. —Tobias Bergemann (talk) 13:32, 6 July 2013 (UTC)
 * (Replying to self.) Stupid me, Join and meet and Join and meet already say this. I was confused by having both pages open side by side. —Tobias Bergemann (talk) 14:33, 6 July 2013 (UTC)
 * And the citation for all of this? I'll say right now that it does not appear to be in Vickers (1989).  Spectral sequence (talk) 15:49, 6 July 2013 (UTC)
 * You are right, I was confused by the presentation in chapters 3.3 and 3.4 in Vickers' book. On page 16 there is the sentence "The significance of this proposition is that if we define ∧ or ∨ independently, as algebraic operators, then it tells us how we have to define ≤ to have any hope of recovering ∧ or ∨ as actual meets or joins." But it never is made explicit. Sorry.
 * There is however Theorem 2.10 on page 40 in
 * —Tobias Bergemann (talk) 17:07, 6 July 2013 (UTC)
 * Indeed, but absorbency is one of the conditions in D&P (L4/L4∂) and not mentioned here. Spectral sequence (talk) 17:17, 6 July 2013 (UTC)
 * You are right. While the absorption laws are not used to prove 2.10(ii) and 2.10(iii), this isn't stated explicitly. So, off-hand, I don't have a citation for these claims. —Tobias Bergemann (talk) 17:48, 6 July 2013 (UTC)
 * Part of the problem is that the statement in such broad generality is not true. For example, consider a three element set {x,y,z} and define a meet operation ∘ by x∘x=x, y∘y=y, z∘z=z and x∘y=y∘x=x, y∘z=z∘y=y.  This is commutative, associative and idempotent but does not define an order: defining u≤v ⇔ u∘v=u implies that x≤y and y≤z but does not imply that x≤z.  The best that one can say is that ≤ is a pre-order.  I expect, but have not worked out the details, that the transitive closure of ≤ is an order ≪ and the meet operation ∧ associated with ≪ is the smallest meet semilattice containing the original operation °.  So in terms of partial binary operations ordered by refinement and preorders again ordered by refinement there would be a Galois connection in which the closed sets were the semilattices and the orders.  However I have no reference for any of that.  Spectral sequence (talk) 09:22, 7 July 2013 (UTC)
 * I'm not sure I understand your example. For the operation you defined to be associative, commutative and idempotent you need x∘z=z∘x=x (because of x∘(y∘z)=x∘y=x and (x∘y)∘z=x∘z), and so you get x≤z because x∘z = x.
 * And the statement in such broad generality most certainly is true:
 * You get reflexivity of ≤ (x≤x for all x) from the idempotency of ∘ (x∘x=x for all x).
 * You get antisymmetry of ≤ (x≤y and y≤x together imply x=y) from the commutativity of ∘:
 * x≤y ⇔ x∘y=x (by definition of ≤)
 * y≤x ⇔ y∘x=y (by definition of ≤)
 * x∘y=y∘x (commutativity of ∘)
 * You get transitivity of ≤ (x≤y and y≤z together imply x≤z) from the associativity of ∘:
 * x≤y ⇔ x∘y=x (by definition of ≤)
 * y≤z ⇔ y∘z=y (by definition of ≤)
 * x∘(y∘z) = x∘y = x
 * (x∘y)∘z = x∘z
 * x∘(y∘z) = (x∘y)∘z (associativity of ∘)
 * Therefore x = x∘z, and so x≤z.
 * To show that x∘y is a lower bound of { x,y } we have to show both x∘y≤x and x∘y≤y, i.e. we have to show both (x∘y)∘x = x∘y and (x∘y)∘y = x∘y. The first is true because (x∘y)∘x = (y∘x)∘x (commutativity of ∘) = y∘(x∘x) (associativity of ∘) = y∘x (idempotency of ∘) = x∘y (commutativity of ∘). The second is true because (x∘y)∘y = x∘(y∘y) (associativity of ∘) = x∘y (idempotency of ∘).
 * To show that x∘y is a/the greatest lower bound of { x,y } we have to show that z≤x and z≤y together imply z≤x∘y:
 * z≤x ⇔ z∘x=z (by definition of ≤)
 * z≤y ⇔ z∘y=z (by definition of ≤)
 * z≤x∘y is true iff z∘(x∘y) = z, by definition of ≤. Now, z∘(x∘y) = (z∘x)∘y (associativity of ∘) = z∘y = z
 * Or perhaps I am even more confused than usual?
 * —Tobias Bergemann (talk) 11:24, 7 July 2013 (UTC)
 * I think there's confusion over what associativity means for partial operations. I'm taking it to mean that if both sides exist then they are equal: you're taking it to mean that if one side exists so does the other and they are equal.  Spectral sequence (talk) 11:55, 7 July 2013 (UTC)

Difference to Infimum and supremum?
What is the difference to Infimum and supremum? Why are there two articles? --Tobias (Talk) 12:07, 18 May 2018 (UTC)


 * Good question! --JBL (talk) 12:10, 18 May 2018 (UTC)


 * Join and meet are operations of an algebraic structure, infimum and supremum are elements of an ordered set. One can talk about the latter without having the former, and one can talk about the former without caring about the induced order. Consider it the Hesperus and Phosphorus of math. ;) Paradoctor (talk) 23:24, 21 February 2019 (UTC)
 * In practice join and meet can be partial operations, and can be used interchangeably with infimum and supremum. Join and meet operations which is total are precisely the topic of the article Semilattice, I believe. 慈居 (talk) 03:43, 10 July 2024 (UTC)

Strict orders not antisymmetric?
Why is antisymmetry shown in the table as not being a property of the strict orders? Though it is trivially true, wouldn't it be at a minimum misleading to place an X in that column?

The note at the bottom of the table states that the check mark indicates the property is "required" by the definition. If I understand correctly, that is saying an X means the property does not hold in general, though it may in certain instances. But that would not be the case here, so it should be checked.

If instead, "required" by the definition meant something along the lines of not being merely a consequence of the definition, then couldn't asymmetry be treated the same way, as it is a consequence of other properties required by the definition? 2601:190:400:40D0:40CC:59A7:150D:A429 (talk) 10:40, 14 January 2022 (UTC)