ideal completion of a poset


Let P be a poset. Consider the set Id⁡(P) of all order ideals of P.

Theorem 1.

Id⁡(P) is an algebraic dcpo, such that P can be embedded in.

Proof.

We shall list, and when necessary, prove the following series of facts which ultimately prove the main assertion. For convenience, write P′=Id⁡(P).

  1. 1.

    P′ is a poset with ≤ defined by set theoretic inclusion.

  2. 2.

    For any x∈P, ↓x∈P′.

  3. 3.

    P can be embedded in P′. The function f:P→P′ defined by f(x)=↓x is order preserving and one-to-one. If x≤y, and a≤x, then a≤y, hence ↓x⊆↓y. If ↓x=↓y, we have that x≤y and y≤x, so x=y, since ≤ is antisymmetric.

  4. 4.

    P′ is a dcpo. Suppose D is a directed setMathworldPlanetmath in P′. Let E=⋃D. For any x,y∈E, x∈I and y∈J for some ideals I,J∈D. As D is directed, there is K∈D such that I⊆K and J⊆K. So x,y∈K and hence there is z∈K⊆E such that x≤z and y≤z. This shows that E is directed. Next, suppose x∈E and y≤x. Then x∈I for some I∈D, so y∈I⊆E as well. This shows that E is a down set. So E is an ideal of P: ⋁D=E∈P′.

  5. 5.

    For every x∈P, ↓x is a compact element of P′. If ↓x≤⋁D, where D is directed in P′, then ↓x⊆⋃D, or x∈⋃D, which implies x∈I for some ideal I∈D. Therefore ↓x⊆I, and ↓x is way below itself: ↓x is compact.

  6. 6.

    P′ is an algebraic dcpo. Let I∈P′. Let C={↓x∣x∈I}. For any x,y∈I, there is z∈I such that x≤z and y≤z. This shows that ↓x≤↓z and ↓y≤↓z in C, so that C is directed. It is easy to see that I=⋁C. Since I is a join of a directed set consisting of compact elements, P′ is algebraic.

This completesPlanetmathPlanetmathPlanetmathPlanetmathPlanetmathPlanetmath the proof. ∎

Definition. Id⁡(P) is called the ideal completion of P.

Remarks.

  • •

    In general, the ideal completion of a poset is not a complete latticeMathworldPlanetmath. It is complete in the sense of being directed complete. This is different from another type of completion, called the MacNeille completion of P, which is a complete lattice.

  • •

    If P is an upper semilatticePlanetmathPlanetmath, then so is Id⁡(P). In fact, the join of any non-empty family of ideals exists. Furthermore, if P has a bottom element 0, then Id⁡(P) is a complete lattice.

    Proof.

    Let S be a non-empty family of ideals in P. Let A be the set of P consisting of all finite joins of elements of those ideals in S, and B=↓A. Clearly, B is a lower set. For every a,b∈B, we have c,d∈A such that a≤c and b≤d. Since c and d are both finite joins of elements of those ideals in S, so is c∨d. Since a≤c∨d and b≤c∨d, B is directed. If I is any ideal larger than any of the ideals in S, clearly A⊆I, since I is directed. So B=↓A⊆↓I=I. Therefore, B=⋁S.

    If 0∈P, then ⟨0⟩, the bottom of Id⁡(P), is the join of the empty family of ideals in P. By this entry (http://planetmath.org/CriteriaForAPosetToBeACompleteLattice), Id⁡(P) is a complete lattice. ∎

  • •

    If P is a lower semilatticePlanetmathPlanetmath, then so is Id⁡(P).

    Proof.

    Let I,J be two ideals in P and K=I∩J. By definition, I and J are non-empty, so let a∈I and b∈J. As P is a lower semilattice, c:=a∧b exists and c≤a and c≤b. So c∈I∩J, and that K=I∩J is non-empty. If x≤y∈K, then x≤y∈I or x∈I. Similarly x∈J. Therefore x∈I∩J=K and K is a lower set. If r,s∈K, then there is u∈I and v∈J such that r,s≤u,v. So r,s≤u∧v and K is directed. This means that I∩J∈Id⁡(P). ∎

References

  • 1 G. Gierz, K. H. Hofmann, K. Keimel, J. D. Lawson, M. W. Mislove, D. S. Scott, Continuous Lattices and Domains, Cambridge University Press, Cambridge (2003).
Title ideal completion of a poset
Canonical name IdealCompletionOfAPoset
Date of creation 2013-03-22 17:03:01
Last modified on 2013-03-22 17:03:01
Owner CWoo (3771)
Last modified by CWoo (3771)
Numerical id 8
Author CWoo (3771)
Entry type Definition
Classification msc 06A12
Classification msc 06A06
Related topic LatticeOfIdeals
Defines ideal completion