Seminar on Set Theory
Hand-in exercise 1, model solution
September 18, 2015
(a) Let x, y ∈ H such that x ≤ y. We know that y ≤ y ∗∗ so by transitivity of ≤ we find
that x ≤ y ∗∗ , and this is equivalent to y ∗ ≤ x∗ .
Alternative. If x ≤ y, then y = x ∨H y. Taking complements, we find that y ∗ =
(x ∨H y)∗ = x∗ ∧H y ∗ , which means exactly that y ∗ ≤ x∗ ,
(b) First of all, notice that x∗ ∈ B for all x ∈ H, since we have x∗∗∗ = x∗ . Also, using
exercise (a) we see that x ≤ y implies y ∗ ≤ x∗ , which in turn implies x∗∗ ≤ y ∗∗ . So the
map H → H : x 7→ x∗∗ preserves order.
Since 0H ≤ 0H , we have 0∗H = (0H ⇒ 0H ) = 1H . Also, there is obviously only one z ∈ H
such that z ∧H 1H = 0H , namely z = 0H . This means that 1∗H = (1H ⇒ 0H ) = 0H .
From these facts we deduce that 0H , 1H ∈ B, so B has a greatest and a least element,
and these are induced from H.
Let us show that for all x, y ∈ B, we have x ∧H y ∈ B. Since x ∧H y ≤ x, we have
(x ∧H y)∗∗ ≤ x∗∗ = x. Similarly, we have (x ∧H y)∗∗ ≤ y. From these it follows that
(x ∧H y)∗∗ ≤ x ∧H y. But we also have x ∧H y ≤ (x ∧H y)∗∗ , so it is indeed the case
that x ∧H y ∈ B. Now clearly, any z ∈ B that is a lower bound of x and y must satisfy
z ≤ x ∧H y. But the latter is itself in B, so we can take x ∧H y to be the infimum of x
and y in B.
Again, let x, y ∈ B be given. Clearly, any z ∈ B that is an upper bound of x and y
must satisfy z ≥ x ∨H y. From this it follows that z = z ∗∗ ≥ (x ∨H y)∗∗ . But (x ∨H y)∗∗ ,
being the pseudocomplement of something, is in B. So we can take (x ∨H y)∗∗ to be
the supremum of x and y in B. We conclude that B is a bounded lattice.
Finally, we have x∗ ∈ B for all x ∈ B ⊂ H, as we already remarked. We have
x∧B x∗ = x∧H x∗ = 0H = 0B and x∨B x∗ = (x∨H x∗ )∗∗ = (x∗ ∧H x∗∗ )∗ = 0∗H = 1H = 1B .
So B is a complemented bounded lattice, i.e. a Boolean algebra.
W
(c) Suppose that H is complete and let X ⊂ B. Then X has a supremum XWin (H, ≤).
Now every z ∈ B that is an upper W bound of X W must certainly satisfy z ≥ X. From
this it follows that z = z ∗∗ ≥ ( X)∗∗ . But (
W ∗∗ X) ∗∗
, being the pseudocomplement of
something, is itself in B. So we can take ( X) to be the supremum of X in B. The
existence of infima can be shown similarly.
(d) We have to prove that
◦
◦ [ ◦
U = U iff U = X− [
X −U .
We will do this by proving that
◦
U =X− [
X −U
1
◦ ◦
and using the fact from topology that if A ⊂ B then A ⊂ B. We notice that
◦ ! ◦ !
a∈ X− [
X −U iff ¬ a∈ [
X −U
iff ¬∃δ > 0(B(a; δ) ⊂ X − U )
iff ∀δ > 0(B(a; δ) ∩ U 6= ∅)
iff a∈U
And therefore RO(X) is the regularization of O(X).
(e) We use the example from (d) to show this. Suppose X = R, equipped with the Euclidean
topology. Let U = (1, 2) and V = (2, 3). Then in O(X) the meet of these opens
is just (1, 2) ∪ (2, 3), which does not contain the point 2. However, the meet in the
regularization looks as follows. First we take the complement in R of (1, 2) ∪ (2, 3),
which is (−∞, 1] ∪ {2} ∪ [3, ∞). The interior of this is (−∞, 1) ∪ (3, ∞), which has [1, 3]
as complement. The interior of this is (1, 3), and that is the meet of U and V in the
regularization of O(R). It follows that the meet in the regularization of O(R) is not
induced from the meet in O(R).
