2

Definition (Dense Boolean SubAlgebra): Let $A$ be a Boolean algebra and $B\subseteq A$ a Boolean subalgebra. We say that $B$ is dense in $A$ if for every $a\in A$ there exist $b\in B$ such that $b \leq a$.

I have found here (https://planetmath.org/booleansubalgebra) stated the following claim (with proof?):

Claim: $B$ is dense in $A$ iff for every $a_1\leq a_2\in A$, there exists $b\in B$ such that $a_1 \leq b \leq a_2$.

The direction $(\Leftarrow)$ is clear, taking $a_1=0$ and $a_2=a$. For $(\Rightarrow)$ the proposed proof is the following:

Proof (?) of ($\Rightarrow$) : Assume $a_1\leq a_2$. Since $B$ is dense in $A$ by hp., we can find $r\in B$ such that $r\leq a_2$. We conclude by picking $b= r\vee a_1$ so that $a_1 \leq b \leq a_2$ as desired.

The problem I see with this proof is that the candidate $b= r\vee a_1$ is not guaranteed to be in $B$ since $a_1$ is not in $B$.

QUESTION: Is the property stated in the link actually true?

  • ) If yes, is the proof above correct and am I missing something obvious?
  • ) If no, what is the correct(ed) property that they might have had in their mind?

1 Answers1

1

The statement is simply incorrect (also, the definition of "dense subalgebra" is wrong: you need to require both $a$ and $b$ to be nonzero, or else trivially every subalgebra is dense, by taking $b=0$). First of all, it would trivially imply that $B$ has to be all of $A$ since you could have $a_1=a_2$. Even if you modify the statement to require $a_1<a_2$ it is still incorrect. For instance, consider the example where $A$ is the power set of some infinite set and $B$ is the subalgebra of finite or cofinite sets. Then $B$ is dense in $A$ but if $S\subseteq T$ are infinite coinfinite sets then there is no element of $B$ in between them.

A correct statement along these lines would be that $B$ is dense in $A$ iff whenever $a_1<a_2$ for $a_1\in B$ and $a_2\in A$, then there exists $b\in B$ such that $a_1<b\leq a_2$. To prove the forward direction for this statement, by density of $B$ there exists $c\in B$ such that $0<c\leq a_2\wedge \neg a_1$, and then you can take $b=a_1\vee c$.

Eric Wofsey
  • 330,363
  • thank you very much for the quick reply. Answer accepted. Do you know if the (correct) property you stated can be proved also in a more general/different setting than Boolean algebras (e.g., Posets, Lattices, Distr. lattices, Vector lattices), i.e., when we do not have the Boolean negation? Or it is something specific of BAs? Thank you again. – albatros Dec 08 '21 at 23:26
  • No interesting generalization comes to mind. For instance, you can easily see that it is not true for totally ordered sets. – Eric Wofsey Dec 08 '21 at 23:46