2
$\begingroup$

Let $L$ be a lattice and let $\textbf{DM}(\cdot)$ denote the Dedekind-MacNeille completion.

Is there a lattice $L$ that is not a quotient of $\textbf{DM}(L)$? And what if we generalise this question to posets: is every poset $P$ a quotient of $\textbf{DM}(L)$? (I just realise whether the second question makes sens depends on the answer to this post.)

$\endgroup$
6
  • 2
    $\begingroup$ Is $\mathbb{Q} $ a quotient of $\mathbb{R}\cup\left\{\pm\infty\right\} $ ? I don't know the answer, but it doesn't look like it works this way. (That said finite posets might be different.) $\endgroup$ Oct 1, 2015 at 12:40
  • $\begingroup$ The DM completion of a finite lattice is the lattice itself. (The DM completion of a poset $P$ is the smallest complete lattice containing $P$ as a subposet. For finite posets, "lattice" and "complete lattice" are equivalent, so the DM completion of a lattice is itself.) $\endgroup$ Oct 1, 2015 at 13:31
  • $\begingroup$ @darijgrinberg For the record, no, there is no order preserving surjection $\phi: \mathbb{R} \cup \{ \pm \infty \} \to \mathbb{Q}$. The image of such a map would land in $[\phi(- \infty), \phi(\infty) ]$, which is not all of $\mathbb{Q}$. There are also no order preserving surjections $\mathbb{R} \to \mathbb{Q}$, although it takes a bit more work. Might be a good problem for an analysis exam... $\endgroup$ Oct 1, 2015 at 15:03
  • 1
    $\begingroup$ @Dominic Do you want the composition $P \to DM(P) \to P$ to be the identity? $\endgroup$ Oct 1, 2015 at 15:03
  • $\begingroup$ For that matter, there is no surjective order-preserving map from the completion of the naturals N to N as a lattice. It should be clear that unbounded and countably (maybe uncountably?) cofinal lattices cannot have such maps from their completions. Gerhard "Where Does The Top Go?" Paseman, 2015.10.01 $\endgroup$ Oct 1, 2015 at 15:56

2 Answers 2

4
$\begingroup$

In order to finally kill every possible version of this question: No, even if $P$ is finite. Let $P$ be two incomparable elements. Then $DM(P)$ is the diamond poset $0 < p,q < 1$. The only maps of posets from $DM(P)$ to $P$ are to send everything to $0$ or everything to $1$. In particular, there is no surjection $DM(P) \to P$ and, thus, $P$ is not a quotient of $DM(P)$.

To summarize comments above: If $L$ is a a finite lattice, then yes, since $L = DM(L)$. If $L$ is an infinite lattice, even a totally ordered one, then no (simplest example is $\mathbb{Z}$). And the example above shows the answer is also "no" for finite posets.

$\endgroup$
4
$\begingroup$

I claim that if $X$ is a poset and $f:DM(X)\rightarrow X$ is a surjective mapping where $f|_{X}$ is the identity mapping, then $X$ is already a complete lattice.

Recall that the Dedekind-Macneille of a poset $X$ is the unique extension of $X$ to a complete lattice $C$ such that for all $c\in C$ there are $R,S\subseteq X$ with $c=\bigvee^{C}R=\bigwedge^{C}S$.

Suppose that $c\in C$. Then there are subsets $R,S\subseteq X$ such that $c=\bigvee^{C}R=\bigwedge^{C}S$. If $r\in R$, then $f(c)\geq f(r)=r$, so $f(c)\geq\bigvee^{C}R=c$ and $f(c)\leq\bigwedge^{C}S=c$. Therefore $f(c)=c$. Thus $c\in X$ after all. Therefore since $X=C$ and $X$ is complete, then $X$ is also complete.

$\endgroup$
1

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.