Let $(\Omega,\leq)$ be a countable linear order. Suppose that for every finite $m \in \mathbb{N}$, and all subsets $S_1$ and $S_2$ of $\Omega$ of order $m$, there is an order-automorphism of $(\Omega,\leq)$ that sends $S_1$ to $S_2$.

Call the order-automorphism group $A$.

Is there a "fundamental theorem" for $(\Omega,\leq)$ in the spirit of the fundamental theorem of projective geometry, of the following form:

  • if $\Omega_1$ and $\Omega_2$ are subsets of $\Omega$, order-isomorphic through an order-isomorphism $\gamma$, then there is an order-automorphism $\overline{\gamma}$ in $A$ which extends $\gamma$.

Of course, this statement is too general -- there should at least be some restrictions on $\Omega_i$ -- but I am looking for a statement "as general as possible."

Second related question: what if we remove the countability condition on $\Omega$ ?

  • 6
    $\begingroup$ The order satisfying your condition for $m=2$ must be dense and have no endpoints, so the only such order (up to isomorphism) is $(\mathbb Q,\leq)$. $\endgroup$ – Wojowu Sep 19 '19 at 13:32

As already observed, you may as well assume $(\Omega,\leq)$ is $(\mathbb{Q},\leq)$. In this case, of course one condition on $\Omega_1$ and $\Omega_2$ ensuring your desired property is finiteness.

So the interesting question is about the case when $\Omega_1$ and $\Omega_2$ are infinite. We can even just consider the special case when $\Omega_1=\Omega_2$.

Call a subset $X$ of $\mathbb{Q}$ good if every order automorphism of $X$ extends to an order automorphism of $\mathbb{Q}$. Then it follows from Theorem 2.19 of this paper by Panagiotopoulos that good subsets are extremely rare. In particular, consider the set of all subsets of $\mathbb{Q}$ as Cantor space. Then it follows from Panagiotopoulos's work that the set of good subsets of $\mathbb{Q}$ is meager (a countable union of nowhere dense sets).

In fact, it's even worse. Given a subset $X$ of $\mathbb{Q}$, we can also view the set $\text{Aut}(X)$ of order automorphisms of $X$ as a Polish space with the pointwise convergence topology. So call a set $X$ really bad if the set of automorphisms of $X$ that extend to automorphisms of $\mathbb{Q}$ is meager in $\text{Aut}(X)$. Then the result is that the set of really bad subsets of $\mathbb{Q}$ is comeager.

Using the language "generic" for properties that happen on a comeager set, the result says: "A generic automorphism of a generic subset of $\mathbb{Q}$ does not extend to an automorphism of $\mathbb{Q}$".

He also proves similar results for pairs of subsets. Moreover, the paper is not about $\mathbb{Q}$, but about general Fraisse limits.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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