1
$\begingroup$

In Topology, Munkres gives the following construction for a minimal uncountable ordered set. Such a set has a largest element; the section obtained from this largest element is uncountable but any other section is countable:

Start with an uncountable set $B$ that is well-ordered under some order $<$

Then consider $A = \{1,2\} \times B$ in the lexicographical order (I can define this if anybody wants). Any $A_{\Omega}$, a section of $A$ (which is defined to be $\{x \in A | x < \Omega \}$), is uncountable for $\Omega$ of the form $(2,y)$ where $y \in B$.

Here is the non-sequitur troubling me:

There exists a smallest $\Omega$ such that $A_{\Omega}$ is uncountable. In other words, for $\beta < \Omega$, $A_{\beta}$ is countable.

How do we know that a smallest $\Omega$ exists?

The set we desire is $A_{\Omega} \cup \{\Omega\}$. It has a largest element $\Omega$ such that $A_{\Omega}$ is uncountable but any other section is countable.

$\endgroup$
  • 2
    $\begingroup$ The smallest uncountable well-ordered set is $A_\Omega,$ not $A_\Omega\cup\{\Omega\}.$ $\endgroup$ – bof May 21 '17 at 10:33
4
$\begingroup$

Since the set $\{1,2\}\times B$ is well ordered, any non-empty subset of this set has a smallest element.

Now consider the subset of $\{1,2\}\times B$ consisting of all those elements whose section is uncountable, clearly, it's non-empty and hence there exists a smallest element, denote it by $\Omega$, hence the result follows!

$\endgroup$
0
$\begingroup$

My opinion of Munkres' method is not high, although he is not illogical.

If $B$ is uncountable and $<$ is a well-order on B then let $C=\{x\in B: \{y|y<x\}$ is uncountable$\}.$ Then

(1) If $C\not =\phi$ then let $x_C=\min C.$ By def'n of $C$ and of $x_C,$ if $y<x_C$ then $\{z|z<y\}$ is countable. So $\{y\in B:y\leq x_0\}$ is a minimal uncountable order.

(2) If $C=\phi$ take $d\not \in B$ and extend $<$ to a well-order on $D=B\cup \{d\}$ by letting $b<d$ for all $b\in B.$ Then $\{y\in D: y<b\}$ is countable for each $b\in B$ but $\{y\in D:y<d\}=B$ is uncountable.

In Set Theory: An Introduction To Independence Proofs, by K. Kunen, we have (not verbatim) : Let $W$ be the set of well-orders that are subsets of $\omega\times \omega.$ (Where $\omega$ is the least infinite ordinal. Recall that any binary relation on a set $S$ is jut a subset of $S\times S.$) Now any well-order is isomorphic to the $\epsilon$-order on a unique ordinal, so we can obtain the set $B$ of all and only those ordinals that are isomorphic to members of $W.$ Then $B$ is in fact the set of all countable ordinals, so $B=\omega_1,$ the least uncountable ordinal .... This is done without AC (the Axiom of Choice). So we do not need AC to prove the existence of an uncountable well-order.

$\endgroup$

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.