6
$\begingroup$

I asked this question on Mathematics Stackexchange but got no answer.

Question. Does $ZFC$ prove that there are non-principal ultrafilters $\mathcal U$ and $\mathcal V$ over $\mathbb N$ such that the respective ultrapowers $\mathbb R^{\mathcal U}$ and $\mathbb R^{\mathcal V}$ are not isomorphic when regarded as extensions of the field $\mathbb R$?

$ZFC+\neg CH$ proves that such ultrafilters do exist. More precisely, $ZFC+\neg CH$ proves the stronger statement that there are $\mathcal U$ and $\mathcal V$ as above such that $\mathbb R^{\mathcal U}$ and $\mathbb R^{\mathcal V}$ are not even isomorphic as fields. On the other hand, $ZFC+CH$ proves that $\mathbb R^{\mathcal U}$ and $\mathbb R^{\mathcal V}$ are always isomorphic as fields.

The question could be asked with $ZFC+CH$ instead of $ZFC$, but, seeing no reason to think that it cannot be settled in $ZFC$, the above formulation seemed more natural to me.

Since the fields $\mathbb R^{\mathcal U}$ and $\mathbb R^{\mathcal V}$ have $2^{2^{\aleph_0}}$ distinct structures of extensions of $\mathbb R$, as shown by Eric Wofsey, a field isomorphism between these two fields will not be $\mathbb R$-linear in general.

The transcendence degree of $\mathbb R^{\mathcal U}$ over $\mathbb R$ is $2^{\aleph_0}$. [Of course the same holds for $\mathbb R^{\mathcal V}$.] Indeed, for $x\in\mathbb R$ define $f_x\in\mathbb R^{\mathbb N}$ by $f_x(n)=e^{e^{nx}}$ [or $f_x(n)=\exp(\exp(nx))$ if you don't like small letters] and let $g_x$ be the image of $f_x$ in $\mathbb R^{\mathcal U}$. Then a straightforward argument shows that the subset $\{g_x\ |\ x>0\}\subset\mathbb R^{\mathcal U}$ is algebraically independent over $\mathbb R$.

As shown by tomasz, $ZFC$ proves that there are models $\mathcal A$ and ultrafilters $\mathcal U$ and $\mathcal V$ as above such that $\mathcal A^{\mathcal U}$ and $\mathcal A^{\mathcal V}$ are not isomorphic.

$\endgroup$
3
  • 1
    $\begingroup$ Rahman. M replaced the tag ultrapowers with set-theory. I rejected the edit because the question asks if certain pairs of ultrapowers are isomorphic. If I had to keep just one tag, I would choose this one. It seems to me the other tags I picked give a more precise idea of the question than the set-theory tag. If you think the ultrapowers tag is inappropriate, thanks for telling me why. $\endgroup$ Commented Aug 6, 2020 at 17:31
  • $\begingroup$ I don't like to argue about things in MO! I just explain you the reason: The question asks explicitly if you can prove a certain thing in ZFC while we know that under the failure of CH the answer is positive. There are certain properties of ultra-filters which are independent from ZFC. In particular an answer may show that there are something combinatotial behind the curation, and the concept of ultrapower is just an instance! This why I prefer set-theory to ultrapowers in such question. $\endgroup$
    – Rahman. M
    Commented Aug 6, 2020 at 19:26
  • $\begingroup$ By the way, I don't think that the ultrapowers is inappropriate, I just do not prefer it to set theory as already mentioned. $\endgroup$
    – Rahman. M
    Commented Aug 6, 2020 at 19:28

0

You must log in to answer this question.