MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am trying to imagine (to some extent, of course) the geometry of ultrapowers of certain 'easy-to-handle' Banach spaces. Let me start with $X = \ell_p$, $p\in (1,\infty)$ or $X=c_0$.

Since the differences between partiuclar choices of ultrafilters are not clear to me enough, let me take any free ultrafilter $\mathcal{U}$ over an arbitrary infinite index set $I$.

So, my beginners question is, does the ultrapower $(X)_\mathcal{U}$ have an unconditional basis?

share|cite|improve this question
For anything interesting to happen, you want a non-principal ultrafilter. Otherwise the ultrapower will just be $X$. – Andrew Parker May 17 '12 at 21:50
Another issue with the generality in which you have posed the question is that if $I$ is very large and there is a measurable cardinal, then the ultrafilter you use on $I$ might be complete beyond the size of $X$, in which case the ultrapower of $X$ will again just be $X$, even though the ultrafilter is nonprincipal. – Joel David Hamkins May 17 '12 at 22:37
Andrew, in this context, "free" means "non-principal". – Bill Johnson May 18 '12 at 6:08
Bill, since "free" was added to the question in the (currently) latest edit, I conjecture that it was added as a result of Andrew's comment. – Andreas Blass May 18 '12 at 12:25
Yes, apologies for not pointing it out. – Alfredo Ortuño May 18 '12 at 15:31
up vote 7 down vote accepted

For $\ell_1$ and $c_0$, the answer to your (implicit) question is easy. If you take a free ultrafilter on the natural numbers (or most any ultrafilter used in Banach space theory) the ultrapower contains $L_1$ (or $C[0,1]$) and hence does not have an unconditional basis.

For $\ell_p$, $1<p\not= 2 < \infty$, the situation might be trickier. Enflo and Rosenthal

Some results concerning Lp(μ)-spaces. J. Functional Analysis 14 (1973), 325–348

proved that if $\mu$ is a finite measure s.t. the density character of $L_p(\mu)$ is at least $\aleph_\omega$, then $L_p(\mu)$ does not embed into a space that has an unconditional basis. Now such an $L_p(\mu)$ embeds into some ultrapower of $\ell_p$ but I don't know off the top of my head that when the density character of $L_p(\mu)$ is equal to $\aleph_\omega$ [EDIT: I meant $\aleph_1$ here] whether (under some or any set theoretic axioms about the size of $\aleph_\omega$ [EDIT: I meant $\aleph_1$ here]) $L_p(\mu)$ embeds into an ultrapower of $\ell_p$ when the ultrafilter is a free ultrafilter on the natural numbers.

share|cite|improve this answer
Thank you. Perhaps I should restrict my attention to ultrafilters over integers only. – Alfredo Ortuño May 18 '12 at 9:04
Bill, I'm confused --- isn't any ultrapower of a Hilbert space also a Hilbert space, and doesn't every Hilbert space have an unconditional basis? – Nik Weaver May 18 '12 at 9:25
Sure, Nik. I should have written $1<p\not= 2 < \infty$. I'll edit. – Bill Johnson May 18 '12 at 10:39
Bill: To embed such $L_p(\mu)$ into an ultrapower of $l_p$ with an ultrafilter on $\omega$ you need (at the very least) that $2^{\aleph_0} > \aleph_\omega$. – Ramiro de la Vega May 18 '12 at 12:33
Sorry, Ramiro; I meant $\aleph_1$ at that point. I'll edit again, but make it clear that it is an edit so that your comment makes sense. – Bill Johnson May 18 '12 at 13:53

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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