4
$\begingroup$

A $P$-point is an ultrafilter $\scr U$on $\omega$ such that for every function $f:\omega\to\omega$ there is $x\in {\scr U}$ such that the restriction $f|_x$ is either constant, or finite-to-one.

A $Q$-point is an ultrafilter $\scr U$on $\omega$ such that for every function $f:\omega\to\omega$ with the property that $f^{-1}(\{m\})$ is finite for each $m\in \omega$, there is $x\in {\scr U}$ such that the restriction $f|_x$ is injective.

$P$-points need not exist, and $Q$-points need not exist.

Question. Is it possible that neither $P$- nor $Q$-points exist?

$\endgroup$

1 Answer 1

12
$\begingroup$

Your definition of P-point should end with "finite-to-one" rather than "injective". As it stands, it defines selective ultrafilters, not P-points.

As far as I know, the consistency of the statement "neither P-points nor Q-points exist" is still an open problem. It is known that this statement implies $2^{\aleph_0}\geq\aleph_3$.

EDIT in response to a request for a reference for the last statement: Although the statement is explicitly somewhere in the literature, I can't find it just now, so instead I'll indicate how it follows from two results that I can find, Theorems 9.25 and 9.27 in my chapter of the Handbook of Set Theory. Neither result is due to me; references are given in the chapter, a pre-publication version of which is available at http://www.math.lsa.umich.edu/~ablass/hbk.pdf .

The results involve the cardinal characteristics cov$(\mathcal B)$ [the smallest number of meager sets to cover $\mathbb R$], $\mathfrak d$ [the smallest number of functions $\omega\to\omega$ to dominate all such functions], and $\mathfrak c$ [the cardinal of the continuum]. It is provable in ZFC that $$ \aleph_1\leq\text{cov}(\mathcal B)\leq\mathfrak d\leq\mathfrak c. $$ Theorem 9.25 says (among other things) that if $\mathfrak d=\mathfrak c$ then there exists a P-point. Theorem 9.27 says (among other things) that if cov$(\mathcal B)=\mathfrak d$ then there exists a Q-point. So in order to have neither a P-point nor a Q-point, you'd need $$ \aleph_1\leq\text{cov}(\mathcal B)<\mathfrak d<\mathfrak c, $$ which implies $\mathfrak c\geq\aleph_3$.

$\endgroup$
2
  • 1
    $\begingroup$ Do you have a reference for the last claim? $\endgroup$
    – Wojowu
    Mar 24, 2019 at 21:17
  • 3
    $\begingroup$ @Wojowu I've edited the answer to include a proof from results with references. $\endgroup$ Mar 25, 2019 at 0:58

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.