7
$\begingroup$

Let $M$ be a transitive model of ZFC (set- or class-sized) and let $\kappa\in M$ be a regular cardinal (in $V$). Let $G$ be $M$-generic for $\operatorname{Add}(\kappa,1)$. Now suppose that there is an $X\subseteq\kappa$ of size $\kappa$ which has no size $\kappa$ subsets in $M$ (for example, we could take $X=G$).

My question concerns which "patterns" can be realized in an $M$-generic on the coordinates in $X$.

Given a subset $Y\subseteq X$, is there an $M$-generic $G'$ for $\operatorname{Add}(\kappa,1)$ such that $G'\cap X=Y$? In other words, if we see $Y$ as a pattern of 0s and 1s, is there a generic $G'$ whose pattern on $X$ matches $Y$?

The requirement that $X$ has no large subsets in $M$ is clearly necessary: if $X'\subseteq X$ is in $M$ and has size $\kappa$, a density argument shows that $G'\cap X'\neq \emptyset$ for any generic $G'$, preventing us from realizing the all 0 pattern $Y=\emptyset$.

Certain patterns can be easy to achieve. For example, if we take $X=G$, then it is simple to realize the all 1 pattern $Y=G$: just take $G'=G$. Similarly, the all 0 pattern $Y=\emptyset$ is realized by $G'=\kappa\setminus G$. This last example also shows that it does not suffice, in general, to simply perform surgery on $G$ on the coordinates in $X$ and that global changes to $G$ might be necessary.

I am particularly interested in the special case when $M$ is the ultrapower by a measure on a measurable cardinal $\lambda$, we have $\kappa=\lambda^+$ and GCH holds. In this situation there is a variety of $M$-generics in $V$, which might make the problem easier.

$\endgroup$
2
  • $\begingroup$ So, in a nutshell, you want to know which subsets of the generic factor as generic for a forcing whose remainder is in itself a Cohen forcing. Right? $\endgroup$
    – Asaf Karagila
    Sep 26, 2016 at 16:17
  • $\begingroup$ @AsafKaragila Maybe? I'm not quite understanding your formulation. Can you explain? $\endgroup$ Sep 28, 2016 at 1:55

1 Answer 1

3
$\begingroup$

In some cases, it is possible to construct a set $X$ for which every pattern $Y\subseteq X$ is realized in the way you describe.

For example, consider the very natural case where $M$ is a countable transitive model of ZFC and $\kappa=\omega$. So we are talking about adding $M$-generic Cohen reals here. Enumerate the dense subsets $D_0,D_1,\ldots$ of $M$, and let us build a generic $G$ by diagonalization, while also constructing $X$ for which any $Y\subseteq X$ is realized. Construct a descending sequence of conditions $p_0\geq p_1\geq\dots$ as follows. Let $p_0$ be any element of $D_0$. Now, let the first element $i_0$ of $X$ be the first unspecified bit of $p_0$. Let $\bar p_1$ extend $p_0$ into $D_1$, and then flip the bit at $i_0$ in $\bar p_1$ and further extend to $p_1$ in $D_1$. So $p_1\in D_1$, extending $p_0$, and flipping the bit at $i_0$ keeps the condition in $D_1$. Continue similarly. At stage $n$, let $i_n$ be the first unspecified bit of $p_n$; first extend to $\bar p_{n+1}$ into $D_{n+1}$, and then flip the bits at all the $i_k$ for $k\leq n$ in any desired pattern, and extend into $D_{n+1}$ again, repeating for each pattern. At the end, let $p_{n+1}$ be the resulting extension of $p_n$, which has all its finite flips on the $i_k$'s in $D_{n+1}$. Let $G$ be the filter generated by the conditions $p_n$, which is $M$-generic since we have met every dense set, and let $X$ be the set of all $i_n$. By design, for any bit-flipping pattern on $X$, the construction exactly arranges that the resulting filter obtained by surgery will also be $M$-generic. Thus, every pattern $Y\subseteq X$ is realized by some $M$-generic surgical modification of $G$.

Update. Let me explain how a similar idea can work higher up, such as in the case of your ultrapower, with some additional assumptions.

Specifically, let's assume that $\kappa$ is a regular cardinal and $M$ is a transitive set or class model of ZFC, with $M^{<\kappa}\subset M$ and $|P(\kappa)^M|=\kappa$, which is true for your ultrapower example at the end of the question, if we have the GCH at your measurable cardinal. These assumptions are enough to construct $M$-generic Cohen sets $G\subset\kappa$ by the usual diagonalization procedure. You enumerate in $V$ the dense sets of $M$, and then construct a descending $\kappa$-sequence that gets inside them.

But let me now assume additionally that the $\Diamond_\kappa$ principle holds in $V$. In this case, I shall construct a set $X\subseteq\kappa$ such that every pattern $Y\subseteq X$ is realized in the way you request. Fix a $\Diamond_\kappa$-sequence $\langle A_\alpha\mid\alpha<\kappa\rangle$. Enumerate the dense sets of $M$ as $D_\alpha$, for $\alpha<\kappa$. At stage $\alpha$ of the construction, we have constructed a descending sequence of conditions $p_\beta$ for $\beta<\alpha$, and we have identified the first $\alpha$ many elements of $X$. Let $\bar p_\alpha$ be the union of the prior conditions, which is a condition in $M$ because $M^{<\kappa}\subset M$. We only do something special if the domain of $\bar p_\alpha$ is $\alpha$ itself, and if furthermore $A_\alpha$ is coding an ordinal $\beta<\kappa$ and a subset $Y$ of the first $\alpha$ members of $X$. In this case, we first extend $\bar p_\alpha$ into $D_\beta$, and then we further extend to a condition $p_\alpha$, so that after flipping the bits on $Y$, the condition remains in $D_\beta$. After this, let the $\alpha^{th}$ member of $X$ simply be the first bit unspecified by $p_\alpha$.

By design, the filter $G$ generated by the descending sequence of conditions $p_\alpha$ is an $M$-generic filter. If $Y\subseteq X$ is any set, then let $G_Y$ be the filter obtained by flipping the bits of $G$ on the coordinates of $Y$. This filter is also $M$-generic, since for any dense set $D_\beta$, there is a stationary set of stages $\alpha$ for which the domain of the conditions up to $\alpha$ is $\alpha$ itself (since that occurs on a club) and where $A_\alpha$ codes $\beta$ and the set $Y\cap\alpha$. In this case, the condition $p_\alpha$ had the property that flipping the bits on $Y$ remained in $D_\beta$, and so $G_Y$ meets $D_\beta$. Thus, any desired pattern can be arranged on the digits of $X$, by choosing $Y$ suitably.

$\endgroup$
4
  • $\begingroup$ Joel, that is a great construction! I particularly like the argument from $\diamondsuit_\kappa$. However, your suggestion provides an $X$ and the generic $G$, while I was hoping that $X$ could be fixed in advance. Perhaps I shall edit the question to make it clear. $\endgroup$ Sep 28, 2016 at 1:53
  • $\begingroup$ Oh, your question was clear, but I didn't have anything to say about that case. I think there can be some bad $X$, which don't realize all patterns, but I don't yet have an argument. $\endgroup$ Sep 28, 2016 at 2:50
  • $\begingroup$ I am also wondering: what if you assume merely that $X$ itself is $M$-generic. In this case, can you realize every pattern $Y\subseteq X$? $\endgroup$ Sep 28, 2016 at 2:51
  • $\begingroup$ Another natural case: assume $X$ decides every set in $M$ on a tail, so that the tail filter of $X$ is an $M$-ultrafilter. At first I thought this might realize all patterns, but I couldn't push an argument through. $\endgroup$ Sep 28, 2016 at 3:06

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.