Given $M$ a transitive class model of ZFC denote $M_\beta$ the sets in $M$ of rank less thank $\beta$.

I'm trying to verify that elements of $(V[G])_\beta$, where $G$ is a generic subset of some forcing poset $P$, "have names in" $V_\beta$ whenever $|P|<\beta$. (I quote "have names in" as I'm not sure if the author of what I'm reading means that every name is in $V_\beta$ or they have at least one name in $V_\beta$.)

I tried to state and solve a simpler problem by attempting to show that every $x\in V[G]$ has a name of cardinality $|x|$. But I'm having trouble there too as I don't really see how to define names for $x$ in the ground model, given that I can't refer to $x$ there. So for example fixing a name $\tau_i$ for every $x_i\in x$ and then creating the name $\{(\tau_i,\mathbb{1})\}$ fails as I don't see how this name could be definible in $V$.

The only other strategy for forming names I've seen is using definability of the forcing relation, but I haven't been able to apply it here.

Thank you for reading.

  • 1
    $\begingroup$ That's a nice question. I'm going to have to try and remember it for when I teach forcing someday. I hope. $\endgroup$ – Asaf Karagila Feb 27 '18 at 5:51
  • $\begingroup$ Re: First paragraph. Let $(\beta+1)^*$ be the canonical $P$-name over $V$ for $\beta+1.$ Let $p\in P\setminus G.$ Let $x=y_G\in (V[G])_{\beta},$ where $y$ is a $P$-name over $V$ Let $y'=y\cup \{( \beta+1)^*,p)\}.$....... Then we have (rank$(y'))^V\geq \beta$ but $y'_G=x.$ $\endgroup$ – DanielWainfleet Feb 27 '18 at 6:09
  • $\begingroup$ Off the top of my head I suggest transfinite induction on $ \alpha <\beta$ for $x\in (V[G])_{\beta}$ and rank $(x)=\alpha,$ employing the Truth Lemma (If $x' $is a $P$-name over $ V$ and $x'_G=x$ then some $p\in G$ forces that rank$(x')=\alpha)$ and the Definability Lemma (That is,$ \{p\in P: p$ forces rank$ (x') =\alpha\}$ is a member of $V).$ $\endgroup$ – DanielWainfleet Feb 27 '18 at 6:30

Your question is a touch confusing; do you mean to say that every element in the ground model has a name in $V[G]$? What have you assumed so far about $V[G]$? What text or notes are you reading from?

If you’re following Kunen’s “Set Theory,” one approach is to start by defining $V[G]$ as the set $\{ \tau_G : \tau \in V^P \}$ where we assume $P \in V$ and $V^P$ denotes the set of all $P$-names which are also in $V$. Here also $\tau_G$ is the image of $\tau$ under a certain evaluation function for names. We then interpret $V[G]$ as an $FL$-structure, where $FL$ is the forcing language, by interpreting names $\tau$ to the corresponding element $\tau_G$ in $V[G]$, and interpreting all other sentences in the obvious way from there. Starting from this side of things, it’s clear that every element of $V[G]$ doesn’t “have” a name in $V$, but rather is the interpretation of a name which happens to be in $V$.

After that, it takes some time to show that $V[G]$ is the smallest transitive model of of ZFC containing both $V$ and $G$, but along the way one can show:

$$rank(\tau_G ) \leq rank(\tau), \forall \tau \in V^P$$

Which is, I think, the detail you are asking about - or at least related. Here is a proof of that fact. Let $y \downarrow$ be the set of all predecessors of $y$ under $\in $.

We proceed by transfinite induction on names, along the well-founded set-like relation $x \in trcl(y)$, which we will denote as $x < y$. Suppose we have shown the theorem for $\sigma < \tau$. We now have: $$ rank(\tau_G) = sup \{ rank(x) + 1 : x \in \tau_G \downarrow\} \\ = sup \{ rank(x) + 1 : x \in \tau_G \}$$ since $V[G]$ is transitive $$ = sup \{ rank(\sigma_G) + 1 : \sigma \in dom(\tau) \}$$ since $x \in \tau_G \rightarrow x = \sigma_G$ for some $\sigma$ in $dom(\tau)$ $$ \leq sup \{ rank(\sigma) + 1 : \sigma \in dom(\tau) \} $$ by inductive hypothesis $$ \leq sup \{ rank(\sigma , p) + 1 : (\sigma, p) \in \tau \} \\ = rank(\tau)$$

  • $\begingroup$ I think the OP is actually trying to show the opposite: that many sets have names of the least possible rank (your argument shows that small rank names can't correspond to large rank objects). $\endgroup$ – Noah Schweber Feb 28 '18 at 22:34
  • 1
    $\begingroup$ Sets in V[G] are, as you say, evaluations of names in V. My question is: Fix $x\in V[G]$ of rank $<\beta$, can you find a name in $V$ of rank $<\beta$ such that $x$ is the evaluation of that name? $\endgroup$ – JKEG Feb 28 '18 at 22:42
  • $\begingroup$ That makes much more sense. Can’t one, given $\tau_G$, find a $\sigma$ of the same rank as $\tau_G$ such that $\sigma_G = \tau_G$ just by reworking the construction of $\tau$ step by step from the ground up, eliminating elements as you go along of the form $( \tau_i, p )$ for any $p$ not in $G$, and keeping elements of the form $(\tau_i, p)$ as you go along for which $p \in G$? Then the rank should be the same and the new name $\sigma$ should still be in $V^P$. $\endgroup$ – Rachael Alvir Feb 28 '18 at 22:52
  • $\begingroup$ @RachaelAlvir I don't see how to do that without referencing $G$, which is necessary for the name to be in $V$. $\endgroup$ – JKEG Mar 18 '18 at 19:57
  • $\begingroup$ The trick is that by working on $\tau$ instead of $\tau_G$, you can reference G. $\tau$ lives outside of your forcing extension, in the “real world” where G is accessible. I’ll try to get around to writing up a more explicit explanation of this. $\endgroup$ – Rachael Alvir Mar 21 '18 at 14:16

This is only a partial result, but it was good enough for my purposes.

Let $G$ be a generic subset of a forcing poset $\mathbb{P}$. Then, for any ordinal $\alpha$ there exists $\beta\geq\alpha$ such that for any $x\in V[G]_\beta$ there is a name $\tau\in V_\beta$ such that $\tau_G=x.$

Define $\phi:V[G]\rightarrow\text{On}$ by $$\phi(x) = \min\{\xi\in\text{On}\mid\text{There is a name }\tau\in V_\xi\text{ such that }\tau_G=x\} $$ and define $\psi:\text{On}\rightarrow\text{On}$ by $$\psi(\xi+1)=\sup\{\phi(x)\mid x\in V[G]_{\xi+1}\}+\psi(\xi)+1 $$at successor ordinals and $\psi(\lambda)=\bigcup_{\xi<\lambda}\psi(\xi)$ for limit $\lambda.$ Being strictly increasing and continuous, $\psi$ has arbitrarily large fixed points.

Now, fix an ordinal $\alpha$ and let $\beta$ be a limit ordinal $\beta≥\alpha$ such that $\beta=\psi(\beta).$ Let $x\in V[G]_\beta.$ There is a $\gamma<\beta$ such that $x\in V[G]_{\gamma+1}$. Then, $$\phi(x)≤\psi(\gamma+1)≤\psi(\beta)=\beta, $$and it follows from the definition of $\phi(x)$ that there is a $\xi≤\beta$ and a name $\tau\in V_\xi\subseteq V_\beta$ such that $\tau_G=x,$ which is what we wanted.


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.