w23-logic-2/inputs/lecture_22.tex
Josia Pietsch ab14be172f
Some checks failed
Build latex and deploy / checkout (push) Failing after 14m16s
some small changes
2024-02-16 19:47:21 +01:00

254 lines
7.6 KiB
TeX

\lecture{22}{2024-01-22}{More Forcing}
\begin{warning}+
Forcing will not be relevant for the exam.
Because of a lack of time, this is more of an outlook
than a thorough presentation of the material.
\end{warning}
For the rest of the section, let us fix
a transitive model $M$ of $\ZFC$
a partial order $\mathbb{P}$
and an $M$-generic filter $g$.
\begin{definition}[$\mathbb{P}$-names]
For an ordinal $\alpha \in M$%
\footnote{Recall that $\Ord_M = \Ord \cap M$.},
let $M^{\mathbb{P}}_\alpha$,
the \vocab{$\mathbb{P}$-names} in $M$ of rank $\le \alpha$,
be defined as follows:
\[
\tau \in M_{\alpha}^{\mathbb{P}} :\iff
\tau \in M \land
\tau \subseteq \mathbb{P} \times \bigcup \{M_\beta^{\mathbb{P}}: \beta < \alpha\},
\]
i.e.~the elements of $\tau \in M_\alpha^{\mathbb{P}}$
are of the form $(p, \sigma)$,
where $p \in \cP$ and $\sigma \in M^{\mathbb{P}}_\beta$
for some $\beta < \alpha$.
Finally $M^{\mathbb{P}} = \bigcup \{M_\alpha^{\mathbb{P}} : \alpha \in M\}$.
\end{definition}
Let $R$ be the relation on $M^{\mathbb{P}}$ defined by
$\sigma R \tau$ iff $\exists p \in \mathbb{P}.~(p,\sigma) \in \tau$.
If $\tau \in M^\mathbb{P}$
and $(p, \sigma) \in \tau$,
then $\sigma \in \{p, \sigma\} \in (p,\sigma) \in \tau$,
so the relation $R$
is well founded.
\begin{definition}
Let $\tau \in M_\alpha^{\mathbb{P}}$.
Then $\tau^g$,
the \vocab{$g$-interpretation of $\tau$},
is defined to be
\[
\{\sigma^g : \exists p \in g .~(p,\sigma) \in \tau\}.
\]
\end{definition}
\begin{definition}
$M[g]$,
the forcing extension of $M$
given by $g$,
is
\[
\{ \tau^g : \tau \in M^{\mathbb{P}}\} .
\]
\end{definition}
\begin{lemma}
$M[g]$ is transitive.
\end{lemma}
\begin{proof}
Trivial!
\end{proof}
\begin{lemma}
$M \cup \{g\} \subseteq M[g]$.
\end{lemma}
\begin{proof}
For all $x \in M$ we need to find a name \vocab{$\check{x}$}
such that $\check{x}^g = x$.
We can recursively (along $\in$) define
\[
\check{x} = \{(p, \check{y}) : p \in \mathbb{P} \land y \in x\}.
\]
By induction, $\check{x} \in M$ for all $x \in M$.
\begin{claim}
$\check{x}^g = x$.
\end{claim}
\begin{subproof}
Recall that $\mathbb{P} \neq \emptyset$.
Inductively, we get
\begin{IEEEeqnarray*}{rCl}
\check{x}^g &=& \{\check{y}^g : \exists p \in g.~(p,\check{y}) \in \check{x}\}\\
&\overset{\text{induction}}{=}& \{y : \exists p \in g.~(p,\check{y}) \in \check{x}\}\\
&\overset{\text{definition of $\check{x}$}}{=}& \{y : y \in x\} = x.
\end{IEEEeqnarray*}
\end{subproof}
So $M \subseteq M[g]$.
We also need a name for $g$.
Let \vocab{$\dot{g}$}$ \coloneqq \{(p, \check{p}) : p \in \mathbb{P}\}$.
Indeed
\begin{IEEEeqnarray*}{rCl}
\dot{g}^g &=& \{\check{p}^g : \exists p \in g.~(p, \check{p}) \in \dot{g}\}\\
&=& \{p : p \in g\} = g.
\end{IEEEeqnarray*}
\end{proof}
\begin{lemma}
\label{lem:mgmodelexfundinfpairunion}
$M[g] \models \AxExt, \AxFund, \AxInf, \AxPair, \AxUnion$.
\end{lemma}
\begin{proof}
\begin{itemize}
\item \AxExt:
The formula $\forall x.~\forall y .~((\forall z \in x.~z \in y \land \forall z \in y.~z \in x) \to x = y)$
is $\Pi_1$, hence it is true in $M[g]$
by \yaref{lem:pi1downardsabsolute}.
\item \AxFund:
Again,
\[
\forall x.~(\exists y \in x .~y = y \to \exists y \in x.~\forall z \in y.~z \not\in x)
\]
is $\Pi_1$.
\item \AxInf
can be written as
\[
\exists x .~(\underbrace{\neq \in x \land \forall y \in x.y \cup \{y\} \in x}_{\Sigma_0}).
\]
We have $ \omega \in M \subseteq M[g]$,
so $M[g] \models \AxInf$.
\item \AxPair:
Let us assume $x,y \in M[g]$,
say $x = \tau^g$ and $y = \sigma^g$.
Let $\pi = \{(p,\tau) : p \in \mathbb{P} \} \cup \{(p,\sigma) : p \in \mathbb{P}\} \in M^{\mathbb{P}}$.
Then $\pi^g = \{\tau^g, \sigma^g\} = \{x,y\}$,
so $\{x,y\} \in M[g]$.
As a $\cL_\in$-statement,
$z = \{x,y\}$ is $\Sigma_0$,
so $M[g] \models \text{``$\{x,y\}$ is the pair of $x$ and $y$''}$.
Hence $M[g] \models \AxPair$.
\item \AxUnion:
Similar to \AxPair.
\end{itemize}
\end{proof}
Still missing are
\begin{itemize}
\item \AxPow,
\item \AxAus,
\item \AxRep,
\item \AxC.
\end{itemize}
\begin{definition}[\vocab{Forcing relation}]
Let $M$ be a countable transitive model of $\ZFC$
and let $\mathbb{P} \in M$ be a partial order.
Let $p \in \mathbb{P}$
and let $\phi$ be a $\cL_{\in }$-formula.
Let $\tau_1,\ldots, \tau_k \in M^{\mathbb{P}}$ be names.
We say that
$p$ \vocab[force]{forces} $\phi(\tau_1,\ldots, \tau_k)$,
\[
p \Vdash^{\mathbb{\cP}}_{M} \phi(\tau_1, \ldots, \tau_k),
\]
if for all $h \subseteq \mathbb{P}$
which are $\mathbb{P}$-generic over $M$
with $p \in h$,
\[
M[h] \models\phi(\tau_1^h, \ldots, \tau_k^h).
\]
\end{definition}
\begin{theorem}
Fix an $\cL_\in$-formula $\phi$.
Then the relation
\[
R = \{(p,\tau_1,\ldots,\tau_k : p \Vdash ^{\mathbb{P}}_M \phi(\tau_1,\ldots,\tau_k)\}
\]
is definable over $M$
(in the parameter $\mathbb{P}$).
\end{theorem}
\begin{proof}
Omitted.
\end{proof}
\begin{theorem}[\vocab{Forcing Theorem}]
Let $M$, $\mathbb{P}$, $g$,
be as above,
let $\phi$ be a formula,
and let $\tau_1, \ldots, \tau_k \in M^{\mathbb{P}}$.
Then the following are equivalent:
\begin{enumerate}[(1)]
\item $M[g] \models \phi(\tau_1^g, \ldots, \tau_k^g)$.
\item There is some $p \in g$ with
\[
p \Vdash^{\mathbb{P}}_M \phi(\tau_1,\ldots, \tau_k).
\]
\end{enumerate}
\end{theorem}
\begin{proof}
Omitted.
\end{proof}
\begin{theorem}
$M[g] \models \ZFC$.
\end{theorem}
\begin{proof}
We have already shown a part of this in
\yaref{lem:mgmodelexfundinfpairunion}.
Let us show that $M[g] \models \AxAus$,
the rest is similar and left as an exercise.%
\footnote{or done next semester in Logic IV!}
Let $\phi$ be a formula,
let $a, x_1,\ldots,x_k \in M[g]$.
We need to see
\[M[g] \models \exists y.~y = \{z \in a : \phi(z, x_1,\ldots,x_k)\}.\]
If suffices to show that there is some $y \in M[g]$
with $y = \{ z \in a : M[g] \models \phi(z, x_1,\ldots,x_k)\}$.
For this, let us construct a name for $y$.
Let $a = \tau^g$, $x_i = \sigma_i^g$.
Let
\[
\pi = \{(p,\rho) : \exists \overline{p} > p.~(\overline{p}, \rho) \in \tau \land
p \Vdash^{\mathbb{P}}_M \phi(\rho, \sigma_1, \ldots, \sigma_k)\}.
\]
We have $\pi \in M$, since the relation
$ \Vdash^{\mathbb{P}}_M$ can be defined in $M$.
Let $z \in a$ such that $M[g] \models \phi(z, x_1,\ldots,x_n)$.
We have $z = \rho^g$
for some $\rho$
and there is $\overline{p} \in g$ with $(\overline{p}, \rho) \in \pi$.
Now $M[g] = \phi(\rho^g, \sigma_1^g,\ldots\sigma_k^g)$.
Let $p' \Vdash^{\mathbb{P}}_M \phi(\rho, \sigma_1,\ldots, \sigma_k)$,
where $p' \in g$.
We have $p', \overline{p} \in g$,
so there is some $p \le p', \overline{p}$ with $p \in g$.
Then $(p,\rho) \in \pi$,
so $\rho^g \in \pi^g$.
This shows that
\[
\{z \in a : M[g] \models \phi(z,x_1,\ldots,x_k)\} \subseteq \pi^g.
\]
The other inclusion is easy.
\end{proof}