content/set-theory/replacement/replacement.tex
1% Part: set-theory2% Chapter: replacement34\documentclass[../../../include/open-logic-chapter]{subfiles}56\begin{document}78\olchapter{sth}{replacement}{Replacement}910\olimport{introduction}11\olimport{strength}12\olimport{extrinsic}13\olimport{limofsize}14\olimport{absinf}15\olimport{ref}16\olimport{refproofs}17\olimport{finiteaxiomatizability}1819\OLEndChapterHook2021\end{document}
content/set-theory/replacement/introduction.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{replacement}{intro}6\olsection{Introduction}78Replacement is the axiom scheme which makes the difference between $\ZF$9and~$\Z$. We helped ourselves to it throughout10\crefrange{sth:ordinals::chap}{sth:spine::chap}. In this chapter, we11will finally consider the question: is Replacement justified? 1213To make the question sharp, it is worth observing that Replacement is really14rather \emph{strong}. We will get a sense of just how strong it is, during this chapter (and again in \olref[sth][card-arithmetic][fix]{sec}). But this will suggest that justification really is required. 1516We will discuss two kinds of justification. Roughly: an \emph{extrinsic} justification is an attempt to justify an axiom by its fruits; an \emph{intrinsic} justification is an attempt to justify an axiom by suggesting that it is vindicated by the mathematical concepts in question. We will get a greater sense of what this means during this chapter, but it is just the tip of an iceberg. For more, see in particular \citeauthor{Maddy1988a} (\citeyear{Maddy1988a} and \citeyear{Maddy1988b}).1718\end{document}
content/set-theory/replacement/strength.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{replacement}{strength}6\olsection{The Strength of Replacement}78We begin with a simple observation about the strength of Replacement: unless we go beyond $\Z$, we cannot prove the existence of any von Neumann9ordinal greater than or equal to $\omega + \omega$. 1011Here is a sketch of12why. Working in~$\ZF$, consider the set $V_{\omega+\omega}$. This set acts13as the domain for a \emph{model} for~$\Z$. To see this, we introduce some notation for the \emph{relativization} of a formula: 14\begin{defn}\ollabel{formularelativization}15 For any set $M$, and any formula $\phi$, let $\phi^M$ be the formula which results by restricting all of $\phi$'s quantifiers to $M$. That is, replace ``$\lexists[x]$'' with ``$(\lexists[x \in M])$'', and16 replace ``$\lforall[x]$'' with ``$(\lforall[x \in M])$''. 17\end{defn}\noindent18It can be shown that, for every axiom $\phi$ of~$\Z$, we have that $\ZF \vdash19\phi^{V_{\omega+\omega}}$. But $\omega+\omega$ is not \emph{in}20$V_{\omega+\omega}$, by \olref[spine][rank]{ordsetrankalpha}. So $\Z$ is21consistent with the non-existence of $\omega+\omega$.2223This is why we said, in \olref[ordinals][replacement]{sec}, that24\olref[ordinals][ordtype]{thmOrdinalRepresentation} cannot be proved25without Replacement. For it is easy, within~$\Z$, to define an26explicit well-ordering which intuitively \emph{should} have order-type27$\omega+\omega$. Indeed, we gave an informal example of this in28\olref[ordinals][idea]{sec}, when we presented the ordering on the29natural numbers given by:30\begin{align*}31 n \lessdot m \text{ iff }&\text{either }n < m\text{ and }m-n\text{ is even,}\\32 & \text{or $n$ is even and $m$ is odd.}33\end{align*}34But if $\omega+\omega$ does not exist, this well-ordering is not35isomorphic to any ordinal. So $\Z$ does \emph{not} prove36\olref[ordinals][ordtype]{thmOrdinalRepresentation}. 3738Flipping things around: Replacement allows us to prove the existence39of $\omega+\omega$, and hence must allow us to prove the existence of40$V_{\omega+\omega}$. And not just that. For \emph{any} well-ordering41we can define, \olref[ordinals][ordtype]{thmOrdinalRepresentation}42tells us that there is some $\alpha$ isomorphic with that43well-ordering, and hence that $V_\alpha$ exists. In a straightforward44way, then, Replacement guarantees that the hierarchy of sets must be45\emph{very tall}. 4647Over the next few sections, and then again in48\olref[card-arithmetic][fix]{sec}, we'll get a better sense of better49just \emph{how} tall Replacement forces the hierarchy to be. The50simple point, for now, is that Replacement really \emph{does} stand in51need of justification!5253\end{document}
content/set-theory/replacement/extrinsic.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{replacement}{extrinsic}6\olsection[Extrinsic Considerations]{Extrinsic Considerations about Replacement}78We start by considering an \emph{extrinsic} attempt to justify9Replacement. Boolos suggests one, as follows. 10\begin{quote}11 [\ldots] the reason for adopting the axioms of replacement is quite12 simple: they have many desirable consequences and (apparently) no13 undesirable ones. In addition to theorems about the iterative14 conception, the consequences include a satisfactory if not ideal15 theory of infinite numbers, and a highly desirable result that16 justifies inductive definitions on well-founded relations.17 \citep[229]{Boolos1971}18\end{quote} 19The gist of Boolos's idea is that we should justify Replacement by its20fruits. And the specific fruits he mentions are the things we have21discussed in the past few chapters. Replacement allowed us to prove22that the von Neumann ordinals were excellent surrogates for the idea23of a well-ordering type (this is our ``satisfactory if not ideal24theory of infinite numbers''). Replacement also allowed us to define25the $V_\alpha$s, establish the notion of rank, and prove26$\in$-Induction (this amounts to our ``theorems about the iterative27conception''). Finally, Replacement allows us to prove the Transfinite28Recursion Theorem (this is the ``inductive definitions on well-founded29relations''). 3031These are, indeed, desirable consequences. But do these desirable32consequences suffice to \emph{justify} Replacement? \emph{No}. Or at33least, not straightforwardly. 3435Here is a simple problem. Whilst we have stated some desirable36consequences of Replacement, we could have obtained many of them via37other means. This is not as well known as it ought to be, though, so we should pause to explain the situation. 3839There is a simple theory of sets, Level Theory, or $\LT$ for short.\footnote{The first versions of $\LT$ are offered by \citet{Montague1965} and \citet{Scott1974}; this was simplified, and given a book-length treatment, by \citet{Potter2004}; and \citet{ButtonLT1} has recently simplified $\LT$ further.} $\LT$'s axioms are just Extensionality, Separation, and the claim that every set is a subset of some \emph{level}, where ``level'' is cunningly defined so that the levels behave like our friends, the $V_\alpha$s. So $\ZF$ proves $\LT$; but $\LT$ is \emph{much} weaker than $\ZF$. In fact, $\LT$ does not give you Pairs, Powersets, Infinity, or Replacement. Let $\Zr$ be the result of adding Infinity and Powersets to $\LT$; this delivers Pairs too, so, $\Zr$ is at least as strong as $\Z$. But, in fact, $\Zr$ is strictly stronger than $\Z$, since it adds the claim that every set has a rank (hence my suggestion that we call it $\Zr$). Indeed, $\Zr$ delivers: a perfectly satisfactory theory of ordinals;40results which stratify the hierarchy into well-ordered stages; a proof41of $\in$-Induction; and a \emph{version} of Transfinite Recursion. 4243In44short: although Boolos didn't know this, all of the desirable45consequences which he mentions could have been arrived at46\emph{without} Replacement; he simply needed to use $\Zr$ rather than $\Z$. 4748(Given all of this, why did we follow the conventional route, of49teaching you $\ZF$, rather than $\LT$ and $\Zr$? There are two reasons. First: for purely historical reasons, starting with $\LT$ is rather nonstandard; we wanted to50equip you to be able to read more standard discussions of set theory. Second: when you are ready to51appreciate $\LT$ and $\Zr$, you can simply read \citealt{Potter2004} and \citealt{ButtonLT1}.)5253Of course, since $\Zr$ is strictly weaker than $\ZF$, there are results which54$\ZF$ proves which $\Zr$ leaves open. So one could try to justify55Replacement on extrinsic grounds by pointing to one of these results.56But, once you know how to use $\Zr$, it is quite hard to find many57examples of things that are (a) settled by Replacement but not58otherwise, and (b) are intuitively true. (For more on this, see59\citealt[\S13.2]{Potter2004}.)6061The bottom line is this. To provide a compelling extrinsic62justification for Replacement, one would need to find a result which63\emph{cannot} be achieved without Replacement. And that's not an easy64enterprise. 6566Let's consider a further problem which arises for any attempt to offer67a purely extrinsic justification for Replacement. (This problem is68perhaps more fundamental than the first.) Boolos does not just point69out that Replacement has many desirable consequences. He also states70that Replacement has ``(apparently) no undesirable'' consequences. But71this parenthetical caveat, ``apparently,'' is surely absolutely72crucial.7374Recall how we ended up here: Na\"ive Comprehension ran into75inconsistency, and we responded to this inconsistency by embracing the76cumulative-iterative conception of set. This conception comes equipped77with a story which, we hope, assures us of its consistency. But if we78cannot justify Replacement from within that story, then we have (as79yet) no reason to believe that $\ZF$ is consistent. Or rather: we have80no reason to believe that $\ZF$ is consistent, apart from the (perhaps81merely contingent) fact that no one has discovered a contradiction82\emph{yet}. In exactly that sense, Boolos's comment seems to come down83to this: ``(apparently) $\ZF$ is consistent''. We should demand84greater reassurance of consistency than this. 8586This issue will affect any \emph{purely} extrinsic attempt to justify87Replacement, i.e., any justification which is couched solely in terms88of the (known) consequences of $\ZF$. As such, we will want to look89for an \emph{intrinsic} justification of Replacement, i.e., a90justification which suggests that the story which we told about sets91somehow ``already'' commits us to Replacement. 9293\end{document}
content/set-theory/replacement/limofsize.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{replacement}{limofsize}6\olsection{Limitation-of-size}78Perhaps the most common attempt to offer an ``intrinsic'' justification of9Replacement comes via the following notion:10\begin{enumerate}11 \item[] \limofsize. Any things form a set, provided that there are12 not too many of them.13\end{enumerate}14This principle will immediately vindicate Replacement. After all, any15set formed by Replacement cannot be any larger than any set from which16it was formed. Stated precisely: suppose you form a set17$\funimage{\tau}{A} = \Setabs{\tau(x)}{x \in A}$ using Replacement;18then $\cardle{\funimage{\tau}{A}}{A}$; so if the !!{element}s of $A$19were not too numerous to form a set, their images are not too numerous20to form $\funimage{\tau}{A}$. 2122The obvious difficulty with invoking \limofsize{} to justify23Replacement is that we have \emph{not} yet laid down any principle24like \limofsize. Moreover, when we told our story about the25cumulative-iterative conception of set in26\crefrange{sth:story::chap}{sth:z::chap}, nothing ever \emph{hinted}27in the direction of \limofsize. This, indeed, is precisely why Boolos28at one point wrote: ``Perhaps one may conclude that there are at least29two thoughts `behind' set theory'' \citeyearpar[p.~19]{Boolos1989}. On30the one hand, the ideas surrounding the cumulative-iterative31conception of set are meant to vindicate~$\Z$. On the other hand,32\limofsize{} is meant to vindicate Replacement. 3334But the issue it is not just that we have thus far been \emph{silent}35about \limofsize. Rather, the issue is that \limofsize{} (as just36formulated) seems to sit quite badly with the cumulative-iterative37notion of set. After all, it mentions nothing about the idea of sets38as formed in \emph{stages}.3940This is really not much of a surprise, given the history of these41``two thoughts'' (i.e., the cumulative-iterative conception of set,42and \limofsize). These ``two thoughts'' ultimately amount to two43rather different projects for blocking the set-theoretic paradoxes.44The cumulative-iterative notion of set blocks Russell's paradox by45saying, roughly: \emph{we should never have expected a Russell set to46exist, because it would not be ``formed'' at any stage}. By contrast,47\limofsize{} is meant to rule out the Russell set, by saying, roughly:48\emph{we should never have expected a Russell set to exist, because it49would have been too big}. 5051Put like this, then, let's be blunt: considered as a reply to the52paradoxes, \limofsize{} stands in need of \emph{much} more53justification. Consider, for example, this version of Russell's54Paradox: \emph{no pug sniffs exactly the pugs which don't sniff55themselves} (see \olref[sth][story][rus]{sec}). If you ask ``why is there no such pug?'', it is not a56good answer to be told that such a pug would have to sniff too many57pugs. So why would it be a good intuitive explanation, of the58non-existence of a Russell set, that it would have to be ``too big''59to exist? 6061In short, it's forgivable if you are a bit mystified concerning the ``intuitive''62motivation for \limofsize. 6364\end{document}
content/set-theory/replacement/absinf.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{replacement}{absinf}6\olsection{Replacement and ``Absolute Infinity''}78We will now put \limofsize{} behind us, and explore a different9family of (intrinsic) attempts to justify Replacement, which do take10seriously the idea of the sets as formed in stages.1112When we first outlined the iterative process, we offered some13principles which explained what happens at each stage. These were14\stageshier, \stagesord, and \stagesacc. Later, we added some15principles which told us something about the number of stages:16\stagessucc{} told us that the process of set-formation never ends,17and \stagesinf{} told us that the process goes through an infinite-th18stage. 1920It is reasonable to suggest that these two latter principles fall out21of some a broader principle, like:22\begin{enumerate}23 \item[] \stagesinex. There are absolutely infinitely many stages;24 the hierarchy is as tall as it could possibly be.25\end{enumerate}26Obviously this is an informal principle. But even if it is not27immediately \emph{entailed} by the cumulative-iterative conception of28set, it certainly seems \emph{consonant} with it. At the very least,29and unlike \limofsize, it retains the idea that sets are formed30stage-by-stage. 3132The hope, now, is to leverage \stagesinex{} into a justification of33Replacement. So let us see how this might be done. 3435In \olref[ordinals][idea]{sec}, we saw that it is easy to36construct a well-ordering which (morally) should be isomorphic to37$\omega+\omega$. Otherwise put, we can easily imagine a stage-by-stage38iterative process, whose order-type (morally) is $\omega+\omega$. As39such, if we have accepted \stagesinex, then we should surely accept40that there is at least an $\omega+\omega$-th stage of the hierarchy,41i.e., $V_{\omega+\omega}$, for the hierarchy surely \emph{could}42continue thus far. 4344This thought generalizes as follows: for any well-ordering, the45process of building the iterative hierarchy should run at least as far46as that well-ordering. And we could guarantee this, just by treating47\olref[ordinals][ordtype]{thmOrdinalRepresentation} as an48\emph{axiom}. This would tell us that any well-ordering is isomorphic49to a von Neumann ordinal. Since each von Neumann ordinal will be equal50to its own rank, \olref[ordinals][ordtype]{thmOrdinalRepresentation}51will then tell us that, whenever we can describe a well-ordering in52our set theory, the iterative process of set building must outrun that53well-ordering. 5455This idea certainly seems like a corollary of \stagesinex.56Unfortunately, if our aim is to extract Replacement from this idea,57then we face a simple, technical, barrier: Replacement is strictly stronger than58\olref[ordinals][ordtype]{thmOrdinalRepresentation}. (This observation is made by 59\citet[\S13.2]{Potter2004}; we will prove it in \olref[sth][replacement][finiteaxiomatizability]{sec}.)6061The upshot is that, if we are going to understand \stagesinex{} in62such a way as to yield Replacement, then it cannot \emph{merely} say63that the hierarchy outruns any well-ordering. It must make a stronger64claim than that. To this end, \cite{Shoenfield:AST} proposed a very65natural strengthening of the idea, as follows: the hierarchy is not66\emph{cofinal} with any set.\footnote{G\"odel seems to have proposed a67similar thought; see \citet[p.~223]{Potter2004}. For discussion of G\"odel and \citeauthor{Shoenfield:AST}, see \citet[90--5]{Incurvati2020}.} In slightly more detail:68if $\tau$ is a mapping which sends sets to stages of the hierarchy,69the image of any set $A$ under $\tau$ does not exhaust the hierarchy.70Otherwise put (schematically): 71\begin{enumerate}72 \item[] \stagescofin. If $A$ is a set and $\tau(x)$ is a stage for73 every $x \in A$, then there is a stage which comes after each74 $\tau(x)$ for $x \in A$.75\end{enumerate}76It is obvious that $\ZF$ proves a suitably formalised version of77\stagescofin. Conversely, we can informally argue that \stagescofin{}78justifies Replacement.\footnote{It would be harder to prove79Replacement using some formalisation of \stagescofin, since $\Z$ on80its own is not strong enough to define the stages, so it is not clear81how one would formalise \stagescofin. One option, though, is to82work in some extension of $\LT$, as discussed in \olref[sth][replacement][extrinsic]{sec}.} For suppose $(\forall x \in A)\lexists![y][\phi(x,y)]$. Then for each $x \in A$, let $\sigma(x)$ be the $y$ such83that $\phi(x,y)$, and let $\tau(x)$ be the stage at which $\sigma(x)$84is first formed. By \stagescofin, there is a stage $V$ such that85$(\forall x \in A)\tau(x)\in V$. Now since each $\tau(x) \in V$ and86$\sigma(x) \subseteq \tau(x)$, by Separation we can obtain $\Setabs{y87\in V}{(\exists x \in A)\sigma(x) = y} = \Setabs{y}{(\exists x \in88A)\phi(x,y)}$.8990\begin{prob}91 Formalize \stagescofin{} within $\ZF$.92\end{prob}9394So \stagescofin{} vindicates Replacement. And it is at least plausible95that \stagesinex{} vindicates \stagescofin. For suppose \stagescofin{}96fails. So the hierarchy is cofinal with some set~$A$, i.e., we have a97map $\tau$ such that for any stage~$S$ there is some $x \in A$ such98that $S \in \tau(x)$. In that case, we do have a way to get a handle99on the supposed ``absolute infinity'' of the hierarchy: it is100\emph{exhausted} by the range of $\tau$ applied to $A$. And that101compromises the thought that the hierarchy is ``absolutely infinite''.102Contraposing: \stagesinex{} entails \stagescofin, which in turn103justifies Replacement.104105This represents a genuinely promising attempt to provide an106intrinsic justification for Replacement. But whether it ultimately107works, or not, we will have to leave to you to decide.108109\end{document}
content/set-theory/replacement/ref.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{replacement}{ref}6\olsection{Replacement and Reflection}78Our last attempt to justify Replacement, via \stagesinex, begins with a deep and lovely result:\footnote{A reminder: all formulas can have parameters (unless explicitly stated otherwise).}9%In this, I will use overlining, such as $x_1, \ldots, x_n$, to10%abbreviate ``$x_1, \ldots, x_n$'': 11\begin{thm}[Reflection Schema]\ollabel{reflectionschema}12For any formula $\phi$:13\[14\forall \alpha \exists \beta > \alpha (\forall x_1 \ldots, x_n \in15V_\beta)(\phi(x_1, \ldots, x_n) \liff \phi^{V_\beta}(x_1, \ldots, x_n))16\]17\end{thm}18\noindent 19As in \olref[sth][replacement][strength]{formularelativization}, $\phi^{V_\beta}$ is the result of restricting every20quantifier in $\phi$ to the set~$V_\beta$. So, intuitively, Reflection21says this: if $\phi$ is true in the entire hierarchy, then $\phi$ is22true in arbitrarily many \emph{initial segments} of the hierarchy. 2324\citet{Montague1961} and \citet{Levy1960} showed that (suitable25formulations of) Replacement and Reflection are equivalent,26modulo~$\Z$, so that adding either gives you~$\ZF$. (We prove these results in \olref[sth][replacement][refproofs]{sec}.) Given this27equivalence, one might hope to justify Reflection and Replacement via28\stagesinex{} as follows: given \stagesinex, the hierarchy should be29very, very tall; so tall, in fact, that nothing we can say about it is30sufficient to bound its height. And we can understand this as the31thought that, if any sentence~$\phi$ is true in the entire hierarchy,32then it is true in arbitrarily many initial segments of the hierarchy.33And that is just Reflection. 3435Again, this seems like a genuinely promising attempt to provide an36intrinsic justification for Replacement. But there is much too much to37say about it here. You must now decide for yourself whether it38succeeds.\footnote{Though you might like to continue by reading \citet[95--100]{Incurvati2020}.}3940\end{document}
content/set-theory/replacement/refproofs.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{replacement}{refproofs}6\olsection{Appendix: Results surrounding Replacement}78In this section, we will prove Reflection within $\ZF$. We will also9prove a sense in which Reflection is equivalent to Replacement. And we10will prove an interesting consequence of all this, concerning the11strength of Reflection/Replacement. \emph{Warning: this is easily the12most advanced bit of mathematics in this textbook.} 1314We'll start with a lemma which, for brevity, employs the notational15device of \emph{overlining} to deal with sequences of variables or16objects. So: ``$\overline{a}_k$'' abbreviates ``$a_{k_1}$, \dots,17$a_{k_n}$'', where $n$ is determined by context.1819\begin{lem}\ollabel{lemreflection}20For each $1 \leq i \leq k$, let $\phi_i(\overline{v}_i, x)$ be a21formula. Then for each $\alpha$22there is some $\beta > \alpha$ such that, for any $\overline{a}_1,23\ldots, \overline{a}_k \in V_\beta$ and each $1 \leq i \leq k$:24\[25 \exists x\phi_i(\overline{a}_i, x) \rightarrow (\exists x \in V_\beta) \phi_i(\overline{a}_i, x)26\]27\end{lem}2829\begin{proof}30We define a term $\mu$ as follows: $\mu(\overline{a}_1, \ldots,31\overline{a}_k)$ is the least stage, $V$, which satisfies all of the32following conditionals, for $1 \leq i \leq k$:33\[34\exists x\phi_i(\overline{a}_i, x) \rightarrow (\exists x \in V) \phi_i(\overline{a}_i, x))35\]36It is easy to confirm that $\mu(\overline{a}_1, \ldots, \overline{a}_k)$ exists for all $\overline{a}_1, \ldots, \overline{a}_k$. Now, using Replacement and our recursion theorem, define:37\begin{align*}38 S_0 & = V_{\alpha+1}\\39 S_{n+1} & = S_n \cup \bigcup40 \Setabs{\mu(\overline{a}_1, \ldots, \overline{a}_k)}41 {\overline{a}_1, \ldots, \overline{a}_k \in S_n} \\42 S &= \bigcup_{m < \omega} S_n.43\end{align*}44Each $S_n$, and hence $S$ itself, is a stage after $V_\alpha$. Now fix45$\overline{a}_1$, \dots,~$\overline{a}_k \in S$; so there is some $n <46\omega$ such that $\overline{a}_1$, \dots, $\overline{a}_k \in S_n$.47Fix some $1 \leq i \leq k$, and suppose that $\exists x48\phi_i(\overline{a}_i,x)$. So $(\exists x \in \mu(\overline{a}_1,49\ldots, \overline{a}_k))\phi_i(\overline{a}_i, x)$ by construction, so50$(\exists x \in S_{n+1})\phi_i(\overline{a}_i, x)$ and hence51$(\exists x \in S)\phi_i(\overline{a}_i, x)$. So $S$ is our $V_\beta$.52\end{proof}53\noindent 54We can now prove \olref[sth][replacement][ref]{reflectionschema} quite55straightforwardly:5657\begin{proof}[Proof] 58Fix $\alpha$. Without loss of generality, we can assume $\phi$'s only59connectives are $\exists$, $\lnot$ and $\land$ (since these are60expressively adequate). Let $\psi_1, \ldots, \psi_k$ enumerate each of61$\phi$'s subformulas according to complexity, so that $\psi_k = \phi$.62By \olref{lemreflection}, there is a $\beta > \alpha$ such that, for63any $\overline{a}_i \in V_\beta$ and each $1 \leq i \leq k$:64\begin{align}\label{reflectionnicelybehaved}65 \exists x\psi_i(\overline{a}_i, x) \rightarrow 66 (\exists x \in V_\beta) \psi_i(\overline{a}_i, x)\tag{*}67\end{align}68By induction on complexity of $\psi_i$, we will show that69$\psi_i(\overline{a}_i) \leftrightarrow70\psi_i^{V_\beta}(\overline{a}_i)$, for any $\overline{a}_i \in71V_\beta$. If $\psi_i$ is atomic, this is trivial. The biconditional72also establishes that, when $\psi_i$ is a negation or conjunction of73subformulas satisfying this property, $\psi_i$ itself satisfies this74property. So the only interesting case concerns quantification. Fix75$\overline{a}_i \in V_\beta$; then:76\begin{align*}77 (\exists x \psi_i(\overline{a}_i, x))^{V_\beta}78 &\text{ iff }79 (\exists x \in V_\beta)\psi_i^{V_\beta}(\overline{a}_i, x)80 &&\text{by definition}\\81 &\text{ iff }82 (\exists x \in V_\beta)\psi_i(\overline{a}_i, x)83 &&\text{by hypothesis}\\84 &\text{ iff }85 \exists x \psi_i(\overline{a}_i, x)86 &&\text{by \eqref{reflectionnicelybehaved}}87\end{align*}88This completes the induction; the result follows as $\psi_k = \phi$.89\end{proof}9091We have proved Reflection in $\ZF$. Our proof essentially92followed \citet{Montague1961}. We now want to prove in $\Z$ that93Reflection entails Replacement. The proof follows \citet{Levy1960},94but with a simplification. 9596Since we are working in $\Z$, we cannot present Reflection in exactly97the form given above. After all, we formulated Reflection using the98``$V_\alpha$'' notation, and that cannot be defined in $\Z$ (see99\olref[sth][spine][zf]{sec}). So instead we will offer an apparently100weaker formulation of Replacement, as follows:101102\begin{defish}103\emph{Weak-Reflection.} For any formula $\phi$, there is a transitive104set $S$ such that $0$, $1$, and any parameters to $\phi$ are105!!{element}s of $S$, and $(\forall \overline{x} \in S)(\phi \liff106\phi^S)$.107\end{defish}108109To use this to prove Replacement, we will first follow \citet[first110part of Theorem 2]{Levy1960} and show that we can ``reflect'' two111formulas at once:112113\begin{lem}[in $\Z + \text{Weak-Reflection}$.]\ollabel{lem:reflect}114For any formulas $\psi, \chi$, there is a transitive set $S$ such that115$0$ and $1$ (and any parameters to the formulas) are !!{element}s of116$S$, and $(\forall \overline{x} \in S)((\psi \liff \psi^S) \land (\chi117\liff \chi^S))$.118\end{lem}119120\begin{proof}121Let $\phi$ be the formula $(z = 0 \land \psi) \lor (z = 1 \land \chi)$. 122123Here we use an abbreviation; we should spell out ``$z = 0$'' as124``$\forall t\, t \notin z$'' and ``$z =1$'' as ``$\forall s(s \in z125\liff \forall t\, t \notin s)$''. But since $0, 1 \in S$ and $S$ is126transitive, these formulas are \emph{absolute} for $S$; that is, they127will apply to the same object whether we restrict their quantifiers to128$S$.\footnote{More formally, letting $\xi$ be either of these129formulas, $\xi(z) \liff \xi^S(z)$.}130131By Weak-Reflection, we have some appropriate $S$ such that:132\begin{align*}133 (\forall z, \overline{x} \in S)(&\phi \liff \phi^S)\\134 \text{i.e. }(\forall z, \overline{x} \in S)(&((z = 0 \land \psi) \lor (z = 1 \land \chi)) \liff {}\\135 &\phantom{(}((z = 0 \land \psi) \lor (z = 1 \land \chi))^S)\\136 \text{i.e. }(\forall z, \overline{x} \in S)(&((z = 0 \land \psi) \lor (z = 1 \land \chi))\liff {}\\137 &\phantom{(}((z = 0 \land \psi^S) \lor (z = 1 \land \chi^S)))\\138 \text{i.e. }(\forall \overline{x} \in S)(&(\psi \liff \psi^S) \land (\chi \liff \chi^S))139\end{align*}140The second claim entails the third because ``$z = 0$'' and ``$z=1$''141are absolute for $S$; the fourth claim follows since $0 \neq 1$.142\end{proof}\noindent We can now obtain Replacement, just by following and simplifying 143\citet[Theorem 6]{Levy1960}:144145\begin{thm}[in $\Z$ + Weak-Reflection]\label{thm:replacement} 146For any formula $\phi(v,w)$, and any $A$, if $(\forall x \in A)\lexists![y][\phi(x,y)]$, then147$\Setabs{y}{(\exists x \in A)\phi(x,y}$ exists.148\end{thm}149150\begin{proof}151Fix $A$ such that $(\forall x \in A)\lexists![y][\phi(x,y)]$, and152define formulas:153\begin{align*}154 \psi &\text{ is } (\phi(x, z) \land A = A)\\155 \chi &\text{ is } \lexists[y][\phi(x, y)]156\end{align*}157Using \olref{lem:reflect}, since $A$ is a parameter to $\psi$, there158is a transitive~$S$ such that $0, 1, A \in S$ (along with any other159parameters), and such that:160\[161 (\forall x,z \in S)((\psi \liff \psi^S) \land (\chi \liff \chi^S))162\]163So in particular:164\begin{align*}165 (\forall x, z \in S)(&\phi(x, z) \liff \phi^S(x, z))\\166 (\forall x \in S)(&\exists y\phi(x, y) \liff (\exists y \in S)\phi^S(x, y)) 167\end{align*}168Combining these, and observing that $A \subseteq S$ since $A \in S$ and $S$ is transitive:169\begin{align*}170 (\forall x \in A)(&\exists y\phi(x, y) \liff (\exists y \in S)\phi(x, y))171\end{align*}172Now $(\forall x \in A)(\lexists![y \in S])\phi(x, y)$, because173$(\forall x \in A)\lexists![y][\phi(x, y)]$. Now Separation yields174$\Setabs{y \in S}{(\exists x \in A) \phi(x, y)} = \Setabs{y}{(\exists175x \in A) \phi(x, y)}$. 176\end{proof}177178\end{document}
content/set-theory/replacement/finiteaxiomatizability.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}4 5\olfileid{sth}{replacement}{finiteaxiomatizability}6\olsection{Appendix: Finite axiomatizability}7We close this chapter by extracting some results from Replacement. The first result is due to \citet{Montague1961}; note that it is not a proof \emph{within} $\ZF$, but a proof \emph{about} $\ZF$:8\begin{thm}\ollabel{zfnotfinitely}9 $\ZF$ is not finitely axiomatizable. More generally: if $\Th{T}$ is finite and $\Th{T} \Proves \ZF$, then $\Th{T}$ is inconsistent.10 11 (Here, we tacitly restrict ourselves to first-order sentences whose only non-logical primitive is $\in$, and we write $\Th{T} \Proves \ZF$ to indicate that $\Th{T} \Proves \phi$ for all $\phi \in \ZF$.)12\end{thm}13\begin{proof}14 Fix finite $\Th{T}$ such that $\Th{T} \Proves \ZF$. So, $\Th{T}$ proves Reflection, i.e.\ \olref[sth][replacement][ref]{reflectionschema}. Since $\Th{T}$ is finite, we can rewrite it as a single conjunction, $\theta$. Reflecting with this formula, $\Th{T} \Proves \exists \beta(\theta \liff \theta^{V_\beta})$. Since trivially $\Th{T} \Proves \theta$, we find that $\Th{T} \Proves \exists \beta\ \theta^{V_\beta}$. 15 16 Now, let $\psi(X)$ abbreviate:17 \[18 \theta^X \land X\text{ is transitive} \land (\forall Y \in X)(Y\text{ is transitive}\lif \lnot \theta^{Y})19 \]20 roughly this says: $X$ is a transitive model of $\theta$, and $\in$-minimal in this regard. Now, recalling that $\Th{T} \Proves \exists \beta\ \theta^{V_\beta}$, by basic facts about ranks within $\ZF$ and hence within $\Th{T}$, we have:21 \begin{equation}22 \Th{T} \Proves \exists M \psi(M). \tag{*}\label{Mpsi}23 \end{equation}24 Using the first conjunct of $\psi(X)$, whenever $\Th{T} \Proves \sigma$, we have that $\Th{T} \Proves \forall X(\psi(X) \lif \sigma^X)$. So, by \eqref{Mpsi}:25 \begin{align*}26 \Th{T} &\Proves \forall X(\psi(X) \lif (\exists N \psi(N))^X)\\27 \intertext{Using this, and \eqref{Mpsi} again:}28 \Th{T} &\Proves \exists M(\psi(M) \land (\exists N \psi(N))^M)29 \intertext{In particular, then:}30 \Th{T} &\Proves \exists M(\psi(M) \land (\exists N \in M)((N\text{ is transitive})^N \land (\theta^N)^M))31 \intertext{So, by elementary reasoning concerning transitivity:}32 \Th{T} &\Proves \exists M(\psi(M) \land (\exists N \in M)(N\text{ is transitive} \land \theta^N))33 \end{align*} 34 So that $\Th{T}$ is inconsistent.\footnote{This ``elementary reasoning'' involves proving certain ``absoluteness facts'' for transitive sets.}35\end{proof}3637Here is a similar result, noted by \citet[223]{Potter2004}:3839\begin{prop}\ollabel{finiteextensionofZ}40 Let $\Th{T}$ extend $\Z$ with finitely many new axioms. If $\Th{T} \Proves \ZF$, then $\Th{T}$ is inconsistent. (Here we use the same tacit restrictions as for \olref{zfnotfinitely}.)41\end{prop}42\begin{proof}43 Use $\theta$ for the conjunction of all of $\Th{T}$'s axioms \emph{except} for the (infinitely many) instances of Separation. Defining $\psi$ from $\theta$ as in \olref{zfnotfinitely}, we can show that $\Th{T} \Proves \exists M \psi(M)$. 44 45 As in \olref{zfnotfinitely}, we can establish the schema that, whenever $\Th{T} \Proves \sigma$, we have that $\Th{T} \Proves \forall X(\psi(X) \lif \sigma^X)$. We then finish our proof, exactly as in \olref{zfnotfinitely}.46 47 However, establishing the schema involves a little more work than in \olref{zfnotfinitely}. After all, the Separation-instances are in $\Th{T}$, but they are not conjuncts of $\theta$. However, we can overcome this obstacle by proving that $\Th{T} \Proves \forall X(X\text{ is transitive} \lif \sigma^X)$, for every Separation-instance $\sigma$. We leave this to the reader. 48\end{proof}49\begin{prob}50 Show that, for every Separation-instance $\sigma$, we have: $\Z \Proves \forall X(X\text{ is transitive} \lif \sigma^X)$. (We used this schema in \olref[sth][replacement][finiteaxiomatizability]{finiteextensionofZ}.)51\end{prob}52\begin{prob}53 Show that, for every $\phi \in \Z$, we have $\ZF \Proves \phi^{V_{\omega+\omega}}$.54\end{prob}55\begin{prob}56 Confirm the remaining schematic results invoked in the proofs of \olref[sth][replacement][finiteaxiomatizability]{zfnotfinitely} and \olref[sth][replacement][finiteaxiomatizability]{finiteextensionofZ}.57\end{prob}5859As remarked in \olref[sth][replacement][absinf]{sec}, this shows that Replacement is strictly stronger than60\olref[ordinals][ordtype]{thmOrdinalRepresentation}. Or, slightly more strictly: if $\Z$ + ``every well-ordering is isomorphic to a unique ordinal'' is consistent, then it fails to prove some Replacement-instance.616263 % By assumption, $\psi^{V_\alpha}$ has the form: 64 % \[65 % (\forall A \in V_\alpha)(\exists S \in V_\alpha)(\forall x \in V_\alpha)(x \in S \liff (\phi^{V_\alpha}(x) \land x \in A))66 % \]67 % To establish this holds, fix $A \in V_\alpha$. Using Separation, obtain: 68 % \[69 % S = \Setabs{x \in A}{\phi^{V_\alpha}(x)}70 % \]71 % Now $S \in V_\alpha$, since $S \subseteq A \in V_\alpha$, and clearly $(\forall x \in V_\alpha)(x \in S \liff (\phi^{V_\alpha}(x) \land x \in A)$.7273\end{document}