content/set-theory/ordinals/ordinals.tex
1% Chapter: Naive23\documentclass[../../../include/open-logic-chapter]{subfiles}45\begin{document}67\olchapter{sth}{ordinals}{Ordinals}89\olimport{introduction}10\olimport{idea}11\olimport{wo}12\olimport{iso}13\olimport{vn}14\olimport{basic}15\olimport{replacement}16\olimport{milestone}17\olimport{ordtype}18\olimport{opps}1920\OLEndChapterHook2122\end{document}
content/set-theory/ordinals/introduction.tex
1% Part: set-theory2% Chapter: ordinals3% Section: introduction45\documentclass[../../../include/open-logic-section]{subfiles}67\begin{document}89\olfileid{sth}{ordinals}{intro}1011\olsection{Introduction}1213In \olref[z][]{chap}, we postulated that there is an infinite-th stage14of the hierarchy, in the form of \stagesinf{} (see also our axiom of15Infinity). However, given \stagessucc{}, we can't stop at the16infinite-th stage; we have to keep going. So: at the next stage after17the first infinite stage, we form all possible collections of sets18that were available at the first infinite stage; and repeat; and19repeat; and repeat; \dots2021Implicitly what has happened here is that we have started to invoke an22``intuitive'' notion of number, according to which there can be23numbers \emph{after} all the natural numbers. In particular, the24notion involved is that of a \emph{transfinite ordinal}. The aim of25this chapter is to make this idea more rigorous. We will explore the26general notion of an ordinal, and then explicitly define certain sets27to be our ordinals. 2829\end{document}
content/set-theory/ordinals/idea.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{ordinals}{idea} 6\olsection{The General Idea of an Ordinal}78Consider the natural numbers, in their usual order:9\begin{center}10 \begin{tikzpicture}11 \foreach \x/\xtext in {0, 1, 2, 3, 4, 5}12 {13 \node (\x a) at (\x, 1) {\small{$\x$}};14 \node (\x a) at (\x.5, 1) {\small{$<$}};15 }16 \node (ldots) at (6, 1) {\small{$\ldots$}};17 \end{tikzpicture}18\end{center}19We call this, in the jargon, an $\omega$-sequence. And indeed, this20general ordering is mirrored in our initial construction of the stages21of the set hierarchy. But, now suppose we move $0$ to the end of this22sequence, so that it comes after all the other numbers:23\begin{center}24 \begin{tikzpicture}25 \foreach \x/\xtext in {1, 2, 3, 4, 5}26 {27 \node (\x a) at (\x, 1) {\small{$\x$}};28 \node (\x a) at (\x.5, 1) {\small{$<$}};29 }30 \node (ldots) at (6, 1) {\small{$\ldots$}};31 \node (bea) at (6.5, 1) {\small{$<$}};32 \node (noa) at (7, 1) {\small{$0$}};33 \end{tikzpicture}34\end{center}35We have the same entities here, but ordered in a fundamentally36different way: our first ordering had no last element; our new37ordering does. Indeed, our new ordering consists of an38$\omega$-sequence of entities ($1, 2, 3, 4, 5, \ldots$), followed by39another entity. It will be an $\omega+1$-sequence.4041We can generate even more types of ordering, using just these42entities. For example, consider all the even numbers (in their natural43order) followed by all the odd numbers (in their natural order):44\begin{center}45 \begin{tikzpicture}46 \node(a1) at (1,1) {\small{$0$}};47 \node(a2) at (1.5,1) {\small{$<$}};48 \node(a3) at (2,1) {\small{$2$}};49 \node(a4) at (2.5,1) {\small{$<$}};50 \node(a5) at (3,1) {\small{$4$}};51 \node(a6) at (3.5,1) {\small{$<$}};52 \node(dots) at (4,1) {\small{$\ldots$}};53 \node(aaoe) at (4.5,1) {\small{$<$}};54 \node(b1) at (5,1) {\small{$1$}};55 \node(b2) at (5.5,1) {\small{$<$}};56 \node(b3) at (6,1) {\small{$3$}};57 \node(b4) at (6.5,1) {\small{$<$}};58 \node(b5) at (7,1) {\small{$\ldots$}};59 \end{tikzpicture}60\end{center}61This is an $\omega$-sequence followed by another $\omega$-sequence; an62$\omega+\omega$-sequence. 6364Well, we can keep going. But what we would like is a general way to65understand this talk about \emph{orderings}. 6667\end{document}
content/set-theory/ordinals/wo.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{ordinals}{wo}6\olsection{Well-Orderings}78The fundamental notion is as follows:910\begin{defn}11The relation $<$ \emph{well-orders} $A$ iff it meets these two conditions:12\begin{enumerate}13 \item $<$ is connected, i.e., for all $a, b \in A$, either $a < b$14 or $a = b$ or $b < a$;15 \item every non-empty subset of $A$ has a $<$-minimal !!{element},16 i.e., if $\emptyset \neq X \subseteq A$ then $(\exists m \in17 X)(\forall z \in X)z \nless m$18\end{enumerate}19\end{defn}2021It is easy to see that three examples we just considered were indeed22well-ordering relations. 2324\begin{prob}25\Olref[sth][ordinals][idea]{sec} presented three example orderings26on the natural numbers. Check that each is a well-ordering.27\end{prob}2829Here are some elementary but extremely important observations30concerning well-ordering.3132\begin{prop}\ollabel{wo:strictorder}33If $<$ well-orders $A$, then every non-empty subset of $A$ has a unique34$<$-least member, and $<$ is irreflexive, asymmetric and transitive.35\end{prop}3637\begin{proof}38If $X$ is a non-empty subset of $A$, it has a $<$-minimal !!{element}39$m$, i.e., $(\forall z \in X)z \nless m$. Since $<$ is connected,40$(\forall z \in X)m \leq z$. So $m$ is the $<$-least !!{element} of $X$.4142For irreflexivity, fix $a \in A$; the $<$-least !!{element} of $\{a\}$ is $a$, so $a \nless a$. For transitivity, if $a < b < c$, then43since $\{a, b, c\}$ has a $<$-least !!{element}, $a < c$. Asymmetry44follows from irreflexivity and transitivity45\end{proof}4647\begin{prop}\ollabel{propwoinduction}48If $<$ well-orders $A$, then for any formula $\phi(x)$:49\[50 \text{if }(\forall a \in A)((\forall b < a)\phi(b) \lif 51 \phi(a))\text{, then }(\forall a \in A)\phi(a).52\]53\end{prop}5455\begin{proof}56We will prove the contrapositive. Suppose $\lnot(\forall a \in57A)\phi(a)$, i.e., that $X = \Setabs{x \in A}{\lnot\phi(x)} \neq58\emptyset$. Then $X$ has an $<$-minimal !!{element}, $a$. So $(\forall59b < a)\phi(b)$ but $\lnot \phi(a)$. \end{proof}\noindent This last60property should remind you of the principle of strong induction on the61naturals, i.e.: if $(\forall n \in \omega)((\forall m < n)\phi(m)62\lif \phi(n))$, then $(\forall n \in \omega)\phi(n)$. And this63property makes well-ordering into a very \emph{robust} notion.\footnote{A reminder: all formulas can have parameters (unless explicitly stated otherwise).} 6465\end{document}
content/set-theory/ordinals/iso.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{ordinals}{iso}6\olsection{Order-Isomorphisms}78To explain \emph{how} robust well-ordering is, we will start by9introducing a method for comparing well-orderings.1011\begin{defn}12A \emph{well-ordering} is a pair $\tuple{A, <}$, such that $<$13well-orders $A$. The well-orderings $\tuple{A, <}$ and $\tuple{B,14\lessdot}$ are \emph{order-isomorphic} {iff} there is !!a{bijection}15$f \colon A \to B$ such that: $x < y$ iff $f(x) \lessdot f(y)$. In16this case, we write $\ordeq{\tuple{A, <}}{\tuple{B, \lessdot}}$, and17say that $f$ is an \emph{order-isomorphism}.18\end{defn}19\noindent20In what follows, for brevity, we will speak of ``isomorphisms'' rather21than ``order-isomorphisms''. Intuitively, isomorphisms are22structure-preserving !!{bijection}s. Here are some simple facts about23isomorphisms.2425\begin{lem}\ollabel{isoscompose}26Compositions of isomorphisms are isomorphisms, i.e.: if $f \colon A27\to B$ and $g \colon B \to C$ are isomorphisms, then $(g \circ f)28\colon A \to C$ is an isomorphism.29\end{lem}30\begin{prob}31 Prove \olref[sth][ordinals][iso]{isoscompose}.32\end{prob}33\begin{proof}34Left as an exercise.35\end{proof}36\begin{cor}\ollabel{ordisoisequiv}37 $\ordeq{X}{Y}$ is an equivalence relation.38\end{cor}39\begin{prop}\ollabel{ordisounique}40If $\tuple{A, <}$ and $\tuple{B, \lessdot}$ are isomorphic41well-orderings, then the isomorphism between them is unique.42\end{prop}4344\begin{proof}45Let $f$ and $g$ be isomorphisms $A \to B$. We will prove the result by induction, i.e.\ using \olref[wo]{propwoinduction}. 46Fix $a\in A$, and suppose (for induction) that $(\forall b < a)f(b) = g(b)$. Fix $x \in B$. 4748If $x \lessdot f(a)$, then $f^{-1}(x) < a$, so $g(f^{-1}(x)) \lessdot49g(a)$, invoking the fact that $f$ and $g$ are isomorphisms. But since50$f^{-1}(x) < a$, by our supposition $x =f(f^{-1}(x)) = g(f^{-1}(x))$.51So $x \lessdot g(a)$. Similarly, if $x \lessdot g(a)$ then $x \lessdot52f(a)$. 5354Generalising, $(\forall x \in B)(x \lessdot f(a) \liff x \lessdot55g(a))$. It follows that $f(a) = g(a)$ by56\olref[sfr][rel][ord]{prop:extensionality-strictlinearorders}. So $(\forall57a \in A)f(a) = g(a)$ by \olref[wo]{propwoinduction}.58\end{proof}\noindent 59This gives some sense that well-orderings are robust. But to continue60explaining this, it will help to introduce some more notation. 6162\begin{defn}63When $\tuple{A, <}$ is a well-ordering with $a \in A$, let $A_a = \Setabs{x \in A}{x64< a}$. We say that $A_a$ is a proper \emph{initial segment} of $A$ (and allow that $A$ itself is an improper initial segment of $A$). Let65$<_a$ be the restriction of $<$ to the initial segment, i.e.,66$\funrestrictionto{\mathord{<}}{A_a^2}$. 67\end{defn}68\noindent69Using this notation, we can state and prove that no well-ordering is70isomorphic to any of its proper initial segments.7172\begin{lem}\ollabel{wellordnotinitial}73If $\tuple{A, <}$ is a well-ordering with $a \in A$, then74$\ordneq{\tuple{A, <}}{\tuple{A_a, <_a}}$ 75\end{lem}7677\begin{proof}78For reductio, suppose $f \colon A \to A_a$ is an isomorphism. Since79$f$ is a bijection and $A_a \subsetneq A$, using \olref[wo]{wo:strictorder} let $b \in A$ be the80$<$-least !!{element} of $A$ such that $b \neq f(b)$. We'll show that81$(\forall x \in A)(x<b \liff x < f(b))$, from which it will follow by82\olref[sfr][rel][ord]{prop:extensionality-strictlinearorders} that $b =83f(b)$, completing the reductio.8485Suppose $x < b$. So $x = f(x)$, by the choice of $b$. And $f(x) <86f(b)$, as $f$ is an isomorphism. So $x < f(b)$.8788Suppose $x < f(b)$. So $f^{-1}(x) < b$, since $f$ is an isomorphism,89and so $f^{-1}(x) = x$ by the choice of $b$. So $x < b$. 90\end{proof}9192Our next result shows, roughly put, that an ``initial segment'' of an93isomorphism is an isomorphism:9495\begin{lem}\ollabel{wellordinitialsegment}96Let $\tuple{A, <}$ and $\tuple{B, \lessdot}$ be well-orderings. If $f97\colon A \to B$ is an isomorphism and $a \in A$, then98$\funrestrictionto{f}{A_{a}} : A_a \to B_{f(a)}$ is an isomorphism.99\end{lem}100101\begin{proof}102Since $f$ is an isomorphism:103 %Since $f$ is an isomorphism, $b < a$ iff $f(b) \lessdot f(a)$, so that 104\begin{align*}105 \funimage{f}{A_a} &= \funimage{f}{\Setabs{x \in A}{x < a}}\\106 &= \funimage{f}{\Setabs{f^{-1}(y) \in A}{f^{-1}(y) < a}} \\107 &= \Setabs{y \in B}{y \lessdot f(a)} \\108 &=B_{f(a)} 109\end{align*}110And $\funrestrictionto{f}{A_a}$ preserves order because $f$ does. 111\end{proof}112113Our next two results establish that well-orderings are always114comparable:115116\begin{lem}\ollabel{lemordsegments}117Let $\tuple{A, <}$ and $\tuple{B, \lessdot}$ be well-orderings. If118$\ordeq{\tuple{A_{a_1}, <_{a_1}}}{\tuple{B_{b_1}, \lessdot_{b_1}}}$119and $\ordeq{\tuple{A_{{a_2}}, <_{a_2}}}{\tuple{B_{{b_2}},120\lessdot_{b_2}}}$, then ${a_1} < {a_2} \text{ iff }{b_1} \lessdot121{b_2}$122\end{lem}123124\begin{proof}125We will prove \emph{left to right}; the other direction is similar.126Suppose both $\ordeq{\tuple{A_{a_1}, <_{a_1}}}{\tuple{B_{b_1},127\lessdot_{b_1}}}$ and $\ordeq{\tuple{A_{{a_2}},128<_{a_2}}}{\tuple{B_{{b_2}}, \lessdot_{b_2}}}$, with $f \colon129A_{{a_2}} \to B_{{b_2}}$ our isomorphism. Let ${a_1} < {a_2}$; then130$\ordeq{\tuple{A_{a_1}, <_{a_1}}}{\tuple{B_{f({a_1})},131\lessdot_{f({a_1})}}}$ by \olref{wellordinitialsegment}. So132$\ordeq{\tuple{B_{b_1}, \lessdot_{b_1}}}{\tuple{B_{f({a_1})},133\lessdot_{f({a_1})}}}$, and so ${b_1} = f({a_1})$ by134\olref{wellordnotinitial}. Now ${b_1} \lessdot {b_2}$ as $f$'s domain135is $B_{{b_2}}$. 136\end{proof}137138\begin{thm}\ollabel{thm:woalwayscomparable}139Given any two well-orderings, one is isomorphic to an initial segment140(not necessarily proper) of the other.141\end{thm}142143\begin{proof}144Let $\tuple{A, <}$ and $\tuple{B, \lessdot}$ be well-orderings. Using145Separation, let146\[147 f = \Setabs{\tuple{a, b} \in A \times B}{148 \ordeq{\tuple{A_a, <_a}}{\tuple{B_b, \lessdot_b}}}.149\]150By \olref{lemordsegments}, $a_1 < a_2$ iff $b_1 \lessdot b_2$ for all151$\tuple{a_1, b_1}, \tuple{a_2, b_2} \in f$. So $f \colon \dom{f} \to152\ran{f}$ is an isomorphism. 153154If $a_2 \in \dom{f}$ and $a_1 < a_2$, then $a_1 \in \dom{f}$ by155\olref{wellordinitialsegment}; so $\dom{f}$ is an initial segment of156$A$. Similarly, $\ran{f}$ is an initial segment of $B$. For reductio,157suppose both are \emph{proper} initial segments. Then let $a$ be the158$<$-least !!{element} of $A \setminus \dom{f}$, so that $\dom{f} =159A_a$, and let $b$ be the $\lessdot$-least !!{element} of $B \setminus160\ran{f}$, so that $\ran{f} = B_b$. So $f \colon A_a \to B_b$ is an161isomorphism, and hence $\tuple{a, b} \in f$, a contradiction.162\end{proof}163164\end{document}
content/set-theory/ordinals/vn.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{ordinals}{vn} 6\olsection[Von Neumann's Construction]{Von Neumann's Construction of the Ordinals}78\olref[sth][ordinals][iso]{thm:woalwayscomparable} gives rise to a9thought. We could introduce certain objects, called \emph{order10types}, to go proxy for the well-orderings. Writing $\ordtype{A, <}$11for the order type of the well-ordering $\tuple{A, <}$, we would hope12to secure the following two principles:13\begin{align*}14 \ordtype{A, <} = \ordtype{B, \lessdot} & 15 \text{ iff } \ordeq{\tuple{A, <}}{\tuple{B, \lessdot}}\\16 \ordtype{A, <} < \ordtype{B, \lessdot}&17 \text{ iff }\ordeq{\tuple{A, <}}{\tuple{B_b, \lessdot_b}}\text{ for some }b \in B18\end{align*}19Moreover, we might hope to introduce order-types \emph{as certain20sets}, just as we can introduce the natural numbers as certain sets. 2122The most common way to do this---and the approach we will follow---is23to define these order-types via certain \emph{canonical} well-ordered24sets. These canonical sets were first introduced by von Neumann:2526\begin{defn}27The set $A$ is \emph{transitive} {iff} $(\forall x \in A)x \subseteq28A$. Then $A$ is an \emph{ordinal} {iff} $A$ is transitive and29well-ordered by $\in$.30\end{defn}31\noindent32In what follows, we will use Greek letters for ordinals. It follows33immediately from the definition that, if $\alpha$ is an ordinal, then34$\tuple{\alpha, \in_\alpha}$ is a well-ordering, where $\in_\alpha =35\Setabs{\tuple{x, y} \in \alpha^2}{x \in y}$. So, abusing notation a36little, we can just say that $\alpha$ \emph{itself} is a37well-ordering. 3839Here are our first few ordinals:40\[41 \emptyset, \{\emptyset\}, 42 \{\emptyset, \{\emptyset\}\}, 43 \{\emptyset, \{\emptyset\}, \{\emptyset, \{\emptyset\}\}\}, \ldots44\]45You will note that these are the first few ordinals that we46encountered in our Axiom of Infinity, i.e., in von Neumann's47definition of $\omega$ (see \olref[sth][z][infinity-again]{sec}). This48is no coincidence. Von Neumann's definition of the ordinals treats49natural numbers as ordinals, but allows for transfinite ordinals too. 5051As always, we can now ask: \emph{are} these the ordinals? Or has von52Neumann simply given us some sets that we can \emph{treat} as the53ordinals? The kinds of discussions one might have about this question54are similar to the discussions we had in \olref[sfr][rel][ref]{sec},55\olref[sfr][arith][ref]{sec},56\olref[sfr][infinite][dedekindsproof]{sec}, and57\olref[sth][z][nat]{sec}, so we will not belabour the point.58Instead, in what follows, we will simply use ``the ordinals'' to speak59of ``the von Neumann ordinals''. 6061\end{document}
content/set-theory/ordinals/basic.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{ordinals}{basic}6\olsection{Basic Properties of the Ordinals}78We observed that the first few ordinals are the natural numbers. The9main reason for developing a theory of ordinals is to extend the10principle of induction which holds on the natural numbers. We will11build up to this via a sequence of elementary results.1213\begin{lem}\ollabel{ordmemberord}14Every !!{element} of an ordinal is an ordinal.15\end{lem}1617\begin{proof}18Let $\alpha$ be an ordinal with $b \in \alpha$. Since $\alpha$ is19transitive, $b \subseteq \alpha$. So $\in$ well-orders $b$ as $\in$20well-orders $\alpha$.2122To see that $b$ is transitive, suppose $x \in c \in b$. So $c \in \alpha$ as $b23\subseteq \alpha$. Again, as $\alpha$ is transitive, $c \subseteq24\alpha$, so that $x \in \alpha$. So $x, c, b \in \alpha$. But $\in$25well-orders $\alpha$, so that $\in$ is a transitive relation on26$\alpha$ by \olref[sth][ordinals][wo]{wo:strictorder}. So since $x27\in c \in b$, we have $x \in b$. Generalising, $c \subseteq b$28\end{proof}2930\begin{cor}\ollabel{ordissetofsmallerord}31$\alpha = \Setabs{\beta \in \alpha}{\beta \text{ is an ordinal}}$, for32any ordinal~$\alpha$33\end{cor}3435\begin{proof}36Immediate from \olref{ordmemberord}.37\end{proof}3839The rough gist of the next two main results,40\olref{ordinductionschema} and \olref{ordtrichotomy}, is that the41ordinals themselves are well-ordered by membership:4243\begin{thm}[Transfinite Induction]\ollabel{ordinductionschema}44For any formula $\phi(x)$: 45\[46 \text{if }\exists \alpha \phi(\alpha)\text{, then }\exists \alpha(\phi(\alpha)47 \land (\forall \beta \in \alpha) \lnot \phi(\beta))48\]49where the displayed quantifiers are implicitly restricted to ordinals.50\end{thm}5152\begin{proof}53%\olref{ordinductionschema1}. 54Suppose $\phi(\alpha)$, for some ordinal $\alpha$. If $ (\forall \beta55\in \alpha) \lnot \phi(\beta)$, then we are done. Otherwise, as56$\alpha$ is an ordinal, it has some $\in$-least !!{element} which is57$\phi$, and this is an ordinal by \olref{ordmemberord}.58% \olref{ordinductionschema2}. Suppose there is some ordinal59% $\gamma$ such that $\lnot\phi(\gamma)$. Then by60% \olref{ordinductionschema1}, there is an $\in$-minimal ordinal61% $\alpha$ for which $\lnot\phi(\alpha)$. So $(\forall \beta <62% \alpha) \phi(\beta)$, rendering the antecedent of the conditional63% false.64\end{proof}65\noindent66Note that we can equally express \olref{ordinductionschema} as the67scheme:68\[69\text{if }\forall \alpha((\forall \beta \in \alpha)\phi(\beta) \lif 70\phi(\alpha))\text{, then }\forall \alpha\phi(\alpha)71\]72just by taking $\lnot\phi(\alpha)$ in \olref{ordinductionschema}, and then performing elementary logical manipulations.7374\begin{thm}[Trichotomy]\ollabel{ordtrichotomy} 75$\alpha \in \beta \lor \alpha = \beta \lor \beta \in \alpha$, for any76ordinals $\alpha$ and $\beta$. 77\end{thm}7879\begin{proof}80The proof is by double induction, i.e., using81\olref{ordinductionschema} twice. Say that $x$ is \emph{comparable}82with $y$ iff $x \in y \lor x = y \lor y \in x$. 8384For induction, suppose that every ordinal in~$\alpha$ is comparable85with \emph{every} ordinal. For further induction, suppose that86$\alpha$ is comparable with every ordinal in~$\beta$. We will show that87$\alpha$ is comparable with~$\beta$. By induction on~$\beta$, it will88follow that $\alpha$ is comparable with every ordinal; and so by89induction on~$\alpha$, \emph{every} ordinal is comparable with90\emph{every} ordinal, as required. It suffices to assume that $\alpha \notin \beta$ and $\beta \notin91\alpha$, and show that $\alpha = \beta$. 9293To show that $\alpha \subseteq \beta$, fix $\gamma \in \alpha$; this94is an ordinal by \olref{ordmemberord}. So by the first induction95hypothesis, $\gamma$ is comparable with $\beta$. But if either $\gamma96= \beta$ or $\beta \in \gamma$ then $\beta \in \alpha$ (invoking the97fact that $\alpha$ is transitive if necessary), contrary to our98assumption; so $\gamma \in \beta$. Generalising, $\alpha \subseteq99\beta$.100101Exactly similar reasoning, using the second induction hypothesis,102shows that $\beta \subseteq \alpha$. So $\alpha = \beta$.103\end{proof}\noindent As such, we will sometimes write $\alpha <\beta$104rather than $\alpha \in \beta$, since $\in$ is behaving as an ordering105relation. There are no deep reasons for this, beyond familiarity, and106because it is easier to write $\alpha \leq \beta$ than $\alpha \in107\beta \lor \alpha = \beta$.\footnote{We could write $\alpha108\mathrel{\underline{\in}} \beta$; but that would be wholly109non-standard.}110111Here are two quick consequences of our last results, the first of112which puts our new notation into action:113114\begin{cor}\ollabel{ordordered}115If $\exists \alpha\phi(\alpha)$, then $\exists \alpha(\phi(\alpha)116\land \forall \beta(\phi(\beta) \lif \alpha \leq \beta))$.117Moreover, for any ordinals $\alpha, \beta, \gamma$, both $\alpha118\notin \alpha$ and $\alpha \in \beta \in \gamma \lif \alpha \in119\gamma$. 120\end{cor}121122\begin{proof}123Just like \olref[wo]{wo:strictorder}.124\end{proof}125126\begin{prob}127Complete the ``exactly similar reasoning'' in the proof of128\olref[sth][ordinals][basic]{ordtrichotomy}.129\end{prob}130131\begin{cor}\ollabel{corordtransitiveord}132$A$ is an ordinal iff $A$ is a transitive set of ordinals.133\end{cor}134135\begin{proof}136\emph{Left-to-right.} By \olref{ordmemberord}. \emph{Right-to-left.}137If $A$ is a transitive set of ordinals, then $\in$ well-orders $A$ by138\olref{ordinductionschema} and \olref{ordtrichotomy}.139\end{proof}140141Now, we glossed 142\olref{ordinductionschema} and \olref{ordtrichotomy} as telling us that $\in$ well-orders the ordinals. However, we 143have to be \emph{very cautious} about this sort of claim, thanks to the144following result:145146\begin{thm}[Burali-Forti Paradox]\ollabel{buraliforti}147There is no set of all the ordinals148\end{thm}149150\begin{proof}151For reductio, suppose $O$ is the set of all ordinals. If $\alpha \in152\beta \in O$, then $\alpha$ is an ordinal, by \olref{ordmemberord}, so153$\alpha \in O$. So $O$ is transitive, and hence $O$ is an ordinal by154\olref{corordtransitiveord}. Hence $O \in O$, contradicting155\olref{ordordered}. 156\end{proof}157\noindent158This result is named after \citeauthor{Burali-Forti1897}. But, it was Cantor in 1899---in a letter to Dedekind---who first saw159clearly the \emph{contradiction} in supposing that there is a set of160all the ordinals. As van161Heijenoort explains:162\begin{quote}163 Burali-Forti himself considered the contradiction as establishing,164 by \emph{reductio ad absurdum}, the result that the natural ordering165 of the ordinals is just a partial ordering.166 \citep[p.~105]{Heijenoort1967}167\end{quote}168Setting Burali-Forti's mistake to one side, we can summarize the foregoing as follows. Ordinals are sets which are individually well-ordered by169membership, and collectively well-ordered by membership (without collectively constituting a set). 170171Rounding this off, here are some more basic properties about the172ordinals which follow from \olref{ordinductionschema} and173\olref{ordtrichotomy}. 174175\begin{prop}176Any strictly descending sequence of ordinals is finite.177\end{prop}178179\begin{proof}180Any infinite strictly descending sequence of ordinals $\alpha_0 > \alpha_1 > \alpha_2 > \ldots$ has no $<$-minimal181member, contradicting \olref{ordinductionschema}.182\end{proof}183184\begin{prop}\ollabel{ordinalsaresubsets}185$\alpha \subseteq \beta \lor \beta \subseteq \alpha$, for any ordinals186$\alpha, \beta$.187\end{prop}188189\begin{proof}190If $\alpha \in \beta$, then $\alpha \subseteq \beta$ as $\beta$ is191transitive. Similarly, if $\beta \in \alpha$, then $\beta \subseteq192\alpha$. And if $\alpha = \beta$, then $\alpha \subseteq \beta$ and193$\beta \subseteq \alpha$. So by \olref{ordtrichotomy} we are done.194\end{proof}195196\begin{prop}\ollabel{ordisoidentity}197$\alpha = \beta$ iff $\ordeq{\alpha}{\beta}$, for any ordinals198$\alpha, \beta$.199\end{prop}200201\begin{proof}202The ordinals are well-orders; so this is immediate from Trichotomy203(\olref[basic]{ordtrichotomy}) and204\olref[iso]{wellordnotinitial}. 205\end{proof}206207\begin{prob}\ollabel{probunionordinalsordinal}208Prove that, if every member of $X$ is an ordinal, then $\bigcup X$ is an ordinal.209\end{prob}210211\end{document}
content/set-theory/ordinals/replacement.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{ordinals}{replacement}6\olsection{Replacement}78In \olref[sth][ordinals][vn]{sec}, we motivated the introduction of ordinals by9suggesting that we could treat them as order-types, i.e., canonical10proxies for well-orderings. In order for that to work, we would need11to prove that \emph{every well-ordering is isomorphic to some12ordinal}. This would allow us to define $\ordtype{A, <}$ as the13ordinal $\alpha$ such that $\tuple{A, <} \isomorphic \alpha$. 1415Unfortunately, we \emph{cannot} prove the desired result only the16Axioms we provided introduced so far. (We will see why in17\olref[replacement][strength]{sec}, but for now the point is: we can't.) We need a18new thought, and here it is:1920\begin{axiom}[Scheme of Replacement]21For any formula $\phi(x, y)$, the following is an axiom:22\begin{quote}23 for any $A$, if $(\forall x \in A)\lexists![y][\phi(x,y)]$, then24 $\Setabs{y}{(\exists x \in A)\phi(x,y)}$ exists.25\end{quote}26\end{axiom}27\noindent28As with Separation, this is a scheme: it yields infinitely many29axioms, for each of the infinitely many different $\phi$'s. And it can30equally well be (and normally is) written down thus:3132\begin{defish}33For any formula $\phi(x,y)$ which does not contain ``$B$'', the following is an axiom:34\[35\forall A[(\forall x \in A)\lexists![y][\phi(x,y)] \lif \exists B\forall y (y \in B \liff (\exists x \in A)\phi(x,y))]36\]37\end{defish}3839On first encounter, however, this is quite a tangled formula. The40following quick consequence of Replacement probably gives a41\emph{clearer} expression to the intuitive idea we are working with:\footnote{A term is an expression which picks out exactly one object (on any completion of its free variables). For example, ``$\emptyset$'' is a term which picks out the empty set; ``$\{x\}$'' is a term which picks out $x$'s singleton (whatever $x$ might be); ``$x \cup y$'' is a term which picks out the union of $x$ and $y$ (whatever they might be).}4243\begin{cor}44For any term $\tau(x)$, and any set $A$, this set exists:45\[46 \Setabs{\tau(x)}{x \in A} = \Setabs{y}{(\exists x \in A)y = \tau(x)}.47\]48\end{cor}4950\begin{proof}51Since $\tau$ is a \emph{term}, $\forall x \lexists![y][\tau(x) = y]$.52A fortiori, $(\forall x \in A)\lexists![y][\tau(x) = y]$. So53$\Setabs{y}{(\exists x \in A)\tau(x) = y}$ exists by Replacement.54\end{proof}55\noindent56This suggests that ``Replacement'' is a good name for the Axiom: given57a set $A$, you can form a new set, $\Setabs{\tau(x)}{x \in A}$, by58replacing every member of $A$ with its image under~$\tau$. Indeed,59following the notation for the image of a set under a function, we60might write $\funimage{\tau}{A}$ for $\Setabs{\tau(x)}{x \in A}$.6162Crucially, however, $\tau$ is a \emph{term}. It need not be (a name63for) a \emph{function}, in the sense of \olref[sfr][fun][rel]{sec},64i.e., a certain set of ordered pairs. After all, if $f$ is a function65(in that sense), then the set $\funimage{f}{A} = \Setabs{f(x)}{x \in66A}$ is just a particular subset of $\ran{f}$, and that is already67guaranteed to exist, just using the axioms of~$\Zminus$.\footnote{Just68consider $\Setabs{y \in \bigcup \bigcup f}{(\exists x \in A)y =69f(x)}$.} Replacement, by contrast, is a \emph{powerful} addition to70our axioms, as we will see in \olref[sth][replacement][]{chap}.717273\end{document}
content/set-theory/ordinals/milestone.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}4\olfileid{sth}{ordinals}{zfm} 5\olsection{$\ZFminus$: a milestone}67The question of how to justify Replacement (if at all) is not8straightforward. As such, we will reserve that for9\olref[replacement][]{chap}. However, with the addition of Replacement, we10have reached another important milestone. We now have all the axioms11required for the theory $\ZFminus$. In detail:1213\begin{defn}14The theory $\ZFminus$ has these axioms: Extensionality, Union, Pairs,15Powersets, Infinity, and all instances of the Separation and16Replacement schemes. Otherwise put, $\ZFminus$ adds Replacement to17$\Zminus$.18\end{defn}19\noindent20This stands for \emph{Zermelo--Fraenkel} set theory (\emph{minus}21something which we will come to later). Fraenkel gets the honour,22since he is credited with the formulation of Replacement in23\citeyear{Fraenkel1922}, although the first precise formulation was24due to \citet{Skolem1922}.2526\end{document}
content/set-theory/ordinals/ordtype.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{ordinals}{ordtype}6\olsection{Ordinals as Order-Types}78Armed with Replacement, and so now working in $\ZFminus$, we can finally prove the result we have been aiming for:910\begin{thm}\ollabel{thmOrdinalRepresentation}11Every well-ordering is isomorphic to a unique ordinal. 12\end{thm}1314\begin{proof}15Let $\tuple{A, <}$ be a well-order. By16\olref[basic]{ordisoidentity}, it is isomorphic to at most one17ordinal. So, for reductio, suppose $\tuple{A, <}$ is not isomorphic to18\emph{any} ordinal. We will first ``make $\tuple{A, <}$ as small as19possible''. In detail: if some proper initial segment $\tuple{A_a,20<_a}$ is not isomorphic to any ordinal, there is a least $a \in A$21with that property; then let $B = A_a$ and $\mathord{\lessdot} =22\mathord{<_a}$. Otherwise, let $B = A$ and $\mathord{\lessdot} =23\mathord{<}$. 2425By definition, every proper initial segment of $B$ is isomorphic to26some ordinal, which is unique as above. So27by Replacement, the following set exists, and is a function:28\[29 f = \Setabs{\tuple{\beta, b}}{b \in B\text{ and }\ordeq{\beta}{\tuple{B_b, \lessdot_b}}}30\]31To complete the reductio, we'll show that $f$ is an isomorphism32$\alpha \to B$, for some ordinal $\alpha$. 3334It is obvious that $\ran{f}35= B$. And by \olref[iso]{lemordsegments}, $f$ preserves ordering,36i.e., $\gamma \in \beta$ iff $f(\gamma) \lessdot f(\beta)$. To show that $\dom{f}$ is an ordinal, by37\olref[basic]{corordtransitiveord} it suffices to show that38$\dom{f}$ is transitive. So fix $\beta \in \dom{f}$, i.e.,39$\ordeq{\beta}{\tuple{B_b, \lessdot_b}}$ for some $b$. If $\gamma \in40\beta$, then $\gamma \in \dom{f}$ by41\olref[iso]{wellordinitialsegment}; generalising, $\beta \subseteq42\dom{f}$.43\end{proof}4445This result licenses the following definition, which we have wanted to46offer since \olref[vn]{sec}:4748\begin{defn}49If $\tuple{A, <}$ is a well-ordering, then its order type,50$\ordtype{A, <}$, is the unique ordinal $\alpha$ such that51$\ordeq{\tuple{A, < }}{\alpha}$.52\end{defn}5354Moreover, this definition licenses two nice principles:5556\begin{cor}\ollabel{ordtypesworklikeyouwant}57Where $\tuple{A, <}$ and $\tuple{B, \lessdot}$ are well-orderings: 58\begin{align*}59 \ordtype{A, <} = \ordtype{B, \lessdot}&\text{ iff }\ordeq{\tuple{A, <}}{\tuple{B, \lessdot}}\\60 \ordtype{A, <} \in \ordtype{B, \lessdot}&\text{ iff }\ordeq{\tuple{A, <}}{\tuple{B_b, \lessdot_b}}\text{ for some }b \in B61\end{align*}62\end{cor}6364\begin{proof}65The identity holds by \olref[basic]{ordisoidentity}. To prove66the second claim, let $\ordtype{A, <} = \alpha$ and $\ordtype{B,67\lessdot} = \beta$, and let $f \colon \beta \to \tuple {B, \lessdot}$68be our isomorphism. Then:69\begin{align*}70 \alpha \in \beta&\text{ iff }\funrestrictionto{f}{\alpha} \colon \alpha \to B_{f(\alpha)}\text{ is an isomorphism}\\71 &\text{ iff }\ordeq{\tuple{A, <}}{\tuple{B_{f(\alpha)}, \lessdot_{f(\alpha)}}}\\72 &\text{ iff }\ordeq{\tuple{A, <}}{\tuple{B_b, \lessdot_b}}\text{ for some $b \in B$}73\end{align*}74by \olref[iso]{ordisounique},75\olref[iso]{wellordinitialsegment}, and76\olref[basic]{ordissetofsmallerord}.77\end{proof}7879\end{document}
content/set-theory/ordinals/opps.tex
1\documentclass[../../../include/open-logic-section]{subfiles}23\begin{document}45\olfileid{sth}{ordinals}{opps}6\olsection{Successor and Limit Ordinals}78In the next few chapters, we will use ordinals a great deal. So it9will help if we introduce some simple notions. 1011\begin{defn}12For any ordinal $\alpha$, its \emph{successor} is $\ordsucc{\alpha}13=\alpha \cup \{\alpha\}$. We say that $\alpha$ is a \emph{successor}14ordinal if $\ordsucc{\beta} = \alpha$ for some ordinal~$\beta$. We15say that $\alpha$ is a \emph{limit} ordinal iff $\alpha$ is neither16empty nor a successor ordinal.17\end{defn}18\noindent19The following result shows that this is the right notion of20\emph{successor}:2122\begin{prop}23For any ordinal $\alpha$:24\begin{enumerate}25 \item $\alpha \in \ordsucc{\alpha}$;26 \item $\ordsucc{\alpha}$ is an ordinal;27 \item there is no ordinal $\beta$ such that $\alpha \in \beta \in28 \ordsucc{\alpha}$.29 \end{enumerate}30\end{prop}3132\begin{proof}33Trivially, $\alpha \in \alpha \cup \{\alpha\} = \ordsucc{\alpha}$.34Equally, $\ordsucc{\alpha}$ is a transitive set of ordinals, and hence35an ordinal by \olref[basic]{corordtransitiveord}.36And it is impossible that $\alpha \in \beta \in \ordsucc{\alpha}$,37since then either $\beta \in \alpha$ or $\beta = \alpha$,38contradicting \olref[basic]{ordordered}.39\end{proof}40\noindent41This also licenses a variant of proof by transfinite induction:4243\begin{thm}[Simple Transfinite Induction]\ollabel{simpletransrecursion}44Let $\phi(x)$ be a formula such that:45\begin{enumerate}46 \item $\phi(\emptyset)$; and47 \item for any ordinal $\alpha$, if $\phi(\alpha)$ then48 $\phi(\ordsucc{\alpha})$; and 49 \item if $\alpha$ is a limit ordinal and $(\forall \beta \in50 \alpha)\phi(\beta)$, then $\phi(\alpha)$.51\end{enumerate}52Then $\forall \alpha \phi(\alpha)$.53\end{thm}5455\begin{proof}56We prove the contrapositive. So, suppose there is some ordinal which57is $\lnot\phi$; let $\gamma$ be the least such ordinal. Then either58$\gamma = \emptyset$, or $\gamma = \ordsucc{\alpha}$ for some $\alpha$59such that $\phi(\alpha)$; or $\gamma$ is a limit ordinal and $(\forall60\beta \in \gamma)\phi(\beta)$.61\end{proof}62\noindent63A final bit of notation will prove helpful later on:6465\begin{defn}\ollabel{defsupstrict}66If $X$ is a set of ordinals, then $\supstrict(X) = \bigcup_{\alpha67\in X} \ordsucc{\alpha}$.68\end{defn}69\noindent70Here, ``lsub'' stands for ``least strict upper bound''.\footnote{Some71books use ``$\text{sup}(X)$'' for this. But other books use72``$\text{sup}(X)$'' for the least \emph{non-strict} upper bound, i.e.,73simply $\bigcup X$. If $X$ has a greatest element, $\alpha$, these74notions come apart: the least \emph{strict} upper bound is75$\ordsucc{\alpha}$, whereas the least \emph{non-strict} upper bound is76just $\alpha$.} The following result explains this:7778\begin{prop}79If $X$ is a set of ordinals, $\supstrict(X)$ is the least ordinal80greater than every ordinal in $X$.81\end{prop}8283\begin{proof}84Let $Y = \Setabs{\ordsucc{\alpha}}{\alpha \in X}$, so that85$\supstrict(X) = \bigcup Y$. Since ordinals are transitive and every86member of an ordinal is an ordinal, $\supstrict(X)$ is a transitive87set of ordinals, and so is an ordinal by88\olref[basic]{corordtransitiveord}. 8990If $\alpha \in X$, then $\ordsucc{\alpha} \in Y$, so $\ordsucc{\alpha}91\subseteq \bigcup Y = \supstrict(X)$, and hence $\alpha \in92\supstrict(X)$. So $\supstrict(X)$ is strictly greater than every93ordinal in $X$.9495Conversely, if $\alpha \in \supstrict(X)$, then $\alpha \in96\ordsucc{\beta} \in Y$ for some $\beta \in X$, so that $\alpha \leq97\beta \in X$. So $\supstrict(X)$ is the \emph{least} strict upper98bound on $X$.99\end{proof}100101\end{document}