%%% This is the scribe notes template for CS611 %%% There are several comments preceded by CS611: and boxed in %%%%'s %%% which indicate where macros should be altered to set up the header %%% for the paper. Your Notes should go at the comment SCRIBE NOTES GO HERE!. %%% In the various .sty files that accompany this .tex file you will %%% find LaTeX macros that make it easier to typeset inference rules %%% and programming language constructs. You must make sure that the %%% file proof.sty is in a path searched by LaTeX when you try to %%% use this file. Take a look to see what macros are available--it %%% will save you time and make the notes look better. Feel free to %%% extend the set of macros--post them to the newsgroup and contact %%% the course staff if you come up with some good ones so they can be %%% added to the template. %%% This template includes examples of how to use some of the macros %%% to give you an idea of how they work. (Delete the examples when %%% you do your scribing.) \documentclass{article} \usepackage{611-lecture} \usepackage{amsmath,amssymb,amsthm,amsfonts,comment} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% %%% CS611: Please fill in these macros as appropriate: \lecture{29} %% Lecture number \title{Propositions as Types} %% Title of lecture %\author{Mia Minnes, Nam Nguyen} %% name of scribe \date{13 November 2006} %% Date of lecture, e.g., 1 January 2001 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % See 611.sty for a variety of macros that will be helpful in % typesetting the lecture. Here are a few of particular interest: % % "x" x in keyword font (e.g., "if", "#t") % _x_ x in italics % \nm{n} n in slanted font (used for abbreviations) % e in angle brackets % \lt less-than sign % \gt greater-than sign % \SB{x} x in semantic brackets % \Tr x{y} x[[y]] with x in calligraphic font % (if x is more than a single character, use \Tr{x}{y}) \renewcommand\emptyset\varnothing \newcommand{\inL}{\ensuremath{\mathsf{inL}}} \newcommand{\inR}{\ensuremath{\mathsf{inR}}} \newcommand\caseof[3]{\mathsf{case}~{#1}~\mathsf{of}~{#2}~|~{#3}} \newtheorem{theorem}{Theorem} \newtheorem{lemma}[theorem]{Lemma} \newtheorem{proposition}[theorem]{Proposition} \newtheorem{corollary}[theorem]{Corollary} \newtheorem{definition}[theorem]{Definition} \newcommand{\Z}{\mathbb{Z}} \newcommand{\group}[1]{\left\langle{#1}\right\rangle} %\theoremstyle{definition} %\newtheorem*{defn}{Definition} \newcommand{\nondet}{\left[\!\kern1pt\right]} \renewcommand\phi\varphi \renewcommand\wp[2]{\mathsf{wp}~{#1}~{#2}} \newcommand\wlp[2]{\mathsf{wlp}~{#1}~{#2}} \renewcommand\({\begin{eqnarray*}} \renewcommand\){\end{eqnarray*}} \newcommand\LOOKUP[2]{\mathrm{LOOKUP}~{#1}~{#2}} \newcommand\UPDATE[3]{\mathrm{UPDATE}~{#1}~{#2}~{#3}} \newcommand\MALLOC[2]{\mathrm{MALLOC}~{#1}~{#2}} \newcommand\EMPTY{\mathrm{EMPTY\mbox{-}STORE}} \renewcommand\dom[1]{\mathrm{dom}\,{#1}} \newcommand\p[2]{\langle{#1},\,{#2}\rangle} \newcommand\bigcdot{\mathrel{\raisebox{1pt}{$\scriptscriptstyle\bullet$}}} \newcommand\holed[1]{[\,#1\,]} \newcommand\hole{\holed\bigcdot} \newcommand\context[1]{E\kern1pt\holed{#1}} \newcommand\contextHole{\context\bigcdot} \newcommand\goesto[2]{\underset{#2}{\overset{#1}\to}} \newcommand\ifthenelse[3]{\mathsf{if\ }#1\mathsf{\ then\ }#2\mathsf{\ else\ }#3} \newcommand\ifpthenelse[3]{\mathsf{ifp\ }#1\mathsf{\ then\ }#2\mathsf{\ else\ }#3} \newcommand\whiledo[2]{\mathsf{while\ }#1\mathsf{\ do\ }#2} \newcommand\letin[3]{\mathsf{let\ }#1 = #2\mathsf{\ in\ }#3} \newcommand\letrec[5]{\mathsf{letrec\ }#1 = #2\mathsf{\ and\ \ldots\ and\ }#3 = #4\mathsf{\ in\ }#5} \newcommand\letrecone[3]{\mathsf{letrec\ }#1 = #2\mathsf{\ in\ }#3} \newcommand\true{\ensuremath{\mathsf{true}}} \newcommand\false{\ensuremath{\mathsf{false}}} \newcommand\error{\ensuremath{\mathsf{error}}} \newcommand\pca[3]{\{#1\}\kern1pt{#2}\kern1pt\{#3\}} \newcommand\states{\Set{St}} \newcommand\rtc{^{\textstyle *}} \newcommand\sat\vDash \newcommand\force\vdash \newcommand\hyphen{\mbox{-}} \newcommand\lookup[2]{\nm{LOOKUP}~#1~\mquote{#2}} \newcommand\update[3]{\nm{UPDATE}~#1~\mquote{#2}~#3} \newcommand\SBk[1]{\SB{#1}k} \newcommand\fix[1]{\mathsf{fix}\,{#1}} \newlength\reasonwidth \setlength\reasonwidth{3cm} \newcommand\reasoning[1]{\def\longest{#1}\settowidth{\reasonwidth}{$\displaystyle\longest$}\addtolength{\reasonwidth}{5mm}} \newcommand\reason[2]{\makebox[\reasonwidth][l]{$\displaystyle{#1}$}\mbox{#2}} \renewcommand\inj[1]{\mathsf{in}_{#1}} \newcommand\proj[1]{\pi_{#1}} \newcommand{\dlt}{\sqsubseteq} \newcommand\floor[1]{\lfloor{#1}\rfloor} \newcommand\cf[1]{[\kern1pt{#1}\kern1pt]} \newcommand\SBpr[1]{\SB{#1}\,\phi\,\rho} \renewcommand\C[3]{\Tr C{#1}\kern1pt{#2}\kern1pt{#3}} \renewcommand\Cr[1]{\C{#1}\Gamma\rho} \newcommand\judge[3]{{#1}\force{#2}:{#3}} \newcommand\Gjudge[2]{\judge\Gamma{#1}{#2}}% \newcommand\forceUSN{\mathrel{\makebox[2pt][l]{$\force$}\raisebox{-3pt}[0pt][0pt]{\tiny{\textit{USN}}}}} \newcommand\judgeUSN[3]{{#1}\forceUSN{#2}:{#3}} \renewcommand\C[3]{\Tr C{#1}\kern1pt{#2}\kern1pt{#3}} \renewcommand\Cr[1]{\C{#1}\Gamma\rho} \newcommand\Irred[1]{\ensuremath{\mathrm{Irred}(#1)}} \newcommand\seq[3]{#1_{#2},\ldots,#1_{#3}} \newcommand\substtwo[5]{\subst{#1}{#2}{#3,\,#4/#5}} \newcommand\substlist[5]{\subst{#1}{#2}{#3,\ldots,#4/#5}} \newcommand\Unify[1]{\mathrm{Unify}(#1)} \newcommand\fa[2]{\forall{#1}\kern1pt.\kern1pt{#2}}% \newcommand\Judge[4]{\judge{#1;\,#2}{#3}{#4}} \newenvironment{proofof}[1]{\addtolength{\topsep}{1mm}\begin{trivlist}\item[]\hspace{\parindent}{\em Proof of #1.}}{\qed\end{trivlist}} \begin{document} \maketitle \section{Intuitionistic Logic and Constructive Mathematics} We have previously observed that several familiar type judgements $\judge{}e\tau$ of the pure simply-typed $\lambda$-calculus correspond to a tautologies of propositional logic: \[ \begin{array}{cc} \mbox{\textit{type judgement}} & \mbox{\textit{propositional tautology}}\\[2pt]\hline\\[-8pt] \judge{}I{\alpha\to\alpha} & P\to P\\ \judge{}K{\alpha\to\beta\to\alpha} & P\to (Q\to P)\\ \judge{}S{(\alpha\to\beta\to\gamma)\to(\alpha\to\beta)\to(\alpha\to\gamma)} & (P\to Q\to R)\to(P\to Q)\to(P\to R)\\ \end{array} \] This is no accident. It turns out that all derivable type judgements $\judge{}e\tau$ (with the empty environment to the left of the turnstile) give propositional tautologies. This is because the typing rules of $\lambda^\rightarrow$ correspond exactly to the proof rules of propositional _intuitionistic logic_. Intuitionistic logic is the basis of _constructive mathematics_. Constructive mathematics takes a much more conservative view of truth than classical mathematics. It is concerned less with _truth_ than with _provability_. Its main proponents were Kronecker and Brouwer around the beginning of the last century. Their views at the time generated great controversy in the mathematical world. In constructive mathematics, not all deductions of classical logic are considered valid. For example, to prove in classical logic that there exists an object having a certain property, it is enough to assume that no such object exists and derive a contradiction. Intuitionists would not consider this argument valid. Intuitionistically, you must actually construct the object and prove that it has the desired property. Intuitionists do not accept the law of double negation: $P\leftrightarrow \neg\neg P$. They do believe that $P\to\neg\neg P$, that is, if $P$ is true then it is not false; but they do not believe $\neg\neg P\to P$, that is, even if $P$ is not false, then that does not automatically make it true. Similarly, intuitionists do not accept the law of the excluded middle $P\vee\neg P$. In order to prove $P\vee\neg P$, you must prove either $P$ or $\neg P$. It may well be that neither is provable, in which case the intuitionist would not accept that $P\vee\neg P$. For intuitionists, the implication $P\to Q$ has a much stronger meaning than merely $\neg P\vee Q$, as in classical logic. To prove $P\to Q$, one must show how to construct a proof of $Q$ from any given proof of $P$. So a proof of $P\to Q$ is a (computable) function from proofs of $P$ to proofs of $Q$. Similarly, to prove $P\wedge Q$, you must prove both $P$ and $Q$; thus a proof of $P\wedge Q$ is a pair consisting of a proof of $P$ and a proof of $Q$. \subsection{Example} Here is an example of a proof that would not be accepted by an intuitionist. \medskip\noindent \textbf{Theorem}\quad There exist irrational numbers $a$ and $b$ such that $a^b$ is rational. \begin{proof} Either $\sqrt 2^{\sqrt 2}$ is rational or not. If it is, take $a=b=\sqrt 2$ and we are done. If it is not, take $a=\sqrt 2^{\sqrt 2}$ and $b=\sqrt 2$; then $a^b=(\sqrt 2^{\sqrt 2})^{\sqrt 2}=\sqrt 2^2=2$, and again we are done. \end{proof} Now an intuitionist would not like this, because we haven't actually constructed a definite $a$ and $b$ with the desired property. We have used the law of the excluded middle, which is cheating. \section{Syntax} Syntactically, formulas $\phi,\psi,\ldots$ of intuitionistic logic look the same as their classical counterparts. At the propositional level, we have propositional variables $P,Q,R,\ldots$ and formulas \( \phi &::=& \top \bnf \bot \bnf P \bnf \phi\to\psi \bnf \phi\vee\psi \bnf \phi\wedge\psi \bnf \neg\phi. \) We might also add a second-order quantifier $\forall P$ ranging over propositions: \( \phi &::=& \cdots \bnf \fa P\phi. \) \section{Natural Deduction (Gentzen, 1943)} Intuitionistic logic uses a sequent calculus to derive the truth of formulas. Assertions are judgements of the form $\seq\phi 1n\force\phi$, which means that $\phi$ can be derived from the assumptions $\seq\phi 1n$. If $\force\phi$ without assumptions, then $\phi$ is a theorem of intuitionistic logic. The system is called _natural deduction_. As we write down the proof rules, it will be clear that they correspond exactly to the typing rules of the pure simply-typed $\lambda$-calculus $\lambda^\to$ (and with quantifiers, System F). We will show them side by side. There are generally _introduction_ and _elimination_ rules for each operator. \[ \begin{array}{ccc} & \mbox{\textit{intuitionistic logic}} & \mbox{\textit{$\lambda^\to$ or System F type system}}\\ \hline\\ \mbox{(axiom)} & \Gamma,\,\phi\force\phi & \judge{\Gamma,\,x:\tau}x\tau\\[1em] \mbox{($\to$-intro)} & \dfrac{\Gamma,\,\phi\force\psi}{\Gamma\force\phi\to\psi} & \dfrac{\judge{\Gamma,\,x:\sigma}e\tau}{\judge\Gamma{(\lam{x:\sigma}e)}{\sigma\to\tau}}\\[1em] \mbox{($\to$-elim)} & \dfrac{\Gamma\force\phi\to\psi \quad \Gamma\force\phi}{\Gamma\force\psi} & \dfrac{\judge\Gamma{e_0}{\sigma\to\tau} \quad \judge\Gamma{e_1}\sigma}{\judge\Gamma{(e_0~e_1)}\tau}\\[1em] \mbox{($\wedge$-intro)} & \dfrac{\Gamma\force\phi\quad\Gamma\force\psi}{\Gamma\force\phi\wedge\psi} & \dfrac{\judge\Gamma{e_1}\sigma\quad\judge\Gamma{e_2}\tau}{\judge\Gamma{(e_1,e_2)}{\sigma*\tau}}\\[1em] \mbox{($\wedge$-elim)} & \dfrac{\Gamma\force\phi\wedge\psi}{\Gamma\force\phi}\quad\dfrac{\Gamma\force\phi\wedge\psi}{\Gamma\force\psi} & \dfrac{\judge\Gamma e{\sigma*\tau}}{\judge\Gamma{\#1\,e}\sigma}\quad\dfrac{\judge\Gamma e{\sigma*\tau}}{\judge\Gamma{\#2\,e}\tau}\\[1em] \mbox{($\vee$-intro)} & \dfrac{\Gamma\force\phi}{\Gamma\force\phi\vee\psi} \quad \dfrac{\Gamma\force\psi}{\Gamma\force\phi\vee\psi} & \dfrac{\judge\Gamma e\sigma}{\judge\Gamma{"inl"_{\sigma+\tau}}e{\sigma+\tau}} \quad \dfrac{\judge\Gamma e\tau}{\judge\Gamma{"inr"_{\sigma+\tau}}e{\sigma+\tau}}\\[1em] \mbox{($\vee$-elim)} & \dfrac{\Gamma\force\phi\vee\psi\quad\Gamma\force\phi\to\chi\quad\Gamma\force\psi\to\chi}{\Gamma\force\chi} & \dfrac{\judge\Gamma e{\sigma+\tau} \quad \judge\Gamma{e_1}{\sigma\arrow\rho} \quad \judge\Gamma{e_2}{\tau\arrow\rho}}{\judge\Gamma{"case"~e_0~"of"~e_1~|~e_2}\rho}\\[1em] \mbox{($\forall$-intro)} & \dfrac{\Gamma,\,P\force\phi}{\Gamma\force\fa P\phi} & \dfrac {\Judge{\Delta,\,\alpha}\Gamma e\tau \quad \alpha\notin\FV\Gamma} {\Judge\Delta\Gamma{(\Lam\alpha e)}{\fa\alpha\tau}}\\[1em] \mbox{($\forall$-elim)} & \dfrac{\Gamma\force\fa P\phi}{\Gamma\force\subst\phi\psi P} & \dfrac {\Judge\Delta\Gamma e{\fa\alpha\tau} \quad \Delta\force\sigma} {\Judge\Delta\Gamma{(e~\sigma)}{\subst\tau\sigma\alpha}} \end{array} \] The $\to$-elimination rule is often called _modus ponens_. \section{The Curry--Howard Isomorphism} The fact that propositions in intuitionistic logic correspond to types in our $\lambda$-calculus type systems is known as the _Curry--Howard isomorphism_ or the _propositions as types_ principle. The analogy is far reaching: \begin{center} \begin{tabular}{ll@{\qquad\qquad}ll} \multicolumn 2 c{\textit{type theory}} & \multicolumn 2 c{\textit{logic}}\\ \hline $\tau$ & type & $\phi$ & proposition\\ $\tau$ & inhabited type & $\phi$ & theorem\\ $e$ & well-typed program & $\pi$ & proof\\ $\to$ & function space & $\to$ & implication\\ * & product & $\wedge$ & conjunction\\ + & sum & $\vee$ & disjunction\\ $\forall$ & type quantifier & $\forall$ & 2nd order quantifier\\ 1 & unit & $\top$ & truth\\ 0 & void & $\bot$ & falsity \end{tabular} \end{center} A proof in intuitionistic logic is a construction, which is essentially a program ($\lambda$-term). Saying that a proposition has an intuitionistic or constructive proof says essentially that the corresponding type is inhabited by a $\lambda$-term. If we are given a well-typed term in System F or $\lambda^{\arrow}$, then its proof tree will look exactly like the proof tree for the corresponding formula in intuitionistic logic. This means that every well-typed program proves something, i.e.\ is a proof in constructive logic. Conversely, every theorem in constructive logic corresponds to an inhabited type. Several automated deduction systems (e.g.\ Nuprl, Coq) are based on this idea. \section{Theorem Proving and Type Checking} We have seen that _type inference_ is the process of inferring a type for a given $\lambda$-term. Under the Curry--Howard isomorphism, this is the same as determining what theorem a given proof proves. Theorem proving, on the other hand, is going in the opposite direction: Given a formula, does it have a proof? Equivalently, given a type, is it inhabited? For example, consider the formula expressing transitivity of implication: \[ \forall P,Q,R\,.~((P\to Q) \wedge (Q \to R))\ \to\ ( P \to R) \] Under the Curry--Howard isomorphism, this is related to the type \[ \forall\alpha,\beta,\gamma\,.~(\alpha\to\beta)*(\beta\to\gamma)\ \to\ (\alpha\to\gamma). \] If we can construct a term of this type, we will have proved the theorem in intuitionistic logic. The program \[ \Lam{\alpha,\beta,\gamma}{\lam {p:(\alpha\to\beta)*(\beta\to\gamma)}{\lam{x:\alpha}{(\#2\,p)~((\#1\,p)~x)}}} \] does it. This is a function that takes a pair of functions as its argument and returns their composition. The proof tree that establishes the typing of this function is essentially an intuitionistic proof of the transitivity of implication. Here is another example. Consider the formula \[ \forall P,Q,R\,.~(P\wedge Q\to R)\ \leftrightarrow\ (P\to Q\to R) \] The double implication $\leftrightarrow$ is an abbreviation for the conjunction of the implications in both directions. It says that the two formulas on either side are propositionally equivalent. The typed expressions corresponding to each side of the formula above are \( \alpha*\beta\to\gamma &\qquad& \alpha\to\beta\to\gamma. \) We know that any term of the first type can be converted to one of the second by _currying_, and we can go in the opposite direction by _uncurrying_. The two $\lambda$-terms that convert a function to its curried form and back constitute a proof of the logical statement. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% \end{document}