Statistiques
| Révision :

root / CSL17 / soundness.tex @ 202

Historique | Voir | Annoter | Télécharger (6,77 ko)

1 166 adas
\section{Soundness}
2 168 adas
\label{sect:soundness}
3 168 adas
4 168 adas
The main result of this section is the following:
5 168 adas
6 168 adas
\begin{theorem}
7 202 adas
	\label{thm:soundness}
8 168 adas
	If $\arith^i$ proves $\forall \vec u^\normal . \forall \vec x^\safe . \exists y^\safe . A(\vec u ; \vec x , y)$ then there is a $\mubci{i-1}$ program $f(\vec u ; \vec x)$ such that $\Nat \models A(\vec u ; \vec x , f(\vec u ; \vec x))$.
9 168 adas
\end{theorem}
10 168 adas
11 171 adas
12 172 adas
13 172 adas
The main problem for soundness is that we have predicates, for example equality, that take safe arguments in our theory but do not formally satisfy the polychecking lemma for $\mubc$ functions.
14 172 adas
For this we will use length-bounded witnessing, borrowing a similar idea from Bellantoni's previous work \cite{Bellantoni95}.
15 172 adas
16 201 adas
17 172 adas
\begin{definition}
18 172 adas
[Length bounded basic functions]
19 172 adas
We define \emph{length-bounded equality}, $\eq(l;x,y)$ as the characteristic function of the predicate:
20 172 adas
\[
21 172 adas
x \mode l = y \mode l
22 172 adas
\]
23 201 adas
which is definable by safe recursion on $l$:
24 201 adas
\[
25 201 adas
\begin{array}{rcl}
26 201 adas
\eq (0 ; x,y) & \dfn & \equivfn (;\bit (0;x),\bit(0;y) ) \\
27 201 adas
\eq (\succ i l; x,y) & \dfn & \cond (; \eq ( u;x,y ) , 0, \equivfn (; \bit (\succ i u ; x ) , \bit (\succ i l ; y ))  )
28 201 adas
\end{array}
29 201 adas
\]
30 202 adas
We also define length-bounded inequality as:
31 202 adas
\[
32 202 adas
\begin{array}{rcl}
33 202 adas
\leqfn (0 ; x ,y) & \dfn & \cimp (; \bit (0;x), \bit (0;y) ) \\
34 202 adas
\leqfn (\succ i l ; x,y) & \dfn & \orfn ( ; <(\bit (\succ i l ; x) , \bit(\succ i l ; y) ) , \andfn (; \equivfn (\bit (\succ i l ; x) , \bit(\succ i l ; y)) , \leqfn (l;x,y ) ) )
35 202 adas
\end{array}
36 202 adas
\]
37 172 adas
\end{definition}
38 172 adas
39 172 adas
\anupam{Do we need the general form of length-boundedness? E.g. the $*$ functions from Bellantoni's paper? Put above if necessary. Otherwise just add sequence manipulation functions as necessary.}
40 172 adas
41 172 adas
Notice that $\eq$ is a polymax bounded polyomial checking function on its normal input, and so can be added to $\mubc$ without problems.
42 172 adas
43 172 adas
\begin{definition}
44 172 adas
	[Length bounded characteristic functions]
45 172 adas
	We define $\mubci{}$ programs $\charfn{}{A} (l , \vec u; \vec x)$, parametrised by a formula $A(\vec u ; \vec x)$, as follows.
46 172 adas
%	If $A$ is a $\Pi_{i}$ formula then:
47 172 adas
	\[
48 172 adas
	\begin{array}{rcl}
49 202 adas
	\charfn{}{s\leq t} (l, \vec u ; \vec x) & \dfn & \leqfn(l;s,t) \\
50 172 adas
	\smallskip
51 202 adas
	\charfn{}{s=t} (l, \vec u ; \vec x) & \dfn & \eq(l;s,t) \\
52 172 adas
	\smallskip
53 202 adas
	\charfn{}{\neg A} (l, \vec u ; \vec x) & \dfn & \neg (;\charfn{}{A}(l , \vec u ; \vec x)) \\
54 172 adas
	\smallskip
55 202 adas
	\charfn{}{A\cor B} (l, \vec u ; \vec x ) & \dfn & \cor (; \charfn{}{A} (l, \vec u ; \vec x , w), \charfn{}{B} (\vec u ;\vec x) ) \\
56 172 adas
	\smallskip
57 202 adas
	\charfn{}{A\cand B} (l, \vec u ; \vec x ) & \dfn & \cand(; \charfn{}{A} (l, \vec u ; \vec x , w), \charfn{}{B} (\vec u ;\vec x) ) \\
58 172 adas
	\smallskip
59 202 adas
	\charfn{}{\exists x^\safe . A(x)} (l, \vec u ;\vec x) & \dfn & \begin{cases}
60 202 adas
	1 & \exists x^\safe . \charfn{}{A(x)} (l, \vec u ;\vec x , x) = 1 \\
61 172 adas
	0 & \text{otherwise}
62 172 adas
	\end{cases} \\
63 172 adas
	\smallskip
64 202 adas
	\charfn{}{\forall x^\safe . A(x)} (l, \vec u ;\vec x ) & \dfn &
65 172 adas
	\begin{cases}
66 173 adas
	0 & \exists x^\sigma. \charfn{}{ A(x)} (l, \vec u; \vec x , x) = 0 \\
67 172 adas
	1 & \text{otherwise}
68 172 adas
	\end{cases}
69 172 adas
	\end{array}
70 172 adas
	\]
71 173 adas
\end{definition}
72 173 adas
73 173 adas
74 176 adas
\begin{proposition}
75 176 adas
	$\charfn{}{A} (l, \vec u ; \vec x)$ is the characteristic function of $A (\vec u \mode l ; \vec x \mode l)$.
76 176 adas
\end{proposition}
77 176 adas
78 173 adas
\begin{definition}
79 173 adas
	[Length bounded witness function]
80 173 adas
	We now define $\Wit{\vec u ; \vec x}{A} (l , \vec u ; \vec x)$ for a $\Sigma_{i+1}$-formula $A$ with free variables amongst $\vec u; \vec x$.
81 173 adas
	\[
82 173 adas
	\begin{array}{rcl}
83 173 adas
	\Wit{\vec u ; \vec x}{A} (l, \vec u ; \vec x , w) & \dfn & \charfn{}{A} (l, \vec u ; \vec x)  \text{ if $A$ is $\Pi_i$} \\
84 173 adas
	\smallskip
85 173 adas
	\Wit{\vec u ; \vec x}{A \cor B} (l,\vec u ; \vec x , \vec w^A , \vec w^B) & \dfn & \cor ( ; \Wit{\vec u ; \vec x}{A} (l,\vec u ; \vec x , \vec w^A) ,\Wit{\vec u ; \vec x}{B} (l,\vec u ; \vec x , \vec w^B)  )  \\
86 173 adas
	\smallskip
87 173 adas
	\Wit{\vec u ; \vec x}{A \cand B} (l,\vec u ; \vec x , \vec w^A , \vec w^B) & \dfn & \cand ( ; \Wit{\vec u ; \vec x}{A} (l,\vec u ; \vec x , \vec w^A) ,\Wit{\vec u ; \vec x}{B} (l,\vec u ; \vec x , \vec w^B)  )  \\
88 173 adas
	\smallskip
89 173 adas
	\Wit{\vec u ; \vec x}{\exists x^\safe . A(x)} (l,\vec u ; \vec x , \vec w , w) & \dfn & \Wit{\vec u ; \vec x , x}{A(x)} ( l,\vec u ; \vec x , w , \vec w )
90 173 adas
	\\
91 173 adas
	\smallskip
92 173 adas
	\Wit{\vec u ; \vec x}{\forall u \leq |t(\vec u;)| . A(x)} (l , \vec u ; \vec x, w) & \dfn &
93 173 adas
	\forall u \leq |t(\vec u;)| . \Wit{u , \vec u ; \vec x}{A(u)} (l, u , \vec u ; \vec x, \beta(u;w) )
94 173 adas
	\end{array}
95 173 adas
	\]
96 173 adas
\end{definition}
97 173 adas
98 173 adas
\anupam{may as well use a single witness variable since need it for sharply bounded quantifiers anyway.}
99 173 adas
100 202 adas
\anupam{sharply bounded case obtained by sharply bounded lemma}
101 202 adas
102 202 adas
103 202 adas
\begin{proposition}
104 202 adas
	If, for some $w$, $\wit{\vec u ; \vec x}{A} (l, \vec u ; \vec x,w) =1$, then $A$ is true.
105 202 adas
	\anupam{check statement, need proof-theoretic version?}
106 202 adas
\end{proposition}
107 202 adas
108 202 adas
In order to prove Thm.~\ref{thm:soundness} we need the following lemma:
109 202 adas
110 202 adas
111 202 adas
\begin{lemma}
112 202 adas
	[Proof interpretation]
113 202 adas
	\label{lem:proof-interp}
114 202 adas
	For any $\arith^i$ proof of a $\Sigma^\safe_i$ sequent $\Gamma \seqar \Delta$, there is a $\mubci{i-1}$ function $f$ such that, for any $l, \vec u , \vec x  , w$, we have:
115 202 adas
	\[
116 202 adas
	\wit{\vec u ; \vec x}{ \wedge \Gamma } (l, \vec u ; \vec x , w)
117 202 adas
	\quad \leq \quad
118 202 adas
	\wit{\vec u ; \vec x}{\vee \Delta} (l, \vec u ; \vec x , f(l, \vec u ; \vec x , w))
119 202 adas
	\]
120 202 adas
	\anupam{maybe want $f(\vec u \mode l ; \vec x \mode l , w)$}
121 202 adas
\end{lemma}
122 202 adas
\begin{proof}
123 202 adas
	We assume the proof, say $\pi$, is in integer positive free-cut free form, by the results from the previous section.
124 202 adas
	This means that the predicate $\charfn{\vec u ; \vec x}{A}$ is defined for each formula $A(\vec u ; \vec x)$ occurring in a proof, so the theorem is well-stated.
125 202 adas
	We define the function $f$ inductively, by considering the various final rules of $\pi$.
126 202 adas
127 202 adas
	\paragraph*{Negation}
128 202 adas
	Can assume only on atomic formulae, so no effect.
129 202 adas
	\paragraph*{Quantifiers}
130 202 adas
	\anupam{Do $\exists$-right and $\forall$-right, left rules are symmetric.}
131 202 adas
132 202 adas
	\paragraph*{Contraction}
133 202 adas
	Left contraction simply duplicates an argument, whereas right contraction requires a conditional on a $\Sigma^\safe_i$ formula.
134 202 adas
135 202 adas
	\paragraph*{Induction}
136 202 adas
	Corresponds to safe recursion on notation.
137 202 adas
\end{proof}
138 202 adas
139 202 adas
We are now ready to prove the soundness theorem.
140 202 adas
141 202 adas
\begin{proof}
142 202 adas
	[Proof of Thm.~\ref{thm:soundness} from Lemma~\ref{lem:proof-interp}]
143 202 adas
	(watch out for dependence on $l$, try do without)
144 202 adas
145 202 adas
	Suppose $\arith^i \proves \forall \vec u^\normal . \exists x^\normal . A(\vec u ; x)$. Then by raising, inversion, free-variable normal forms, we have a proof of,
146 202 adas
	\[
147 202 adas
	\normal (\vec u ) \seqar \exists x^\normal . A(\vec u , x ;)
148 202 adas
	\]
149 202 adas
	whence, by Lemma~\ref{lem:proof-interp}, we have a $\mubci{i-1}$ function $f$ such that:
150 202 adas
	\[
151 202 adas
	\vec u \mode l = \vec a \mode l
152 202 adas
	\quad \implies \quad
153 202 adas
	\wit{\vec u ; }{A} ( l , \vec u , f(l, \vec u;) ) =1
154 202 adas
	\]
155 202 adas
156 202 adas
	Now it suffices to choose an $l$ bigger that both all the $\vec u$ and $f(\vec u)$, which is a polynomial in $\vec u$ by the polymax bounding lemma.
157 202 adas
\end{proof}