27.04.2015 Views

Computability and Logic

Computability and Logic

Computability and Logic

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

274 MONADIC AND DYADIC LOGIC<br />

Our induction hypothesis is that if a 1 ,...,a s , a s+1 <strong>and</strong> b 1 ,...,b s , b s+1 match,<br />

then a 1 ,...,a s , a s+1 satisfy Q in M if <strong>and</strong> only if b 1 ,...,b s , b s+1 satisfy Q in N .<br />

What we want to show is that if b 1 ,...,b s , b s+1 satisfy Q in N for all b s+1 in N,<br />

then a 1 ,...,a s , a s+1 satisfy Q in M for all a s+1 in M. Therefore it will be enough<br />

to show that if a 1 ,...,a s <strong>and</strong> b 1 ,...,b s match, where s < r, then for any a s+1 in M<br />

there is a b s+1 in N such that a 1 ,...,a s , a s+1 <strong>and</strong> b 1 ,...,b s , b s+1 match.<br />

In the degenerate case where a s+1 is identical with one of the previous a i , we may<br />

simply take b s+1 to be identical with the corresponding b i . In the non-degenerate case,<br />

a s+1 belongs to some equivalence class C <strong>and</strong> is distinct from any <strong>and</strong> all previous a i<br />

that belong to C. Let the number of such a i be t (where possibly t = 0), so that there<br />

are at least t + 1 elements in C, counting a s+1 . To ensure matching, it will suffice<br />

to choose b s+1 to be some element of C that is distinct from any <strong>and</strong> all previous b i<br />

that belong to C. Since a 1 ,...,a s <strong>and</strong> b 1 ,...,b s match, the number of such b i will<br />

also be t. Since t ≤ s < r, <strong>and</strong> there are at least t + 1 ≤ r elements in C, there will<br />

be at least that many elements of C in N, <strong>and</strong> so we can find an appropriate b s+1 ,to<br />

complete the proof.<br />

21.10 Corollary. If a sentence involving no nonlogical symbols (but only identity) is<br />

satisfiable, then it has a model of size no greater than r, where r is the number of variables<br />

in the sentence.<br />

21.11 Corollary. If a sentence of monadic logic involving only one variable is satisfiable,<br />

then it has a model of size no greater than 2 k , where k is the number of monadic<br />

predicates in the sentence.<br />

Proofs: These are simply the cases k = 0 <strong>and</strong> r = 1 of Lemma 21.9.<br />

Proof of Lemma 21.7: This is immediate from Corollary 21.11 <strong>and</strong> the following, which<br />

is a kind of normal form theorem.<br />

21.12 Lemma. Any sentence of monadic logic without identity is logically equivalent<br />

to one with the same predicates <strong>and</strong> only one variable.<br />

Proof: Call a formula clear if in any subformula ∀xB(x) or∃xB(x) that begins<br />

with a quantifier, no variable other than the variable x attached to the quantifier<br />

appears in F. Thus ∀x∃y(Fx & Gy) is not clear, but ∀xFx& ∃ yGy is clear. To prove<br />

the lemma, we show how one can inductively associate to any formula A of monadic<br />

logic without identity an equivalent formula A © with the same predicates that is clear<br />

(as in our example the first formula is equivalent to the second). We then note that any<br />

clear sentence is equivalent to the result of rewriting all its variables to be the same (as<br />

in our example the second sentence is equivalent to ∀zFz& ∃zGz). The presence of<br />

identity would make such clearing impossible. (There is no clear sentence equivalent<br />

to ∀x∃yx≠ y, for instance.)<br />

To an atomic formula we associate itself. To a truth-functional compound of formulas<br />

to which clear equivalents have been associated, we associate the same truthfunctional<br />

compound of those equivalents. Thus (B ∨ C) © is B © ∨ C © , for instance,<br />

<strong>and</strong> analogously for &. The only problem is how to define the associate (∃xB(x)) ©

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!