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.

23<br />

Arithmetical Definability<br />

Tarski’s theorem tells us that the set V of (code numbers of) first-order sentences of the<br />

language in arithmetic that are true in the st<strong>and</strong>ard interpretation is not arithmetically<br />

definable. In section 23.1 we show that this negative result is poised, so to speak, between<br />

two positive results. One is that for each n the set V n of sentences of the language of<br />

arithmetic of degree of complexity n that are true in the st<strong>and</strong>ard interpretation is<br />

arithmetically definable (in a sense of degree of complexity to be made precise). The<br />

other is that the class {V} of sets of natural numbers whose one <strong>and</strong> only member is V<br />

is arithmetically definable (in a sense of arithmetical definability for classes to be made<br />

precise). In section 23.2 we take up the question whether the class of arithmetically<br />

definable sets of numbers is an arithmetically definable class of sets. The answer is<br />

negative, according to Addison’s theorem. This result is perhaps most interesting on<br />

account of its method of proof, which is a comparatively simple application of the method<br />

of forcing originally devised to prove the independence of the continuum hypothesis in<br />

set theory (as alluded to in the historical notes to Chapter 18).<br />

23.1 Arithmetical Definability <strong>and</strong> Truth<br />

Throughout this chapter we use L <strong>and</strong> N for the language of arithmetic <strong>and</strong> its<br />

st<strong>and</strong>ard interpretation (previously called L* <strong>and</strong> N *), <strong>and</strong> V for the set of code<br />

numbers of first-order sentences of L ture in N . It will be convenient to work with<br />

a version of logic in which the only operators are ∼ <strong>and</strong> ∨ <strong>and</strong> ∃ (& <strong>and</strong> ∀ being<br />

treated as unofficial abbreviations). We measure the ‘complexity’ of a sentence by<br />

the number of occurrences of logical operators ∼ <strong>and</strong> ∨ <strong>and</strong> ∃ in it. (Our results<br />

do, however, go through for other reasonable notions of measures of complexity: see<br />

the problems at the end of the chapter). By V n we mean the set of code numbers of<br />

first-order sentences of L of complexity ≤n that are true in N .<br />

We are going to be discussing natural numbers, sets of natural numbers, <strong>and</strong> sets<br />

of sets of natural numbers. To keep the levels straight, we generally use numbers for<br />

the natural numbers, sets for the sets of natural numbers, <strong>and</strong> classes for the sets of<br />

sets of natural numbers. We write L c for the expansion of L by adding a constant c,<br />

<strong>and</strong> Na c for the expansion of N that assigns c as denotation the number a. Then a<br />

set S of numbers is arithmetically definable if <strong>and</strong> only if there is a sentence F(c)of<br />

L c such that S is precisely the set of a for which F(c) is true in Na c. Analogously,<br />

we write L G for the expansion of L by adding a one-place predicate G, <strong>and</strong> NA G for<br />

286

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

Saved successfully!

Ooh no, something went wrong!