27.04.2015 Views

Computability and Logic

Computability and Logic

Computability and Logic

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

180 PROOFS AND COMPLETENESS<br />

possible, so that no strictly shorter derivation would be a counterexample. So suppose<br />

that Ɣ ∪{∼A}⇒ is the sequent derived in such a shortest possible counterexample.<br />

We ask by what rule the last step Ɣ ∪{∼A}⇒ could have been justified.<br />

Could it have been (R0)? If that were so, the counterexample would simply be<br />

the one-step derivation of {∼A}⇒{∼A}, <strong>and</strong> we would have Ɣ = ∅, ={∼A}.<br />

The sequent Ɣ ⇒{A}∪ for which supposedly no derivation exists would then just<br />

be ⇒{A, ∼A}. But there is a derivation of this sequent, in two steps, starting with<br />

{A}⇒{A} by (R0) <strong>and</strong> proceeding to ⇒{A, ∼A} by (R2a). So (R0) is excluded,<br />

<strong>and</strong> Ɣ ∪{∼A}⇒ must have been inferred from some earlier step or steps by one<br />

of the other rules.<br />

Could it have been (R3)? If that were so, the counterexample would be a derivation<br />

of<br />

where the last step was obtained from<br />

Ɣ ∪{∼A}⇒{(B ∨ C)}∪ ′<br />

Ɣ ∪{∼A}⇒{B, C}∪ ′ .<br />

But then, since the derivation down to this last-displayed sequent is too short to be a<br />

counterexample, there will be a derivation of<br />

<strong>and</strong> by applying (R3) we can then get<br />

Ɣ ⇒{A}∪{B, C}∪ ′ ,<br />

Ɣ ⇒{A}∪{(B ∨ C)}∪ ′ ,<br />

which is precisely what we are supposed not to be able to get in the case of a<br />

counterexample to the lemma. Thus (R3) is excluded. Moreover, every case where<br />

∼A is not an entering sentence is excluded for entirely similar reasons.<br />

There remain to be considered three cases where ∼A is an entering sentence. One<br />

case where ∼A enters arises when Ɣ ∪{∼A}⇒ is obtained by (R1) from Ɣ ′ ⇒ ′ ,<br />

where Ɣ ′ is a subset of Ɣ not containing ∼A <strong>and</strong> ′ is a subset of . But in this case<br />

Ɣ ⇒{A}∪ equally follows by (R1) from Ɣ ′ ⇒ ′ , <strong>and</strong> we have no counterexample.<br />

If ∼A enters when Ɣ ∪{∼A}⇒ is obtained by (R2b), the premiss must be<br />

Ɣ ⇒{A}∪ itself or Ɣ ∪{∼A}⇒{A}∪, <strong>and</strong> in the latter case, since the derivation<br />

of the premiss is too short to be a counterexample, there must exist a derivation<br />

of Ɣ ⇒{A}∪{A}∪ or Ɣ ⇒{A}∪; so we have no counterexample.<br />

The other case where ∼A enters arises when ∼A is of the form ∼B(s) <strong>and</strong> the last<br />

lines of the derivation are<br />

using (R8b).<br />

Ɣ ∪{∼B(t)}⇒<br />

Ɣ ∪{s = t, ∼B(s)}⇒

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

Saved successfully!

Ooh no, something went wrong!