22.01.2014 Views

Metatheory - University of Cambridge

Metatheory - University of Cambridge

Metatheory - University of Cambridge

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

3. Normal forms 26<br />

4a: B ∗ contains a disjunction. In that case, since B ∗ is in DNF, B ∗ is<br />

(B 1 ∨ B 2 ), for some B 1 and B 2 in DNF that are both disjunctivelyshorter<br />

than B ∗ . Next, observe:<br />

(B ∧ C )<br />

⊨<br />

⊨<br />

⊨ ((B 1 ∨ B 2 ) ∧ C )<br />

⊨ ((B 1 ∧ C ) ∨ (B 2 ∧ C ))<br />

Now, (B 1 ∧ C ) is disjunctively-shorter than A; so by hypothesis,<br />

there is some DNF sentence D which is exactly as disjunctivelylong<br />

as (B 1 ∧ C ) and such that D ⊨ (B 1 ∧ C ). Similarly, there<br />

is some DNF sentence E which is exactly as disjunctively-long as<br />

(B 2 ∧ C ) and such that E ⊨ (B 2 ∧ C ). So A ⊨ (D ∨ E), which is<br />

in DNF and exactly as disjunctively-long as A.<br />

4b: C ∗ contains a disjunction. Exactly similar to Subcase 4a; I leave the<br />

details as an exercise.<br />

4c: neither B ∗ nor C ∗ contains a disjunction. Then A ⊨ (B ∗ ∧ C ∗ ),<br />

which is already in DNF and is exactly as long as A itself.<br />

This completes the induction.<br />

⊨<br />

Again, this Lemma implicitly provides an algorithm; this time, for pulling<br />

disjunctions outside the scope <strong>of</strong> any conjunction, by repeatedly applying the<br />

Distribution Laws:<br />

((A ∨ B) ∧ C )<br />

(A ∧ (B ∨ C ))<br />

⊨<br />

⊨<br />

⊨<br />

⊨<br />

⊨ ((A ∧ C ) ∨ (B ∧ C ))<br />

⊨ ((A ∧ B) ∨ (A ∧ C ))<br />

Applying these to our earlier example, we get, successively:<br />

((A ∧ (B ∨ ¬C)) ∨ ((¬D ∨ ¬E) ∧ (D ∨ E)))<br />

(((A ∧ B) ∨ (A ∧ ¬C)) ∨ ((¬D ∨ ¬E) ∧ (D ∨ E)))<br />

(((A ∧ B) ∨ (A ∧ ¬C)) ∨ (((¬D ∨ ¬E) ∧ D) ∨ ((¬D ∨ ¬E) ∧ E)))<br />

(((A ∧ B) ∨ (A ∧ ¬C)) ∨ (((¬D ∧ D) ∨ (¬E ∧ D)) ∨ ((¬D ∧ E) ∨ (¬E ∧ E))))<br />

And this is in DNF, as may become clearer if we temporarily allow ourselves<br />

to remove brackets, following the notational conventions <strong>of</strong> fx C§10.3:<br />

(A ∧ B) ∨ (A ∧ ¬C) ∨ (¬D ∧ D) ∨ (¬E ∧ D) ∨ (¬D ∧ E) ∨ (¬E ∧ E)<br />

Together, then, these Lemmas implicitly provide us with an algorithm for computing<br />

an equivalent sentence, in DNF. Moreover, this algorithm – although<br />

lengthy – was much more efficient at generating a DNF sentence than the<br />

truth-table method would have been. (The truth table for our example would<br />

have 32 lines, 22 <strong>of</strong> which are true.)<br />

It remains to assemble these Lemmas, into a pro<strong>of</strong> <strong>of</strong> the DNF Theorem:<br />

Pro<strong>of</strong> <strong>of</strong> DNF Theorem via substitution. Let A be any TFL sentence.<br />

Lemma 3.2 tells us that there is a sentence, B, satisfying (dnf1), such that<br />

A ⊨ B.<br />

Lemma 3.3 tells us that there is a sentence, C , satisfying (dnf1) and (dnf2),<br />

such that B ⊨ C .<br />

Lemma 3.4 tells us that there is a sentence, D, in DNF, such that C ⊨ D.<br />

And, by the transitivity <strong>of</strong> entailment, A ⊨ D.<br />

■<br />

⊨<br />

⊨<br />

⊨<br />

⊨<br />

⊨<br />

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

Saved successfully!

Ooh no, something went wrong!