* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download PDF
Survey
Document related concepts
Bayesian inference wikipedia , lookup
Model theory wikipedia , lookup
Intuitionistic logic wikipedia , lookup
Laws of Form wikipedia , lookup
Mathematical logic wikipedia , lookup
Georg Cantor's first set theory article wikipedia , lookup
Mathematical proof wikipedia , lookup
Gödel's incompleteness theorems wikipedia , lookup
Law of thought wikipedia , lookup
Curry–Howard correspondence wikipedia , lookup
Quantum logic wikipedia , lookup
Propositional calculus wikipedia , lookup
Natural deduction wikipedia , lookup
Transcript
substitution theorem for propositional logic∗ CWoo† 2013-03-22 3:57:08 In this entry, we will prove the substitution theorem for propositional logic based on the axiom system found here. Besides the deduction theorem, below are some additional results we will need to prove the theorem: 1. If ∆ ` A → B and Γ ` B → C, then ∆, Γ ` A → C. 2. ∆ ` A and ∆ ` B iff ∆ ` A ∧ B. 3. ` A ↔ A. 4. ` A ↔ ¬¬A (law of double negation). 5. ⊥→ A (ex falso quodlibet) 6. ∆ ` A implies ∆ ` B iff ∆ ` A → B. The proofs of these results can be found here. Theorem 1. (Substitution Theorem) Suppose p1 , . . . , pm are all the propositional variables, not necessarily distinct, that occur in order in A, and if B1 , . . . , Bm , C1 , . . . , Cm are wff ’s such that ` Bi ↔ Ci , then ` A[B1 /p1 , . . . , Bm /pm ] ↔ A[C1 /p1 , . . . , Cm /pm ] where A[X1 /p1 , . . . , Xm /pm ] is the wff obtained from A by replacing pi by the wff Xi via simultaneous substitution. Proof. We do induction on the number n of → in wff A. If n = 0, A is either a propositional variable, say p, or ⊥, which respectively means that A[B/p] ↔ A[C/p] is either B ↔ C or ⊥↔⊥. The former is the assumption and the latter is a theorem. ∗ hSubstitutionTheoremForPropositionalLogici created: h2013-03-2i by: hCWooi version: h42534i Privacy setting: h1i hResulti h03B05i † This text is available under the Creative Commons Attribution/Share-Alike License 3.0. You can reuse this document or portions thereof only if you do so under terms that are compatible with the CC-BY-SA license. 1 Suppose now A has n + 1 occurrences of →. We may write A as X → Y uniquely by unique readability. Also, both X and Y have at most n occurrences of →. Let A1 be A[B1 /p1 , . . . , Bm /pm ] and A2 be A[C1 /p1 , . . . , Cm /pm ]. Then A1 is X1 → Y1 and X2 → Y2 , where X1 is X[B1 /p1 , . . . , Bk /pk ], Y1 is Y [Bk+1 /pk+1 , . . . , Bm /pm ], X2 is X[C1 /p1 , . . . , Ck /pk ], and Y2 is Y [Ck+1 /pk+1 , . . . , Cm /pm ]. Then ` X1 ↔ X2 (1) by induction ` X1 → X2 and ` X2 → X1 (2) by 2 above ` Y1 ↔ Y2 (3) by induction ` Y1 → Y2 and ` Y2 → Y1 (4) by 2 above since A1 is X1 → Y1 A1 ` X1 → Y1 (5) by applying 1 to ` X2 → X1 and (5) A1 ` X2 → Y1 (6) by applying 1 to (6) and ` Y1 → Y2 A1 ` X2 → Y2 (7) by the deduction theorem ` A1 → A2 (8) by a similar reasoning as above ` A2 → A1 (9) by applying 2 to (8) and (9) ` A1 ↔ A2 (10) As a corollary, we have Corollary 1. If ` B ↔ C, then ` A[B/s(p)] ↔ A[C/s(p)], where p is a propositional variable that occurs in A, s(p) is a set of positions of occurrences of p in A, and the wff A[X/s(p)] is obtained by replacing all p that occur in the positions in s(p) in A by wff X. Proof. For any propositional variable q not being replaced, use the corresponding theorem ` q ↔ q, and then apply the substitution theorem. Remark. What about ` B[A/p] ↔ C[A/p], given ` B ↔ C? Here, B[A/p] and C[A/p] are wff’s obtained by uniform substitution of p (all occurrences of p) in B and C respectively. Since B[A/p] ↔ C[A/p] is just (B ↔ C)[A/p], an 2 instance of the schema B ↔ C by assumption, the result follows directly if we assume B ↔ C is a theorem schema. Using the substitution theorem, we can easily derive more theorem schemas, such as 7. (A → B) ↔ (¬B → ¬A) (Law of Contraposition) 8. A → (¬B → ¬(A → B)) 9. ((A → B) → A) → A (Peirce’s Law) Proof. 7. Since (¬B → ¬A) → (A → B) is already a theorem schema, we only need to show ` (A → B) → (¬B → ¬A). By law of double negation (4 above) and the substitution theorem, it is enough to show that ` (¬¬A → ¬¬B) → (¬B → ¬A). But this is just an instance of an axiom schema. Combining the two schemas, we get ` (A → B) ↔ (¬B → ¬A). 8. First, observe that A, A → B ` B by modus ponens. Since ` B ↔ ¬¬B, we have A, A → B ` ¬¬B by the substitution theorem. So A, A → B, ¬B `⊥ by the deduction theorem, and A, ¬B ` (A → B) →⊥ by the deduction theorem again. Apply the deduction two more times, we get ` A → (¬B → ¬(A → B)). 9. To show ` ((A → B) → A) → A, it is enough to show ` ¬A → ¬((A → B) → A) by 7 and modus ponens, or ¬A ` ¬((A → B) → A) by the deduction theorem. Now, since ` X ∧ Y ↔ ¬(X → ¬Y ) (as they are the same thing, and because C ↔ C is a theorem schema), by the law of double negation and the substitution theorem, ` X ∧ ¬Y ↔ ¬(X → Y ), and we have ` (A → B) ∧ ¬A ↔ ¬((A → B) → A). So to show ¬A ` ¬((A → B) → A), it is enough to show ¬A ` (A → B) ∧ ¬A, which is enough to show that ¬A ` A → B and ¬A ` ¬A, according to a meta-theorem found here. To show ¬A ` A → B, it is enough to show {¬A, A} ` B, and A, ¬A, ⊥, ⊥→ B, B is such a deduction. The second statement ¬A ` ¬A is clear. As an application, we prove the following useful meta-theorems of propositional logic: Proposition 1. There is a wff A such that ∆ ` A and ∆ ` ¬A iff ∆ `⊥ Proof. Assume the former. Let E1 be a deduction of A from ∆ and E2 a deduction of ¬A from ∆, then E1 , E2 , ⊥ is a deduction of ⊥ from ∆. Conversely, assume the later. Pick any wff A (if necessary, pick ⊥). Then ⊥→ A by ex falso quodlibet. By modus ponens, we have ∆ ` A. Similarly, ∆ ` ¬A. Proposition 2. If ∆, A ` B and ∆, ¬A ` B, then ∆ ` B 3 Proof. By assumption, we have ∆ ` A → B and ∆ ` ¬A → B. Using modus ponens and the theorem schema (A → B) → (¬B → ¬A), we have ∆ ` ¬B → ¬A, or ∆, ¬B ` ¬A. Similarly, ∆ ` ¬B → ¬¬A. By the law of double negation and the substitution theorem, we have ∆ ` ¬B → A, or ∆, ¬B ` A. By the previous proposition, ∆, ¬B `⊥, or ∆ ` ¬¬B. Applying the substitution theorem and the law of double negation, we have ∆ ` B. 4