The Calculus of Computation: Decision Procedures with Applications to Verification 2. First-Order Logic (FOL) by Aaron Bradley Zohar Manna Springer 2007 2- 1 2- 2 First-Order Logic (FOL) quantifiers existential quantifier ∃x.F [x] Also called Predicate Logic or Predicate Calculus “there exists an x such that F [x]” FOL Syntax universal quantifier ∀x.F [x] “for all x, F [x]” variables x, y, z, · · · a b c constants , , , · · · FOL formula literal, application of logical connectives f g h functions , , , · · · (¬, ∨ , ∧ , → , ↔ ) to formulae, terms variables, constants or or application of a quantifier to a formula n-ary function applied to n terms as arguments a, x, f (a), g(x, b), f (g(x, g(b))) predicates p, q, r, · · · atom ⊤, ⊥, or an n-ary predicate applied to n terms literal atom or its negation p(f (x), g(x, f (x))), ¬p(f (x), g(x, f (x))) Note: 0-ary functions: constant 0-ary predicates: P, Q, R,... 2- 3 2- 4 Example: FOL formula Translations of English Sentences into FOL ◮ ∀x. p(f (x), x) → (∃y. p(f (g(x, y)), g(x, y))) ∧ q(x, f (x)) The length of one side of a triangle is less than the sum of the | {z } lengths of the other two sides G | {z } F ∀x, y, z. triangle(x, y, z) → length(x) < length(y)+length(z) The scope of ∀x is F . ◮ Fermat’s Last Theorem. The scope of ∃y is G. The formula reads: ∀n. integer(n) ∧ n > 2 “for all x, → ∀x, y, z. if p(f (x), x) integer(x) ∧ integer(y) ∧ integer(z) then there exists a y such that ∧ x > 0 ∧ y > 0 ∧ z > 0 n n n p(f (g(x, y)), g(x, y)) and q(x, f (x))” → x + y 6= z 2- 5 2- 6 FOL Semantics Example: F : p(f (x, y), z) → p(y, g(z, x)) An interpretation I : (DI , αI ) consists of: Interpretation I : (DI , αI ) ◮ Domain DI DI = Z = {· · · , −2, −1, 0, 1, 2, ···} integers non-empty set of values or objects αI : {f 7→ +, g 7→ −, p 7→>} cardinality |DI | finite (eg, 52 cards), Therefore, we can write countably infinite (eg, integers), or uncountably infinite (eg, reals) FI : x + y > z → y > z − x ◮ Assignment αI (This is the way we’ll write it in the future!) ◮ each variable x assigned value xI ∈ DI Also ◮ each n-ary function f assigned αI : {x 7→ 13, y 7→ 42, z 7→ 1} n fI : DI → DI Thus In particular, each constant a (0-ary function) assigned value FI : 13 + 42 > 1 → 42 > 1 − 13 aI ∈ DI Compute the truth value of F under I ◮ each n-ary predicate p assigned 1. I |= x + y > z since 13 + 42 > 1 n pI : DI →{true, false} 2. I |= y > z − x since 42 > 1 − 13 In particular, each propositional variable P (0-ary predicate) 3. I |= F by 1, 2, and → assigned truth value (true, false) F is true under I 2- 7 2- 8 Semantics: Quantifiers Example For Q, the set of rational numbers, consider x variable. x-variant of interpretation I is an interpretation J : (DJ , αJ ) such FI : ∀x. ∃y. 2 × y = x that ◮ DI = DJ Compute the value of FI (F under I ): ◮ αI [y]= αJ [y] for all symbols y, except possibly x Let v J1 : I ⊳ {x 7→ v} J2 : J1 ⊳ {y 7→ } That is, I and J agree on everything except possibly the value of x 2 x-variant of I y-variant of J1 Denote J : I ⊳ {x 7→ v} the x-variant of I in which αJ [x] = v for for v ∈ Q. some v ∈ DI . Then Then ◮ I |= ∀x. F iff for all v ∈ DI , I ⊳ {x 7→ v} |= F v ◮ I |= ∃x. F iff there exists v ∈ DI s.t. I ⊳ {x 7→ v} |= F 1. J2 |= 2 × y = x since 2 × 2 = v 2. J1 |= ∃y. 2 × y = x 3. I |= ∀x. ∃y. 2 × y = x since v ∈ Q is arbitrary 2- 9 2- 10 Satisfiability and Validity Second case F is satisfiable iff there exists I s.t. I |= F 1. I 6|= ∀x. p(x) assumption F is valid iff for all I , I |= F 2. I |= ¬∃x. ¬p(x) assumption 3. I ⊳ {x 7→ v} 6|= p(x) 1and ∀, for some v ∈ DI F is valid iff ¬F is unsatisfiable 4. I 6|= ∃x. ¬p(x) 2and ¬ 5. I ⊳ {x 7→ v} 6|= ¬p(x) 4and ∃ 6. I ⊳ {x 7→ v} |= p(x) 5and ¬ Example: F : (∀x. p(x)) ↔ (¬∃x. ¬p(x)) valid? Suppose not. Then there is I s.t. 3 and 6 are contradictory. 0. I 6|= (∀x. p(x)) ↔ (¬∃x. ¬p(x)) Both cases end in contradictions for arbitrary I ⇒ F is valid. First case 1. I |= ∀x. p(x) assumption 2. I 6|= ¬∃x. ¬p(x) assumption 3. I |= ∃x. ¬p(x) 2and ¬ 4. I ⊳ {x 7→ v} |= ¬p(x) 3and ∃, for some v ∈ DI 5. I ⊳ {x 7→ v} |= p(x) 1and ∀ 4 and 5 are contradictory. 2- 11 2- 12 Example: Prove Example: Show F : p(a) → ∃x. p(x) isvalid. F : (∀x. p(x, x)) → (∃x. ∀y. p(x, y)) is invalid. Assume otherwise. Find interpretation I such that 1. I 6|= F assumption I |= ¬[(∀x. p(x, x)) → (∃x. ∀y. p(x, y))] 2. I |= p(a) 1and → 3. I 6|= ∃x. p(x) 1and → i.e. 4. I ⊳ {x 7→ αI [a]} 6|= p(x) 3and ∃ I |= (∀x. p(x, x)) ∧ ¬(∃x. ∀y. p(x, y)) 2 and 4 are contradictory. Thus, F is valid. Choose DI = {0, 1} pI = {(0, 0), (1, 1)} i.e. pI (0, 0) and pI (1, 1) are true pI (1, 0) and pI (1, 0) are false I falsifying interpretation ⇒ F is invalid. 2- 13 2- 14 Safe Substitution F σ Rename x by x′: ′ ′ Example: replace x in ∀x by x and all free x in the scope of ∀x by x . ′ ′ scope of ∀x ∀x. G[x] ⇔ ∀x . G[x ] z }| { F : (∀x. p(x, y) ) → q(f (y), x) Same for ∃x ր տ ր տ bound by ∀x free free free ∃x. G[x] ⇔ ∃x′. G[x′] free(F )= {x, y} where x′ is a fresh variable substitution Proposition (Substitution of Equivalent Formulae) σ : {x 7→ g(x), y 7→ f (x), q(f (y), x) 7→ ∃x. h(x, y)} σ : {F1 7→ G1, · · · , Fn 7→ Gn} F σ? 1. Rename s.t. for each i, Fi ⇔ Gi F ′ : ∀x′. p(x′, y) → q(f (y), x) ↑ ↑ If F σ a safe substitution, then F ⇔ F σ where x′ is a fresh variable 2. F ′σ : ∀x′. p(x′, f (x)) → ∃x. h(x, y) 2- 15 2- 16 Formula Schema Substitution σ of H Formula σ : {F1 7→ ,..., Fn 7→ } (∀x. p(x)) ↔ (¬∃x. ¬p(x)) mapping place holders Fi of H to FOL formulae, Formula Schema (obeying the side conditions of H) H1 : (∀x. F ) ↔ (¬∃x. ¬F ) Proposition (Formula Schema) ↑ place holder If H is valid formula schema and Formula Schema (with side condition) σ is a substitution obeying H’s side conditions H2 : (∀x. F ) ↔ F provided x ∈/ free(F ) then Hσ is also valid. Valid Formula Schema Example: H is valid iff valid for any FOL formula Fi obeying the side H : (∀x. F ) ↔ F provided x ∈/ free(F ) isvalid conditions σ : {F 7→ p(y)} obeys the side condition Example: H1 and H2 are valid. Therefore Hσ : ∀x. p(y) ↔ p(y) isvalid 2- 17 2- 18 Proving Validity of Formula Schema Normal Forms Example: Prove validity of 1. Negation Normal Forms (NNF) H : (∀x. F ) ↔ F provided x ∈/ free(F ) Augment the equivalence with (left-to-right) Proof by contradiction. Consider the two directions of ↔ . ¬∀x. F [x] ⇔ ∃x. ¬F [x] First case: ¬∃x. F [x] ⇔ ∀x. ¬F [x] 1. I |= ∀x. F assumption 2. I 6|= F assumption Example 3. I |= F 1, ∀, since x 6∈ free(F ) 4. I |= ⊥ 2, 3 G : ∀x. (∃y. p(x, y) ∧ p(x, z)) → ∃w.p(x, w) . Second Case: 1. ∀x. (∃y. p(x, y) ∧ p(x, z)) → ∃w. p(x, w) 1. I 6|= ∀x. F assumption 2. ∀x. ¬(∃y. p(x, y) ∧ p(x, z)) ∨ ∃w. p(x, w) 2. I |= F assumption F1 → F2 ⇔ ¬F1 ∨ F2 3. I |= ∃x. ¬F 1 and ¬ 4. I |= ¬F 3, ∃, since x 6∈ free(F ) 3. ∀x. (∀y. ¬(p(x, y) ∧ p(x, z))) ∨ ∃w. p(x, w) 5. I |= ⊥ 2, 4 ¬∃x. F [x] ⇔ ∀x. ¬F [x] 4. ∀x. (∀y. ¬p(x, y) ∨ ¬p(x, z)) ∨ ∃w. p(x, w) H Hence, is a valid formula schema. 2- 19 2- 20 2. Prenex Normal Form (PNF) 2. Rename quantified variables to fresh names All quantifiers appear at the beginning of the formula F2 : ∀x. (∀y. ¬p(x, y) ∨ ¬p(x, z)) ∨ ∃w. p(x, w) ↑ in the scope of ∀x Q1x1 · · · Qnxn. F [x1, · · · , xn] 3. Remove all quantifiers to produce quantifier-free formula where Qi ∈ {∀, ∃} and F is quantifier-free. F3 : ¬p(x, y) ∨ ¬p(x, z) ∨ p(x, w) Every FOL formula F can be transformed to formula F ′ in PNF 4. Add the quantifiers before F3 s.t. F ′ ⇔ F . F4 : ∀x. ∀y. ∃w. ¬p(x, y) ∨ ¬p(x, z) ∨ p(x, w) Example: Find equivalent PNF of Alternately, ′ F : ∀x. ¬(∃y. p(x, y) ∧ p(x, z)) ∨ ∃y. p(x, y) F4 : ∀x. ∃w. ∀y. ¬p(x, y) ∨ ¬p(x, z) ∨ p(x, w) ↑ to the end of the formula Note: In F2, ∀y is in the scope of ∀x, therefore the order of quantifiers must be ···∀x ···∀y · · · 1.