proof complexity
play

Proof Complexity Olaf Beyersdorff School of Computing University - PowerPoint PPT Presentation

Proof Complexity Olaf Beyersdorff School of Computing University of Leeds, UK 1 Outline of this tutorial Tour of proof systems Resolution Frege and beyond Cutting Planes . . . Relations to other areas Separation of


  1. Proof Complexity Olaf Beyersdorff School of Computing University of Leeds, UK 1

  2. Outline of this tutorial Tour of proof systems ◮ Resolution ◮ Frege and beyond ◮ Cutting Planes ◮ . . . Relations to other areas ◮ Separation of complexity classes ◮ Analysis of SAT algorithms ◮ Proof search – Automatizability ◮ First-Order Logic – Bounded Arithmetic ◮ Further topics 2

  3. A Tour of Proof Systems 3

  4. Proof Systems Definition (Cook, Reckhow 79) A proof system for a language L is a function f with rng ( f ) = L . If f ( w ) = x , then w is called an f -proof of x ∈ L . ◮ correctness: rng ( f ) ⊆ L ◮ completeness: L ⊆ rng ( f ) ◮ efficiency: proofs should be easy to check, i.e. f should be easy to compute. ◮ Most research in proof complexity has studied propositional proof systems where L = TAUT . 4

  5. A First Example: Truth Tables A proof system for TAUT � ϕ if α is a truth table for ϕ with all entries 1 TT ( α, ϕ ) = p ∨ ¬ p otherwise. Why is this not a good proof system? ◮ Most proofs are exponentially long in the size of the formula. 5

  6. A First Example: Truth Tables A proof system for TAUT � ϕ if α is a truth table for ϕ with all entries 1 TT ( α, ϕ ) = p ∨ ¬ p otherwise. Why is this not a good proof system? ◮ Most proofs are exponentially long in the size of the formula. ◮ We look for proof systems with shorter proofs. 5

  7. The Most Studied Proof System: Resolution ◮ Introduced by Blake 1937, Davis & Putnam 1960, and Robinson 1965 ◮ Resolution proofs operate with clauses. ◮ Refutation system ◮ only one rule C ∨ p D ∨ ¬ p C ∨ D ◮ many subsystems studied: tree-like, regular . . . 6

  8. Complexity of Resolution First historical lower bound: ◮ Pigeonhole principle: n + 1 pigeons cannot sit in n holes ◮ CNF formulation PHP n + 1 n � x i , j for all pigeons i ∈ [ n + 1 ] j ∈ [ n ] ¬ x i 1 , j ∨ ¬ x i 2 , j for all distinct i 1 , i 2 ∈ [ n + 1 ] and j ∈ [ n ] ◮ PHP n + 1 requires Resolution refutations of size 2 Ω( n ) . [Haken 85] n Many strong lower bounds ◮ Combinatorial principles: ordering principle, . . . ◮ Graph-theoretic principles: Tseitin formulas, pebbling . . . ◮ Random 3-CNF’s are hard for Resolution. [Beame et al. 98] 7

  9. A Strong System: Frege p 1 → ( p 2 → p 1 ) Axioms ( p 1 → p 2 ) → ( p 1 → ( p 2 → p 3 )) → ( p 1 → p 3 ) p 1 → p 1 ∨ p 2 p 2 → p 1 ∨ p 2 ( p 1 → p 3 ) → ( p 2 → p 3 ) → ( p 1 ∨ p 2 → p 3 ) ( p 1 → p 2 ) → ( p 1 → ¬ p 2 ) → ¬ p 1 ¬¬ p 1 → p 1 p 1 ∧ p 2 → p 1 p 1 ∧ p 2 → p 2 p 1 → p 2 → p 1 ∧ p 2 p 1 p 1 → p 2 Modus Ponens p 2 8

  10. Frege Proofs A Frege proof of a formula ϕ is a sequence ( ϕ 1 , . . . , ϕ n = ϕ ) of propositional formulas such that for i = 1 , . . . , n : ◮ ϕ i is a substitution instance of an axiom, or ◮ ϕ i was derived by modus ponens from ϕ j , ϕ k with j , k < i . 9

  11. Frege Proofs A Frege proof of a formula ϕ is a sequence ( ϕ 1 , . . . , ϕ n = ϕ ) of propositional formulas such that for i = 1 , . . . , n : ◮ ϕ i is a substitution instance of an axiom, or ◮ ϕ i was derived by modus ponens from ϕ j , ϕ k with j , k < i . Major open problem Show non-trivial lower bounds on the size of Frege proofs. 9

  12. Restrictions and Extensions of Frege Systems Bounded-depth Frege Allow only formulas of logical depth d in the proof for a given constant d . Extended Frege EF Abbreviations for complex formulas: p ≡ ϕ , where p is a new propositional variable. Frege systems with substitution SF ϕ Substitution rule: σ ( ϕ ) for arbitrary substitutions σ Extensions of EF Let Φ be a polynomial-time computable set of tautologies. EF + Φ : Φ as axiom schemes 10

  13. Reductions between Proof Systems Definition (Cook, Reckhow 79, Krajíˇ cek, Pudlák 89) Let f and g be proof systems for L . ◮ f simulates g , if for any g -proof w there is an f -proof w ′ of length | w ′ | = | w | O ( 1 ) s.t. f ( w ′ ) = g ( w ) . ◮ If w ′ is computable from w in polynomial time, then f p-simulates g . ◮ f and g are (p-)equivalent if they (p-)simulate each other. 11

  14. Reductions between Proof Systems Definition (Cook, Reckhow 79, Krajíˇ cek, Pudlák 89) Let f and g be proof systems for L . ◮ f simulates g , if for any g -proof w there is an f -proof w ′ of length | w ′ | = | w | O ( 1 ) s.t. f ( w ′ ) = g ( w ) . ◮ If w ′ is computable from w in polynomial time, then f p-simulates g . ◮ f and g are (p-)equivalent if they (p-)simulate each other. Definition (Krajíˇ cek, Pudlák 89) A proof system f for L is (p)-optimal if f (p-)simulates every proof system for L . 11

  15. Simulations Between Proof Systems Theorem (Cook, Reckhow 79) All Frege systems are polynomially equivalent. Theorem (Krajíˇ cek, Pudlák 89) Every proof system is simulated by a proof system of the form EF + Φ . Problem (Krajíˇ cek, Pudlák 89) Do optimal proof systems exist? 12

  16. The Propositional Sequent Calculus ◮ Historically one of the first and best analyzed proof systems [Gentzen 35] ◮ basic objects: sequents ϕ 1 , . . . , ϕ m ⊢ ψ 1 , . . . , ψ k . ◮ Sequents of the form A ⊢ A , 0 ⊢ , ⊢ 1 are called initial sequents. ◮ An LK -proof of a propositional formula ϕ is a derivation of the sequent ⊢ ϕ from initial sequents by the following rules. 13

  17. Rules of LK Γ ⊢ ∆ Γ ⊢ ∆ (weakening) A , Γ ⊢ ∆ Γ ⊢ ∆ , A Γ 1 , A , B , Γ 2 ⊢ ∆ Γ ⊢ ∆ 1 , A , B , ∆ 2 (exchange) Γ 1 , B , A , Γ 2 ⊢ ∆ Γ ⊢ ∆ 1 , B , A , ∆ 2 Γ 1 , A , A , Γ 2 ⊢ ∆ Γ ⊢ ∆ 1 , A , A , ∆ 2 (contradiction) Γ 1 , A , Γ 2 ⊢ ∆ Γ ⊢ ∆ 1 , A , ∆ 2 Γ ⊢ ∆ , A A , Γ ⊢ ∆ ( ¬ introduction) ¬ A , Γ ⊢ ∆ Γ ⊢ ∆ , ¬ A A , Γ ⊢ ∆ A , Γ ⊢ ∆ Γ ⊢ ∆ , A Γ ⊢ ∆ , B ( ∧ rules) A ∧ B , Γ ⊢ ∆ B ∧ A , Γ ⊢ ∆ Γ ⊢ ∆ , A ∧ B A , Γ ⊢ ∆ B , Γ ⊢ ∆ Γ ⊢ ∆ , A Γ ⊢ ∆ , A Γ ⊢ ∆ , B ∨ A ( ∨ rules) A ∨ B , Γ ⊢ ∆ Γ ⊢ ∆ , A ∨ B Γ ⊢ ∆ , A A , Γ ⊢ ∆ (cut rule) Γ ⊢ ∆ 14

  18. A robust proof system: Frege/LK Proposition (Cook, Reckhow 79) Frege systems and the propositional sequent calculus LK are polynomially equivalent. 15

  19. Polynomially Bounded Proof Systems Polynomial Bounds on Proofs A proof system f for L is polynomially bounded if there exists a polynomial p such that every x ∈ L has an f -proof of size ≤ p ( | x | ) . 16

  20. Polynomially Bounded Proof Systems Polynomial Bounds on Proofs A proof system f for L is polynomially bounded if there exists a polynomial p such that every x ∈ L has an f -proof of size ≤ p ( | x | ) . Examples ◮ The standard proof system for SAT is polynomially bounded: � ϕ if α is a satisfying assignment for ϕ sat ( α, ϕ ) = p otherwise. ◮ The truth-table system is not a polynomially bounded proof system for TAUT. 16

  21. The Cook-Reckhow Theorem Question Is there a polynomially bounded proof system for TAUT? Theorem (Cook, Reckhow 79) A language L has a polynomially bounded proof system if and only if L ∈ NP . For propositional proof systems TAUT has a polynomially bounded proof system if and only if NP = coNP. 17

  22. Cook’s Programme Separate NP from coNP (and hence P and NP) by showing super-polynomial lower bounds to the size of proofs in all propositional proof systems. 18

  23. Cook’s Programme Separate NP from coNP (and hence P and NP) by showing super-polynomial lower bounds to the size of proofs in all propositional proof systems. Showing lower bounds for a system P means finding an infinite family θ n of propositional tautologies s.t. ◮ | θ n | = n O ( 1 ) ; ◮ θ n requires super-polynomial size proofs in P . 18

  24. Cook’s Programme Separate NP from coNP (and hence P and NP) by showing super-polynomial lower bounds to the size of proofs in all propositional proof systems. Showing lower bounds for a system P means finding an infinite family θ n of propositional tautologies s.t. ◮ | θ n | = n O ( 1 ) ; ◮ θ n requires super-polynomial size proofs in P . ◮ Better: . . . exponential size proofs. 18

  25. Cook’s Programme Separate NP from coNP (and hence P and NP) by showing super-polynomial lower bounds to the size of proofs in all propositional proof systems. Showing lower bounds for a system P means finding an infinite family θ n of propositional tautologies s.t. ◮ | θ n | = n O ( 1 ) ; ◮ θ n requires super-polynomial size proofs in P . ◮ Better: . . . exponential size proofs. Even better ◮ Find a sequence of polynomially constructible formulas which require long proofs. ◮ This is usually the case: take θ n as the propositional formalization of some combinatorial principle. ◮ Find a large set of formulas (e.g. random 3-CNF) which require long proofs. 18

  26. Cook’s Programme Separate NP from coNP (and hence P and NP) by showing super-polynomial lower bounds to the size of proofs in all propositional proof systems. 19

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend