User:Bynne
Frege Systems
[edit]In propositional calculus a Frege system is an implicationally complete deduction system with a set of logical connectives and a set of sound inference rules.
Implicational completeness
[edit]We say that a deduction system is implicationally complete iff, for every formula that is a tautology, there exists a derivation from the axioms using the rules of the system.
Formal Definition
[edit]Let be a set of boolean connectives on which we will define our system, let be a set of sound inference rules. We denote the deduction system by , if is implicationally complete we call it a Frege system.
F-Proof
[edit]Let and be formulas. Let be a sequence of formulas such that we can derive from using a rule from , and such that we can derive from using a rule from . We call , an -proof/derivation of the formula from .
Refutational completeness
[edit]We define a Frege system to be refutationally complete if for every formula that is a contradiction, there exists an -derivation of False from .
Length of formulas
[edit]We define the length of a -proof to be the number of applications of rules in the proof.
Examples
[edit]- Common axioms are:
- Common rules are:
- (Modus Ponens rule).
- (Resolution rule).
Rules are interpreted as follows, the two statements above the line entails the one below the line.
- Resolution is not a Frege system because it is not implicationally complete, i.e. we cannot conclude from - However adding the weakening rule: makes it implicationally complete and hence a Frege System.
- Resolution is refutationally complete.
- Natural deduction is a Frege system
- Gentzen with cut is a Frege system
- Frege's propositional calculus
Properties
[edit]- A Frege proof can be converted to a sequent proof with at most a polynomial increase in length.
- The minimal number of rounds in the prover-adversary game needed to prove a tautology is proportional to the logarithm of the minimal number of steps in a Frege proof of .
- Any Frege system is equivalent to a Gentzen with cut system, in the sense that there exists a translation of any frege proof/refutation to a Gentzen proof/refutation, with atmost a polynomial increase in length.
- Same holds true for Natural deduction
-
Proof strengths of different systems
References
[edit]- Buss, S. R. Pitassi, T. (1998). "Resolution and the weak pigeonhole", "Ann. Scuola Norm. Sup. Pisa".
- Cook, S. Reckhow, R. (1974). "On the lengths of proofs in the propositional calculus: preliminary"
- Pudlák, P. Buss, S. R. (1995). "How to lie without being (easily) convicted and the lengths of proofs in propositional calculus", "Ann. Scuola Norm. Sup. Pisa".
Further reading
[edit]- MacKay, D. J. (2008). "Information Theory, Inference, and Learning Algorithms"