Bounded Arithmetic in Free Logic Yoriyuki Yamagata CTFM, 2013/02/20 - PowerPoint PPT Presentation
Bounded Arithmetic in Free Logic Yoriyuki Yamagata CTFM, 2013/02/20 Busss theories 2 Language of Peano Arithmetic + # a # b = 2 | | BASIC axioms PIND , , ( ) 2
Bounded Arithmetic in Free Logic Yoriyuki Yamagata CTFM, 2013/02/20
π Bussβs theories π 2 β’ Language of Peano Arithmetic + β#β β a # b = 2 π β | π | β’ BASIC axioms β’ PIND π¦ π΅ , Ξ β Ξ , π΅ ( π¦ ) 2 π΅ 0 , Ξ β Ξ , π΅ ( π’ ) π , i.e. has π -alternations of where π΅ π¦ β Ξ£ π bounded quantifiers βπ¦ β€ π’ , βπ¦ β€ π’ .
π PH and Bussβs theories π 2 1 = π 2 2 = π 2 3 = β¦ π 2 Implies π ) = β¦ π = β‘ ( ππ ) = β‘ ( Ξ£ 2 We can approach (non) collapse of PH from (non) collapse of hierarchy of Bussβs theories (PH = Polynomial Hierarchy)
Our approach π by GΓΆdel incompleteness theorem β’ Separate π 2 β’ Use analogy of separation of π½Ξ£ π
Separation of π½Ξ£ π β¦ π½Ξ£ 3 β’ Con(I Ξ£ 2 ) β π½Ξ£ 2 β’ Con I Ξ£ 2 β π½Ξ£ 1
π Consistency proof inside π 2 β’ Bounded Arithmetics generally are not capable to prove consistency. β π 2 does not prove consistency of Q (Paris, Wilkie) β π 2 does not prove bounded consistency of 1 (PudlΓ‘k) π 2 π does not prove consistency the πΆ π π fragement β π 2 β1 (Buss and IgnjatoviΔ ) of π 2
Buss and IgnjatoviΔ (1995) β¦ 3 β’ πΆ 3 b β Con( π 2 β1 ) π 2 β 2 β’ πΆ 2 b β Con( π 2 β1 ) π 2 β 1 β’ πΆ 1 b β Con( π 2 β1 ) π 2
Whereβ¦ π β π·π·π· π β’ πΆ π π β proofs β consistency of πΆ π π β proofs : the proofs by πΆ π π -formule β πΆ π π : Ξ£ 0 π ) β¦ Formulas generated from Ξ£ π π by π ( Ξ£ π β πΆ π Boolean connectives and sharply bounded quantifiers. β1 β’ π 2 π β Induction free fragment of π 2
Ifβ¦ π β’ πΆ i b β Con π 2 β1 , j > i π 2 Then, Bussβs hierarchy does not collapse.
β1 inside π 2 π Consistency proof of π 2 Problem β’ No truth definition, because β’ No valuation of terms, because β’ The values of terms increase exponentially β’ E.g. 2#2#2#2#2#...#2 π world, terms do not have values a priori . In π 2 β’ Thus, we must prove the existence of values in proofs. β’ We introduce the predicate πΉ which signifies existence of values.
Our result(2012) β¦ 5 β’ 3 β Con( π 2 β1 πΉ ) π 2 β 4 β’ 2 β Con( π 2 β1 πΉ ) π 2 β 3 β’ 1 β Con( π 2 β1 πΉ ) π 2
Whereβ¦ β’ π β π·π·π· π β consistency of π -normal proofs β π -normal proofs : the proofs by π -normal formulas β π -normal formulas: Formulas in the form: βπ¦ 1 β€ π’ 1 βπ¦ 2 β€ π’ 2 β¦ π π¦ π β€ π’ π π π¦ π+1 β€ π’ π+1 . π΅ (β¦ ) Where π΅ is quantifier free
Whereβ¦ β1 πΉ β’ π 2 π πΉ β Induction free fragment of π 2 β have predicate πΉ which signifies existence of values β’ Such logic is called Free logic
π πΉ (Language) π 2 Predicates β’ =, β€ , πΉ Function symbols β’ Finite number of polynomial functions Formulas β’ Atomic formula, negated atomic formula β’ π΅ β¨ πΆ , π΅ β§ πΆ β’ Bounded quantifiers
π πΉ (Axioms) π 2 β’ πΉ -axioms β’ Equality axioms β’ Data axioms β’ Defining axioms β’ Auxiliary axioms
Idea behind axiomsβ¦ β π = π Because there is no guarantee of πΉπ Thus, we add πΉπ in the antecedent πΉπ β π = π
E-axioms β’ πΉπΉ π 1 , β¦ , π π β πΉπ π β’ π 1 = π 2 β πΉπ π β’ π 1 β π 2 β πΉπ π β’ π 1 β€ π 2 β πΉπ π β’ Β¬ π 1 β€ π 2 β πΉπ π
Equality axioms β’ πΉπ β π = π β’ πΉπΉ π β , π β = π β πΉ π β = πΉ π
Data axioms β’ β πΉπΉ β’ πΉπ β πΉπ‘ 0 π β’ πΉπ β πΉπ‘ 1 π
Defining axioms πΉ π£ π 1 , π 2 , β¦ , π π = π’ ( π 1 , β¦ , π π ) π£ π = 0, π , π‘ 0 π , π‘ 1 π πΉπ 1 , β¦ , πΉπ π , πΉπ’ π 1 , β¦ , π π β πΉ π£ π 1 , π 2 , β¦ , π π = π’ ( π 1 , β¦ , π π )
Auxiliary axioms π = π β π # π = π # π πΉπ # π , πΉπ # π , π = | π | β π # π = π # π
PIND-rule π -formulas where π΅ is an Ξ£ π
π πΉ Bootstrapping π 2 π πΉ β’ Tot( πΉ ) for any πΉ , π β₯ 0 I. π 2 π πΉ β’ BASIC β , equality axioms β II. π 2 π πΉ β’ predicate logic β III. π 2 π βPIND β π πΉ β’ Ξ£ π IV. π 2
Theorem (Consistency) π+2 β’ i β Con( π 2 β1 πΉ ) π 2
Valuation trees Ο -valuation tree bounded by 19 Ο(a)=2, Ο(b)=3 a=2 a#a=16 b=3 a#a+b=19 π€ π # π + π , π β 19 19 π π€ π’ , π β π£ π is Ξ£ 1
Bounded truth definition (1) β’ π π£ , π’ 1 = π’ 2 , π β def βπ β€ π£ , π€ π’ 1 , π β π£ π β§ π€ π’ 1 , π β π£ π β’ π π£ , π 1 β§ π 2 , π β def π π£ , π 1 , π β§ π π£ , π 2 , π β’ π π£ , π 1 β¨ π 2 , π β def π π£ , π 1 , π β¨ π π£ , π 2 , π
Bounded truth definition (2) β’ π π£ , βπ¦ β€ π’ , π ( π¦ ) , π β def βπ β€ π£ , π€ π’ , π β π£ π β§ βπ β€ π , π π£ , π π¦ , π π¦ β¦ π β’ π π£ , βπ¦ β€ π’ , π ( π¦ ) , π β def βπ β€ π£ , π€ π’ , π β π£ π β§ βπ β€ π , π ( π£ , π π¦ , π [ π¦ β¦ π ]) π , π π£ , π is Ξ£ π+1 π Remark: If π is Ξ£ π
induction hypothesis π£ : enough large integer π : node of a proof of 0=1 Ξ π β Ξ π : the sequent of node π π : assignment π π β€ π£ βπ£ β² β€ π£ β π , { βπ΅ β Ξ π π π£ β² , π΅ , π β [ βπΆ β Ξ r , π ( π£ β² β π , πΆ , π ) ]}
Conjecture β1 πΉ is weak enough β’ π 2 π+2 can prove π -consistency of π 2 β1 πΉ β π 2 β1 πΉ is strong enough β’ While π 2 π πΉ can interpret π 2 π β π 2 β’ Conjecture β1 πΉ is a good candidate to separate π 2 π and π 2 π+2 . π 2
Future works β’ Long-term goal π β’ πβCon(π 2 β1 πΉ )? π 2 β’ Short-term goal π πΉ β Simplify π 2
Publications β’ Bounded Arithmetic in Free Logic Logical Methods in Computer Science Volume 8, Issue 3, Aug. 10, 2012
Recommend
More recommend
Explore More Topics
Stay informed with curated content and fresh updates.