On p-Optimal Proof Systems and Logics for PTIME

2011 IEEE 26th Annual Symposium on Logic in Computer Science

2011

Self Cite

There are standard logics DTC, TC, and LFP capturing the complexity classes L, NL, and P on ordered structures, respectively. In [4] we have shown that LFP inv , the "order-invariant least fixed-point logic LFP," captures P (on all finite structures) if and only if there is a listing of the Psubsets of the set TAUT of propositional tautologies. We are able to extend the result to listings of the L-subsets (NL-subsets) of TAUT and the logic DTC inv (TC inv ). As a byproduct we get that LFP inv captures P if DTC inv captures L.Furthermore, we show that the existence of a listing of the L-subsets of TAUT is equivalent to the existence of an almost space optimal algorithm for TAUT. To obtain this result we have to derive a space version of a theorem of Levin on optimal inverters. I . INTRODUCTIONIt is well-known that for standard complexity classes C (as L, NL and P) the existence of a logic capturing C is equivalent to the existence of a listing (or effective enumeration) of the classes of structures closed under isomorphism in C by means of Turing machines of type C that decide them (or equivalently, to such a listing of the classes of graphs closed under isomorphism). Recently [4] we have shown for C = P that such a listing exists if there is a listing of the P-subsets of the set TAUT of propositional tautologies; more explicitly, a listing of the subsets in P of TAUT by means of polynomial time Turing machines deciding them. Even more, it is shown in [4] that the following two statements are equivalent:(i) There is a listing of the P-subsets of TAUT.(ii) The logic LFP inv , the "order-invariant least fixed-point logic LFP," 1 captures P. The starting point for the investigations which led to this paper was the observation (cf. Proposition II.3) that one gets a listing of the P-subsets of TAUT if one assumes that there is a listing of its L-subsets. In particular, then LFP inv captures P. Thus, we asked ourselves whether then we even get that DTC inv , the "order-invariant deterministic transitive closure logic DTC," 1 captures L.At the end we realized that the equivalence of (i) and (ii) lifts to their L-analogues and to their NL-analogues. In 1 It is well-known that LFP captures P on ordered structures and that deterministic transitive closure logic DTC captures L on ordered structures. particular, LFP inv captures P if DTC inv captures L. Note that it is not known whether the existence of a logic capturing P is implied by the existence of a logic capturing L.A more general notion of listing turned out to be helpful. For complexity classes C and C we consider listings of the C-subsets of TAUT by means of Turing machines of type C ; we write LIST(C, TAUT, C ) if such a listing exists. For the classes P and NP such listings were already considered and put to good use by Sadowski in [16]. This more general notion is also meaningful in the context of logics (for P and NP already remarked in [3]); loosely speaking, if LIST(C, TAUT, C ), then in the order-invariant logic corresponding to C we can axiomatize the classe...

“…In [4] we have shown that the logic LFP inv is a P-bounded logic for P if and only if LIST(P, TAUT, P). Here, among others, we show the L-analogue and the NL-analogue of this result.…”

Section: Listings and Logicsmentioning

confidence: 98%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Listings and Logics

2011 IEEE 26th Annual Symposium on Logic in Computer Science

2011

Self Cite

“…We show for arbitrary Q that the existence of hard sets for all algorithms is equivalent to the existence of an effective enumeration of all polynomial time decidable subsets of Q, a property which has turned out to be useful in various contexts (cf. [12,2,3]). We analyze what Messner's result means for proof systems.…”

Section: Introductionmentioning

confidence: 99%

Hard Instances of Algorithms and Proof Systems

Lecture Notes in Computer Science

Müller

2012

Self Cite

Abstract. Assuming that the class Taut of tautologies of propositional logic has no almost optimal algorithm, we show that every algorithm A deciding Taut has a polynomial time computable sequence witnessing that A is not almost optimal. The result extends to every Π p t -complete problem with t ≥ 1; however, we show that assuming the Measure Hypothesis there is a problem which has no almost optimal algorithm but has an algorithm without hard sequences.

“…In [3] we showed that TAUT has a p-optimal proof system if and only if a certain logic L ≤ is a P-bounded logic for P (=PTIME). The equivalence in the first line of Theorem 2 is the nondeterministic version of this result; in fact, an immediate consequence of it states that TAUT has an optimal proof system if and only if L ≤ is an NP-bounded logic for P (a concept that we will introduce in Section 6).…”

Section: P-gödelmentioning

confidence: 99%

On Slicewise Monotone Parameterized Problems and Optimal Proof Systems for TAUT

2010

Computer Science Logic

Self Cite

Abstract. For a reasonable sound and complete proof calculus for first-order logic consider the problem to decide, given a sentence ϕ of first-order logic and a natural number n, whether ϕ has no proof of length ≤ n. We show that there is a nondeterministic algorithm accepting this problem which, for fixed ϕ, has running time bounded by a polynomial in n if and only if there is an optimal proof system for the set TAUT of tautologies of propositional logic. This equivalence is an instance of a general result linking the complexity of so-called slicewise monotone parameterized problems with the existence of an optimal proof system for TAUT.