ECCC-Report TR05-077https://eccc.weizmann.ac.il/report/2005/077Comments and Revisions published for TR05-077en-usTue, 19 Jul 2005 10:56:05 +0300
Paper TR05-077
| On a D-N-optimal acceptor for TAUT |
Zenon Sadowski
https://eccc.weizmann.ac.il/report/2005/077The notion of an optimal acceptor for TAUT (the optimality
property is stated only for input strings from TAUT) comes from the line
of research aimed at resolving the question of whether optimal
propositional proof systems exist. In this paper we introduce two new
types of optimal acceptors, a D-N-optimal acceptor and an N-D-optimal
acceptor for TAUT. A deterministic algorithm recognizing TAUT is a
D-N-optimal acceptor if no other nondeterministic algorithm accepting
TAUT has more than a polynomial speed-up over its running time on
instances from TAUT.
We further develop the earlier observed connection between optimal
acceptors, optimal propositional proof systems, and the structure of
easy subsets of TAUT. Namely, we prove that the existence of a
D-N-optimal acceptor for TAUT is equivalent to the existence of an
optimal and automatizable propositional proof system and to the
existence of a suitable recursive presentation of the class of all
NP-easy (acceptable by nondeterministic polynomial time machines)
subsets of TAUT. Additionally, we show that the question of whether
every proof system is weakly automatizable is equivalent to the question
of whether every disjoint NP-pair is P-separable.
Tue, 19 Jul 2005 10:56:05 +0300https://eccc.weizmann.ac.il/report/2005/077