File(s) not publicly available
A Fully Abstract Relational Model of Syntactic Control of Interference
presentation
posted on 2023-06-07, 23:55 authored by Guy McCuskerIn an effort to investigate some linear analogue of the conjecture P $
eq$ NP, that is DLIN $
eq$ NLIN (deterministic/nondeterministic linear time complexity on random-access machines), this paper aims at being a step in the precise classification of the many NP-complete problems which are seemingly not NLIN-complete. We define and discuss the complexity class LIN-LOCAL -- the class of problems linearly reducible to problems defined by Boolean local constraints -- as well as its planar restriction LIN-PLAN-LOCAL. We show that both "local" classes are computationally robust and give some arguments for the conjecture that the following "linear hierarchy" is strict at each level: DLIN $\subseteq$ LIN-PLAN-LOCAL $\subseteq$ LIN-LOCAL $\subseteq$ NLIN. After showing that SAT and PLAN-SAT are complete in classes LIN-LOCAL and LIN-PLAN-LOCAL, respectively, we prove that some unexpected problems that involve some seemingly global constraints indeed belong to those classes and are complete for them. More precisely, we show that VERTEX-COVER and many similar problems involving cardinality constraints are LIN-LOCAL-complete. Our most striking and most technical result is that PLAN-HAMILTON -- the planar version of the Hamiltonian problem, a problem that involves a connectivity constraint of solutions -- is LIN-PLAN-LOCAL and even is LIN-PLAN-LOCAL-complete. Further, since our linear-time reductions also turn out to be parsimonious, they yield new DP-completeness results for UNIQUE-PLAN-HAMILTON and UNIQUE-PLAN-VERTEX-COVER.
eq$ NP, that is DLIN $
eq$ NLIN (deterministic/nondeterministic linear time complexity on random-access machines), this paper aims at being a step in the precise classification of the many NP-complete problems which are seemingly not NLIN-complete. We define and discuss the complexity class LIN-LOCAL -- the class of problems linearly reducible to problems defined by Boolean local constraints -- as well as its planar restriction LIN-PLAN-LOCAL. We show that both "local" classes are computationally robust and give some arguments for the conjecture that the following "linear hierarchy" is strict at each level: DLIN $\subseteq$ LIN-PLAN-LOCAL $\subseteq$ LIN-LOCAL $\subseteq$ NLIN. After showing that SAT and PLAN-SAT are complete in classes LIN-LOCAL and LIN-PLAN-LOCAL, respectively, we prove that some unexpected problems that involve some seemingly global constraints indeed belong to those classes and are complete for them. More precisely, we show that VERTEX-COVER and many similar problems involving cardinality constraints are LIN-LOCAL-complete. Our most striking and most technical result is that PLAN-HAMILTON -- the planar version of the Hamiltonian problem, a problem that involves a connectivity constraint of solutions -- is LIN-PLAN-LOCAL and even is LIN-PLAN-LOCAL-complete. Further, since our linear-time reductions also turn out to be parsimonious, they yield new DP-completeness results for UNIQUE-PLAN-HAMILTON and UNIQUE-PLAN-VERTEX-COVER.
History
Publication status
- Published
Publisher
Springer-VerlagPublisher URL
Volume
LecturPages
15.0Presentation Type
- paper
Event name
Computer Science LogicEvent location
Edinburgh, ScotlandEvent type
conferenceISBN
3-540-44240-5Department affiliated with
- Informatics Publications
Full text available
- No
Peer reviewed?
- Yes
Legacy Posted Date
2012-02-06Usage metrics
Categories
No categories selectedKeywords
Licence
Exports
RefWorks
BibTeX
Ref. manager
Endnote
DataCite
NLM
DC