co-NP

co-NP co-NP is a central complexity class in theoretical computer science that contains decision problems whose complements lie in NP. It relates to foundational questions about Alan Turing, Stephen Cook, Leonid Levin, and the development of complexity theory at institutions such as Princeton University, University of Toronto, and Moscow State University. co-NP has connections to major results and conjectures involving the P versus NP problem, the Cook–Levin theorem, and topics studied at conferences like STOC and FOCS.

Definition

co-NP is the class of decision languages L for which the complement language (Σ* \ L) is in NP, where NP was formalized by Stephen Cook and Leonid Levin. Formally, a language L is in co-NP if there exists a nondeterministic polynomial-time verifier that accepts precisely the complements of instances in L, analogous to how NP is defined by certificate verification used in the Cook–Levin theorem and related to notions from Gödel and Turing's foundational work. Many developments around the definition arose in research groups at MIT, Bell Labs, and IBM Research during the 1970s and 1980s, overlapping with work by researchers at Bellcore and universities like Berkeley.

Relationship to NP and other complexity classes

The relationship between co-NP and NP is a focal question tied to the P versus NP problem and to hierarchies such as the Polynomial Hierarchy (PH) developed in part by Michael Sipser and Avi Wigderson. It is known that NP = co-NP would imply collapses in PH akin to consequences shown in proofs by researchers at Carnegie Mellon University and University of California, San Diego. Complexity classes that interact with co-NP include P, PSPACE, EXP, Σ2^P and Π2^P, and results connecting them have been pursued by groups at Stanford University, Harvard University, and Princeton University. Important separations or equalities among these classes are central to award-winning work recognized by prizes such as the Gödel Prize and the Fields Medal recipients who have contributed to complexity theory.

Complements and closure properties

By definition, co-NP is closed under complementation relative to NP: if L ∈ co-NP then its complement is in NP. Closure properties under operations such as union, intersection, and complementation relate to classic results proved using reductions developed by researchers at Rutgers University and Cornell University. co-NP is closed under polynomial-time many-one reductions when considered as a target class for completeness, and its behavior under oracle constructions was explored in oracle separation results by Bennett and Gill and later by theorists at UC Berkeley and Columbia University. The class is contained in PSPACE and contained or equal relations with EXP have been studied in the context of complete problems by teams at Microsoft Research and Google Research.

Complete problems

Several natural problems are complete for co-NP under polynomial-time many-one reductions; classic examples include the complement of Boolean satisfiability problem (the tautology problem) and certain validity problems in propositional logic studied in the tradition of Hilbert, Frege, and modern proof authors at EPFL and University of Oxford. Other co-NP-complete languages arise from complementing NP-complete problems such as variants of Hamiltonian cycle and Graph coloring decision problems, with reductions formalized in textbooks and surveys originating from groups at McGill University and University of Waterloo. The study of hardness for co-NP intersects work on circuit complexity linked to Valiant and algebraic complexity results from researchers at Caltech.

Proof systems and co-NP in proof complexity

Proof complexity investigates the lengths and structures of proofs for tautologies and related co-NP languages, connecting to systems such as Frege systems, Resolution, and bounded-depth proof systems explored by researchers associated with IAS and Weizmann Institute. Lower bounds on proof size for co-NP-complete problems have been central in separating propositional proof systems and are tied to results by Razborov, Krajíček, and groups at Princeton University and Rutgers University. Connections to interactive proofs like IP and probabilistically checkable proofs (PCP) show how co-NP sits within the landscape of verifiable proof systems studied at Tel Aviv University and University of Chicago.

Open problems and major conjectures

Major open problems include whether NP equals co-NP and implications for the P versus NP problem, a Clay Millennium Prize problem contextualized by institutions such as Clay Mathematics Institute and discussed at venues like ICM. Other conjectures concern separations between co-NP and higher levels of the Polynomial Hierarchy, tradeoffs in proof complexity proposed by researchers like Sanjeev Arora and Shafi Goldwasser, and whether specific co-NP-complete problems admit subexponential algorithms, topics pursued at labs including Google DeepMind and Facebook AI Research. Progress on these questions continues to be a focal point at conferences such as CCC and in journals published by ACM and IEEE.

Category:Computational complexity theory