Boolean satisfiability problem

Boolean satisfiability problem
Name	Boolean satisfiability problem
Field	Theoretical computer science
Introduced	1960s
Notable	Stephen Cook, Richard Karp

Boolean satisfiability problem

The Boolean satisfiability problem asks whether a formula in propositional logic can be assigned truth values that make the formula true, a question foundational to theoretical computer science, mathematical logic, and computational complexity. Originating in early work by logicians and formalized in the 20th century, it became central after landmark results on NP-completeness and has driven advances in automated reasoning, hardware verification, cryptography, and artificial intelligence.

Definition and Problem Variants

The canonical decision form asks, given a propositional formula in Boolean variables, whether there exists an assignment of Godel, Turing-style variables (interpreted as true/false) satisfying the formula. Important syntactic restrictions define variants such as conjunctive normal form (CNF), disjunctive normal form (DNF), k-CNF, and k-SAT; classic references include results by Stephen Cook and Richard Karp. Specific named cases include 2-SAT (polynomial-time solvable) and 3-SAT (first shown NP-complete), while restricted classes such as Horn-SAT, XOR-SAT, and read-once formulas admit specialized algorithms; contributors include Martin Davis, Hilary Putnam, and George Logician. Formulas may be represented as clauses, literals, and variables, and variants consider satisfiability under quantification (QSAT/Quantified Boolean Formula) introduced in complexity studies tied to Alan Turing and later formalized by researchers at institutions such as Princeton University and Stanford University.

Complexity and NP-Completeness

The decision problem was proven NP-complete in the seminal work of Stephen Cook and independently connected by reductions catalogued by Richard Karp, establishing SAT as the canonical NP-complete problem. This classification implies polynomial-time many-one reductions from every problem in NP to SAT, a fact used in completeness proofs across theoretical computer science at places like Bell Labs and MIT. Complexity-theoretic consequences tie SAT to major conjectures and results including the P versus NP problem, completeness frameworks at UC Berkeley, and the polynomial hierarchy studied by theorists affiliated with Princeton University and University of Chicago. Hardness results use reductions from problems such as Hamiltonian cycle and clique captured in Karp’s 21 NP-complete problems, linking SAT to classical combinatorial problems in the literature of Carnegie Mellon University and Cornell University.

Algorithms and Solving Techniques

Algorithmic approaches bifurcate between complete decision procedures and incomplete heuristics. Backtracking search, exemplified by DPLL (Davis–Putnam–Logemann–Loveland) originating from Martin Davis and Hilary Putnam and refined by others at IBM research labs, underlies most exact SAT solvers; modern enhancements include unit propagation, pure literal elimination, and clause learning studied at Stanford University and Microsoft Research. Local search and stochastic methods, popularized in surveys from Sandia National Laboratories and Bell Labs, include GSAT and WalkSAT used in practice by researchers at Carnegie Mellon University and University of California, Santa Barbara. Parameterized complexity and fixed-parameter tractable algorithms link SAT to research at University of Vienna and Weizmann Institute of Science, while approximation and probabilistic algorithms connect to results by scholars at Harvard University and University of Chicago.

SAT Solvers and Implementations

Industrial-strength SAT solvers such as Chaff, MiniSat, Glucose, and CryptoMiniSat emerged from academic groups at Princeton University, Dresden University of Technology, and University of Birmingham, and have been applied in industrial settings at Intel, Google, and Qualcomm. Competitions like the SAT Competition and SAT Race, organized by communities including EPFL and CNRS, drive benchmarking and innovation; solver development benefits from open-source projects hosted by teams at GitHub and academic labs such as University of Oxford and ETH Zurich. Domain-specific solvers integrate SAT into SMT (satisfiability modulo theories) frameworks developed by groups at Microsoft Research and SRI International, and incremental and parallel SAT solving have been advanced by collaborations involving Lawrence Livermore National Laboratory and IBM Research.

Applications and Reductions

SAT serves as a universal target for reductions from combinatorial and algebraic decision problems, enabling encodings of graph coloring, scheduling, integer factoring decision variants, and model checking used by practitioners at Bell Labs, Siemens, and Siemens AG. In hardware verification and electronic design automation, teams at Intel and ARM Holdings employ SAT solvers for equivalence checking and test generation, while software verification groups at Microsoft Research and Google translate program analysis problems into SAT or SMT instances. Cryptanalysis and security research at NIST and NSA have exploited SAT encodings for key-recovery tasks, and planning and AI research at DeepMind and Stanford University use SAT-based encodings to solve constrained search problems. Reductions to SAT are standard pedagogical tools in courses at MIT and Carnegie Mellon University.

Extensions and Related Problems

Extensions include Quantified Boolean Formula (QBF), which generalizes SAT with alternating quantifiers and is complete for PSPACE, a subject of study at University of Toronto and University of Cambridge; Max-SAT and Weighted-SAT optimize the number or weight of satisfied clauses and are applied in operations research groups at INRIA and EPFL; and SAT modulo theories (SMT) integrates domain theories and is advanced by research teams at Microsoft Research and Delft University of Technology. Related decision and optimization problems include constraint satisfaction problem (CSP), graph homomorphism, and model counting (#SAT) investigated at University of Illinois Urbana-Champaign and Weizmann Institute of Science, each spawning specialized algorithms and complexity classifications pursued across international research centers.

Category:Theoretical computer science