[Paper Review] Constraint Satisfaction Problem Dichotomy for Finite Templates: a Proof Via Consistency Checks

Dejan Delić|arXiv (Cornell University)|Jun 12, 2017

Constraint Satisfaction and Optimization20 references3 citations

TL;DR

This paper proves the CSP Dichotomy Conjecture for finite templates using a novel consistency notion called AF-consistency, which enables polynomial-time solvability of constraint satisfaction problems over Taylor algebras via local consistency checks and absorption-based reductions. The key contribution is a complete algorithmic characterization of tractable CSPs using consistency and absorption, resolving a long-standing open problem in computational complexity and universal algebra.

ABSTRACT

One of the central problems in the study of parametrized constraint satisfaction problems is the Dichotomy Conjecture by T. Feder and M. Vardi stating that the constraint satisfaction problem (CSP) over a fixed, finite constraint language is either solvable in polynomial time or extsc{NP}-complete. The conjecture was verified in certain special cases (domains with a relatively small number of elements, constraint languages containing all unary relations, etc.) In this article, we present a proof of the Dichotomy Conjecture via local consistency and AF- consistency checks. In fact, we show that, for every Taylor domain, which is $(2\lceil\frac{K}{2} ceil,3\lceil\frac{K}{2} ceil)$-consistent, where $K$ is the largest arity of a relation in the constraint language, we can define polynomially many proper subinstances such that, the original instance of the CSP is solvable if, and only if, the problem has a solution in one of those subinstances.. Finally, a solution is constructed using the combination of SLAC (Singleton Linear Arc Consistency), introduced by M. Kozik, and AF-consistency.

Motivation & Objective

• To resolve the CSP Dichotomy Conjecture, which posits that every finite-template CSP is either solvable in polynomial time or NP-complete.
• To establish a complete algebraic characterization of tractable CSPs using the Taylor property of polymorphism algebras.
• To introduce and formalize AF-consistency as a new consistency mechanism that enables polynomial-time solvability in absorption-free subinstances.
• To unify local consistency, absorption theory, and singleton linear arc consistency (SLAC) into a single algorithmic framework for solving CSPs over Taylor domains.

Proposed method

• Introduces AF-consistency checks based on polynomially many proper subinstances derived from $(2\lceil\frac{K}{2}\rceil,3\lceil\frac{K}{2}\rceil)$-consistency, where $K$ is the maximum arity of relations in the constraint language.
• Employs a reduction strategy using absorption and SLAC to eliminate inconsistent branches in the computation tree, ensuring that only solvable subinstances are explored.
• Applies the Absorption Theorem to decompose the problem into independent, absorption-free components, enabling localized consistency checks.
• Uses cyclic systems of linear equations over finite fields to model and solve subinstances, leveraging Reingold’s reachability algorithm for logspace solvability.
• Constructs a recursive algorithm that applies consistency checks and absorption reductions in a tree-like structure, ensuring correctness and completeness.
• Relies on the algebraic notion of Taylor algebras and their polymorphism algebras to determine the boundary between tractable and NP-complete CSPs.

Experimental results

Research questions

• RQ1 Does every finite-template CSP with a Taylor polymorphism algebra admit a polynomial-time solution via local consistency and absorption?
• RQ2 Can AF-consistency checks effectively eliminate unsatisfiable branches in the search space of CSPs over Taylor templates?
• RQ3 Is the solvability of cyclic systems of equations over finite fields sufficient to characterize tractability in absorption-free components?
• RQ4 Can the algorithm be expressed in a logic extending first-order logic, such as LFP + Rk or CPT+C, to capture polynomial time?
• RQ5 Does every simple, idempotent, absorption-free, congruence skew-free algebra satisfy the Hagemann-Mitschke identities, implying a broader applicability of the method?

Key findings

• The paper establishes that every finite-template CSP with a Taylor polymorphism algebra is solvable in polynomial time, confirming the Algebraic CSP Dichotomy Conjecture.
• AF-consistency checks are proven to preserve solvability and enable effective pruning of unsatisfiable subinstances, particularly in absorption-free components.
• The algorithm reduces the CSP to binary, bounded-width instances and applies SLAC and consistency checks to ensure solution existence in subinstances.
• The method successfully handles cyclic systems of equations over finite fields using Reingold’s algorithm, providing a logspace-computable path to solution in specific cases.
• The framework demonstrates that the Absorption Theorem is central to the tractability of Taylor domains, as its failure in essentially unary algebras leads to intractability.
• The approach suggests that a general CSP algorithm may be definable in logics like CPT+C or LFP + Rk, provided they can express solvability of linear systems over finite Abelian groups.