Polynomial-time solvable #CSP problems via algebraic models and Pfaffian circuits. (May 2016)

• Record Type: Journal Article
• Authors: Margulies, S.; Morton, J.
• Is Part Of: Journal of symbolic computation. Volume 74(2016)
• Page Start: 152
• Page End: 180
• Publication Date: 2016-05
• Subjects: Dichotomy theorems -- Gröbner bases -- Computer algebra -- #CSP -- Polynomial ideals; Mathematics -- Data processing -- Periodicals; Numerical analysis -- Data processing -- Periodicals; Automatic programming (Computer science) -- Periodicals; Mathématiques -- Informatique -- Périodiques; Analyse numérique -- Informatique -- Périodiques; Programmation automatique -- Périodiques; Automatic programming (Computer science); Mathematics -- Data processing; Numerical analysis -- Data processing; Periodicals; Electronic journals; 510.285
• DOI: 10.1016/j.jsc.2015.06.008
• Languages: English
• ISSNs: 0747-7171

Abstract

Abstract: A Pfaffian circuit is a tensor contraction network where the edges are labeled with basis changes in such a way that a very specific set of combinatorial properties is satisfied. By modeling the permissible basis changes as systems of polynomial equations, and then solving via computation, we are able to identify classes of 0/1 planar #CSP problems solvable in polynomial time via the Pfaffian circuit evaluation theorem (a variant of L. Valiant's Holant Theorem). We present two different models of 0/1 variables, one that is possible under a homogeneous basis change, and one that is possible under a heterogeneous basis change only. We enumerate a collection of 1, 2, 3, and 4-arity gates/cogates that represent constraints, and define a class of constraints that is possible under the assumption of a "bridge" between two particular basis changes. We discuss the issue of planarity of Pfaffian circuits, and demonstrate possible directions in algebraic computation for designing a Pfaffian tensor contraction network fragment that can simulate aswap gate/cogate. We conclude by developing the notion of a decomposable gate/cogate, and discuss the computational benefits of this definition.