Combinatorial Search With Picat

TL;DR

Picat is a new logic programming language that offers advanced features like arrays, maps, and tabling, enabling efficient solving of combinatorial search problems and outperforming existing planners in benchmarks.

Contribution

Introduces Picat, a logic programming language with novel features and a planning module that significantly improves performance on combinatorial search benchmarks.

Findings

Picat outperforms ASP and PDDL planners on recent benchmarks.

Tabling and resource-bounded search enhance planning efficiency.

Picat's features simplify modeling and solving combinatorial problems.

Abstract

Picat, a new member of the logic programming family, follows a different doctrine than Prolog in offering the core logic programming concepts: arrays and maps as built-in data types; implicit pattern matching with explicit unification and explicit non-determinism; functions for deterministic computations; and loops for convenient scripting and modeling purposes. Picat provides facilities for solving combinatorial search problems, including a common interface with CP, SAT, and MIP solvers, tabling for dynamic programming, and a module for planning. Picat's planner module, which is implemented by the use of tabling, has produced surprising and encouraging results. Thanks to term-sharing and resource-bounded tabled search, Picat overwhelmingly outperforms the cutting-edge ASP and PDDL planners on the planning benchmarks used in recent ASP competitions.