LeoPARD — A Generic Platform for the Implementation of Higher-Order Reasoners

Abstract

Abstract. LeoPARD supports the implementation of knowledge representation and reasoning tools for higher-order logic(s). It combines a sophisticated data structure layer (polymorphically typed λ-calculus with nameless spine notation, explicit substitutions, and perfect term sharing) with an ambitious multi-agent blackboard architecture (supporting prover parallelism at the term, clause, and search level). Further features of LeoPARD include a parser for all TPTP dialects, a command line interpreter, and generic means for the integration of external reasoners.

Mathematical Software – ICMS 2016

“…In its core, Leo-III is a new higher-order automated theorem prover based on the associated system platform LeoPARD [WSB15]. LeoPARD is a framework for deduction systems (implemented in Scala) providing sophisticated term, search, and indexing data structures for typed λ-terms, as well as an generic agent-based blackboard architecture.…”

Leo-III Version 1.1 (System description)

“…Leo-III is implemented in Scala 2 based on the associated system platform LeoPARD [36]. The latter provides a reusable framework and infrastructure for higher-order deduction systems, consisting of fundamental generic data structures for typed λ-terms, indexing means, a generic agent-based blackboard architecture, parser, and proof printer.…”

“…The underlying architecture of Leo-III employs a blackboard data structure which agents collaboratively use for finding a proof. The work of the agents is thereby divided in transactional tasks and organized as auctions, in which it is decided what tasks to be executed next in case of interference [35,36]. In Leo-III agents can be utilized in two different ways: First, top down, where one agent implements a sequential loop and all remaining agents perform subsidiary or computationally heavy tasks in parallel.…”

“…Agents do not communicate directly with other agents, but do so indirectly by manipulating the data in the blackboard. The Leo-III blackboard [36] is a collection of globally shared and accessible data structures any agent can query and manipulate at any time in parallel. The set of data structures itself is not fixed, as any object implementing the DataStore interface can be added, even during execution.…”

“…It is the successor of the well-known LEO-II prover [8], whose development significantly influenced the build-up of the TPTP THF infrastructure [32]. Leo-III exemplarily utilizes and instantiates the associated LeoPARD system platform [36] for higherorder (HO) deduction systems implemented in Scala. The prover makes use of LeoPARD's sophisticated data structures and implements its own reasoning logic on top, inter alia as agents in LeoPARD's provided blackboard architecture.…”