Higher-order Theorem Prover Research Articles

Graph Representations for Higher-Order Logic and Theorem Proving

This paper presents the first use of graph neural networks (GNNs) for higher-order proof search and demonstrates that GNNs can improve upon state-of-the-art results in this domain. Interactive, higher-order theorem provers allow for the formalization of most mathematical theories and have been shown to pose a significant challenge for deep learning. Higher-order logic is highly expressive and, even though it is well-structured with a clearly defined grammar and semantics, there still remains no well-established method to convert formulas into graph-based representations. In this paper, we consider several graphical representations of higher-order logic and evaluate them against the HOList benchmark for higher-order theorem proving.

Higher-order theorem proving and its applications

Abstract Automated theorem proving systems validate or refute whether a conjecture is a logical consequence of a given set of assumptions. Higher-order provers have been successfully applied in academic and industrial applications, such as planning, software and hardware verification, or knowledge-based systems. Recent studies moreover suggest that automation of higher-order logic, in particular, yields effective means for reasoning within expressive non-classical logics, enabling a whole new range of applications, including computer-assisted formal analysis of arguments in metaphysics. My work focuses on the theoretical foundations, effective implementation and practical application of higher-order theorem proving systems. This article briefly introduces higher-order reasoning in general and presents an overview of the design and implementation of the higher-order theorem prover Leo-III. In the second part, some example applications of Leo-III are discussed.

Reducing Higher-Order Theorem Proving to a Sequence of SAT Problems

We describe a complete theorem proving procedure for higher-order logic that uses SAT-solving to do much of the heavy lifting. The theoretical basis for the procedure is a complete, cut-free, ground refutation calculus that incorporates a restriction on instantiations. The refined nature of the calculus makes it conceivable that one can search in the ground calculus itself, obtaining a complete procedure without resorting to meta-variables and a higher-order lifting lemma. Once one commits to searching in a ground calculus, a natural next step is to consider ground formulas as propositional literals and the rules of the calculus as propositional clauses relating the literals. With this view in mind, we describe a theorem proving procedure that primarily generates relevant formulas along with their corresponding propositional clauses. The procedure terminates when the set of propositional clauses is unsatisfiable. We prove soundness and completeness of the procedure. The procedure has been implemented in a new higher-order theorem prover, Satallax, which makes use of the SAT-solver MiniSat. We also describe the implementation and give several examples. Finally, we include experimental results of Satallax on the higher-order part of the TPTP library.

Generating Invariant-Based Certificates for Embedded Systems

Automatic verification tools, such as model checkers and tools based on static analysis or on abstract interpretation, have become popular in software and hardware development. They increase confidence and potentially provide rich feedback. However, with increasing complexity, verification tools themselves are more likely to contain errors. In contrast to automatic verification tools, higher-order theorem provers use mathematically founded proof strategies checked by a small proof checker to guarantee selected properties. Thus, they enjoy a high level of trustability. Properties of software and hardware systems and their justifications can be encapsulated into a certificate, thereby guaranteeing correctness of the systems, with respect to the properties. These results offer a much higher degree of confidence than results achieved by verification tools. However, higher-order theorem provers are usually slow, due to their general and minimalistic nature. Even for small systems, a lot of human interaction is required for establishing a certificate. In this work, we combine the advantages of automatic verification tools (i.e., speed and automation) with those of higher-order theorem provers (i.e., high level of trustability). The verification tool generates a certificate for each invocation. This is checked by the higher-order theorem prover, thereby guaranteeing the desired property. The generation of certificates is much easier than producing the analysis results of the verification tool in the first place. In our work, we are able to create certificates that come with an algorithmic description of the proof of the desired property as justification. We concentrate on verification tools that generate invariants of systems and certify automatically that these do indeed hold. Our approach is applied to the certification of the verdicts of a deadlock-detection tool for an asynchronous component-based language.

Quantified Multimodal Logics in Simple Type Theory

We present an embedding of quantified multimodal logics into simple type theory and prove its soundness and completeness. A correspondence between QKπ models for quantified multimodal logics and Henkin models is established and exploited. Our embedding supports the application of off-the-shelf higher-order theorem provers for reasoning within and about quantified multimodal logics. Moreover, it provides a starting point for further logic embeddings and their combinations in simple type theory.

Higher-order aspects and context in SUMO

This article addresses the automation of higher-order aspects in expressive ontologies such as the suggested upper merged ontology SUMO. Evidence is provided that modern higher-order automated theorem provers like LEO-II can be fruitfully employed for the task. A particular focus is on embedded formulas (formulas as terms), which are used in SUMO, for example, for modeling temporal, epistemic, or doxastic contexts. This modeling is partly in conflict with SUMO’s assumption of a bivalent, classical semantics and it may hence lead to counterintuitive reasoning results with automated theorem provers in practice. A solution is proposed that maps SUMO to quantified multimodal logic which is in turn modeled as a fragment of classical higher-order logic. This way automated higher-order theorem provers can be safely applied for reasoning about modal contexts in SUMO.Our findings are of wider relevance as they analogously apply to other expressive ontologies and knowledge representation formalisms.

Higher-Order Aspects and Context in SUMO

This article addresses the automation of higher-order aspects in expressive ontologies such as the Suggested Upper Merged Ontology SUMO. Evidence is provided that modern higher-order automated theorem provers like LEO-II can be fruitfully employed for the task. A particular focus is on embedded formulas (formulas as terms), which are used in SUMO, for example, for modeling temporal, epistemic, or doxastic contexts. This modeling is partly in conflict with SUMO's assumption of a bivalent, classical semantics and it may hence lead to counter-intuitive reasoning results with automated theorem provers in practice. A solution is proposed that maps SUMO to quantified multimodal logic which is in turn modeled as a fragment of classical higher-order logic. This way automated higher-order theorem provers can be safely applied for reasoning about modal contexts in SUMO. Our findings are of wider relevance as they analogously apply to other expressive ontologies and knowledge representation formalisms.

Certifying compilers using higher-order theorem provers as certificate checkers

Correct software requires compilers to work correctly. Especially code generation can be an error prone task, since it potentially uses sophisticated algorithms to produce efficient code.In this pa...

The Watson Theorem Prover

Watson is a general-purpose system for formal reasoning. It is an interactive equational higher-order theorem prover. The higher-order logic supported by the prover is distinctive in being type free (it is a safe variant of Quine's i>NF). Watson allows the development of automated proof strategies, which are represented and stored by the prover in the same way as theorems. The mathematical foundations of the prover and the way these are presented to a user are discussed. The paper also contains discussions of experiences with the prover and relations of the prover to other systems.

Accelerating tableaux proofs using compact representations

In this article a modified form of tableau calculus, calledTableau Graph Calculus, is presented for overcoming the well-known inefficiencies of the traditional tableau calculus to a large extent. This calculus is based on a compact representation of analytic tableaux by using graph structures calledtableau graphs. These graphs are obtained from the input formula in linear time and incorporate most of the rule applications of normal tableau calculus during the conversion process. The size of this representation is linear with respect to the length of the input formula and the graph closely resembles the proof tree of tableau calculi thus retaining the naturalness of the proof procedure. As a result of this preprocessing step, tableau graph calculus has only a single rule which is repeatedly applied to obtain a proof. Many optimizations for the applications of this rule, to effectively prune the search space, are presented as well. Brief details of an implemented prover called FAUST, embedded within the higher-order theorem prover called HOL, are also given. Applications of FAUST within a hardware verification context are also sketched.