Ekaterina Komendantskaya

Coalgebraic logic programming: from Semantics to Implementation

Coinductive definitions, such as that of an infinite stream, may often be described by elegant logic programs, but ones for which SLD-refutation is of no value as SLD-derivations fall into infinite loops. Such definitions give rise to questions of lazy corecursive derivations and parallelism, as execution of such logic programs can have both recursive and corecursive features at once. Observational and coalgebraic semantics have been used to study them abstractly. The programming developments have often occurred separately and have usually been implementation-led. Here, we give a coherent semantics-led account of the issues, starting with abstract category theoretic semantics, developing coalgebra to characterize naturally arising trees and proceeding towards implementation of a new dialect, CoALP, of logic programming, characterised by guarded lazy corecursion and parallelism.

Coalgebraic Derivations in Logic Programming

Coalgebra may be used to provide semantics for SLD-derivations, both finite and infinite. We first give such semantics to classical SLD-derivations, proving results such as adequacy, soundness and completeness. Then, based upon coalgebraic semantics, we propose a new sound and complete algorithm for parallel derivations. We analyse this new algorithm in terms of the Theory of Observables, and we prove correctness and full abstraction results.

Inductive and Coinductive Components of Corecursive Functions in Coq

In Constructive Type Theory, recursive and corecursive definitions are subject to syntactic restrictions which guarantee termination for recursive functions and productivity for corecursive functions. However, many terminating and productive functions do not pass the syntactic tests. Bove proposed in her thesis an elegant reformulation of the method of accessibility predicates that widens the range of terminative recursive functions formalisable in Constructive Type Theory. In this paper, we pursue the same goal for productive corecursive functions. Notably, our method of formalisation of coinductive definitions of productive functions in Coq requires not only the use of ad-hoc predicates, but also a systematic algorithm that separates the inductive and coinductive parts of functions.

Proof-Pattern Recognition and Lemma Discovery in ACL2

We present a novel technique for combining statistical machine learning for proof-pattern recognition with symbolic methods for lemma discovery. The resulting tool, ACL2(ml), gathers proof statistics and uses statistical pattern-recognition to pre-processes data from libraries, and then suggests auxiliary lemmas in new proofs by analogy with already seen examples. This paper presents the implementation of ACL2(ml) alongside theoretical descriptions of the proof-pattern recognition and lemma discovery methods involved in it.

Machine Learning in Proof General: Interfacing Interfaces

We present ML4PG - a machine learning extension for Proof General. It allows users to gather proof statistics related to shapes of goals, sequences of applied tactics, and proof tree structures from the libraries of interactive higher-order proofs written in Coq and SSReflect. The gathered data is clustered using the state-of-the-art machine learning algorithms available in MATLAB and Weka. ML4PG provides automated interfacing between Proof General and MATLAB/Weka. The results of clustering are used by ML4PG to provide proof hints in the process of interactive proof development.

Coalgebraic semantics for derivations in logic programming

Every variable-free logic program induces a PfPf-coalgebra on the set of atomic formulae in the program. The coalgebra p sends an atomic formula A to the set of the sets of atomic formulae in the antecedent of each clause for which A is the head. In an earlier paper, we identified a variable-free logic program with a PfPf -coalgebra on Set and showed that, if C(PfPf) is the cofree comonad on PfPf, then given a logic program P qua PfPf -coalgebra, the corresponding C(PfPf)- coalgebra structure describes the parallel and-or derivation trees of P. In this paper, we extend that analysis to arbitrary logic programs. That requires a subtle analysis of lax natural transformations between Poset-valued functors on a Lawvere theory, of locally ordered endofunctors and comonads on locally ordered categories, and of coalgebras, oplax maps of coalgebras, and the relationships between such for locally ordered endofunctors and the cofree comonads on them.

Using Structural Recursion for Corecursion

We propose a (limited) solution to the problem of constructing stream values defined by recursive equations that do not respect the guardedness condition. The guardedness condition is imposed on definitions of corecursive functions in Coq, AGDA, and other higher-order proof assistants. In this paper, we concentrate in particular on those non-guarded equations where recursive calls appear under functions. We use a correspondence between streams and functions over natural numbers to show that some classes of non-guarded definitions can be modelled through the encoding as structural recursive functions. In practice, this work extends the class of stream values that can be defined in a constructive type theory-based theorem prover with inductive and coinductive types, structural recursion and guarded corecursion.

Coalgebraic semantics for parallel derivation strategies in logic programming

Logic programming, a class of programming languages based on first-order logic, provides simple and efficient tools for goal-oriented proof-search. Logic programming supports recursive computations, and some logic programs resemble the inductive or coinductive definitions written in functional programming languages. In this paper, we give a coalgebraic semantics to logic programming. We show that ground logic programs can be modelled by either PfPf -coalgebras or Pf Listcoalgebras on Set. We analyse different kinds of derivation strategies and derivation trees (proof-trees, SLD-trees, and-or parallel trees) used in logic programming, and show how they can be modelled coalgebraically.

Proof Relevant Corecursive Resolution

Resolution lies at the foundation of both logic programming and type class context reduction in functional languages. Terminating derivations by resolution have well-defined inductive meaning, whereas some non-terminating derivations can be understood coinductively. Cycle detection is a popular method to capture a small subset of such derivations. We show that in fact cycle detection is a restricted form of coinductive proof, in which the atomic formula forming the cycle plays the role of coinductive hypothesis.

Connectionist Representation of Multi-Valued Logic Programs

Holldobler and Kalinke showed how, given a propositional logic program P , a 3-layer feedforward artificial neural network may be constructed, using only binary threshold units, which can compute the familiar immediate-consequence operator TP associated with P . In this chapter, essentially these results are established for a class of logic programs which can handle many-valued logics, constraints and uncertainty; these programs therefore represent a considerable extension of conventional propositional programs. The work of the chapter basically falls into two parts. In the first of these, the programs considered extend the syntax of conventional logic programs by allowing elements of quite general algebraic structures to be present in clause bodies. Such programs include many-valued logic programs, and semiring-based constraint logic programs. In the second part, the programs considered are bilattice-based annotated logic programs in which body literals are annotated by elements drawn from bilattices. These programs are well-suited to handling uncertainty. Appropriate semantic operators are defined for the programs considered in both parts of the chapter, and it is shown that one may construct artificial neural networks for computing these operators. In fact, in both cases only binary threshold units are used, but it simplifies the treatment conceptually to arrange them in so-called multiplication and addition units in the case of the programs of the first part.