We present the first sound and complete axiomatization of infinite trace semantics for generative probabilistic transition systems. Our approach is categorical, and we build on recent results on proper functors over convex sets. At the core of our proof is a characterization of infinite traces as the final coalgebra of a functor over convex algebras. Somewhat surprisingly, our axiomatization of infinite trace semantics coincides with that of finite trace semantics, even though the techniques used in the completeness proof are significantly different.

We forge connections between the theory of fractal sets obtained as attractors of iterated function systems and process calculi. To this end, we reinterpret Milner's expressions for processes as contraction operators on a complete metric space. When the space is, for example, the plane, the denotations of fixed point terms correspond to familiar fractal sets. We give a sound and complete axiomatization of fractal equivalence, the congruence on terms consisting of pairs that construct identical self-similar sets in all interpretations. We further make connections to labelled Markov chains and to invariant measures. In all of this work, we use important results from process calculi. For example, we use Rabinovich's completeness theorem for trace equivalence in our own completeness theorem. In addition to our results, we also raise many questions related to both fractals and process calculi.

This paper is a contribution to neural network semantics, a foundational framework for neuro-symbolic AI. The key insight of this theory is that logical operators can be mapped to operators on neural network states. In this paper, we do this for a neural network learning operator. We map a dynamic operator [φ] to iterated Hebbian learning, a simple learning policy that updates a neural network by repeatedly applying Hebb's learning rule until the net reaches a fixed-point. Our main result is that we can "translate away" [φ]-formulas via reduction axioms. This means that completeness for the logic of iterated Hebbian learning follows from completeness of the base logic. These reduction axioms also provide (1) a human-interpretable description of iterated Hebbian learning as a kind of plausibility upgrade, and (2) an approach to building neural networks with guarantees on what they can learn.

The Vietoris space of compact subsets of a given Hausdorff space yields an endofunctor V on the category of Hausdorff spaces. Vietoris polynomial endofunctors on that category are built from V, the identity and constant functors by forming products, coproducts and compositions. These functors are known to have terminal coalgebras and we deduce that they also have initial algebras. We present an analogous class of endofunctors on the category of extended metric spaces, using in lieu of V the Hausdorff functor ℋ. We prove that the ensuing Hausdorff polynomial functors have terminal coalgebras and initial algebras. Whereas the canonical constructions of terminal coalgebras for Vietoris polynomial functors takes ω steps, one needs ω + ω steps in general for Hausdorff ones. We also give a new proof that the closed set functor on metric spaces has no fixed points.

Abstract Against the backdrop of the ever-improving Natural Language Inference (NLI) models, recent efforts have focused on the suitability of the current NLI datasets and on the feasibility of the NLI task as it is currently approached. Many of the recent studies have exposed the inherent human disagreements of the inference task and have proposed a shift from categorical labels to human subjective probability assessments, capturing human uncertainty. In this work, we show how neither the current task formulation nor the proposed uncertainty gradient are entirely suitable for solving the NLI challenges. Instead, we propose an ordered sense space annotation, which distinguishes between logical and common-sense inference. One end of the space captures non-sensical inferences, while the other end represents strictly logical scenarios. In the middle of the space, we find a continuum of common-sense, namely, the subjective and graded opinion of a “person on the street.” To arrive at the proposed annotation scheme, we perform a careful investigation of the SICK corpus and we create a taxonomy of annotation issues and guidelines. We re-annotate the corpus with the proposed annotation scheme, utilizing four symbolic inference systems, and then perform a thorough evaluation of the scheme by fine-tuning and testing commonly used pre-trained language models on the re-annotated SICK within various settings. We also pioneer a crowd annotation of a small portion of the MultiNLI corpus, showcasing that it is possible to adapt our scheme for annotation by non-experts on another NLI corpus. Our work shows the efficiency and benefits of the proposed mechanism and opens the way for a careful NLI task refinement.

An old theorem of Adámek constructs initial algebras for sufficiently cocontinuous endofunctors via transfinite iteration over ordinals in classical set theory. We prove a new version that works in constructive logic, using "inflationary" iteration over a notion of size that abstracts from limit ordinals just their transitive, directed and well-founded properties. Borrowing from Taylor's constructive treatment of ordinals, we show that sizes exist with upper bounds for any given signature of indexes. From this it follows that there is a rich class of endofunctors to which the new theorem applies, provided one admits a weak form of choice (WISC) due to Streicher, Moerdijk, van den Berg and Palmgren, and which is known to hold in the internal constructive logic of many kinds of topos.

Do state-of-the-art models for language understanding already have, or can they easily learn, abilities such as boolean coordination, quantification, conditionals, comparatives, and monotonicity reasoning (i.e., reasoning about word substitutions in sentential contexts)? While such phenomena are involved in natural language inference (NLI) and go beyond basic linguistic understanding, it is unclear the extent to which they are captured in existing NLI benchmarks and effectively learned by models. To investigate this, we propose the use of semantic fragments—systematically generated datasets that each target a different semantic phenomenon—for probing, and efficiently improving, such capabilities of linguistic models. This approach to creating challenge datasets allows direct control over the semantic diversity and complexity of the targeted linguistic phenomena, and results in a more precise characterization of a model's linguistic behavior. Our experiments, using a library of 8 such semantic fragments, reveal two remarkable findings: (a) State-of-the-art models, including BERT, that are pre-trained on existing NLI benchmark datasets perform poorly on these new fragments, even though the phenomena probed here are central to the NLI task; (b) On the other hand, with only a few minutes of additional fine-tuning—with a carefully selected learning rate and a novel variation of “inoculation”—a BERT-based model can master all of these logic and monotonicity fragments while retaining its performance on established NLI benchmarks.

Abstract This paper studies fundamental questions concerning category-theoretic models of induction and recursion. We are concerned with the relationship between well-founded and recursive coalgebras for an endofunctor. For monomorphism preserving endofunctors on complete and well-powered categories every coalgebra has a well-founded part, and we provide a new, shorter proof that this is the coreflection in the category of all well-founded coalgebras. We present a new more general proof of Taylor’s General Recursion Theorem that every well-founded coalgebra is recursive, and we study conditions which imply the converse. In addition, we present a new equivalent characterization of well-foundedness: a coalgebra is well-founded iff it admits a coalgebra-to-algebra morphism to the initial algebra.

Abstract We add MostXareYto the syllogistic logic of AllXareYand SomeXareY. We prove soundness, completeness, and decidability in polynomial time. Our logic has infinitely many rules, and we prove that this is unavoidable.

We present a new proof system for equality of terms which present elements of the final coalgebra of a finitary set functor. This is most important when the functor is finitary, and we improve on logical systems which have already been proposed in several papers. Our contributions here are (1) a new logical rule which makes for proofs which are somewhat easier to find, and (2) a soundness/completeness theorem which works for all finitary functors, in particular removing a weak pullback preservation requirement that had been used previously. Our work is based on properties of precongruence relations and also on a new parametrized coinduction principle.

Our paper is actually two contributions in one. First, we argue that IBM's Jeopardy! playing machine needs a formal semantics. We present several arguments as we discuss the system. We also situate the work in the broader context of contemporary AI. Our second point is that the work in this area might well be done as a broad collaborative project. Hence our "Blue Sky'' contribution is a proposal to organize a polymath-style effort aimed at developing formal tools for the study of state of the art question-answer systems, and other large scale NLP efforts whose architectures and algorithms lack a theoretical foundation.

We consider infinite sequences of symbols, also known as streams, and the decidability question for equality of streams defined in a restricted format. (Some formats lead to undecidable equivalence problems.) This restricted format consists of prefixing a symbol at the head of a stream, of the stream function `zip', and recursion variables. Here `zip' interleaves the elements of two streams alternatingly. The celebrated Thue- Morse sequence is obtained by the succinct `zip-specification' M = 0 : X X = 1 : zip(X, Y) Y = 0 : zip(Y, X) The main results are as follows. We establish decidability of equivalence of zip-specifications, by employing bisimilarity of observation graphs based on a suitably chosen cobasis. Furthermore, our analysis, based on term rewriting and coalgebraic techniques, reveals an intimate connection between zip-specifications and automatic sequences. This leads to a new and simple characterization of automatic sequences. The study of zip-specifications is placed in a wider perspective by employing observation graphs in a dynamic logic setting, yielding yet another alternative characterization of automatic sequences. By the first characterization result, zip-specifications can be perceived as a term rewriting syntax for automatic sequences. For streams σ the following are equivalent: (a) σ can be specified using zip; (b) σ is 2-automatic; and (c) σ has a finite observation graph using the cobasis (hd, even, odd). Here even and odd are defined by even(a : s) = a : odd(s), and odd(a : s) = even(s). The generalization to zip-k specifications (with zip-k interleaving k streams) and to k-automaticity is straightforward. As a natural extension of the class of automatic sequences, we also consider `zip-mix' specifications that use zips of different arities in one specification. The corresponding notion of automaton employs a state-dependent input-alphabet, with a number representation (n)A = dm... d0where the base of digit di is determined by the automaton A on input di-1... d0. Finally we show that equivalence is undecidable for a simple extension of the zip-mix format with projections analogous to even and odd.

The aim of this tutorial is to present the area of coalgebra to people interested in the kinds of semantic modeling that is prominent at TARK. Coalgebra is a general study of a great many kinds of models, and these include type spaces and Kripke models, and many others. But the theory is not overly general, it is not a theory of absolutely everything. The tutorial is designed to be a short introduction to a substantial technical field, bearing in mind that this is nearly impossible. It is also intended to bring together literatures from theoretical computer science and game theory.

This article provides sound and complete logical systems for several fragments of English which go beyond syllogistic logic in that they use verbs as well as other limited syntactic material: universally and existentially quantified noun phrases, building on the work of Nishihara, Morita and Iwata (1990, Systems and Computers in Japan, 21, 96–111); complemented noun phrases, following our Moss (2007, Syllogistic Logic with Complements); and noun phrases which might contain relative clauses, recursively, based on McAllester and Givan (1992, Artifical Intelligence, 56, 1–20). The logics are all syllogistic in the sense that they do not make use of individual variables. Variables in our systems range over nouns, and in the last system, over verbs as well.

This article has two purposes. The first is to present a final coalgebra construction for finitary endofunctors on Set that uses a certain subset L* of the limit L of the first ω terms in the final sequence. L* is the set of points in L which arise from all coalgebras using their canonical morphisms into L, and it was used earlier for different purposes in Kurz and Pattinson (2005, Mathematical Structures in Computer Science, 15, 543–473). Viglizzo (2005, PhD Dessertation, Indiana University) showed that the same set L* carried a final coalgebra structure for functors in a certain inductively defined family. Our first goal is to generalize this to all finitary endofunctors; the result is implicit in Worrell (2005, Theoritical Computer Science, 338, 184–199). The second goal is to use the final coalgebra construction to propose coalgebraic logics similar to those in Lawrence S. Moss (1999, Annals of Pure and Applied Logic, 96, 277–317) but for all finitary endofunctors F on Set. This time one can dispense with all conditions on F, construct a logical language ℒF directly from it, and prove that two points in a coalgebra satisfy the same sentences of ℒF iff they are identified by the final coalgebra morphism. The language ℒF is very spare, having no boolean connectives. This work on ℒF is thus a re-working of coalgebraic logic for finitary functors on sets.

§1. Introduction. Our understanding of Nature comes in layers, so should the development of logic. Classic logic is an indispensable part of our knowledge, and its interactions with computer science have recently dramatically changed our life. A new layer of logic has been developing ever since the discovery of quantum mechanics. G. D. Birkhoff and von Neumann introduced quantum logic in a seminal paper in 1936 [1]. But the definition of quantum logic varies among authors (see [2]). How to capture the logic structure inherent in quantum mechanics is very interesting and challenging. Given the close connection between classical logic and theoretical computer science as exemplified by the coincidence of computable functions through Turing machines, recursive function theory, and λ-calculus, we are interested in how to gain some insights about quantum logic from quantum computing. In this note we make some observations about quantum logic as motivated by quantum computing (see [5]) and hope more people will explore this connection. The quantum logic as envisioned by Birkhoff and von Neumann is based on the lattice of closed subspaces of a Hilbert space, usually an infinite dimensional one. The quantum logic of a fixed Hilbert space ℍ in this note is the variety of all the true equations with finitely many variables using the connectives meet, join and negation. Quantum computing is theoretically based on quantum systems with finite dimensional Hilbert spaces, especially the states space of a qubit ℂ2. (Actually the qubit is merely a convenience.

The articles in this part of this issue of the journal are a selection of the papers originally presented to the Fifth Workshop on Coalgebraic Methods in Computer Science. CMCS was held in April 2002, in Grenoble, France as a satellite conference of ETAPS. The conference proceedings were published in Electronic Notes in Theoretical Computer Science65 (1). There were fifteen contributed papers and two invited talks. Six papers were selected by the Program Committee after a meeting for consideration of this issue, and revised versions were then sent to referees. In consultation with them, I selected the four published here. Most of them were later revised, so, in general, the papers appearing here are improved versions of the conference versions.

Abstract We study logical systems for reasoning about equations involving recursive definitions. In particular, we are interested in “propositional” fragments of the functional language of recursion FLR [18, 17], i.e., without the value passing or abstraction allowed in FLR. The “pure,” propositional fragment FLR0 turns out to coincide with the iteration theories of [1]. Our main focus here concerns the sharp contrast between the simple class of valid identities and the very complex consequence relation over several natural classes of models.

We present an implemented unification-based parser for relational grammars developed within the stratified feature grammar (SFG) framework, which generalizes Kasper-Rounds logic to handle relational grammar analyses. We first introduce the key aspects of SFG and a lexicalized, graph-based variant of the framework suitable for implementing relational grammars. We then describe a head-driven chart parser for lexicalized SFG. The basic parsing operation is essentially ordinary feature-structure unification augmented with an operation of label unification to build the stratified features characteristic of SFG.

Motivated by ideas from the study of abstract data types, the authors show how to interpret non-well-founded sets as fixed points of continuous transformations of an initial continuous algebra. They consider a preordered structure closely related to the set HF of well-founded, hereditarily finite sets. By taking its ideal completion, the authors obtain an initial continuous algebra in which they are able to solve all of the usual systems of equations that characterize hereditarily finite, non-well-founded sets. In this way, they are able to obtain a structure which is isomorphic to HF/sub 1/, the non-well-founded analog to HF.>