We construct a realizability model of linear dependent type theory from a linear combinatory algebra. Our model motivates a number of additions to the type theory. In particular, we add a universe with two decoding operations: one takes codes to cartesian types and the other takes codes to linear types. The universe is impredicative in the sense that it is closed under both large cartesian dependent products and large linear dependent products. We also add a rule for injectivity of the modality turning linear terms into cartesian terms. With all of the additions, we are able to encode (linear) inductive types. As a case study, we consider the type of lists over a linear type, and demonstrate that our encoding has the relevant uniqueness principle. The construction of the realizability model is fully formalized in the proof assistant Rocq.

Substructural type systems provide the ability to speak about resources . By enforcing usage restrictions on inputs to computations they allow programmers to reify limited system units–such as memory–in types. We demonstrate a new form of resource reasoning founded on constraining outputs and explore its utility for practical programming. In particular, we identify a number of disparate programming features explored largely in the security literature as various fragments of our unified framework. These encompass capabilities, quantitative information leakage, sandboxing in the style of the Linux seccomp interface, authorization protocols, and more. We furthermore explore its connection to conventional input-based resource reasoning, casting it as an internal treatment of the constructive Kripke semantics of substructural logics. We verify the capability, quantity, and protocol safety of our system through a single logical relations argument. In doing so, we take the first steps towards obtaining the ultimate multitool for security reasoning.

Many proof assistants allow the use of features and axioms that increase their expressive power. However, these extensions must be used with care, as some combinations are known to lead to logical inconsistencies. Therefore, proof assistants include mechanisms that track which extensions are used in a proof development or module, ensuring that incompatible extensions are not used simultaneously. Unfortunately, existing extension tracking mechanisms are external to the type system. This means that we cannot specify precisely which extensions a definition depends on. Having the ability to write more precise specifications means we are not picking an overapproximation of the extensions needed, which prevents reusing definitions in the presence of incompatible extensions. Furthermore, we cannot refer to definitions that use incompatible extensions even if they are never used in inconsistent ways. The reasoning principles of one extension therefore cannot be used as a metatheory to reason about the properties of an incompatible extension. In this report, I explore the use of the Dependent Calculus of Indistinguishability (DCOI) by Liu et al. for extension tracking. DCOI is a dependent type system with dependency tracking, where terms and variables are assigned dependency levels alongside their types. These dependency levels form a lattice that describes which levels are permitted to access what. To instead track extensions, each set of extensions would correspond to a dependency level, and the lattice would describe how extensions are permitted to interact.

We present a new design for an algebraic simplification library structured around concepts from universal algebra: theories, models, homomorphisms, and universal properties of free algebras and free extensions of algebras. The library’s dependently typed interface guarantees that both built-in and user-defined simplification modules are terminating, sound, and complete with respect to a well-specified class of equations. We have implemented the design in the Idris 2 and Agda dependently typed programming languages and shown that it supports modular extension to new theories, proof extraction and certification, goal extraction via reflection, and interactive development.

In this work, we explore Landin’s Knot, which is understood as a pattern for encoding general recursion, including non-termination, that is possible after adding higher-order references to an otherwise terminating language. We observe that this isn’t always true – higher-order references, by themselves, don’t lead to non-termination. The key insight is that Landin’s Knot relies not primarily on references storing functions, but on unrestricted quantification over a function’s environment. We show this through a closure converted language, in which the function’s environment is made explicit and hides the type of the environment through impredicative quantification. Once references are added, this impredicative quantification can be exploited to encode recursion. We conjecture that by restricting the quantification over the environment, higher-order references can be safely added to terminating languages, without resorting to more complex type systems such as linearity, and without restricting references from storing functions.

The Dependent Calculus of Indistinguishability (DCOI) uses dependency tracking to identify irrelevant arguments and uses indistinguishability during type conversion to enable proof irrelevance, supporting run-time and compile-time irrelevance with the same uniform mechanism. DCOI also internalizes reasoning about indistinguishability through the use of a propositional equality type indexed by an observer level. As DCOI is a pure type system, prior work establishes only its syntactic type safety, justifying its use as the basis for a programming language with dependent types. However, it was not clear whether any instance of this system would be suitable for use as a type theory for theorem proving. Here, we identify a suitable instance DCOI ω , which has an infinite predicative universe hierarchy. We show that DCOI ω is logically consistent, normalizing, and that type conversion is decidable. We have mechanized all results using the Coq proof assistant.

Ordered, linear, and other substructural type systems allow us to expose deep properties of programs at the syntactic level of types. In this paper, we develop a family of unary logical relations that allow us to prove consequences of parametricity for a range of substructural type systems. A key idea is to parameterize the relation by an algebra, which we exemplify with a monoid and commutative monoid to interpret ordered and linear type systems, respectively. We prove the fundamental theorem of logical relations and apply it to deduce extensional properties of inhabitants of certain types. Examples include demonstrating that the ordered types for list append and reversal are inhabited by exactly one function, as are types of some tree traversals. Similarly, the linear type of the identity function on lists is inhabited only by permutations of the input. Our most advanced example shows that the ordered type of the list fold function is inhabited only by the fold function.

This paper presents preliminary work on a general system for integrating dependent types into substructural type systems such as linear logic and linear type theory. Prior work on this front has generally managed to deliver type systems possessing either syntax or semantics inclusive of certain practical applications, but has struggled to combine these all in one and the same system. Toward resolving this difficulty, I propose a novel categorical interpretation of substructural dependent types, analogous to the use of monoidal categories as models of linear and ordered logic, that encompasses a wide class of mathematical and computational examples. On this basis, I develop a general framework for substructural dependent type theories, and proceed to prove some essential metatheoretic properties thereof. As an application of this framework, I show how it can be used to construct a type theory that satisfactorily addresses the problem of effectively representing cut admissibility for linear sequent calculus in a logical framework.

We combine dependent types with linear type systems that soundly and completely capture polynomial time computation. We explore two systems for capturing polynomial time: one system that disallows construction of iterable data, and one, based on the LFPL system of Martin Hofmann, that controls construction via a payment method. Both of these are extended to full dependent types via Quantitative Type Theory, allowing for arbitrary computation in types alongside guaranteed polynomial time computation in terms. We prove the soundness of the systems using a realisability technique due to Dal Lago and Hofmann. Our long-term goal is to combine the extensional reasoning of type theory with intensional reasoning about the resources intrinsically consumed by programs. This paper is a step along this path, which we hope will lead both to practical systems for reasoning about programs’ resource usage, and to theoretical use as a form of synthetic computational complexity theory .

In type systems with dependency tracking, programmers can assign an ordered set of levels to computations and prevent information flow from high-level computations to the low-level ones. The key notion in such systems is indistinguishability : a definition of program equivalence that takes into account the parts of the program that an observer may depend on. In this paper, we investigate the use of dependency tracking in the context of dependently-typed languages. We present the Dependent Calculus of Indistinguishability (DCOI), a system that adopts indistinguishability as the definition of equality used by the type checker. DCOI also internalizes that relation as an observer-indexed propositional equality type, so that programmers may reason about indistinguishability within the language. Our design generalizes and extends prior systems that combine dependency tracking with dependent types and is the first to support conversion and propositional equality at arbitrary observer levels. We have proven type soundness and noninterference theorems for DCOI and have developed a prototype implementation of its type checker.

Deep learning, despite its remarkable achievements, is still a young field. Like the early stages of many scientific disciplines, it is marked by the discovery of new phenomena, ad-hoc design decisions, and the lack of a uniform and compositional mathematical foundation. From the intricacies of the implementation of backpropagation, through a growing zoo of neural network architectures, to the new and poorly understood phenomena such as double descent, scaling laws or in-context learning, there are few unifying principles in deep learning. This thesis develops a novel mathematical foundation for deep learning based on the language of category theory. We develop a new framework that is a) end-to-end, b) unform, and c) not merely descriptive, but prescriptive, meaning it is amenable to direct implementation in programming languages with sufficient features. We also systematise many existing approaches, placing many existing constructions and concepts from the literature under the same umbrella. In Part I we identify and model two main properties of deep learning systems parametricity and bidirectionality by we expand on the previously defined construction of actegories and Para to study the former, and define weighted optics to study the latter. Combining them yields parametric weighted optics, a categorical model of artificial neural networks, and more. Part II justifies the abstractions from Part I, applying them to model backpropagation, architectures, and supervised learning. We provide a lens-theoretic axiomatisation of differentiation, covering not just smooth spaces, but discrete settings of boolean circuits as well. We survey existing, and develop new categorical models of neural network architectures. We formalise the notion of optimisers and lastly, combine all the existing concepts together, providing a uniform and compositional framework for supervised learning.

Adjoint logic is a general approach to combining multiple logics with different structural properties, including linear, affine, strict, and (ordinary) intuitionistic logics, where each proposition has an intrinsic mode of truth. It has been defined in the form of a sequent calculus because the central concept of independence is most clearly understood in this form, and because it permits a proof of cut elimination following standard techniques. In this paper we present a natural deduction formulation of adjoint logic and show how it is related to the sequent calculus. As a consequence, every provable proposition has a verification (sometimes called a long normal form). We also give a computational interpretation of adjoint logic in the form of a functional language and prove properties of computations that derive from the structure of modes, including freedom from garbage (for modes without weakening and contraction), strictness (for modes disallowing weakening), and erasure (based on a preorder between modes). Finally, we present a surprisingly subtle algorithm for type checking.

We present ${𝜆}_{\text{ert}}$, a type theory supporting refinement types with explicit proofs. Instead of solving refinement constraints with an SMT solver like DML and Liquid Haskell, our system requires and permits programmers to embed proofs of properties within the program text, letting us support a rich logic of properties including quantifiers and induction. We show that the type system is sound by showing that every refined program erases to a simply-typed program, and by means of a denotational semantics, we show that every erased program has all of the properties demanded by its refined type. All of our proofs are formalised in Lean 4.

We present a type theory combining both linearity and dependency by stratifying typing rules into a level for logics and a level for programs. The distinction between logics and programs decouples their semantics, allowing the type system to assume tight resource bounds. A natural notion of irrelevancy is established where all proofs and types occurring inside programs are fully erasable without compromising their operational behavior. Through a heap-based operational semantics, we show that extracted programs always make computational progress and run memory clean. Additionally, programs can be freely reflected into the logical level for conducting deep proofs in the style of standard dependent type theories. This enables one to write resource safe programs and verify their correctness using a unified language.

We investigate containers and polynomial functors in Quantitative Type Theory, and give initial algebra semantics of inductive data types in the presence of linearity. We show that reasoning by induction is supported, and equivalent to initiality, also in the linear setting.

Mechanisation of programming language research is of growing interest, and the act of mechanising type systems and their metatheory is generally becoming easier as new techniques are invented. However, state-of-the-art techniques mostly rely on structurality of the type system — that weakening, contraction, and exchange are admissible and variables can be used unrestrictedly once assumed. Linear logic, and many related subsequent systems, provide motivations for breaking some of these assumptions. We present a framework for mechanising the metatheory of certain substructural type systems, in a style resembling mechanised metatheory of structural type systems. The framework covers a wide range of simply typed syntaxes with semiring usage annotations, via a metasyntax of typing rules. The metasyntax for the premises of a typing rule is related to bunched logic, featuring both sharing and separating conjunction, roughly corresponding to the additive and multiplicative features of linear logic. We use the uniformity of syntaxes to derive type system-generic renaming, substitution, and a form of linearity checking.