A Logic for Reasoning About LF Specifications

TL;DR

This paper introduces L_{LF}, a logic designed to formalize and reason about properties of dependently typed LF specifications, supporting complex quantification, case analysis, and inductive reasoning.

Contribution

It presents a novel logic and proof system for LF specifications, enabling formal reasoning about their properties with support for inductive and case analysis reasoning.

Findings

Formalization of LF properties using L_{LF}

Proof system supports inductive and case analysis reasoning

Application to type uniqueness in simply-typed lambda calculus

Abstract

We present a logic named L_{LF} whose intended use is to formalize properties of specifications developed in the dependently typed lambda calculus LF. The logic is parameterized by the LF signature that constitutes the specification. Atomic formulas correspond to typing derivations relative to this signature. The logic includes a collection of propositional connectives and quantifiers. Quantification ranges over expressions that denote LF terms and LF contexts. Quantifiers of the first variety are qualified by simple types that describe the functional structure associated with the variables they bind; deeper, dependency related properties are expressed by the body of the formula. Context-level quantifiers are qualified by context schemas that identify patterns of declarations out of which actual contexts may be constructed. The semantics of variable-free atomic formulas is articulated via the derivability in LF of the judgements they encode. Propositional constants and connectives are understood in the usual manner and the meaning of quantifiers is explicated through substitutions of expressions that adhere to the type qualifications. The logic is complemented by a proof system that enables reasoning that is sound with respect to the described semantics. The main novelties of the proof system are the provision for case-analysis style reasoning about LF judgements, support for inductive reasoning over the heights of LF derivations and the encoding of LF meta-theorems. The logic is motivated by the paradigmatic example of type assignment in the simply-typed lambda calculus and the proof system is illustrated through the formalization of a proof of type uniqueness for this calculus.