paperbot · PL 论文追踪

全部论文 待读 收藏

P2 A graded dependent type system with a usage-aware semantics

一个带使用感知语义的分级依赖类型系统
POPL 5(POPL)2021匹配度 2/10 · 论文属于纯类型论与形式语义研究,关注分级依赖类型和资源核算,与用户核心方向(LLM/Agent 与静态分析、浮点程序分析、贝叶斯程序分析及其交叉)没有直接关联。仅在最宽泛的“程序语言理论/形式化验证”层面上相关,但未涉及 LLM、Agent、工业落地或相关 benchmark。引用 34 inbox
Pritam Choudhury, Harley Eades III, Richard A. Eisenberg, Stephanie Weirich
crossref: 2026-09-10

AI 速览

一句话:提出 GraD 分级依赖类型系统,并给出基于堆的使用感知语义与资源核算可靠性证明。

分级类型论(graded type theory)能够跟踪并推理类型系统中的资源使用,但标准操作语义通常不感知资源,使资源核算的正确性难以直接说明。本文提出 GraD:一种新的分级依赖类型系统,涵盖函数、张量积、加和类型与单位类型,并为其设计基于堆的操作语义。作者证明了一个可靠性定理,表明系统能正确核算资源使用。由此可推出若干性质,包括标准类型可靠性、无关(irrelevant)资源不干扰计算,以及线性资源的单指针性质。该工作为在 Haskell 等实用语言中整合线性、无关性与依赖类型奠定了基础。

问题
分级类型论可跟踪资源使用,但标准操作语义不感知资源,缺少对资源核算正确性的语义保证。
方法
提出 GraD 分级依赖类型系统,包含函数、张量积、加和与单位类型,并给出基于堆的操作语义,证明资源使用核算的可靠性定理。
评估
以形式化定义与定理证明为评估,证明 soundness theorem 并由此推导标准类型可靠性、无关资源不干扰与线性资源单指针等性质。
原文摘要(Abstract)

Graded Type Theory provides a mechanism to track and reason about resource usage in type systems. In this paper, we develop GraD, a novel version of such a graded dependent type system that includes functions, tensor products, additive sums, and a unit type. Since standard operational semantics is resource-agnostic, we develop a heap-based operational semantics and prove a soundness theorem that shows correct accounting of resource usage. Several useful properties, including the standard type soundness theorem, non-interference of irrelevant resources in computation and single pointer property for linear resources, can be derived from this theorem. We hope that our work will provide a base for integrating linearity, irrelevance and dependent types in practical programming languages like Haskell.

链接与引用

DOI 原文 · PDF(开放获取) ·

BibTeX
@article{paperbot1149,
  title = {A graded dependent type system with a usage-aware semantics},
  author = {Pritam Choudhury and Harley Eades III and Richard A. Eisenberg and Stephanie Weirich},
  journal = {Proceedings of the ACM on Programming Languages},
  volume = {5},
  number = {POPL},
  year = {2021},
  doi = {10.1145/3434331}
}