arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2401.16233cs.PL
收藏

基于模块归纳的Dafny增量式证明开发

Incremental Proof Development in Dafny with Module-Based Induction

Son Ho, Clément Pit-Claudel

AI总结:

针对Dafny等高自动化定理证明器迭代修改时验证时间波动大、证明失败难调试的问题,提出用Dafny模块编码类Coq归纳原理,提升归纳数据结构相关证明的稳定性与可维护性。

摘要:

Dafny这类高度自动化的定理证明器让用户只需少量工作就能证明简单性质,便于快速搭建证明框架。但缺点是这类证明器让用户对证明搜索的控制度很低,意味着证明编写迭代过程中固有的微小改动,往往会导致验证时间出现不可预测的波动,最终引发难以诊断的证明失败。有时这种高自动化的优势会变成弊端:它不像Coq那样会提前中断并向用户展示未解决的目标,反而会让证明逐渐变得不稳定,直到验证时间暴增。此时由于缺乏可探查的证明上下文,用户往往要经历痛苦的调试过程。本文展示了如何利用Dafny模块编码类Coq的归纳原理,大幅提升关于归纳数据结构的证明的稳定性与可维护性。

英文摘要:

Highly automated theorem provers like Dafny allow users to prove simple properties with little effort, making it easy to quickly sketch proofs. The drawback is that such provers leave users with little control about the proof search, meaning that the small changes inherent to the iterative process of writing a proof often lead to unpredictable variations in verification time, and eventually hard-to-diagnose proof failures. This sometimes turns the boon of high automation into a curse, as instead of breaking early and showing unsolved goals to the user like in Coq, proofs tend to gradually become unstable until their verification time explodes. At this point, the absence of a proof context to investigate often leaves the user to a painful debugging session. In this paper, we show how to use Dafny modules to encode Coq-like induction principles to dramatically improve the stability and maintainability of proofs about inductive data structures.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑