Abstract
This entry extends the AFP formalization Correctness and Complexity of BMSSP in Isabelle/HOL, which formalizes the bounded multi-source shortest path algorithm of Duan, Mao, Mao, Shu, and Yin.
The extension adds a charged direct-insert relation that records both the range charged by the recurrence and the semantic output returned by each recursive child. It proves that least costs are realized by verified-correct runs, exhibits a closed branching instance, and records a machine-checked totality obstruction for the older diagonal exact-concrete path-family relation. Its main theorem,
path_nonvac_runtime_bigo_card_V_unconditional, proves without run-existence, cost-bound, or closed-bound assumptions that the least charged cost on the growing unit-weight path family is \(O(|V_n| \log^{2/3}|V_n|)\) when the internal schedule is evaluated at \(n^2\). This is a cost-relation theorem, not a machine-step bound for the SML executable from the base entry.License
Note
Generative AI was used for proof engineering.
Topics
Related publications
- Duan, R., Mao, J., Mao, X., Shu, X., & Yin, L. (2025). Breaking the Sorting Barrier for Directed Single-Source Shortest Paths. Proceedings of the 57th Annual ACM Symposium on Theory of Computing, 36–44. https://doi.org/10.1145/3717823.3718179