Abstract
This development formalizes the bounded multi-source shortest path
(BMSSP) algorithm underlying the deterministic directed single-source
shortest path algorithm of Duan, Mao, Mao, Shu, and Yin [2]. It
proves three explicit claims. First, the executable bmssp_distances
computes the exact reachable shortest-distance map for well-formed
f
inite natural weighted digraphs. Second, a separate costed BMSSP
recurrence satisfies the O(mlog2/3 n) bound under the abstract In
sert/BatchPrepend/Pull operation-cost interface. Third, the bucketed
partition operations realize the primitive costs required by that interface.
The formalization does not state a single generated-code machine-step
theorem for the executable. The entry also exports executable SML
and includes a checked worked shortest-path example whose evaluated
output is [(0,0),(1,3),(2,5),(3,8),(4,6)]
License
Note
Codex with gpt 5.5 on xhigh was used to help with proof engineering
Topics
Session BMSSP_Correctness
- BMSSP_Correctness
- BMSSP_Algorithm_Correctness
- BMSSP_Shortest_Path_Lemmas
- BMSSP_Base_Case
- BMSSP_Partition_Interface
- BMSSP_Bucketed_Partition_Internal
- BMSSP_Bucketed_Partition
- BMSSP_Code_Export
- BMSSP_Executable_Base_Case
- BMSSP_Unique_Shortest_Tree
- BMSSP_Find_Pivots_Core
- BMSSP_Find_Pivots
- BMSSP_Pull_Minimum
- BMSSP_Partition_Pull_Bridge
- BMSSP_Initialization
- BMSSP_Concrete_Step
- BMSSP_Concrete_Top_Level
- BMSSP_Recursive
- BMSSP_Operational_Pull
- BMSSP_Complexity
- BMSSP_Range_Costed
- BMSSP_Exact_Range_Costed
- BMSSP_Direct_Insert_Costed
- BMSSP_Strict_Tie_Breaking
- BMSSP_Partition_Data_Structure
- BMSSP_Exact_Concrete_Cost
- BMSSP_Top_Level_Bounds
- BMSSP_Executable_Refinement_Internal
- BMSSP_Executable_Refinement
- BMSSP_Executable_Headline
- BMSSP_Runtime_Instance
- BMSSP_Bucketed_Cost_Bridge
- BMSSP_Path_Family
- BMSSP_Bucketed_Runtime