Efficient decrease-key support remains an important practical concern in priority queues used by shortest-path and graph-processing algorithms. Although duplicate-entry heaps avoid explicit decrease-key operations and are simple to implement, they may accumulate many stale entries when tentative priorities are repeatedly improved. This paper presents FastHeapDict, an indexed binary min-heap based on permanent slot indexing. Instead of repeatedly locating keys through hash-table lookups during heap repair, FastHeapDict assigns each key a stable integer slot and stores that slot directly in heap entries. As a result, position updates during sift-up and sift-down operations are performed through integer indexing while preserving the usual indexed-heap interface. We evaluate FastHeapDict in both Python and C++20 using Dijkstra's algorithm as the primary experimental workload. The benchmarks include constructed dense stress graphs, randomized dense graph variants, and comparisons against duplicate-entry heaps and existing priority-queue alternatives. Across the tested Python workloads, FastHeapDict consistently outperforms the duplicate-entry heapq baseline and the tested heapdict and pqdict implementations. In the C++20 implementation, FastHeapDict also achieves strong constant-factor improvements over std::priority_queue duplicates, std::set-based decrease-key, and std::multiset duplicates. These results indicate that permanent slot indexing can provide a compact and effective implementation strategy for indexed priority queues in workloads where priority updates are frequent.

翻译：暂无翻译