Uloženo v:
| Hlavní autor: | |
|---|---|
| Médium: | Recurso digital |
| Jazyk: | angličtina |
| Vydáno: |
Zenodo
2025
|
| Témata: | |
| On-line přístup: | https://doi.org/10.5281/zenodo.17018248 |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
Obsah:
- <p><strong>This work presents a polynomial-time structural beam search algorithm for the Traveling Salesman Problem (TSP). The method combines bounded beam expansion with a restricted 2-opt repair step, ensuring polynomial bounds on both time and space. Experimental validation was carried out on symmetric, asymmetric, and blocked matrices for problem sizes up to n = 100. For n ≤ 15, the algorithm’s outputs were verified against the Held–Karp exact method, with complete agreement. Beyond this range, stress tests demonstrated empirical polynomial behavior, with runtime growth around O(n^{3.4}) and memory growth between O(n^{1.2}) and O(n^{1.7}). These results highlight the scalability and reliability of the approach, providing a new structural heuristic framework for combinatorial optimization</strong></p>