Library Header Image
LSE Research Online LSE Library Services

The probabilistic relationship between the assignment and travelling salesman problems

Frieze, Alan and Sorkin, Gregory B. ORCID: 0000-0003-4935-7820 (2006) The probabilistic relationship between the assignment and travelling salesman problems. SIAM Journal on Computing, 36 (5). pp. 1435-1452. ISSN 0097-5397

Full text not available from this repository.
Identification Number: 10.1137/S0097539701391518


We consider the gap between the cost of an optimal assignment in a complete bipartite graph with random edge weights, and the cost of an optimal traveling salesman tour in a complete directed graph with the same edge weights. Using an improved “patching” heuristic, we show that with high probability the gap is $O((\ln n)^2/n)$, and that its expectation is $\Omega(1/n)$. One of the underpinnings of this result is that the largest edge weight in an optimal assignment has expectation $\Theta(\ln n / n)$. A consequence of the small assignment–TSP gap is an $e^{\tilde{O}(\sqrt{n})}$‐time algorithm which, with high probability, exactly solves a random asymmetric traveling salesman instance. In addition to the assignment–TSP gap, we also consider the expected gap between the optimal and second‐best assignments; it is at least $\Omega(1/n^2)$ and at most $O(\ln n/n^2)$.

Item Type: Article
Official URL:
Additional Information: © 2007 Society for Industrial and Applied Mathematics
Divisions: Management
Subjects: Q Science > QA Mathematics
Date Deposited: 13 Apr 2011 13:45
Last Modified: 16 May 2024 00:31

Actions (login required)

View Item View Item