Performance Guarantees for the TSP with a Parameterized Triangle Inequality

  • Authors:
  • Michael A. Bender;Chandra Chekuri

  • Affiliations:
  • -;-

  • Venue:
  • WADS '99 Proceedings of the 6th International Workshop on Algorithms and Data Structures
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

We consider the approximability of the tsp problem in graphs that satisfy a relaxed form of triangle inequality. More precisely, we assume that for some parameter Τ ≥ 1, the distances satisfy the inequality dist(x, y) ≤ Τċ(dist(x, z)+dist(z, y)) for every triple of vertices x, y, and z. We obtain a 4Τ approximation and also show that for some Ɛ 0 it is np-hard to obtain a (1+ ƐΤ) approximation. Our upper bound improves upon the earlier known ratio of (3Τ2=2+Τ2) [1] for all values of Τ