Papers
Topics
Authors
Recent
Search
2000 character limit reached

Efficient constructions of convex combinations for 2-edge-connected subgraphs on fundamental classes

Published 24 Nov 2018 in cs.DM, cs.DS, and math.CO | (1811.09906v3)

Abstract: Finding the exact integrality gap $\alpha$ for the LP relaxation of the 2-edge-connected spanning multigraph problem (2EC) is closely related to the same problem for the Held-Karp relaxation of the metric traveling salesman problem (TSP). While the former problem seems easier than the latter, since it is less constrained, currently the upper bounds on the respective integrality gaps for the two problems are the same. An approach to proving integrality gaps for both of these problems is to consider fundamental classes of extreme points. For 2EC, better bounds on the integrality gap are known for certain important special cases of these fundamental points. For example, for half-integer square points, the integrality gap is between $\frac{6}{5}$ and $\frac{4}{3}$. Our main result is to improve the approximation factor to $\frac{9}{7}$ for 2EC for these points. Our approach is based on constructing convex combinations and our key tool is the top-down coloring framework for tree augmentation, whose flexibility we employ to exploit beneficial properties in both the initial spanning tree and in the input graph. We also show how these tools can be tailored to the closely related problem of uniform covers for which the proofs of the best-known bounds do not yield polynomial-time algorithms. Another key ingredient is to use a rainbow spanning tree decomposition, which allows us to obtain a convex combination of spanning trees with particular properties

Authors (2)
Citations (6)

Summary

No one has generated a summary of this paper yet.

Paper to Video (Beta)

No one has generated a video about this paper yet.

Whiteboard

No one has generated a whiteboard explanation for this paper yet.

Open Problems

We haven't generated a list of open problems mentioned in this paper yet.

Continue Learning

We haven't generated follow-up questions for this paper yet.

Collections

Sign up for free to add this paper to one or more collections.