Proceedings of the Twenty-Ninth Annual ACM Symposium on Theory of Computing - STOC '97 1997
DOI: 10.1145/258533.258667
|View full text |Cite
|
Sign up to set email alerts
|

A polylog(n)-competitive algorithm for metrical task systems

Abstract: We present a randomized on-line algorithm for the Metrical Tti System problem that achieves a competitive ratio of O(log6 n) for arbitrary metric spaces, against art oblivious adversary. This is the first algorithm to achieve a sublinear competitive ratio for all mernc spaces. Our algorithm uses a recent result of Bart.al[Bar96] thatan arbitrarymetric space can be probabilistically approximated by a set of metric spaces called "k-hierarchical well-separated trees" (k-HST'S). Indeed, the main technical result o… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2
1

Citation Types

0
115
0

Year Published

2000
2000
2006
2006

Publication Types

Select...
5
2
1

Relationship

2
6

Authors

Journals

citations
Cited by 78 publications
(115 citation statements)
references
References 14 publications
0
115
0
Order By: Relevance
“…1 In fact, an algorithm for the r -unfair ratio on the uniform metric space is at the heart of the best algorithm known for the MTS problem on arbitrary metric spaces (Bartal et al, 1997) (together with an approximation of arbitrary spaces by hierarchical ones (Bartal, 1996)). …”
Section: The Mts Problem and The Competitive Ratiomentioning
confidence: 99%
“…1 In fact, an algorithm for the r -unfair ratio on the uniform metric space is at the heart of the best algorithm known for the MTS problem on arbitrary metric spaces (Bartal et al, 1997) (together with an approximation of arbitrary spaces by hierarchical ones (Bartal, 1996)). …”
Section: The Mts Problem and The Competitive Ratiomentioning
confidence: 99%
“…A randomized online algorithm A is called r-competitive if there exists some constant c such that for any task sequence σ, E[cost A (σ)] ≤ r · cost Opt (σ) + c, where cost A (σ) is the random variable of the cost for serving σ by A, and cost Opt (σ) is the optimal (offline) cost for serving σ. The current best online algorithm for the MTS problem in n-point metric spaces is O(log 2 n log log n) competitive [16,15] (an improvement of [5,3]). Both papers [5,16] actually solve the MTS problem for ultrametrics, and then reduce arbitrary metric spaces to ultrametrics using Theorem 1.…”
Section: Applicationsmentioning
confidence: 99%
“…The current best online algorithm for the MTS problem in n-point metric spaces is O(log 2 n log log n) competitive [16,15] (an improvement of [5,3]). Both papers [5,16] actually solve the MTS problem for ultrametrics, and then reduce arbitrary metric spaces to ultrametrics using Theorem 1. We next show that path embedding suffices: Proposition 1.…”
Section: Applicationsmentioning
confidence: 99%
“…A powerful technique that has been successfully used recently in this context is to embed the given metric space in a simpler metric space such that the distances are approximately preserved in the embedding. New and improved algorithms have resulted from this idea for several important problems [1,2,7,11,12,20]. Tree metrics are a very natural class of simple metric spaces since many algorithmic problems become tractable on them.…”
Section: Introductionmentioning
confidence: 99%