Abstract
We present a near linear time algorithm for constructing hierarchical nets in finite metric spaces with constant doubling dimension. This data-structure is then applied to obtain improved algorithms for the following problems: Approximate nearest neighbor search, well-separated pair decomposition, spanner construction, compact representation scheme, doubling measure, and computation of the (approximate) Lipschitz constant of a function. In all cases, the running (preprocessing) time is near-linear and the space being used is linear.
Original language | English |
---|---|
Pages | 150-158 |
Number of pages | 9 |
DOIs | |
State | Published - 2005 |
Externally published | Yes |
Event | 21st Annual Symposium on Computational Geometry, SCG'05 - Pisa, Italy Duration: 6 Jun 2005 → 8 Jun 2005 |
Conference
Conference | 21st Annual Symposium on Computational Geometry, SCG'05 |
---|---|
Country/Territory | Italy |
City | Pisa |
Period | 6/06/05 → 8/06/05 |
Keywords
- Approximate distance oracle
- Approximate nearest neighbor search
- Compact representation scheme
- Doubling metrics
- Spanners
- Well separated pair decomposition