You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Current implementation uses the Packed Hilbert R-tree algorithm. I've had some recent discussions indicating that depending on data might produce trees that have significant overlaps and as such do not have optimal query performance.
For this to be interesting the resulting tree must work as is with current query implementation and I believe there is a good chance it could.
Note that OMT will likely make tree build time more expensive but since the intent of FlatGeobuf with spatial index is write once read many that is not a bad thing.
The text was updated successfully, but these errors were encountered:
The Priority R-Tree: A Practically Efficient and Worst-Case Optimal R-Tree Lars Arge, Mark de Berg, Herman Haverkort, and Ke Yi Proceedings of the 2004 ACM SIGMOD International Conference on Management of Data (SIGMOD '04), Paris, France, June 2004, 347-358. Journal version in ACM Transactions on Algorithms.
Current implementation uses the Packed Hilbert R-tree algorithm. I've had some recent discussions indicating that depending on data might produce trees that have significant overlaps and as such do not have optimal query performance.
Specifically the Overlap Minimizing Top-down (OMT) seem promising. See http://ftp.informatik.rwth-aachen.de/Publications/CEUR-WS/Vol-74/files/FORUM_18.pdf.
For this to be interesting the resulting tree must work as is with current query implementation and I believe there is a good chance it could.
Note that OMT will likely make tree build time more expensive but since the intent of FlatGeobuf with spatial index is write once read many that is not a bad thing.
The text was updated successfully, but these errors were encountered: