algorithm - How to do high dimensional range query with fixed range? -


i have 10^4 points in 7 dimensional space. application, need make ~10^6 range queries on input find points lie inside given range. in application, queries use same range size. appropriate data structure problem?

kd-tree seems apt, 7 dimensions , small output size linear in time complexity queries. solution range trees, seems complex construct small number of input in application. also, don't see of these structures using fact range constant size advantage. example, if 1d problem, queries asking points lying inside range of size 10 example, @ different places along number line.

well, late answer, dont know if s usefull now. can use fixed height fqts (fhfqt or fhqt) [http://users.dcc.uchile.cl/~rbaeza/ftp/fqtrees.ps.gz] or fixed query arrays (fqa) [http://www.dcc.uchile.cl/~gnavarro/ps/mtap01.pdf]. these structures work in kind of similarity searching. beside that, need use pivot selection method incremental 1 results. assume using euclidian distance , distances histogram gaussian distribution. sorry english...


Comments

Popular posts from this blog

java - Intellij Synchronizing output directories .. -

git - Initial Commit: "fatal: could not create leading directories of ..." -