[SERVER-691] n-dimensional geospatial search Created: 02/Mar/10 Updated: 28/Dec/23 |
|
| Status: | Backlog |
| Project: | Core Server |
| Component/s: | Geo, Index Maintenance |
| Affects Version/s: | None |
| Fix Version/s: | None |
| Type: | New Feature | Priority: | Major - P3 |
| Reporter: | Eliot Horowitz (Inactive) | Assignee: | Backlog - Query Integration |
| Resolution: | Unresolved | Votes: | 50 |
| Labels: | qi-geo | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Issue Links: |
|
||||||||
| Assigned Teams: |
Query Integration
|
||||||||
| Participants: | |||||||||
| Description |
|
for 2d you cna use http://www.mongodb.org/display/DOCS/Geospatial+Indexing |
| Comments |
| Comment by Vlad Radu [ 29/Aug/19 ] |
|
X-tree would be a good candidate for this. |
| Comment by cristian lorenzetto [ 03/May/18 ] |
|
I m interessed in this feature . In my case for 1d vectors |
| Comment by Stav Yagev [ 26/May/16 ] |
|
This could be super useful for machine learning / classification problems! |
| Comment by Hans Kristian [ 17/Nov/13 ] |
|
Also related to this one: https://jira.mongodb.org/browse/SERVER-9220 |
| Comment by Nick Shaw [ 15/Oct/13 ] |
|
What gets me is that Thermopylae Sciences have largely solved this problem for 10-gen: http://www.slideshare.net/nknize/mongo-sv-knizefinal Mongos licensing states that "The goal of the server license is to require that enhancements to MongoDB be released to the community." http://www.mongodb.org/about/licensing/ Am I missing something here? |
| Comment by Alex [ 11/Oct/13 ] |
|
Very interested in such a feature to build a backend for a 3d world |
| Comment by Michaël Pico [ 11/Mar/13 ] |
|
I would also be very interested by this feature especially for 3D rendering (where you need 4 dimensional storage with the normalisation parameter !) |
| Comment by Jamil Bou Kheir [ 05/May/12 ] |
|
Thanks Noeska. I could implement nearest neighbor in non-realtime MapReduce but it would really help to have native DB implementation of this to be fast. Even simple n-dimension Euclidean distance would suffice for now. http://en.wikipedia.org/wiki/Euclidean_distance. Maybe I can hack it in on my own. Let the Source be with me... |
| Comment by Noeska Smit [ 05/May/12 ] |
|
Hi Jamil, I pasted my kd-tree python code for you here: http://pastebin.com/5e9AAQ7F |
| Comment by Jamil Bou Kheir [ 05/May/12 ] |
|
+1. N-Dimensional distance queries would be highly useful for similarity and other clustering applications! Noeska – how are you implementing your KDtree in a collection? |
| Comment by Arun Vijayan [ 07/Mar/12 ] |
|
N-dim spatial index will be awesome for clustering algorithms. Please prioritize this. |
| Comment by Noeska Smit [ 13/Sep/11 ] |
|
This would be a great feature indeed. For now I implemented this by building a k-d Tree and storing that in a collection for fast graph traversal. |
| Comment by Yuriy Bogdanov [ 13/Sep/11 ] |
|
+1 for n-dim. |
| Comment by Edmunds Kalnins [ 13/Sep/11 ] |
|
+1 for n-dimentional. We use geospatial index with bit interleaving to turn 5 numbers into 2 and then sort by those. n-dimensional index would make this much more precise and simpler. |
| Comment by Ian Mercer [ 08/Sep/11 ] |
|
+1 but don't stop at 3. I want to be able to search multi-dimensional spaces with an arbitrary number of dimensions. |
| Comment by John Hurliman [ 03/Mar/10 ] |
|
I would be interested in 3D geospatial searches for our virtual world backend (http://openmetaverse.googlecode.com/). We're currently using 2D MySQL and additional filtering in PHP, and considering PostgreSQL or other alternatives such as MongoDB. |