[
https://issues.apache.org/jira/browse/LUCENE-6450?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14527392#comment-14527392
]
Michael McCandless commented on LUCENE-6450:
--------------------------------------------
This new approach is nice!
I don't fully understand all the geo math, but I think I get the gist:
you recursively approximate the target shape using smaller and smaller
ranges from the morton encoding, and then record when that z-shape is
fully within the query and avoid the post-filtering for those ranges.
This visits fewer terms than the original patch, which did just a
single range that can (w/ the right 'adversary') visit a great many
false terms.
It's impressive how fast this is, without using any NumericField
prefix terms. I think we can explore that later and we should commit
this approach now ..
Maybe add a test case w/ more data, e.g. randomized test? It could
index a bunch of random points, and then run random rects/shapes and
do the "dumb slow check every single doc" check and confirm query hits
agree.
> Add simple encoded GeoPointField type to core
> ---------------------------------------------
>
> Key: LUCENE-6450
> URL: https://issues.apache.org/jira/browse/LUCENE-6450
> Project: Lucene - Core
> Issue Type: New Feature
> Affects Versions: Trunk, 5.x
> Reporter: Nicholas Knize
> Priority: Minor
> Attachments: LUCENE-6450-5x.patch, LUCENE-6450-TRUNK.patch,
> LUCENE-6450.patch, LUCENE-6450.patch, LUCENE-6450.patch, LUCENE-6450.patch
>
>
> At the moment all spatial capabilities, including basic point based indexing
> and querying, require the lucene-spatial module. The spatial module, designed
> to handle all things geo, requires dependency overhead (s4j, jts) to provide
> spatial rigor for even the most simplistic spatial search use-cases (e.g.,
> lat/lon bounding box, point in poly, distance search). This feature trims the
> overhead by adding a new GeoPointField type to core along with
> GeoBoundingBoxQuery and GeoPolygonQuery classes to the .search package. This
> field is intended as a straightforward lightweight type for the most basic
> geo point use-cases without the overhead.
> The field uses simple bit twiddling operations (currently morton hashing) to
> encode lat/lon into a single long term. The queries leverage simple
> multi-phase filtering that starts by leveraging NumericRangeQuery to reduce
> candidate terms deferring the more expensive mathematics to the smaller
> candidate sets.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: dev-h...@lucene.apache.org
