[ 
https://issues.apache.org/jira/browse/LUCENE-7147?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15215890#comment-15215890
 ] 

Dawid Weiss commented on LUCENE-7147:
-------------------------------------

Any spatial tree is typically to speed up collision/overlap checking -- my 
experience was not with information retrieval but car navigation systems, but 
it's essentially similar underlying theory. Anyway, this is the part I have a 
hard time picturing in my head (note the "simple logic" part...). Since Ryan 
mentioned there is a visualization of that I thought I'd ask for a screenshot, 
that's all.
bq. Here is a patch which adds the following simple logic: if the rect of the 
subtree does not cross over any axis of the bbox, and none of the corners of 
the rect are in the circle, then the rect is disjoint from the circle. Note 
that while the meridian axis of the bbox is simple (it is just the longitude of 
the circle center), the latitude is a little more complicated. This is the 
latitude at which the bbox longitudes intersect the circle, which is not the 
latitude of the circle center (as it would be in normal 2D), except when the 
center is on the equator.



> Improve disjoint check for geo distance query traversal
> -------------------------------------------------------
>
>                 Key: LUCENE-7147
>                 URL: https://issues.apache.org/jira/browse/LUCENE-7147
>             Project: Lucene - Core
>          Issue Type: Improvement
>            Reporter: Ryan Ernst
>         Attachments: LUCENE-7147.patch
>
>
> When doing geo distance queries, it is important to avoid traversing subtrees 
> which do not contain any relevant points. We currently have checks which 
> compare the bbox of the query to the bounds of the subtree. However, it is 
> possible for a subtree to overlap the bbox, but still not intersect the 
> query. This issue is to improve that check to avoid unnecessary traversals.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to