[ https://issues.apache.org/jira/browse/LUCENE3814?page=com.atlassian.jira.plugin.system.issuetabpanels:commenttabpanel&focusedCommentId=13817007#comment13817007
]
David Smiley commented on LUCENE3814:

The pertinent code is actually in Spatial4j now and it was recently fixed. When Lucenespatial
updates the Spatial4j version (within a month I figure), I'll close this issue.
> Manhattan distance function is incorrect, not absolute distance between coordinates
> 
>
> Key: LUCENE3814
> URL: https://issues.apache.org/jira/browse/LUCENE3814
> Project: Lucene  Core
> Issue Type: Bug
> Components: modules/spatial
> Affects Versions: 4.0ALPHA
> Reporter: Neil Hooey
> Assignee: David Smiley
> Labels: distance, geometric
>
> The Lucene vectorDistance() function's Manhattan distance function is incorrect.
> Wikipedia says: http://en.wikipedia.org/wiki/Manhattan_distance
> "Taxicab geometry, blahblahblah, is a form of geometry in which the usual distance function
or metric of Euclidean geometry is replaced by a new metric in which the distance between
two points is the sum of the *absolute differences* of their coordinates."
> The Lucene function isn't taking the absolute value before subtracting the vector coordinates.
> I don't have a patch, but the offending code is here:
> {code}
> // lucene/contrib/spatial/src/java/org/apache/lucene/spatial/DistanceUtils.java
> } else if (power == 1.0) {
> for (int i = 0; i < vec1.length; i++) {
> result += vec1[i]  vec2[i];
> }
> {code}
> It just needs to use Math.abs() when subtracting the coordinates.

This message was sent by Atlassian JIRA
(v6.1#6144)

To unsubscribe, email: devunsubscribe@lucene.apache.org
For additional commands, email: devhelp@lucene.apache.org
