[jira] [Created] (MATH-1096) implementation of smallest enclosing ball algorithm sometime fails

["Luc Maisonobe (JIRA)" <ji...@apache.org>]

Luc Maisonobe created MATH-1096:
-----------------------------------

             Summary: implementation of smallest enclosing ball algorithm sometime fails
                 Key: MATH-1096
                 URL: https://issues.apache.org/jira/browse/MATH-1096
             Project: Commons Math
          Issue Type: Bug
    Affects Versions: 3.3
            Reporter: Luc Maisonobe
            Assignee: Luc Maisonobe
            Priority: Blocker


The algorithm for finding the smallest ball is designed in such a way the radius should be strictly increasing at each iteration.

In some cases, it is not true and one iteration has a smaller ball. In most cases, there is no consequence, there is just one or two more iterations. However, in rare cases discovered while testing 3D, this generates an infinite loop.

Some very short offending cases have already been identified and added to the test suite. These cases are currently deactivated in the main repository while I am already working on them. The test cases are

* WelzlEncloser2DTest.testReducingBall
* WelzlEncloser2DTest.testLargeSamples
* WelzlEncloser3DTest.testInfiniteLoop
* WelzlEncloser3DTest.testLargeSamples



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)