You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@commons.apache.org by "Gilles (JIRA)" <ji...@apache.org> on 2015/06/21 19:56:00 UTC

[jira] [Commented] (MATH-1238) NoBracketingException send with valid brackets

    [ https://issues.apache.org/jira/browse/MATH-1238?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14595148#comment-14595148 ] 

Gilles commented on MATH-1238:
------------------------------

This behaviour stems from the particular semantics of NaN (all comparisons with NaN being "false").
The library lets NaN values propagate... until some code gives up.

The error message might indeed be misleading but not more than what would have been reported later on (i.e. in this case, if the "brent" method were passed a NaN as one of its argument).


> NoBracketingException send with valid brackets
> ----------------------------------------------
>
>                 Key: MATH-1238
>                 URL: https://issues.apache.org/jira/browse/MATH-1238
>             Project: Commons Math
>          Issue Type: Bug
>    Affects Versions: 3.5
>            Reporter: charles sitbon
>            Priority: Minor
>
> The brent solver sometimes send a NoBracketingException with valid brackets : example for a beta distribution with large Beta (285) and alpha around 15
> Exception in thread "main" org.apache.commons.math3.exception.NoBracketingException: function values at endpoints do not have different signs, endpoints: [0, 1], values: [-0.15, 0.85]
> the exception is actually caused by the yinitial which is Nan



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