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 2012/12/13 14:34:12 UTC

[jira] [Commented] (MATH-914) Inconsistent multi-start randomization (optimizers)

    [ https://issues.apache.org/jira/browse/MATH-914?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13531008#comment-13531008 ] 

Gilles commented on MATH-914:
-----------------------------

In revision 1421287, I've marked (with "XXX") the code that should be revised.
I don't think that it is a blocking issue (it is also present in the "optimization" package), but anyone is welcome to provide a patch if a fix is quickly found.

                
> Inconsistent multi-start randomization (optimizers)
> ---------------------------------------------------
>
>                 Key: MATH-914
>                 URL: https://issues.apache.org/jira/browse/MATH-914
>             Project: Commons Math
>          Issue Type: Bug
>    Affects Versions: 3.0, 3.1
>            Reporter: Gilles
>            Priority: Minor
>             Fix For: 3.2
>
>
> In class "o.a.c.m.optim.BaseMultiStartMultivariateOptimizer", the "starting points" generator is passed at construction. But random initial guesses must fulfill the bound constraint and be somehow related to the user-supplied initial guess; and those are passed to the "optimize" method and thus can change from one call to the other, leading to inconsistent (and probably useless) multi-starts.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira