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 2012/04/25 23:50:17 UTC

[jira] [Updated] (MATH-783) Inconsistent assumptions about tolerances in "PowellOptimizer"

     [ https://issues.apache.org/jira/browse/MATH-783?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Gilles updated MATH-783:
------------------------

    Attachment: MATH-783.patch

The attached patch proposes to fix this problem by using a custom checker that will bypass the default stopping criterion ("BrentOptimizer") in the line search.
                
> Inconsistent assumptions about tolerances in "PowellOptimizer"
> --------------------------------------------------------------
>
>                 Key: MATH-783
>                 URL: https://issues.apache.org/jira/browse/MATH-783
>             Project: Commons Math
>          Issue Type: Improvement
>    Affects Versions: 3.0
>            Reporter: Gilles
>            Assignee: Gilles
>             Fix For: 3.1
>
>         Attachments: MATH-783.patch
>
>
> Internally, the algorithm in "PowellOptimizer" (package "optimization.direct") uses "BrentOptimizer" (from package "optimization.univariate") to perform a "line" search. The problem is that the default stopping criterion of "BrentOptimizer" refers to the domain values while the default criterion in "PowellOptimizer" refers to function values.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira