[
https://issues.apache.org/jira/browse/MATH-1092?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13917636#comment-13917636
]
Bruce A Johnson commented on MATH-1092:
---------------------------------------
Another constructor so one could do:
line = new LineSearch(this, relativeTolerance,
absoluteTolerance,initialStep);
where initialStep is the value set with BracketingStep in
NonLinearConjugateGradientOptimizer
I presume the BracketingStep parameter was added for a reason to
NonLinearConjugateGradientOptimizer, but we've lost this capability in the new
code. Setting a smaller initial step (than 1.0) is clearly necessary for
actual use of my application, but I'm not sure I can easily represent this need
in a unit test case.
> NonLinearConjugateGradientOptimizer's Line search is a gradient search
> returns obviously suboptimal point.
> ----------------------------------------------------------------------------------------------------------
>
> Key: MATH-1092
> URL: https://issues.apache.org/jira/browse/MATH-1092
> Project: Commons Math
> Issue Type: Bug
> Reporter: Ajo Fod
> Attachments: MATH-1092.patch
>
>
> In package : org.apache.commons.math3.optim.nonlinear.scalar.gradient
> In a minimization problem, a line search should not return a point where the
> value is greater than the values at the edges of the interval. The line
> search violates this obvious requirement by focusing solely on solving for
> gradient=0 and ignoring the value.
> Moreover LineSearchFunction is something that can be used in other contexts,
> so perhaps this should be a standalone class.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
