[
https://issues.apache.org/jira/browse/MATH-428?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12923819#action_12923819
]
Gilles commented on MATH-428:
-----------------------------
I hadn't looked at "MultiDirectional.java"... :-(
I don't quite get what is going on inside its {{iterateSimplex}} method: It
contains a loop that checks convergence, using the same convergence checker as
in the loop within {{doOptimize}}! Why should the convergence criteria be the
same?
If the checker is really the same, then this poses a problem (wrt the new
design) since there is again a mix between optimization and simplex
"management")...
> Simplify "DirectSearchOptimizer"
> --------------------------------
>
> Key: MATH-428
> URL: https://issues.apache.org/jira/browse/MATH-428
> Project: Commons Math
> Issue Type: Improvement
> Reporter: Gilles
> Assignee: Gilles
> Priority: Minor
> Fix For: 3.0
>
> Attachments: opt_direct.tar.gz
>
>
> This issue refers to classes in package {{optimization.direct}}.
> Currently the class {{NelderMead}} inherits from {{DirectSearchOptimizer}}.
> However the method {{doOptimize}} is implemented in
> {{DirectSearchOptimizer}}. This is backwards from the intended design (where
> an "optimizer" is defined as a class that implements a specific algorithm
> within {{doOptimize}}). According to this "terminology",
> {{DirectSearchOptimizer}} is the optimizer whereas {{NelderMead}} could be
> considered as a specific way to construct a simplex. Indeed, that's what
> seems intended since it overrides the abstract method {{iterateSimplex}}.
> I suggest to create 2 classes that will make it clear the separation between
> the "optimizer" and the "simplex manager".
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
