[
https://issues.apache.org/jira/browse/MATH-1026?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13752393#comment-13752393
]
Evan Ward commented on MATH-1026:
---------------------------------
bq. What is the advantage over inheriting the test and overriding a
"getOptimizer()" method?
My goal was to run all of the test cases with different configurations of the
same optimizer. The two options I saw were to add a parameter to each test
case, or to create a separate class for each optimizer configuration. Since the
second option required me to duplicate the GN specific test cases in
GaussNewtonOptimizerTest, I chose the first option. It is also a nice example
of one way to use the new stateless optimizer interface. I don't have a strong
preference on how the tests are written, just that all of the optimization
algorithms are tested.
As for the tests, I only made one change that caused a failing test case to
pass, which I brought up on the ML. The rest of the assertions are preserved in
the new code. All of the changes I did make were to use the new interfaces,
and to share common code between the test cases. (I did add some new
assertions, as commented in the code.) I didn't think that re-factoring the
test cases would be an issue for the ML, but now I know. :)
bq. I'm off for a week now.
I'm off for two. :)
> Separate Optimization Problem from Algorithm
> --------------------------------------------
>
> Key: MATH-1026
> URL: https://issues.apache.org/jira/browse/MATH-1026
> Project: Commons Math
> Issue Type: Improvement
> Reporter: Evan Ward
> Attachments: opt.tar.gz
>
>
> See discussion on the mailing list starting with:
> http://www.mail-archive.com/[email protected]/msg39681.html
--
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
