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 2010/07/25 23:26:50 UTC
[jira] Created: (MATH-395) Bugs in "BrentOptimizer"
Bugs in "BrentOptimizer"
------------------------
Key: MATH-395
URL: https://issues.apache.org/jira/browse/MATH-395
Project: Commons Math
Issue Type: Bug
Affects Versions: 2.1
Reporter: Gilles
Assignee: Gilles
I apologize for having provided a buggy implementation of Brent's optimization algorithm (class "BrentOptimizer" in package "optimization.univariate").
The unit tests didn't show that there was something wrong, although (from the "changes.xml" file) I discovered that, at the time, Luc had noticed something weird in the implementation's behaviour.
Comparing with an implementation in Python, I could figure out the fixes. I'll modify "BrentOptimizer" and add a test. I also propose to change the name of the unit test class from "BrentMinimizerTest" to "BrentOptimizerTest".
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (MATH-395) Bugs in "BrentOptimizer"
Posted by "Phil Steitz (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MATH-395?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Phil Steitz updated MATH-395:
-----------------------------
Fix Version/s: 2.2
> Bugs in "BrentOptimizer"
> ------------------------
>
> Key: MATH-395
> URL: https://issues.apache.org/jira/browse/MATH-395
> Project: Commons Math
> Issue Type: Bug
> Affects Versions: 2.1
> Reporter: Gilles
> Assignee: Gilles
> Fix For: 2.2
>
>
> I apologize for having provided a buggy implementation of Brent's optimization algorithm (class "BrentOptimizer" in package "optimization.univariate").
> The unit tests didn't show that there was something wrong, although (from the "changes.xml" file) I discovered that, at the time, Luc had noticed something weird in the implementation's behaviour.
> Comparing with an implementation in Python, I could figure out the fixes. I'll modify "BrentOptimizer" and add a test. I also propose to change the name of the unit test class from "BrentMinimizerTest" to "BrentOptimizerTest".
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Resolved: (MATH-395) Bugs in "BrentOptimizer"
Posted by "Gilles (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MATH-395?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Gilles resolved MATH-395.
-------------------------
Resolution: Fixed
Last bug fixed in revision 980032.
[This revision also contains the modifications due to the changes in "AbstractUnivariateRealOptimizer".]
The test comparing with Python has been removed because a tracing of the execution paths (in Python and Java) showed that the remaining discrepancies were due to different values being used for the "golden ratio" constant.
> Bugs in "BrentOptimizer"
> ------------------------
>
> Key: MATH-395
> URL: https://issues.apache.org/jira/browse/MATH-395
> Project: Commons Math
> Issue Type: Bug
> Affects Versions: 2.1
> Reporter: Gilles
> Assignee: Gilles
> Fix For: 2.2
>
>
> I apologize for having provided a buggy implementation of Brent's optimization algorithm (class "BrentOptimizer" in package "optimization.univariate").
> The unit tests didn't show that there was something wrong, although (from the "changes.xml" file) I discovered that, at the time, Luc had noticed something weird in the implementation's behaviour.
> Comparing with an implementation in Python, I could figure out the fixes. I'll modify "BrentOptimizer" and add a test. I also propose to change the name of the unit test class from "BrentMinimizerTest" to "BrentOptimizerTest".
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Commented: (MATH-395) Bugs in "BrentOptimizer"
Posted by "Gilles (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MATH-395?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12892283#action_12892283 ]
Gilles commented on MATH-395:
-----------------------------
Bugs corrected in revision 979257.
Not resolving yet because the implementation still does not behave as the Python one. I've added a unit test that indicates the discrepancies (with "XXX" markers).
> Bugs in "BrentOptimizer"
> ------------------------
>
> Key: MATH-395
> URL: https://issues.apache.org/jira/browse/MATH-395
> Project: Commons Math
> Issue Type: Bug
> Affects Versions: 2.1
> Reporter: Gilles
> Assignee: Gilles
> Fix For: 2.2
>
>
> I apologize for having provided a buggy implementation of Brent's optimization algorithm (class "BrentOptimizer" in package "optimization.univariate").
> The unit tests didn't show that there was something wrong, although (from the "changes.xml" file) I discovered that, at the time, Luc had noticed something weird in the implementation's behaviour.
> Comparing with an implementation in Python, I could figure out the fixes. I'll modify "BrentOptimizer" and add a test. I also propose to change the name of the unit test class from "BrentMinimizerTest" to "BrentOptimizerTest".
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.