[Insight-developers] my change to 1+1ES optimizer breaks some tests

[Lorensen, William E (Research)]

You should communicate with the test authors to see if the new results are OK.

Thanks for fixing this by the way,

Bill

-----Original Message-----
From: insight-developers-bounces at itk.org
[mailto:insight-developers-bounces at itk.org]On Behalf Of Zachary Pincus
Sent: Wednesday, March 02, 2005 12:56 PM
To: insight-developers at itk.org
Subject: [Insight-developers] my change to 1+1ES optimizer breaks some
tests


Hello,

I committed my proposed changes to the  
itkOnePlusOneEvolutionaryOptimizer yesterday. This had the (somewhat  
expected) effect of breaking a few of the tests.

The tests listed below all broke yesterday, and all depend on the 1+1ES  
optimizer. I could go through the tests and re-calibrate them to the  
the new optimizer to stop the breakage. This, for example, would  
clearly be OK for the first test listed, which breaks by a matter of  
five or ten edge pixels out of several hundred. The second test listed  
breaks by a little bit more than that though... so I'm not sure if I  
should just blindly go through "fixing" tests, or if I should  
communicate with the test authors about recalibrating the tests to the  
modified (and almost certainly more correct)  
itkOnePlusOneEvolutionaryOptimizer class.

Any thoughts?

Zach

Broken tests for which I claim responsibility
---------------------------------------------
http://www.itk.org/Testing/Dashboard/20050302-0500-Nightly/TestDetail/ 
__Examples_Segmentation_GeodesicActiveContourShapePriorCallosumTest.html

http://www.itk.org/Testing/Dashboard/20050302-0500-Nightly/TestDetail/ 
__Examples_Registration_ImageRegistration14Test.html

http://www.itk.org/Testing/Dashboard/20050302-0500-Nightly/TestDetail/ 
__Testing_Code_Numerics_Statistics_itkGoodnessOfFitMixtureModelCostFunct 
ionTest.html

http://www.itk.org/Testing/Dashboard/20050302-0500-Nightly/TestDetail/ 
__Testing_Code_Algorithms_itkSpatialObjectToImageRegistrationTest.html

_______________________________________________
Insight-developers mailing list
Insight-developers at itk.org
http://www.itk.org/mailman/listinfo/insight-developers