BUG: Fix machine precision errors in eigen value calculation #355
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
If the segmentation in the dataset represents a 2D segmentation, the smallest of the 3 eigenvectors should be 0. But because a transform to physical coordinates (taking into account rotation and scaling), this can actually compute very small (~1e-16) negative numbers. This in turn causes an error in numpy.sqrt. This commit implements a 3-fold fix. 1) Because we do not need the eigen vectors, we can also calculate the eigenvalues using just the scaled coordinates, without rotation. In case of a 2D segmentation, this then results in correct calculation of 0 for the smallest eigen value. 2) After computation of eigen values. They are checked if they are negative, but > -1e-10. If that is the case, a warning is logged and they are rounded to 0. 3) Add additional checks to ensure the eigenvectors are not < 0 (effectively < -1e-10) in the feature functions. If that is the case, a warning is logged and nan is returned.
|
fixes #354 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
If the segmentation in the dataset represents a 2D segmentation, the smallest of the 3 eigenvectors should be 0.
But because a transform to physical coordinates (taking into account rotation and scaling), this can actually compute very small (~1e-16) negative numbers.
This in turn causes an error in numpy.sqrt. This commit implements a 3-fold fix.