[Work Log] Building reference implementation of full-tree covariance

September 17, 2013
Project Tulips
Subproject Data Association v3
Working path projects/​tulips/​trunk/​src/​matlab/​data_association_3
SVN Revision 15229
Unless otherwise noted, all filesystem paths are relative to the "Working path" named above.
Building reference version of `build_attachment_covariance()`. It does full recursion for every element, taking \(O(n^2 \log n)\) time.

Running now with model_type = 1; estimated runtime: 45 minutes. (compare this to less than 5 seconds in the fast version).

Results don't match.

Issue 1: topology indices array was reversed; handled inheritance wrongly. Issue 2: forgot about start_index; should add to all curve indices.

Results now match when model type = 1.

Matching results suggest that the covariance matrix is correctly implemented as designed, but why are we getting negative eigenvalues?

Next: re-run with model_type = 3.

Posted by Kyle Simek
blog comments powered by Disqus