CGAL users discussion list

[Pierre Alliez <>]

sorry - I was too fast fixing the bug yesterday - there is no bug in
fact (but I have updated the comment)

if all three eigenvalues are equal then it is fully degenerate and any
plane that goes through the centroid achieves the same approximation
(after all I could let the general case handle it as well).

if the two smallest eigenvalues are equal then it is also degenerate
(albeit less) and any plane that goes through the best fitting line
works - and I let the general case pick one of this plane.

pierre

Pierre Alliez
INRIA Sophia Antipolis - Mediterranee 
Project-team GEOMETRICA 
http://www-sop.inria.fr/members/Pierre.Alliez/
Tel: +33 4 92 38 76 77
Fax: +33 4 97 15 53 95



杨成林 a écrit :
> Hello
> I reported a bug in PCA component in Feb. In CGAL 3.5 beta1 I have
> seen the change, but I don't think the problem is fixed.
> It is about line 709-710 in PCA_util.h of CGAL 3.4. The condition is
> too strong for regular case. Now in CGAL 3.5 beta1, it is line
> 708-709. I think the condition is still too strong, but for degenerate
> case. To confirm degenerate case, only "eigen_values[1] ==
> eigen_values[2]" is needed. "eigen_values[0] != eigen_values[2]" is
> not a guarantee of regular case.
>
> -- 
> 杨成林