CGAL users discussion list

[Andre Massing <>]

Hi all,

I am working on intersection of overlapping meshes, used in a certain finite 
element method. Since the upcoming CGAL 3.5 release will offer an additional 
package implementing an AABB Tree 
http://www.cgal.org/Manual/3.5/doc_html/cgal_manual/AABB_tree/Chapter_main.html
I am wondering  how it compares to
http://www.cgal.org/Manual/3.5/doc_html/cgal_manual/Box_intersection_d/Chapter_main.html
 
.
Since I am going to intersect two meshes (which could contain many cells) I 
am 
thinking to use the bipartite version of the CGAL::box_intersection_d 
function. 
http://www.cgal.org/Manual/3.5/doc_html/cgal_manual/Box_intersection_d_ref/Function_box_self_intersection_d.html#Cross_link_anchor_1587

Would this function be (theoretically) faster than building up a tree of the 
first mesh and query each bbox of the 2nd mesh via
http://www.cgal.org/Manual/3.5/doc_html/cgal_manual/AABB_tree_ref/Class_AABB_tree.html
by simply iterating through the  bounding box list of the 2nd mesh?
I could imagine that in the bipartite case some performance tweaks are used?

BTW does the new AABB tree build up on the former package or is this module a 
completly independent implementation?

Any pointers/hints are really appreciated :-)

Best,
Andre