CGAL users discussion list

[kyewong <>]

Hi, I've met an issue(probably a numerical issue) in CGAL::intersection.
When I let the data type of vertex coordinates of Polygon_2 to be "double",
and there may be more than 6 digits after the decimal point (eg,
2666.666748), calling CGAL::intersection would cause an error in the "sweep"
function in the Basic_sweep_line_2.h file. 
However if I reduce the precision (eg, 2666.67), CGAL::intersection works
very well.
The kernel I use is CGAL::Cartesian<double>.
I'm wondering if it is caused by numerical issue of the "double" type. If
yes, how could I rapidly modify the kernal? Shall I use the 
CGAL::Exact_predicates_exact_constructions_kernel or change "double" to
"float" ? Since my program relies on CGAL very much and is of relatively big
size, I would like to minimize the changes to related definitions of the
variables.
Could anyone help me please? Thanks very much!


--
View this message in context: 
http://cgal-discuss.949826.n4.nabble.com/Issue-on-CGAL-Polygon-2-Intersection-tp3877407p3877407.html
Sent from the cgal-discuss mailing list archive at Nabble.com.