CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

I think everything is fine.
I even did the computation with intervals to get bounds on the error and 
it is not that bad

[0.52614566421975906;0.52614566421975928] 
[0.85039325614585848;0.85039325614585892] 
[0.0014317557027606011;0.0014317557027606021]


Using this code:

#include <CGAL/Exact_predicates_inexact_constructions_kernel.h>
#include <iostream>
#include <vector>

typedef CGAL::Exact_predicates_inexact_constructions_kernel K;
typedef CGAL::Simple_cartesian<CGAL::Interval_nt<> > EK;

int main()
{
  CGAL::Cartesian_converter<K,EK> to_exact;

  K::Point_3 p1(-119.452959304302,54.727712609640498,888);
  K::Point_3 p2(-119.27352872835201,54.731184848162897,820);
  K::Point_3 p3(-119.369047730614,55.029360094556502,678);



  EK::Vector_3 vi = CGAL::cross_product(to_exact(p2-p1), to_exact(p3-p1) );
  K::Vector_3 v = CGAL::unit_normal(p1, p2, p3);

  std::cout << v << "\n";
  std::cout << std::setprecision(17) << vi/CGAL::sqrt(vi*vi) << "\n";

  std::cout << v * v << std::endl;
}

Sebastien.



On 09/12/2016 06:56 PM, chrism wrote:
> Agreed, that produces 1. However, the plane that unit normal corresponds to
> is an almost vertical plane. There is only 200m of elevation change between
> the min and max points, and it is not vertical.
>
> https://www.wolframalpha.com/input/?i=plane+through+(-119.452959304302,54.727712609640498,888)+and+normal+%5B0.52614566421975917,0.8503932561458587,0.0014317557027606017%5D
>
> This is what I don't understand.
>
>
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/Wrong-unit-normal-with-CGAL-unit-normal-tp4662261p4662270.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.