CGAL users discussion list

[<>]

Hello,
I've been pointed to CGAL in order to do some exact geometrical computations.
Thus, I am relatively new to the whole framework and am currently having some
problems with numerical types. I hope someone can help me in sorting them out.
Given an arrangement of circles, my goal is to compute a point within each 
face
of the given arrangement. When converting the arising values to doubles at 
some
point in the computation, the algorithm works fine but if I try to use exact
values until the end I'm running into trouble with the numerical types:

At first, I need to work with the source and the target point of the halfedges
of each face of the arrangement and, if applicable, the center of the
supporting circle. Unfortunately they are based on different numerical types
(_One_root_point_2<Quotient<MP_Float>> vs. Quotient<MP_Float>). Thus, I cannot
easily convert between the two types.

Secondly, I need to compute the unit vector of a given vector. This also seems
to be a problem, since I can only easily obtain the squared length of a vector
which is of a typefor which the sqrt function is not defined.

Is there a way to carry out my computations with exact types until the end
(after having computed the faces and their vertices, I only need basic 2D
algebra, orientation tests and 90° rotations), or do I have to convert back to
doubles?

(Exact_circular_kernel_2 is used throughout the algorithm)

bye
Dennis