mirror of https://github.com/CGAL/cgal
37 lines
2.1 KiB
TeX
37 lines
2.1 KiB
TeX
\section{todo open questions:}
|
|
|
|
\begin{itemize}
|
|
\item Some of the functors in PolynomialTraits can be implemented using some more basic ones.
|
|
On the other hand, the way they should be implemented may depend on the way the polynomial is
|
|
represented and on the cost of the other functors.
|
|
I therefore decided to put them into the PolynomialTraits.
|
|
On the other hand one could decide to provide a very basic PolynomialTraits
|
|
(just constructor, access to coefficients, and properties (degree)) and call the current
|
|
PolynomialTraits a PolynomialToolBox. But I don't think that this is going to be very efficient
|
|
or used.
|
|
|
|
\item Note that the numbering of variables in \ccc{Exponent_vector} is inconsistent
|
|
with the one in \ccc{Polynomial_d} and \ccc{PolynomialTraits_d}. \\
|
|
TODO: resolve this conflict.
|
|
|
|
\item Should the concept define the total order of Polynomials, i.e. how two polynomials are compared?
|
|
|
|
\item Note that a Polynomial is also a Model of an algebraic structure. Therefore a Polynomial will have a
|
|
valid \ccc{Algebraic_structure_traits}. \\
|
|
Question: should concept \ccc{PolynomialTraits} refine \ccc{AlgebraicStructureTraits}.\\
|
|
However, the \ccc{Polynomial_traits} could easily derive from \ccc{Algebraic_structure_traits}.\\
|
|
see also package: \ccc{Algebraic_foundations}.
|
|
|
|
\item \ccc{PolynomialTraits_d::Evaluate}: take \ccc{Innermost_coefficient_type} as argument type only.
|
|
this relates \ccc{Coercion_traits} \ccc{Algebraic_foundations}
|
|
|
|
\item This is just the general concept for \ccc{Polynomial_d}. \\
|
|
I plan to propose additional concepts for \ccc{Polynomial_1},\ccc{Polynomial_2} and maybe \ccc{Polynomial_3}.
|
|
These concepts will refine \ccc{Polynomial_d}, i.e. fix the constant \ccc{Polynomial_traits_d::d} to the appropriate value
|
|
and introduce extra functors as \ccc{Sign_at}, for univariate bivariate polynomials.
|
|
|
|
\item Do we need a SignAt for multivariate polynomials?
|
|
\item see in uni/bivariate\_polynomial\_hacks.h of NiX whether something is useful.
|
|
\item document functor Move
|
|
\end{itemize}
|