Generic bivariate multi-point evaluation, interpolation and modular composition with precomputation
Résumé
If $\mathbb{K}$ is a large enough field and $\mathcal{P} \subset \mathbb{K}^2$ is a fixed, generic set of points, which is available for precomputation, we show how to compute all the evaluations of any dense polynomial $f$ on $\mathcal{P}$ in quasi-linear time.
Similarly, in quasi-linear time then given interpolation constraints on $\mathcal{P}$ and a target $y$-degree, we compute an $f$ having those evaluations on $\mathcal{P}$ and at most that $y$-degree.
Our genericity assumption is explicit and we prove most point sets over a large enough field satisfy it.
If $\mathcal{P}$ violates the assumption our algorithms still work and the performance degrades smoothly according to a distance from being generic.
We apply the same technique to modular composition: fix a square-free $G \in \mathbb{K}[x]$ and generic $R \in \mathbb{K}[x]$ both available for precomputation, we then input $f \in \mathbb{K}[x,y]$ and output $f(x, R(x)) ~\mathrm{rem}~ G \in \mathbb{K}[x]$ in quasi-linear time in the size of $f, G, R$.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...