English

A Quadratic-Time Algorithm for General Multivariate Polynomial Interpolation

Numerical Analysis 2017-10-31 v1

Abstract

For m,nNm,n \in \mathbb{N}, m1m\geq 1 and a given function f:RmRf : \mathbb{R}^m\longrightarrow \mathbb{R} the polynomial interpolation problem (PIP) is to determine a \emph{generic node set} PRmP \subseteq \mathbb{R}^m and the coefficients of the uniquely defined polynomial QR[x1,,xm]Q\in\mathbb{R}[x_1,\dots,x_m] in mm variables of degree deg(Q)nN\mathrm{deg}(Q)\leq n \in \mathbb{N} that fits ff on PP, i.e., Q(p)=f(p)Q(p) = f(p), pP\forall\, p \in P. We here show that in general, i.e., for arbitrary m,nNm,n \in \mathbb{N}, m1m \geq 1, there exists an algorithm that determines PP and computes the N(\mboxm,n)=#PN(\mbox{m,n})=\#P coefficients of QQ in O(N(\mboxm,n)2)\mathcal{O}\big(N(\mbox{m,n})^2\big) time using O(\mboxmN(\mboxm,n))\mathcal{O}\big(\mbox{m}N(\mbox{m,n})\big) storage, without inverting the occurring Vandermonde matrix. We provide such an algorithm, termed PIP-SOLVER, based on a recursive decomposition of the problem and prove its correctness. Since the present approach solves the PIP without matrix inversion, it is computationally more efficient and numerically more robust than previous approaches. We demonstrate this in numerical experiments and compare with previous approaches based on matrix inversion and linear systems solving.

Keywords

Cite

@article{arxiv.1710.10846,
  title  = {A Quadratic-Time Algorithm for General Multivariate Polynomial Interpolation},
  author = {M. Hecht and B. L. Cheeseman and K. B. Hoffmann and I. F. Sbalzarini},
  journal= {arXiv preprint arXiv:1710.10846},
  year   = {2017}
}